Handbook On The Shadow Economy

Handbook on the Shadow Economy
Friedrich Schneider - 9781848443358

Downloaded from Elgar Online at 03/27/2017 09:18:32PM
via University of Melbourne
M2678 - SCHNEIDER TEXT.indd i 27/07/2011 16:16

M2678 - SCHNEIDER TEXT.indd ii 27/07/2011 16:16

Handbook on the
Shadow Economy
Edited by
Friedrich Schneider
Professor of Economics and Public Finance, Johannes Kepler
University of Linz, Austria
Edward Elgar
Cheltenham, UK • Northampton, MA, USA

M2678 - SCHNEIDER TEXT.indd iii 27/07/2011 16:16

© Friedrich Schneider 2011
All rights reserved. No part of this publication may be reproduced, stored in a

retrieval system or transmitted in any form or by any means, electronic,
mechanical or photocopying, recording, or otherwise without the prior
permission of the publisher.
Published by
Edward Elgar Publishing Limited
The Lypiatts
15 Lansdown Road
Cheltenham
Glos GL50 2JA
UK
Edward Elgar Publishing, Inc.

William Pratt House
9 Dewey Court
Northampton
Massachusetts 01060
USA
A catalogue record for this book

is available from the British Library
Library of Congress Control Number: 2011924162
ISBN 978 1 84844 335 8
Typeset by Servis Filmsetting Ltd, Stockport, Cheshire

Printed and bound by MPG Books Group, UK

01
M2678 - SCHNEIDER TEXT.indd iv 27/07/2011 16:16

Contents
List of contributors vii
Introduction 1
Friedrich Schneider
PART I SIZE AND DEVELOPMENT OF THE SHADOW

ECONOMIES ALL OVER THE WORLD
1 Shadow economies all over the world: new estimates for

162 countries from 1999 to 2007 9
Friedrich Schneider, Andreas Buehn and Claudio E. Montenegro
2 Survey on the shadow economy and undeclared work in
OECD countries 78
Lars P. Feld and Friedrich Schneider
3 The size and development of the shadow economy in India:
a first attempt at a public choice explanation 131
Kausik Chaudhuri and Friedrich Schneider
4 Size, development and perception of the shadow economy in
Switzerland 150
Christoph A. Schaltegger
PART II REGIONAL VARIATION IN THE SIZE AND

DEVELOPMENT OF THE SHADOW ECONOMY
5 Regional variations in the nature of the shadow economy:

evidence from a survey of 27 European Union member states 177
Colin C. Williams and Jan Windebank
6 Regional patterns of the shadow economy: modelling issues
and evidence from the European Union 201
Helmut Herwartz, Friedrich Schneider and Egle Tafenau

M2678 - SCHNEIDER TEXT.indd v 27/07/2011 16:16

vi Handbook on the shadow economy
PART III SHADOW ECONOMY, ILLICIT WORK AND

RELATED ACTIVITIES
7 The shadow economy and do-it-yourself activities: what do

we know? 261
Andreas Buehn and Alexander Karmann
8 The shadow economy in the residential construction sector 293
Christopher Bajada
9 Who is working illicitly and why? Insights from
representative survey data in Germany 324
Dominik H. Enste
PART IV TAX MORALE AND THE SHADOW ECONOMY
10 Tax morale, tax evasion and the shadow economy 347

Gebhard Kirchgässner
11 The link between the intrinsic motivation to comply and
compliance behaviour: a critical appraisal of existing
evidence 375
Martin Halla
12 Deterrence policy and the size of the shadow economy in
Germany: an institutional and empirical analysis 409
Lars P. Feld, Andreas J. Schmidt and Friedrich Schneider
PART V CORRUPTION AND THE SHADOW ECONOMY
13 The impact of institutions on the shadow economy and

corruption: a latent variables approach 443
Axel Dreher, Christos Kotsogiannis and Steve McCorriston
14 Shadow economy, voice and accountability, and corruption 469
Benno Torgler, Friedrich Schneider and Alison Macintyre
Index 503

M2678 - SCHNEIDER TEXT.indd vi 27/07/2011 16:16

Contributors
Christopher Bajada is an Associate Professor of Economics and Associate
Dean at the University of Technology Sydney, Australia. He started his
teaching career at the University of New South Wales, from which he
holds a PhD. He has taught economics in a variety of undergraduate and
post-graduate courses. Chris’s research is primarily in applied macroeco-
nomics, with a special interest in the underground economy. His research
expertise in this field has attracted national publicity and he has been fea-
tured on numerous radio and television programmes as well as in the print
media. He has worked with the Australian Taxation Office as a member of
the Cash Economy Task Force and has served as a member of the Council
of the Economic Society of Australia.
Andreas Buehn studied economics at Technische Universität Dresden,
Germany. Since 2010 he has been a postdoctoral research assistant,
Chair of Monetary Economics at Technische Universität Dresden and
visiting researcher at Georgia State University, USA. Formerly he was a
visiting scholar at the Office of Tax Policy Research at the University of
Michigan (September–December 2010), research assistant at Technische
Universität Dresden (2005–2010) and lecturer at Sofia University St.
Kliment Ohridski (2005–2010). His research interests are public econom-
ics, especially tax evasion, shadow economics and do-it-yourself activi-
ties, international economics, especially illegal trade and smuggling and
applied econometrics.
Kausik Chaudhuri is currently Senior Lecturer at Leeds University Business
School, UK, and Professor at Madras School of Economics, India. Prior
to this, Dr Chaudhuri was at Indira Gandhi Institute of Development
Research, India. He has held many visiting positions at institutions
including the University of Sydney, University of New South Wales,
National Institute of Public Finance and Policy and the Indian Statistical
Institute. After his PhD he worked at the World Bank as a consultant.
Dr Chaudhuri’s current research interests are in development econom-
ics, public policy, political economy and applied econometrics. He has
published around 35 papers in refereed journals including the Journal
of Development Economics, Journal of Banking and Finance, Journal of
vii

M2678 - SCHNEIDER TEXT.indd vii 27/07/2011 16:16

viii Handbook on the shadow economy
Development Studies, Social Indicators Research, Applied Economics,

Economic Record, Economic Letters, Journal of Policy Modeling and the
B.E. Journal of Macroeconomics. At present, he is associate editor of India
Macroeconomics Annual and on the editorial board of Macroeconomics
and Finance in Emerging Market Economies.
Axel Dreher is Professor of Development Economics and International
Economics at the Georg-August University Goettingen, Germany,
and director of the Center for European, Governance and Economic
Development Research (cege) at the University of Goettingen. Dreher is
Affiliated Professor at the KOF Swiss Economic Institute and Research
Fellow at IZA and CESifo. He has published more than 50 articles in ref-
ereed journals, such as the Journal of Development Economics, European
Economic Review, International Economic Review and the Journal of
Money, Credit and Banking. He is editor of the Review of International
Organizations and author of the KOF Index of Globalization. Most
of his current research is in the fields of political economy, economic
development and international economics.
Dominik H. Enste is Professor for Economics, Ethical and Behavioral
Economics at the University of Applied Science Cologne and lecturer
at the University of Cologne, Germany. Since April 2003, he has acted
as senior economist and project manager at the Cologne Institute for
Economic Research. His main research focus is on institutional and
ethical economics/business ethics and behavioral economics. He studied
economics, sociology and economic psychology (1990–1996) at univer-
sities in Cologne, Dublin and Fairfax (Virginia). After completing his
studies, he worked at the Economic Policy Department of the University
of Cologne (1996–2001) where he also completed his thesis on ‘Shadow
economy and institutional change’. He was an HR Manager for two years
at the Gerling Insurance Group (2001–2003).
Lars P. Feld is a Full Professor of Economics, specialising in economic
policy at the University of Freiburg; Director of the Walter Eucken
Institute; member of the Scientific Advisory Board to the German
Federal Finance Ministry; member of the German Academy of Sciences
Leopoldina; and member of the Kronberger Kreis; research associate
at the Centre for European Economic Research (ZEW) Mannheim,
Germany. He gained his Master in Economics (Dipl.-Volksw.) in 1993
from the University of Saarland, Saarbrücken; and his PhD in 1999 and
Habilitation in 2002 from the University of St. Gallen. He was Professor
of Economics at the University of Marburg from 2002 to 2006; and
Professor of Economics at the University of Heidelberg from 2006 to 2010.

M2678 - SCHNEIDER TEXT.indd viii 27/07/2011 16:16

Contributors ix
His research interests include public economics, public choice, economic

policy, law and economics and fiscal psychology.
Martin Halla is currently an assistant professor at the University of Linz,
Austria and research affiliate of the Institute for the Study of Labor (IZA)
in Bonn, Germany. He received a PhD in economics in July 2007 from
the University of Linz, Germany. In 2007, he was visiting scholar at the
Institute for International Economic Studies at Stockholm University and
in 2009 at the Department of Economics at the University of California,
Berkeley. His general research interests are applied microeconometrics,
family economics, labour economics, and law and economics. In 2008,
he was awarded the prize for the best paper by a young economist at
the conference of the European Association of Labour Economists in
Amsterdam. His research has been published in the Scandinavian Journal
of Economics, Public Choice, European Journal of Political Economy and
Kyklos, among others.
Helmut Herwartz has been Professor of Econometrics at Christian
Albrechts University at Kiel, Germany, since 2003. He received his doc-
toral degree in 1995 at Humboldt University. His research interests cover
robust inference, state space modelling and applied financial and macro-
economic analysis and forecasting. He has published numerous articles in
leading econometric journals, for instance, the Journal of Econometrics,
International Journal of Forecasting, Journal of International Money and
Finance, Computational Statistics and Data Analysis.
Alexander Karmann studied mathematics at the University of Erlangen-
Nuernberg, Germany, and musics/concert piano at the Konversatorium
Nuernberg, Germany. He gained his PhD in 1979 and Habilitation in
1983 at the University of Karlsruhe, Germany. Since 1993 he has been
Chair of Monetary Economics at Technische Universitaet Dresden. He
was formerly assistant lecturer at the University of Karlsruhe, Germany
(1975–1986), Professor of Economics at University of Hamburg, Germany
(1986–1993). His research interests include monetary economics, shadow
economy and DIY, financial crises and health economics.
Gebhard Kirchgässner has been Professor of Economics and Econometrics
and Director of the Swiss Institute for International Economics and
Applied Economic Research at the University of St Gallen, Switzerland,
since 1992. He gained his PhD in 1976 at the University of Constance,
Germany, Habilitation in 1981, at the University of Constance, Germany,
and at the Swiss Federal Institute of Technology, Zürich in 1982. He was
Professor of Economics, especially Public Finance, at the University of
Osnabrück, 1984–1992; President of the European Public Choice Society,

M2678 - SCHNEIDER TEXT.indd ix 27/07/2011 16:16

x Handbook on the shadow economy
2003–2005; and President of the Economic Advisory Board of the Swiss

Government, 2003–2007. He has been President of the Swiss Society of
Economics and Statistics since 2008. His main research areas are modern
political economy, in particular, federalism and direct democracy, envi-
ronmental economics and methodological problems of the social sciences.
Christos Kotsogiannis is an associate professor in Economics at the
University of Exeter, UK, and a CESifo research fellow. His research
is in the field of public economics, applied microeconomics and politi-
cal economy, and he has had work published in journals such as the
American Economic Review, Journal of Public Economics, Journal of Public
Economic Theory, Journal of Urban Economics and International Tax and
Public Finance.
Alison Macintyre is a researcher at Queensland University of Technology,
Brisbane, Australia, where she is working towards her PhD on national
pride/national identity and public good problems. She is interested in
environmental and behavioural economics, and has previously explored
the relationship between institutional quality, social capital and environ-
mental outcomes.
Steve McCorriston is Professor of Economics and currently Head of
the Department of Economics at the University of Exeter, UK. His
main research interests lie in the area of public policy issues with par-
ticular reference to trade policy, foreign direct investment and agricul-
tural economics. He has published widely, including in the European
Economic Review, American Journal of Agricultural Economics, Review
of International Economics, and the Journal of Public Economic Theory.
He was awarded the prize for ‘Quality of Policy Contribution’ by the
European Association of Agricultural Economists in 2008 for the paper
(joint with MacLaren) on state trading that appeared in the European
Economic Review in 2007. He has acted as consultant for the OECD, the
FAO and the UK government.
Claudio E. Montenegro is an economist and statistician with the World
Bank’s Development Research Group. He is also an ad-honorem profes-
sor in the University of Chile’s Economic Department. He has held several
positions at the University of Chile in Santiago, the Inter-American
Development Bank and the World Bank. He has a Master’s in Economics
from the University of Maryland at College Park, USA, and a Master’s
in Statistics from George Washington University, USA. His research
interests are labour economics, international trade, poverty and applied
statistical methods. His research has been published in several books and
academic journals.

M2678 - SCHNEIDER TEXT.indd x 27/07/2011 16:16

Contributors xi
Christoph A. Schaltegger has been Professor of Economics at the

University of Lucerne, Switzerland, since 2010. He earned his PhD in
economics from the University of Basel, Switzerland, in 2003 and his
Habilitation from the University of St Gallen, Switzerland, in 2009. He
is also a lecturer in public finance at the University of St Gallen and
the University of Basel. From 2008 until 2010 he was member of the
board of economiesuisse, the umbrella organization representing the
Swiss economy. Prior to joining economiesuisse in 2008, Christoph was
a personal adviser to the Swiss Minister of Finance, Federal Councillor
Hans-Rudolf Merz.
Friedrich Schneider is Professor of Economics and Public Finance at the
Johannes Kepler University of Linz, Austria. He obtained his PhD in
Economics from the University of Konstanz in 1976 and has since held
numerous visiting and honorary positions at a number of universities. He
was Dean and Vice President of the University of Linz and President of
the Austrian Economic Association the German Economic Association
(Verein für Socialpolitik). He has published extensively in leading eco-
nomics journals including American Economic Review, Quarterly Journal
of Economics, Economic Journal and Journal of Economic Literature. He
has also published numerous books including Readings in Public Choice
and Constitutional Political Economy (together with Charles K. Rowley)
and The Economics of the Hidden Economy (editor of two volumes).
Andreas J. Schmidt completed his undergraduate studies as Diplom-
Volkswirt with a specialization in public finance and economic policy
at the University of Cologne, Germany, and at Trinity College, Dublin,
Ireland. He started his graduate studies at the Phillips-University
Marburg and is currently continuing his PhD studies at the University
of Cologne, Germany. Currently, he is acting as the managing director
of the Center for Macroeconomic Research (CMR) at the Faculty of
Business, Economics and Social Sciences of the University of Cologne.
His main research interests are copyright law and compliance, tax evasion
and deterrence measures as well as behavioral aspects of public goods
provision.
Egle Tafenau is currently a reasearch assistant at the Christian Albrechts
Universität at Kiel, Germany. In 2010 she gained her PhD in econom-
ics from the University of Tartu, Estonia. Her main research field is new
economic geography, especially the effects of policies on spatial distribu-
tion of economic activity and on social welfare. She is also interested in
questions of regional shadow economies and economic growth. In her
publications she has analysed the patterns of international trade.

M2678 - SCHNEIDER TEXT.indd xi 27/07/2011 16:16

xii Handbook on the shadow economy
Benno Torgler is a Professor of Economics in the School of Economics

and Finance, Queensland University of Technology, Brisbane, Australia.
He is also a research fellow of the Center for Research in Economics,
Management and the Arts (CREMA) in Switzerland, a CESifo Research
Affiliate (CESifo Research Network), an Executive Board Member of
the National Centre for Econometric Research (NCER) in Australia and
the editor of Economic Analysis & Policy. He has written a large number
of journal articles and books in the area of tax compliance and illegal
activities.
Colin C. Williams is Professor of Public Policy and Director of the Centre
for Regional Economic and Enterprise Development (CREED) in the
Management School at the University of Sheffield, UK. His background
lies in economic geography and, reflecting this, his research interests are
in theorising the shadow economy and understanding possible policy
responses at various spatial scales. His recent books include: The Informal
Economy in Developed Nations (2010, Routledge), The Hidden Enterprise
Culture (2008, Edward Elgar) and Rethinking the Future of Work (2007,
Routledge).
Jan Windebank is Professor of French and European Societies and Head
of the Department of French Studies in the School of Modern Languages
and Linguistics at the University of Sheffield, UK. Her research interests
are in the gender divisions of labour in Europe in general, and theorising
the configuration of the informal economy more particularly. She has
published extensively on these topics in journals across the disciplines of
economics, sociology, social policy, geography, planning and business and
management studies.

M2678 - SCHNEIDER TEXT.indd xii 27/07/2011 16:16

Introduction
Friedrich Schneider
The goal of this handbook on the shadow economy is to provide the

reader with some of the latest developments in the size and development
of the shadow economies, the driving factors of the shadow economy and
the interaction of the shadow economy, with tax morale, government
institutions and corruption. In the last ten years some progress has been
made with respect to estimating the size and development of the shadow
economies using one method over 160 countries from the end of the 1990s
up to 2007/08. This is discussed in Part I. In Part II, the regional variation
in the size and development of the shadow economy is demonstrated. Part
III deals with the shadow economy and illicit work and related activities.
In particular, the interaction between do-it-yourself (DIY) activities and
the shadow economy is shown. Part IV deals with the difficult issue of tax
morale and shadow economy, and Part V, the relation between corruption
and the shadow economy and other government institutions is shown.
Chapter 1, by Friedrich Schneider, Andreas Buehn and Claudio
Montenegro, deals with shadow economies all over the world with new
estimates for 162 countries from 1999 to 2007. This chapter presents esti-
mations of the shadow economies for 162 countries, including developing,
Eastern European, Central Asian and high income OECD countries over
1999 to 2006/07. According to the authors’ estimations, the average size of
the shadow economy (as a percent of ‘official’ GDP) in 2006 in 98 devel-
oping countries was 38.7 per cent, in 21 Eastern European and Central
Asian (mostly transition) countries 38.1 per cent and in 25 high income
OECD countries 18.7 per cent. The authors find that an increased burden
of taxation, combined with labour market regulations and the quality of
public goods and services as well as the state of the ‘official’ economy are
the driving forces of the shadow economy.
In Chapter 2, Lars P. Feld and Friedrich Schneider provide a survey
on the shadow economy and undeclared work in OECD countries.
The authors start from the observation that in most OECD countries
the policy instrument of choice to prevent people from working in the
shadows has been deterrence. While deterrence policy is well-founded

M2678 - SCHNEIDER TEXT.indd 1 27/07/2011 16:16

2 Handbook on the shadow economy
from a theoretical point of view, the empirical evidence on its success is

weak: compared to the impact of tax morale, deterrence is quantitatively
less important. Moreover, tax policies and state regulation increase the
shadow economy. The discussion of the recent literature underlines
that economic opportunities, the overall situation of the labour market
and unemployment are substitutive processes and are crucial for an
understanding of the dynamics of the shadow economy.
In Chapter 3, Kausik Chaudhuri and Friedrich Schneider provide a case
study of the size and development of the shadow economy in India. Their
analysis shows an increasing trend in the size of the hidden economy over
the years, from 8.9 per cent (1960/61) to about 23 per cent (1997/98). Given
this development of the shadow economy, they try to explain the growth
in the size of the hidden economy in India by political and institutional
factors. Their results demonstrate that per capita newspaper circulation
exerts a negative significant impact on the growth in the size of the hidden
economy. This growth is around 1 per cent significantly lower in the
election year compared to a no-election year.
In Chapter 4, Christoph A. Schaltegger provides new evidence on
the size and development and perception of the shadow economy in
Switzerland, another case study. Schaltegger states that in Switzerland
the size of the shadow economy is traditionally estimated to be below 10
per cent of the official GDP. However, the extent of the shadow economy
has nearly tripled during the last 30 years. Against this background, Swiss
politics has intensified its efforts to tackle moonlighting. Nevertheless,
according to surveys, the Swiss population does not perceive the shadow
economy as a major problem in comparison to other challenges, even
though a majority of the population categories moonlighting as a severe
offence. In any case, a successful strategy to fight the black economy
has to cope with the causes. Pursuit and higher penalty clauses fight the
symptoms but are limited in scope. Recent studies point out that the insti-
tutional framework represents an important factor which should not be
neglected.
Chapter 5, by Colin C. Williams and Jan Windebank, presents regional
variations in the nature of the shadow economy: evidence from a survey
of 27 European Union member states. In their chapter the authors show
that there are considerable regional variations in the nature of the shadow
economy in European countries and that an overall aggregated value may
be specially misleading, if there is regional variation. For example, in the
south of Italy there is a shadow economy of 30 per cent, while in the north
of Italy it is 12 per cent. The aim of their chapter is to move beyond depict-
ing the shadow economy as possessing universal characteristics and logics
and to begin charting its spatially variable configuration and character.


Introduction 3
To achieve this, the results of the first extensive survey of the nature of the
shadow economy in the western world are reported involving 26,659 face-
to-face interviews in 27 European Union countries. This reveals marked
regional and urban–rural variations in the nature of the shadow economy.
In southern European and East–Central European nations, as well as
urban areas, a greater share of all shadow work is wages employment and
the shadow economy is more likely to be conducted for chiefly economic
rationales. Meanwhile, in Nordic and Continental European nations and
rural areas more shadow work is carried out on an own-account basis and
a higher proportion undertaken for closer social relations and for reasons
other than purely monetary gain. The outcome clearly demonstrates that
it is necessary to move beyond theorisations that impose universal charac-
teristics and logics on to the shadow economy and towards an understand-
ing of the nature of work in the shadow economy sensitive to its spatially
variable meanings.
In Chapter 6, Regional Patterns of Shadow Economy: Modelling Issues
and Evidence from the European Union, written by Helmut Herwartz,
Friedrich Schneider and Egle Tafenau, for the first time a multiple indi-
cators multiple causes approach, amended to include spatial effects, is
adopted to estimate the extent of the shadow economy in the European
Union at the NUTS 2 regional level. The authors find that in the year 2004
the shadow economy was smallest in regions of the Netherlands, below 10
per cent, while Polish regions had the largest share of shadow economy,
around 30 per cent. Their results are, in general, consistent with country
level estimates from earlier studies. The variation of the extent of the
shadow economy is, in some countries, considerable. Thus policy measures
against shadow activities should take specific institutional situations into
account. Moreover, in implementing the regional policies of the European
Union, interactions with the shadow economy should be considered.
Chapter 7, written by Andreas Buehn and Alexander Karmann, deals
with shadow economy and do-it-yourself activities. The authors begin by
explaining that while much is known about the shadow economies around
the world, DIY activities have attracted less attention. Consequently, the
relationship between the shadow economy and DIY activities is almost
unexplored. Their chapter tries to fill this gap. Using appropriate determi-
nants, they estimate a structural equation model for the shadow economy
and DIY activities in Germany. Their results suggest that institutional
factors determine the shadow economy, while DIY activities respond
to individual constraints. Goods and services produced in the DIY
economy are not complemented by the demand for services in the shadow
economy. Rather, it seems that the shadow economy and DIY activities
are substitutive to each other.


Chapter 8, the contribution of Christopher Bajada, deals with the

shadow economy in the residential construction sector. A growing body
of literature on the shadow economy has focused on aggregate country-
level estimates, but little is known of the industry level measures that com-
prise the aggregate size of the shadow economy. Using the results from a
comprehensive telephone survey of home builders, Bajada provides the
first known estimates of the shadow economy in the construction sector
in Australia – a sector which has been under significant scrutiny by the
Australian Taxation Office in recent years. From these survey results, he
finds that there is considerable shadow economy activity in the residential
construction sector and that both builders and owners alike initiate nego-
tiation involving cash transactions to reduce transaction costs. The survey
results provide a number of reasons for builders to engage in shadow
economy activity and the means to reduce it.
In Chapter 9, Dominik H. Enste asks who is working illicitly and why?
Enste starts with the remark that some economic activities such as illicit
work, take place beyond official rules and generate added value, which is
not registered in official national accounts. In Germany, 20 per cent of
the population has worked illicitly during the last 12 months and 30 per
cent have employed someone illicitly. In more than 4 million households,
a person, who is not officially registered is, for example, caring for the
elderly. This adds up to 150 billion Euros of value added not covered in the
statistics. The fact that people do not act according to the law reveals that
their acceptance of official institutions and formal norms is rather low.
This may in turn give an insight into the necessity of reforms in the institu-
tional framework to decrease the attractiveness and acceptance of deviant
(shadow economy) behaviour. In order to investigate this phenomenon,
Enste undertook an analysis, which focuses on the results of representative
surveys, asking for people’s attitudes towards illicit work. One of his most
important results is that perceived norms in the social environment play
by far the most important role if one offers or demands illicit work oneself.
In Chapter 10, Gebhard Kirchgässner considers tax morale, tax evasion
and the shadow economy. Kirchgässner asks under what conditions
is moral justification of taxation possible? This question does not only
interest philosophers and economists from a scientific point of view but
has considerable practical relevance because the willingness of citizens to
pay taxes may depend upon whether they consider taxation to be morally
justified or not. Kirchgässner first considers theoretical arguments on the
role of tax morale, and when tax evasion might be considered as justified
by citizens. Then he asks how tax morale can be measured. Next, he dis-
cusses the role of tax morale in the shadow economy, before determinants
of tax morale and empirical results for the impact of tax morale and tax


Introduction 5
compliance are discussed. For a high tax morale, institutional and cultural
factors are at least as important as economic incentives.
In Chapter 11, Martin Halla investigates the link between the intrinsic
motivation to comply and compliance behaviour. Halla starts with the
observation that recent economic literature emphasises the importance
of moral considerations to explain compliance behaviour with respect
to underground activities such as tax evasion. A considerable amount of
effort has been made to identify factors (both at individual and country
levels) that affect the intrinsic motivation to comply. However, the causal
link between the intrinsic motivation to comply and actual compliance
behaviour has not been established yet. Halla provides a clear discussion
of the underlying identification problem and suggests (potentially) feasible
empirical strategies to uncover a causal effect.
In Chapter 12, by Lars P. Feld, Andreas J. Schmidt and Friedrich
Schneider, the authors show that the traditional Allingham–Sandmo-type
theory of tax evasion stresses deterrence: higher expected punishment is
alleged to cause lower tax evasion and undeclared work. Hence, deterrence
measures appear to be the most favoured policy instruments in almost all
OECD countries. However, when it comes to field data outside the USA,
empirical evidence on this relation remains rather scarce. In this chapter,
the authors survey the legal environment of deterrence policy in Germany
and present a new long-run time series of measures for deterrence as well
as for the shadow economy in Germany. These two sets of time series data
are connected by an econometric analysis on their causal relationship
using Granger causality tests.
In Chapter 13, Axel Dreher, Christos Kotsogiannis and Steve
McCorriston deal with the impact of institutions of the shadow economy
and corruption. The authors first analyse a simple model that captures the
relationship between institution quality, the shadow economy and cor-
ruption. They demonstrate that an improvement in institutional quality
reduces the shadow economy and affects the corruption market. The exact
relationship between corruption and institutional quality is, however,
ambiguous and depends on the relative effectiveness of the institutional
quality in the shadow and corruption markets. The predictions of the
model are empirically tested, by means of structural equation modelling
that treats the shadow economy and the corruption market as latent
variables, using data from OECD countries. The results show that an
improvement in institutional quality reduces the shadow economy directly
and corruption both directly and indirectly.
In Chapter 14, Benno Torgler, Friedrich Schneider and Alison
Macintyre provide an analysis of the shadow economy, voice and account-
ability and corruption. The authors use an international data set to


empirically analyse how governance or institutional quality affects par-

ticipation in the shadow economy. They focus on two key factors, namely
voice and accountability and corruption to estimate the relationship as
these factors are particularly important for understanding the level of the
changes in underground activities. The results of this investigation provide
additional empirical evidence to the literature, confirming the connection
between institution, quality and the shadow economy. The authors then
extend the analysis to include tax performance and find that lower institu-
tional quality not only affects the extent of the shadow economy but also
reduces the tax effort.


PART I
Size and development of the shadow

economies all over the world



1 Shadow economies all over the
world: new estimates for 162
countries from 1999 to 20071
Friedrich Schneider, Andreas Buehn and
Claudio E. Montenegro
1.1 INTRODUCTION
Activities associated with shadow economies are facts of life around the
world. Most societies attempt to control these activities through various
measures such as punishment, prosecution, economic growth or educa-
tion. To more effectively and efficiently allocate resources, it is crucial for
a country to gather information about the extent of the shadow economy,
its magnitude, who is engaged in underground activities, and the fre-
quency of these activities. Unfortunately, it is very difficult to get accurate
information about shadow economy activities, including the goods and
labour involved, because individuals engaged in these activities do not
wish to be identified. Hence, doing research in this area can be considered
a scientific passion for ‘knowing the unknown’.
Although substantial literature2 exists on single aspects of the hidden
or shadow economy and comprehensive surveys have been written by
Schneider and Enste (2000) and Feld and Schneider (2009; Chapter 2, this
volume), the subject is still quite controversial as there are disagreements
about the definition of shadow economy activities, estimation procedures
utilized and the use of their estimates in economic and policy analysis.3
Nevertheless, there are some indications that the shadow economy has
grown around the world, but little is known about the development and
the size of the shadow economies in developing Eastern European and
Central Asian (mostly former transition) countries, and high income
OECD countries over the period 1999 to 2006/2007. The period was
chosen as it has the most comprehensive data availability. This study is an
attempt to fill this gap by using the same estimation technique and almost
the same data sample used in Schneider and Buehn (2009) and Schneider
and Enste (2000).


The goal of this chapter is twofold: (1) to undertake the challenging

task of estimating the shadow economy for 162 countries in various stages
of development and located in several regions throughout the world4
and (2) to provide some insights about the main causes of the shadow
economy. To our knowledge, such an attempt has not been undertaken
so far; hence, we provide a unique database of the size and trends of the
shadow economy in 162 countries over the period 1999 to 2006/2007. This
is an improvement on previous work – we used the MIMIC (Multiple
Indicators Multiple Causes) estimation method for all countries, thus cre-
ating a unique data set that allows us to compare shadow economy data.
The rest of the chapter is organized as follows. In Section 1.2 we make
an attempt to define the shadow economy. This section also includes some
theoretical considerations about its determinants. Section 1.3 presents
the econometric estimation results and the calculation of the size of the
shadow economy in 162 countries over the period 1999 to 2006/2007,
depending on data availability. In Section 1.4, a summary is given and
some policy conclusions are drawn. Finally, Appendix 1.1 presents the
currency demand method approach; Appendix 1.2 presents the variable
definitions and data sources; Appendix 1.3 presents the descriptive statis-
tics; Appendix 1.4 presents additional empirical specifications for the sub-
samples of transition and high-income OECD countries; and Appendix
1.5 presents the ranking for the 162 countries in alphabetic order.
1.1.1 Summary of Results
According to our analysis, the shadow economy has reached remarkable

proportions, with a weighted average value of 17.2 per cent of official
GDP over 162 countries between 1999 and 2006/2007. The unweighted
average size of the shadow economies in the 162 selected countries
(developing Eastern European and Central Asian, as well as high-income
OECD countries) decreased from 34.1 per cent of official GDP in 1999
to 31.0 per cent of official GDP in 2007. Comparing results across the
four different specifications we calibrated, it turns out that the variation
in the estimates is relatively low across all countries. Each model predicts
a similarly sized shadow economy for each country and our results are
quite robust for most of the countries over the period 1999 to 2006/2007.
Our results further show that the driving forces of the shadow economy
include an increased burden of taxation, labour market regulations, the
quality of public goods and services and the state of the ‘official’ economy.
According to specification 3 in Table 1.1 (page 23) – the empirical model
covering a broad set of countries and all important driving forces of the
shadow economy – reducing the tax burden is the best policy measure to


Shadow economies all over the world: new estimates 11
reduce the shadow economy, followed by a lessening of fiscal and busi-

ness regulation. The estimated coefficients indicate that a unit improve-
ment of these driving forces reduces the shadow economy by 0.15 and
0.08 units, respectively. However, the relative importance of these driving
forces differs significantly across various country groups, as detailed in the
results section of the chapter.
1.2 THEORETICAL CONSIDERATIONS
This section makes an attempt to define the shadow economy and offers
theoretical considerations about the shadow economy’s most important
determinants. It addresses the difficulty encountered when attempting to
decide whether a variable is a cause or indicator of the shadow economy.
Although this section refers to various articles from the literature, it does
not review the literature comprehensively.5 Rather, it draws the most
important explanations and findings from the literature and uses them as
inputs to the choice of variables (causes and indicators) in the empirical
models.
1.2.1 Defining the Shadow Economy
Most authors trying to measure the shadow economy face the difficulty of
how to define it. One commonly used working definition is all currently
unregistered economic activities that contribute to the officially calculated
(or observed) gross national product.6 Smith (1994, p. 18) defines it as
‘market-based production of goods and services, whether legal or illegal,
that escapes detection in the official estimates of GDP.’ Or to put it in
another way, one of the broadest definitions of it includes ‘those economic
activities and the income derived from them that circumvent or otherwise
avoid government regulation, taxation or observation.’7
In this chapter, the following more specific definition of the shadow
economy is used:8 the shadow economy includes all market-based legal
production of goods and services that are deliberately concealed from
public authorities for any of the following reasons:
1. to avoid payment of income, value added or other taxes,

2. to avoid payment of social security contributions,
3. to avoid having to meet certain legal labour market standards, such as
minimum wages, maximum working hours, safety standards, and
4. to avoid complying with certain administrative procedures, such as
completing statistical questionnaires or other administrative forms.


This chapter utilizes a more precise definition of the shadow economy so

as not to deal with typical underground, classical economic crime activi-
ties, which are all illegal actions that fit the characteristics of crimes like
burglary, robbery, drug dealing and so forth. Also, this chapter does not
focus on tax evasion or tax compliance due to time and length constraints,
and the fact that tax evasion is a subject on which a lot of research has
already been undertaken.9
1.2.2 Main Causes of the Shadow Economy
Tax and social security contribution burdens

In almost all studies it has been ascertained that the overall tax and social
security contribution burdens are among the main causes for the existence
of the shadow economy.10 Since taxes affect labour–leisure choices, and
also stimulate labour supply in the shadow economy, the distortion of the
overall tax burden is a major concern for economists. The bigger the differ-
ence between the total cost of labour in the official economy and the after-
tax earnings (from work), the greater the incentive to avoid this difference
and to work in the shadow economy. Since this difference depends largely
on social security burden/payments and the overall tax burden, the latter
are key features of the existence and the increase of the shadow economy.
Empirical results showing the influence of the tax burden on the shadow
economy are provided in the studies of Schneider (1994a, 2000, 2004,
2005, 2007) and Johnson, Kaufmann and Zoido-Lobatón (1998a, 1998b);
they all found statistically significant evidence for the influence of taxation
on the shadow economy. This strong influence of indirect and direct taxa-
tion on the shadow economy is further demonstrated by discussing empiri-
cal results in the case of Austria and the Scandinavian countries. For
Austria, the driving force for shadow economic activities is the direct tax
burden (including social security payments); it has the biggest influence,
followed by the intensity of regulation and complexity of the tax system.
A similar result has been found by Schneider (1986) for the Scandinavian
countries Denmark, Norway and Sweden. In all three countries various
tax variables – average direct tax rate, average total tax rate (indirect and
direct tax rate) and marginal tax rates – have the expected positive effect
on currency demand and are highly statistically significant. These findings
are supported by studies by Kirchgässner (1983, 1984) for Germany and
by Klovland (1984) for Norway and Sweden.
The concrete measurement of the tax and social security contribution
burdens is not easy to define because the tax and social security systems
are vastly different among countries. In order to have some general
comparable proxies for this, we use the following causal variables:


1. Share of direct taxes: direct taxes as a proportion of overall taxation

(positive sign expected),
2. Size of government: general government final consumption expen-
ditures (percentage of GDP, which includes all government current
expenditures for purchases of goods and services; positive sign
expected), and
3. Fiscal freedom, a subcomponent of the Heritage Foundation’s
economic freedom index, which measures the fiscal burden in an
economy; that is, top tax rates on individual and corporate income.
The index ranges from 0 to 100, where 0 is least fiscal freedom and 100
maximum degree of fiscal freedom (negative sign expected).
Intensity of regulations
Increased intensity of regulations is another important factor that reduces
the freedom of choice for individuals engaged in the official economy.
Regulations include labour market regulations (e.g. minimum wages or
dismissal protections), trade barriers (e.g. import quotas), and labour
market restrictions for foreigners (e.g. restrictions regarding the free
movement of foreign workers). Johnson, Kaufmann and Zoido-Lobatón
(1998b) find significant overall empirical evidence of the influence of
labour regulations on the shadow economy; and the impact is clearly
described and theoretically derived in other studies, for example, for
Germany (Deregulation Commission 1991). Regulations lead to a sub-
stantial increase in labour costs in the official economy, but since most of
these costs can be shifted to the employees, these costs provide another
incentive to work in the shadow economy, where they can be avoided.
Their empirical evidence supports the model of Johnson, Kaufmann,
and Shleifer (1997), which predicts, inter alia, that countries with more
general regulation of their economies tend to have a higher share of
the unofficial economy in total GDP. Johnson, Kaufmann and Zoido-
Lobatón (1998b) conclude that it is the enforcement of regulation which
is the key factor for the burden levied on firms and individuals, and not
the overall extent of regulation – mostly not enforced – which drives firms
into the shadow economy. Friedman et al. (2000) reach a similar conclu-
sion. In their study, every available measure of regulation is significantly
correlated with the share of the unofficial economy and the estimated sign
of the relationship between their measures of regulation and the shadow
economy is unambiguously positive: more regulation is associated with
a larger shadow economy. These findings show that governments should
put more emphasis on improving enforcement of laws and regulations,
rather than increasing their number. Some governments, however, prefer
this policy option (more regulations and laws), when trying to reduce the


shadow economy, mostly because it leads to an increase in power for the

bureaucrats, to a higher rate of employment in the public sector, is easier
to implement, is easily perceived and thus positively rewarded by the
public.
To measure the intensity of regulation or the impact of regulation on
the decision of whether to work in the official or unofficial economy, this
study uses business freedom as a causal variable. Business freedom is a
subcomponent of the Heritage Foundation’s economic freedom index; it
measures the time and efforts of business activity. It ranges from 0 to 100,
where 0 is least business freedom and 100 maximum business freedom
(negative sign expected).
Public sector services

Shadow economy growth can lead to reduced state revenues, which in
turn reduce the quality and quantity of publicly provided goods and
services. Ultimately, this can lead to an increase in the tax rate for firms
and individuals in the official sector, often combined with a deterioration
in the quality of the public goods (such as public infrastructure) and of
the administration, resulting in even stronger incentives to participate in
the shadow economy. Johnson, Kaufmann and Zoido-Lobatón (1998a,
b) present a simple model of this relationship. Their findings show that
smaller shadow economies appear in countries with higher tax revenues
achieved by lower tax rates, fewer laws and regulations and less cor-
ruption. Countries where the rule of law is respected and upheld, and
financed by tax revenues, also have smaller shadow economies. Transition
countries have higher levels of regulation leading to a significantly higher
incidence of bribery, higher effective taxes on official activities and a large
discretionary regulatory framework, consequently resulting in a higher
shadow economy. Their overall conclusion is that
wealthier countries of the OECD, as well as some in Eastern Europe, find them-
selves in the ‘good equilibrium’ of relatively low tax and regulatory burden,
sizeable revenue mobilization, good rule of law and corruption control, and
a [relatively] small unofficial economy. By contrast, a number of countries in
Latin American and the former Soviet Union exhibit characteristics consistent
with a ‘bad equilibrium’: tax and regulatory discretion and burden on the firm
is high, the rule of law is weak, and there is a high incidence of bribery and
thus a relatively high share of activities in the unofficial economy. (Johnson,
Kaufmann and Zoido-Lobatón, 1998a, p. 1)
The provision and, especially, the quality of public sector services is also
a crucial causal variable for people’s decision to work or not work in the
shadow economy. To capture this effect, we have the following variable:


government effectiveness from the World Bank’s Worldwide Governance

Indicators. It captures perceptions of the quality of public services, the
quality of the civil service and the degree of its independence from political
pressures, the quality of policy formulation and implementation and the
credibility of the government’s commitment to such policies. The scores of
this index fall between –2.5 and 12.5 with higher scores corresponding to
better outcomes (negative sign expected).
Official economy
As demonstrated in a number of studies (Bajada and Schneider, 2005;
Enste and Schneider, 2006; Feld and Schneider, 2009; Chapter 2, this
volume), the situation of the official economy also plays a crucial role
in people’s decision to work or not to work in the shadow economy. In
a booming official economy, people have a lot of opportunities to earn
a good salary and ‘extra money’ in the official economy. This is not the
case in an economy facing a recession; more people try to compensate
their income losses from the official economy through additional shadow
economy activities.11 In order to capture this, we will use the following
variables:
1. GDP per capita: GDP per capita based on purchasing power parity
(PPP), measured in constant 2005 US$. PPP as gross domestic
product converted to international dollars using PPP rates (negative
sign expected),
2. Unemployment rate: unemployment, total (as a percentage of total
labour force). Unemployment refers to the share of labour force that
is without work but available for and seeking employment (positive
sign expected).
1.2.3 Indicators of the Shadow Economy
Since the shadow economy cannot be directly measured, we have to use

indicators that best capture and reflect the characteristics of shadow
economy activities. Here, we use the following indicators:
Monetary indicators
To avoid leaving traces of their transactions, people engaged in shadow
economy activities primarily use cash. Hence, most shadow economy
activities are reflected in an additional use of cash (or currency). To take
this into account, we use M0 over M1 as an indicator: M0 corresponds to
the currency outside the banks and for M1, the usual definition is M0 plus
deposits.


Labour market indicators

Shadow economy activities are also reflected in labour market indicators.
We use the following two:
1. Labour force participation rate: labour force participation rate is the

proportion of the population that is economically active, all people
who supply labour for the production of goods and services during a
specified period.
2. Growth rate of the total labour force: total labour force compromises
people aged 15 years and older who meet the International Labor
Organisation’s (ILO) definition of the economically active popula-
tion: all people who supply labour for the production of goods and
services during a specified period.
State of the official economy

Also, shadow economy activities are reflected in the state of the official
economy. We use the the annual growth rate of the GDP per capita as an
indicator variable. GDP per capita is gross domestic product converted to
international dollars using PPP rates, divided by the population.
1.2.4 The Problem of Identifying Indicator versus Cause Variables
Finally, we want to explicitly mention that when using the MIMIC

method, there is no clear division between causal variables, which directly
influence (drive) the shadow economy, and indicator variables, in which
shadow economy activities are reflected. In other words, one caveat of
the MIMIC method is that, unfortunately, there is not a clear-cut divi-
sion (or theoretically oriented guiding rule) between indicator and causal
variables. For example, when the economy is in a recession with high
unemployment, people have a stronger incentive to work in the shadow
economy; this may be seen as a causal variable, but GDP per capita and
other measures are also used as indicator variables, in which shadow
economy activities are reflected. Hence, we recognize that there is some
arbitrariness whether to use a certain variable as causal or indicator. In
this chapter, we try to be consistent, but we admit that we use GDP per
capita, for instance, as a causal variable in some cases, and as an indicator
variable in other cases (specifications 6 and 7 presented in Appendix 1.4).
The reasoning here is that we use GDP per capita as a causal control vari-
able in the specifications with a relatively heterogeneous sample, that is,
in specifications considering developing countries and the comprehensive
sample of 151/120 countries. We use the growth rate of GDP per capita
as indicator in these specifications and in the specification considering the


transition countries (specification 5). The remaining two specifications

considering the high-income OECD countries (specifications 6 and 7 pre-
sented in Appendix 1.4) use the GDP per capita as an indicator. Given that
the OECD countries are relatively homogeneous, the GDP per capita is
not necessarily required as a causal control variable in these specifications.
1.3 THE SIZE OF THE SHADOW ECONOMY IN

162 COUNTRIES
1.3.1 Econometric Methodology
Estimating the size and trend of a shadow economy is a difficult and chal-
lenging task. Methods – designed to estimate the size and trend of the
shadow economy – such as the currency demand approach or the electric-
ity approach consider just one indicator that ‘must’ capture all effects of
the shadow economy. However, it is obvious that shadow economy effects
show up simultaneously in the production, labour and money markets.
An even more important critique is that the causes that determine the
size of the shadow economy are taken into account only in some of the
monetary approach studies that usually consider one cause, the burden of
taxation. The empirical method used in this chapter is different: it is based
on the statistical theory of unobserved variables, which considers multiple
causes and indicators of the phenomenon to be measured, that is, it expli-
citly considers multiple causes leading to the existence and growth of the
shadow economy, as well as the multiple effects of the shadow economy
over time.12 In particular, we use a MIMIC model – a particular type of a
structural equations model (SEM) – to analyze and estimate the shadow
economies of 162 countries around the world.13
The main idea behind SEM is to examine the relationships among unob-
served variables with respect to the relationships among a set of observed
variables by using the covariance information of the latter. In particular,
SEM compare a sample covariance matrix, that is the covariance matrix
of the observed variables, with the parametric structure imposed on it by a
hypothesized model.14 The relationships among the observed variables are
described in terms of their covariances and it is assumed that they are gen-
erated by (a usually smaller number of) unobserved variables. In MIMIC
models, the shadow economy is the unobserved variable and is analyzed
with respect to its relationship to the observed variables using the covari-
ance matrix of the latter. For this purpose, the unobserved variable is first
linked to the observed indicator variables in a factor analytical model,
also called a measurement model. Second, the relationships between the


unobserved variable and the observed explanatory (causal) variables are

specified through a structural model. Thus, a MIMIC model is the simulta-
neous specification of a factor model and a structural model. In this sense,
the MIMIC model tests the consistency of a ‘structural’ theory through
data and is thus a rather confirmatory than exploratory technique. In
fact, in a confirmatory factor analysis a model is constructed in advance;
whether an unobserved (latent) variable or factor influences an observed
variable is specified by the researcher, and parameter constraints are often
imposed. Thus, an economic theory is tested by examining the consist-
ency of actual data with the hypothesized relationships between observed
(measured) variables and the unobserved variable.15 Such a confirmatory
factor analysis has two goals: (1) estimating the parameters (coefficients,
variances, etc.), and (2) assessing the fit of the model. Applying this to
the shadow economy research, these two goals mean: (a) measuring the
relationships of a set of observed causes and indicators to the shadow
economy (latent variable), and (b) testing if the researcher’s theory or the
derived hypotheses, as a whole, fit the data used.
Formally, the MIMIC model consists of two parts: the structural equa-
tion model and the measurement model. The structural equation model is
given by:
h 5 grx 1 ß, (1.1)
where xr 5 (x1, x2, . . ., xq) is a (1 3 q) vector and each xi, i 5 1, . . ., q

is a potential cause of the latent variable h and gr 5 (g1, g2, . . ., gq) is a
(1 3 q) vector of coefficients describing the relationships between the
latent variable and its causes. Thus, the latent variable h is determined
by a set of exogenous causes. Since these causes only partially explain the
latent variable h, the error term ß represents the unexplained component.
The variance of ß is denoted by y. F is the (q 3 q) covariance matrix
of the causes x. The measurement model represents the link between the
latent variable and its indicators, that is the latent variable determines its
indicators. The measurement model is specified by:
y 5 lh 1 e, (1.2)
where yr 5 (y1, y2, . . ., yp) is a (1 3 p) vector of several indicator varia-

bles. l is the vector of regression coefficients, and er is a (1 3 p) vector of
white noise disturbances. Their (p 3 p) covariance matrix is given by Qe.
Figure 1.1 shows the structure of the MIMIC model using a path diagram.
Using Equation (1.1) in Equation (1.2) yields a reduced form multivari-
ate regression model where the endogenous variables yj, j 5 1, . . ., p are the


Causes Indicators
x1 γ1 ς λ1 y1 ε1
γ2 λ2
x2 η y2 ε2
γq λp
…
…
xq yp εp
Figure 1.1 General structure of a MIMIC model
latent variable h’s indicators and the exogenous variables xi, i 5 1, . . ., q

its causes. This model is given by:
y 5 Px 1 z, (1.3)
where P 5 lgr is a matrix with rank equal to 1 and z 5 lß 1 e. The

error term z in Equation (1.3) is a (p 3 1) vector of linear combina-
tions of the white noise error terms ß and e from the structural equation
and the measurement model, i.e. z ~ (0, W). The covariance matrix W is
given by Cov (z) 5 E [ (lß 1 e) (lß 1 e) r ] 5 llry 1 Qe and is similarly
constrained like P. The identification and estimation of the model there-
fore requires the normalization of one of the elements of the vector l to
an a priori value (Bollen, 1989). From Equations (1.1) and (1.2) we can
derive the MIMIC model’s covariance matrix S (q) . This matrix describes
the relationship between the observed variables in terms of their covari-
ances. Decomposing the matrix yields the structure between the observed
variables and the latent variable. This covariance matrix is given by:
l (grFg 1 y) 1 Qe
S (q) 5 a b,
lgrF
(1.4)
Fglr F
where S (q) is a function of the parameters l and g and of the covari-

ances contained in F, Qe and y. If the hypothesized model is correct and
the parameters are known, the population covariance matrix S would be
exactly reproduced by estimation of the model, that is S will equal S (q) .
In practice, one does, however, not know either the population variances
and covariances, or the parameters, but uses the sample covariance matrix
of the observed variables, that is of y (vector of indicators) and x (vector of
causes), and sample estimates of the unknown parameters for estimation
of the model. The goal of the estimation procedure then is to estimate the
parameters and covariances that produce an estimate for S (q) , S^ 5 S (q^ )


that is as close as possible to the sample covariance matrix of the observed

causes and indicators. The function that measures how close a given S* is
to the sample covariance matrix S is called fitting function F (S; S*) . The
most widely used fitting function for SEM is the maximum likelihood
(ML) function:
FML 5 log 0 S (q) 0 1 tr [ SS21 (q) ] 2 log 0 S 0 2 (p 1 q) , (1.5)
where log 0 0 is the log of the respective matrix’s determinant and (p 1 q) is

the number of observable variables. In general, no closed form or explicit
solution for the structural parameters that minimize FML exists. Hence, the
estimates that minimize the fitting function are derived applying iterative
numerical procedures (see Appendix 4C in Bollen, 1989, for details).
In summary, the first step in the MIMIC model estimation is to confirm
the hypothesized relationships between the shadow economy (the latent
variable) and its causes and indicators. Once the relationships are identi-
fied and the parameters estimated, the MIMIC model results are used to
calculate the MIMIC index. However, this analysis provides only relative
estimates, not absolute, of the size of the shadow economy. Therefore, an
additional procedure, benchmarking or calibration procedure, is required
in order to calculate absolute values of the size of the shadow economy.
These values are presented in Subsection 1.3.3. The next subsection first
presents the MIMIC model estimation results.
1.3.2 Econometric Results
Remarks about the different estimation specifications

As mentioned in the Introduction, one of the major goals is to use a
coherent data set for a maximum number of countries to produce consist-
ent data of the size and trend of the shadow economies of these countries.
Doing this, we face the problem that there may still be data limitations
and due to this, we present in Table 1.1 four different estimation specifica-
tions.16 It is interesting to see which variables turn out to be significant,
especially if one uses subsamples of countries, where more and different
causal variables are available. This is the reason why we have two speci-
fications for the developing countries (covering in one case 98, in another
case 88 countries) and two specifications for samples of 120 and 151
countries. Consistent estimation for 120 and 151 countries is provided in
specification 3 and 4 in Table 1.1, from which we can also calculate the
size and trend of the shadow economy. The ideal situation would be if a
large data set (many causal and indicator variables) were available for
all countries over the entire period 1996 to 2007. Unfortunately, this is


not the case and this is (as already argued) the sole reason for the differ-
ent specifications. The sources and definitions of the variables we have
used in the estimations are elaborated in Appendix 1.2. Appendix 1.3
presents the descriptive statistics of the variables for each of the estimated
specifications.
Econometric findings
The results of our MIMIC model estimations are presented in Table 1.1.
For the total sample two estimations are shown, one for the 151 countries
over 1996 to 2007 and, with more causal variables, one sample for 120
countries over 1996 to 2006. In addition to the total sample estimations,
econometric estimations using the MIMIC approach (latent estimation
approach) are presented for 88 and 98 developing countries over the
period 1994 to 2006.17 This grouping was necessary because the available
data are different across countries and time periods. For the developing
countries, two estimations, with and without the direct tax burden rate
as causal variable are presented; without the direct tax burden rate the
number of developing countries increased from 88 to 98. Appendix 1.4
presents additional specifications for 21 Eastern European and Central
Asian (mostly former transition) countries, and 25 high-income OECD
countries. For the high-income OECD countries, one specification is esti-
mated over the period 1996 to 2006 and one over the period 1996 to 2007.
For the 21 Eastern European and Central Asian countries, the estimation
was done over the period 1994 to 2006. For the total sample of 120 and 151
countries, we use data for the period from 1996 to 2006/07.
For the developing countries, we use the following six cause variables:
(1) share of direct taxation (direct taxes in per cent of overall taxation);
(2) size of government (general government final consumption expendi-
ture, as a percentage of GDP) as proxy for indirect taxation and a vari-
able; (3) fiscal freedom (an index consisting of top individual income tax
rate, top individual corporate tax rate and total tax revenues as percent-
age of GDP) as three tax burden variables in a wide sense; (4) regulatory
intensity for state regulation; (5) the business freedom index (which is
composed of the following components: time to open a business, finan-
cial costs to start a business, minimum capital stock to start a business
and costs for obtaining a licence); and (6) the state of economy with the
two variables: the unemployment rate and GDP per capita. As indicator
variables, we use growth rate of GDP per capita, the labour force partici-
pation rate (people over 15 economically active as a percentage of total
population), and as currency we use M0 divided by M1.18 For the total
sample of 151 countries, we use as cause variables the size of the govern-
ment, the unemployment rate, government effectiveness and the GDP per


capita. As indicator variables, we use currency (M0 over M1), the growth
rate of GDP per capita and the labour force participation rate. For the
120 countries, we have additional causal variables. Here we include the
size of the government, the fiscal freedom index, the share of direct taxa-
tion, the business freedom index, the unemployment rate, government
effectiveness and the GDP per capita. As indicator variables, we use cur-
rency (M0 over M1), the growth rate of GDP per capita and the growth
rate of total labour force.
The estimation results for the 88 developing countries, including the
direct tax burden over the period 1994 up to 2006 are shown in specifica-
tion 1, and the estimation results for the 98 developing countries (exclud-
ing direct taxation) over the same period are shown in specification 2. In
both estimations, all estimated coefficients of the cause variables have the
theoretically expected signs. Except for the unemployment rate, all other
cause variables are statistically significant, at least at the 90-per cent con-
fidence level. The share of direct taxation and the size of government are
highly statistically significant, as well as the fiscal freedom and the business
freedom variable. Also, the GDP per capita is, in both equations, highly
statistically significant with the expected negative sign. In reference to the
indicator variables, the labour force participation rate and the growth rate
of GDP per capita are, in both equations, highly statistically significant.
The test statistics are also quite satisfactory.
In specifications 3 and 4, we present two estimations for samples of 120
and 151 countries, respectively. In specification 4, we present the results
of 151 countries estimated over the period 1996 to 2007. Turning first to
the causal variables, we see that the size of government has the expected
positive sign and is highly statistically significant. The same holds true for
the two variables which describe the state of the economy, the unemploy-
ment variable, statistically significant with a positive sign, and GDP per
capita, which is highly statistically significant with the expected negative
sign. With respect to the indicator variables, the growth rate of GDP per
capita and the labour force participation rate have the expected signs and
are highly statistically significant. If we reduce this sample to 120 coun-
tries, we can include more causal variables and the results are presented
in specification 3. Here, we see that we have three variables capturing
the burden of taxation: the size of government, fiscal freedom and share
of direct taxation. All three have the expected signs and are statistically
significant. As regulatory variables we have business freedom and govern-
ment effectiveness which, again, have the expected negative signs and are
statistically significant. For the state of the economy, we have the unem-
ployment rate, which is not statistically significant, and GDP per capita
with the expected negative sign, which is highly statistically significant.


Table 1.1 MIMIC model estimation results
Independent Specification 1 Specification 2 Specification 3 Specification 4

variables 88 Developing 98 Developing 120 Countries 151 Countries
Countries Countries (1996–2006) (1996–2007)
(1994–2006) (1994–2006)
Causal variables
Size of 0.15 0.14 0.10 0.05
government (5.57)*** (5.97)*** (3.77)*** (2.64)***
Share of direct 0.06 0.05
taxation (2.57)** (2.39)**
Fiscal freedom −0.03 −0.06 −0.04
(1.69)* (2.90)*** (2.08)**
Business freedom −0.05 −0.05 −0.04
(2.33)** (2.18)** (1.84)*
Unemployment −0.00 0.01 0.02 0.04
rate (0.06) (0.67) (0.89) (2.08)**
GDP per capita −0.26 −0.27 −0.33 −0.38
(6.87)*** (8.79)*** (9.15)*** (15.89)***
Government −0.04 −0.05
effectiveness (2.11)** (2.64)***
Indicator variables
Growth rate of −1.39 −1.01 −0.99 −0.79
GDP per (6.70)*** (7.88)*** (8.42)*** (10.93)***
capita
Labor force 0.02 0.05 −0.19
participation (0.14) (0.59) (3.15)***
rate
Growth rate of −0.16
labour force (1.76)*
Currency 1 1 1 1
Statistical tests
RMSEA 0.03 0.03 0.02 0.03
(p-value) (0.99) (0.99) (1.00) (1.00)
Chi-square 44.43 38.70 51.82 29.95
(p-value) (0.02) (0.00) (0.03) (0.00)
AGFI 0.98 0.98 0.98 0.99
Degrees of 27 20 35 13
freedom
Number of 741 1045 942 1563
observations
Note: Absolute z-statistics in parentheses. ***, **, * denote significance at the 1, 5 and
10% significance levels. All variables are used as their standardized deviations from the
mean. According to the MIMIC models identification rule (see also Section 1.3.1), one
indicator has to be fixed to an a priori value. We have consistently chosen the currency
variable. The degrees of freedom are determined by 0.5(p 1 q)(p 1 q1 1) – t; with
p 5 number of indicators; q 5 number of causes; t 5 the number for free parameters.


For the indicator variables, we have the same three (currency defined as
M0 over M1), labour force participation rate and GDP per capita, the
latter two being highly statistically significant and showing the expected
sign.
Summarizing the econometric (MIMIC) results, we can say that for all
groups of countries, the theoretical considerations of the causes of the
shadow economy in Section 1.2 behave according to our expectations.
Tax burden variables (direct and/or indirect and/or overall tax burden)
as well as indices measuring the fiscal freedom in a country are driving
forces for the growth of the shadow economy in all three types of coun-
tries. The same can be said about the measures of regulation (measured
with the business freedom variable, the economic freedom variable and
regulatory quality), and about the measures of the official economy, the
unemployment rate and, for the developing countries, GDP per capita.
However, the estimated coefficients are quite different in magnitude
from one specification to the next. For example, the coefficient on fiscal
freedom is twice the size in specification 5 (see Appendix 1.4) as it is in
specification 3 and the difference in the coefficient of the unemployment
rate is also significant between specifications. Because it is rather difficult
to come up with an explanation for the exact differences in the magnitude
of the coefficients, we only present a general interpretation for this obser-
vation. With respect to the indices measuring regulation in one way or the
other, that is the fiscal freedom and business/economic freedom indices,
our results suggest that regulation is a much more important determinant
in developed and transition countries than in developing ones. It seems
that – for the reason that the burden of regulation is on average higher in
developed and transition countries as more rules, regulations and admin-
istrative procedures are in place – the importance of regulation being a
determinant of the shadow economy increases with the level of develop-
ment. In contrast, in developing countries where regulation is often less
burdensome, the coefficients of the fiscal and business freedom indices
are much smaller and hence regulation is a less important determinant of
the shadow economy. Regarding the unemployment rate, the results are
comparable. It does not influence the shadow economies in developing
countries (specifications 1 and 2), but determine the shadow economies
in transition and OECD countries (specifications 5 and 6/7 in Appendix
1.4, respectively). It seems that higher unemployment rates due to, on
average, more regulated and hence less flexible labour markets, signifi-
cantly contribute to the size and trend of the shadow economies in OECD
countries. In developing countries, however, unemployment is not a sig-
nificant determinant of the shadow economy. In these countries, income
earned in the shadow economy guarantees the subsistence of families.


Finally, comparing specifications 5 and 7 in Appendix 1.4 it turns out

that the unemployment rate is a more important determinant in OECD
than in transition countries.
The estimation results further show a slightly different impact of
‘policy’ causal variables compared to non-policy ‘economic’ causal
variables across the different groups of countries. In general, economic
variables, that is the level of development and the state of the economy
measured by the GDP per capita and the unemployment rate, are very
important determinants of the shadow economy. The estimated coef-
ficients indicate that an improvement of economic conditions would
reduce the size of the shadow economy. Of course, for the unemployment
rate this is only true for transition and highly developed OECD countries
(see Appendix 1.4). Comparing the impact of policy variables, such as
the different measures of the tax burdern and regulation on the shadow
economy, across the estimated specifications also reveals interesting
results. For example, one could expect that a reduction of the regulatory
burden and improvement of business/economic freedom in transition and
highly developed OECD countries (see Appendix 1.4), leads to a much
higher reduction of the shadow economy than it would in developing
countries (which is clearly indicated by the much larger coefficients of
these variables). Fiscal freedom, however, is similarily important across
all groups of countries.
The actual interpretation of the estimation parameters is straightfor-
ward and similar to that of regression coefficients in conventional regres-
sion analysis. Their magnitude shows the resulting change of the shadow
economy for a unit change in a causal variable, all other variables being
equal. Thus according to specification 1, a 1 per cent reduction of the
size of government, the proxy for the burden of indirect taxation, would
on average reduce the shadow economy by 0.14 per cent in developing
countries. In transition countries the 1 per cent reduction of the size of
government reduces the shadow economy by 0.18 per cent (see Appendix
1.4). This means that reducing the burden of indirect taxation in develop-
ing and transition countries by 1 per cent would on average reduce the
shadow economy from 38.6 and 38.1 per cent in 2006 to 38.4 and 37.9
per cent in 2007. An improvement in the measures reflecting regulatory
burden in these countries, that is the business and economic freedom
indices of the Heritage Foundation, by one unit reduces the shadow
economy by 0.05 per cent in developing countries and 0.09 per cent in
transition countries. This effect is stronger in developed countries. In these
countries, an improvement in the business environment – measured by the
business freedom index of the Heritage Foundation – by one unit reduces
the shadow economy by 0.23 per cent. Thus, in developed countries the


shadow economy would on average decrease from 18.7 per cent in 2006 to
18.4 per cent in 2007.
1.3.3 The Size of the Shadow Economies for 162 Countries from 1999 to
2006/07
The estimated MIMIC coefficients allow us to determine only relatively

estimated sizes of the shadow economy, which describe the pattern of
the shadow economy in a particular country over time. In order to cal-
culate the size and trend of the shadow, we must convert the MIMIC
index into ‘real world’ figures measured as percentage of official GDP.
This final step requires an additional benchmarking or calibration pro-
cedure. Unfortunately, no consensus exists in the literature as to which
benchmarking procedure should be utilized. The methodology we use
was promoted by Dell’Anno and Schneider (2006), Dell’Anno (2007) and
Dell’Anno and Solomon (2008). In the first step, the MIMIC model index
of the shadow economies is calculated using the Structural Equation (1.1),
that is by multiplying the coefficients of the significant causal variables
with the respective time series. For the numerical example of specification
1 the structural equation is given as:
|
ht 5 0.14 # x1t 2 0.06 # x2t 2 0.05 # x3t 2 0.27 # x4t19 (1.6)
Second, this index is converted into absolute values of the shadow econo-
mies, which take up a base value in a particular base year. The base values
necessary for this final step of the calibration procedure are from the year
2000 and taken from Schneider (2007), who presents estimates of the
shadow economies in 145 countries around the world using the MIMIC
and the currency demand approach.20 Thus, the size of the shadow
economy h^ t at time t is given as:
|
ht
h^ t 5 | h*2000, (1.7)
h2000
where |ht denotes the value of the MIMIC index at t according to Equation
(1.6), |
h2000 is the value of this index in the base year 2000, and h*2000 is
the exogenous estimate (base value) of the shadow economies in 2000.
Applying this benchmarking procedure, the final estimates of the shadow
economies are calculated for each specification 1 to 7.21
Due to shortcomings in the MIMIC and currency demand methods,
comparisons of geographically and developmentally different countries


are not precise, especially with respect to the ranking and size of the
shadow economies in these countries over time.22 Due to these short-
comings, a detailed discussion of the (relative) ranking of the size of the
shadow economies is not conducted.
Developing countries23
Two different sets of estimates are presented for the developing countries
due to the fact that the direct taxation variable was only available for a
smaller country sample (88 developing countries instead of 98); the calibra-
tion of the size and trend of the shadow economy of the developing coun-
tries is done for both sets of estimations. In Table 1.2, the size and trend of
the shadow economy of 88 developing countries are presented – ordered
with respect to the size of the shadow economy – using the MIMIC estima-
tion for the developing countries with the direct taxation, specification 1. It
thus includes a direct measure of the tax burden, in addition to the rather
indirect tax burden measure, size of government, which we solely use in
specification 2. Although including direct taxation reduces the sample size
by 10 countries, specification 1 is superior to specification 2 because it has
been shown in various studies that the direct tax burden is a major driving
force for the shadow economy. Hence, if possible, this variable should be
included in an empirical model measuring the shadow economy.
The sizes of the shadow economies of those 88 countries are in
both samples – calculated according to specification 1 and 2 – quite
similar. According to specification 1, the average size – taking the simple
unweighted mean – of the shadow economy of these 88 developing coun-
tries was 36.2 per cent in 1999 and modestly decreased to 34.2 per cent in
the year 2006. The lowest size of the average shadow economy over the
period 1999 to 2006 includes China, Singapore and Vietnam; the middle
position includes Jamaica, Bangladesh and Papua New Guinea with
35.8, 35.9 and 35.9 per cent. The highest shadow economies include Peru,
Panama and Bolivia with 59.0, 63.9 and 66.9 per cent.
In Table 1.3 we present the size of the shadow economy in 98 developing
countries (excluding the direct taxation variable in the MIMIC estima-
tion). If we consider the trend of the simple unweighted average of these
98 countries over time, in the year 1999 the size was 37.0 per cent and
modestly decreased to 35.1 per cent in the year 2006. The three countries
with the smallest shadow economies are China, Singapore and Vietnam
with an average country size of 12.8, 13.0 and 15.2 per cent respectively.24
The middle of the distribution includes Cape Verde, Jamaica and Nepal
with an average size of 35.7, 35.7 and 36.6 per cent of GDP. The highest
shadow economies include Peru, Panama and Bolivia with a size of 58.7,
63.5 and 66.6 per cent of GDP.


Table 1.2 Ranking of 88 developing countries according to size of the

shadow economy
No. Country Years Country

Average
1999 2000 2001 2002 2003 2004 2005 2006
1 China 13.0 13.1 13.1 13.1 12.9 12.6 12.3 12.1 12.8
2 Singapore 13.0 13.1 13.4 13.4 13.1 12.7 12.5 12.5 13.0
3 Vietnam 15.7 15.6 15.6 15.4 15.4 15.2 14.8 – 15.4
4 Mongolia 18.2 18.4 18.4 18.1 17.9 17.4 16.8 – 17.9
5 Bahrain 18.7 18.4 18.5 18.4 18.0 17.6 17.2 – 18.1
6 Saudi Arabia 18.7 18.4 18.9 18.9 18.0 17.6 17.2 – 18.2
7 Iran, Islamic 18.4 18.9 19.0 18.5 18.3 17.9 17.9 17.7 18.3
Rep.
8 Jordan 19.5 19.4 19.3 19.1 19.0 18.4 17.6 17.7 18.7
9 Oman 19.5 18.9 18.8 18.9 19.0 18.6 18.0 – 18.8
10 Syrian Arab 19.0 19.3 19.2 19.2 19.2 19.4 18.8 – 19.2
Republic
11 Indonesia 19.3 19.4 19.5 19.6 19.7 19.9 19.1 – 19.5
12 Chile 19.8 19.8 19.8 20.1 19.7 19.4 19.0 18.8 19.5
13 Kuwait 21.6 21.7 21.7 21.8 21.4 21.0 20.7 20.5 21.3
14 Israel 22.2 21.9 22.4 22.9 22.5 21.8 21.2 21.0 22.0
15 India 23.3 23.1 22.9 22.6 22.2 21.9 21.6 21.4 22.4
16 Mauritius 23.4 23.1 22.7 22.8 22.8 22.8 22.8 22.7 22.9
17 Argentina 25.2 25.4 26.2 25.7 25.0 24.5 24.2 – 25.2
18 Costa Rica 25.8 26.2 26.8 27.1 26.9 26.5 25.8 25.4 26.3
19 United Arab 26.5 26.4 27.1 27.5 26.5 25.5 25.1 – 26.4
Emirates
20 Yemen, Rep. 27.9 27.4 27.6 27.7 27.5 27.3 26.8 – 27.5
21 Malta 27.5 27.1 27.9 27.6 28.1 28.2 27.6 27.6 27.7
22 South Africa 28.6 28.4 28.5 28.3 28.5 28.0 27.4 27.0 28.1
23 Cyprus 29.4 28.7 28.9 29.2 29.2 28.3 28.3 28.4 28.8
24 Lao PDR 30.9 30.6 30.0 29.8 29.2 28.9 28.2 – 29.6
25 Mexico 30.5 30.1 30.1 30.2 30.1 29.9 29.3 – 30.0
26 Namibia 31.9 31.4 31.2 30.4 30.4 29.3 28.8 – 30.5
27 Lesotho 31.4 31.3 31.2 31.1 31.2 30.7 30.2 29.4 30.8
28 Malaysia 31.7 31.1 32.0 31.9 31.8 31.4 31.0 – 31.6
29 Dominican 32.2 32.1 32.4 32.3 32.3 31.7 30.9 31.1 31.9
Republic
30 Fiji 32.6 33.6 33.3 32.5 32.3 31.4 31.1 – 32.4
31 Cameroon 33.5 32.8 33.2 33.1 33.1 32.8 32.2 – 32.9
32 Algeria 34.8 34.1 34.2 34.0 32.8 32.2 30.9 30.7 33.0
33 Botswana 33.6 33.4 33.6 33.5 33.3 32.8 32.7 – 33.3
34 Lebanon 33.8 34.1 34.4 34.1 33.9 33.5 33.3 33.4 33.8
35 Venezuela, RB 33.4 33.6 33.7 34.5 35.6 34.1 32.7 – 33.9


Table 1.2 (continued)

Average
1999 2000 2001 2002 2003 2004 2005 2006
36 Sudan 34.1 – – – – – – – 34.1
37 Ecuador 35.7 34.4 34.9 34.4 34.4 33.5 32.2 – 34.2
38 Kenya 34.5 34.3 34.2 35.2 36.0 35.1 33.8 32.7 34.5
39 Egypt, Arab 34.8 35.1 35.4 35.4 34.9 34.6 34.1 33.5 34.7
Rep.
40 Togo 35.0 35.1 35.7 34.6 35.3 35.2 35.4 – 35.2
41 Morocco 36.6 36.4 35.9 35.5 35.0 35.0 34.8 – 35.6
42 Mauritania 35.9 36.1 36.1 35.7 35.9 35.1 34.5 – 35.6
43 Jamaica 36.3 36.4 36.1 36.1 35.9 35.6 34.9 34.8 35.8
44 Bangladesh 35.9 35.6 35.3 36.1 36.4 36.3 35.5 – 35.9
45 Papua New 35.7 36.1 – – – – – – 35.9
Guinea
46 Trinidad and – – 37.2 36.7 36.3 35.7 35.2 35.0 36.0
Tobago
47 Cape Verde – – – – – – 36.8 35.6 36.2
48 Nepal 36.9 36.8 37.0 37.2 36.8 36.9 36.1 36.1 36.7
49 Pakistan 37.3 36.8 37.4 37.3 36.7 36.1 36.5 35.9 36.7
50 Tunisia 38.5 38.4 38.3 38.5 38.0 37.3 36.4 36.1 37.7
51 Colombia 39.5 39.1 39.0 39.0 38.6 37.7 36.8 35.8 38.2
52 Paraguay 38.5 39.8 39.3 39.7 38.6 37.8 37.7 36.9 38.6
53 Burundi 38.8 38.9 39.1 39.2 39.4 39.5 39.7 – 39.2
54 Ethiopia 39.9 40.3 39.5 40.2 40.8 39.1 37.6 – 39.6
55 Brazil 40.6 39.8 39.7 39.8 40.2 39.3 39.0 – 39.8
56 Mozambique 41.0 40.3 40.6 40.1 40.0 39.7 39.6 – 40.2
57 Rwanda 40.2 40.3 40.7 39.7 40.8 40.3 39.6 – 40.2
58 Madagascar 39.9 39.6 39.6 42.4 42.0 40.8 39.6 – 40.6
59 Niger 41.4 41.9 41.0 40.1 39.5 40.8 39.9 – 40.7
60 Burkina Faso 41.0 41.4 41.4 41.7 41.2 41.0 40.6 – 41.2
61 Swaziland 43.6 41.4 41.2 40.9 40.7 – – – 41.6
62 Malawi 40.5 40.3 41.7 42.7 42.2 42.1 41.9 – 41.6
63 Mali 42.5 42.3 41.1 41.3 41.8 41.5 41.6 41.3 41.7
64 Philippines 43.8 43.3 43.0 42.4 41.9 41.0 40.1 39.6 41.9
65 Guinea 43.0 42.8 42.6 42.4 42.0 41.8 41.0 – 42.2
66 Ghana 42.5 41.9 41.5 42.4 43.3 42.7 42.0 – 42.3
67 Côte d’Ivoire 42.0 43.2 42.8 43.6 43.7 43.4 43.4 43.9 43.3
68 Uganda 43.6 43.1 43.3 43.9 43.5 43.5 42.9 42.4 43.3
69 Sierra Leone 44.6 44.0 43.4 42.8 42.7 43.0 – – 43.4
70 Sri Lanka 44.9 44.6 44.5 44.0 43.1 43.0 42.7 44.4 43.9
71 Chad 46.8 46.2 45.5 45.4 44.4 41.2 41.9 – 44.5
72 Senegal 45.0 45.1 44.2 – – – – – 44.8



Average
1999 2000 2001 2002 2003 2004 2005 2006
73 Nicaragua 45.6 45.2 45.1 45.3 45.2 45.0 44.1 44.0 44.9
74 El Salvador 46.3 46.3 46.3 46.4 46.0 45.8 45.4 45.2 46.0
75 Central African – – 45.1 45.5 46.4 46.3 46.9 – 46.0
Republic
76 Congo, Rep. 50.3 48.2 48.1 48.0 47.5 47.7 45.9 – 48.0
77 Congo, 48.0 – – – – – – – 48.0
Dem. Rep.
78 Benin 50.8 49.8 49.9 49.7 49.5 49.4 48.3 – 49.6
79 Honduras 50.2 49.6 50.2 50.2 49.9 49.2 48.6 – 49.7
80 Zambia 50.3 50.0 49.7 50.0 49.9 50.4 49.3 48.8 49.8
81 Uruguay 50.4 51.1 51.4 52.1 51.1 49.1 47.7 47.1 50.0
82 Myanmar 51.9 52.6 51.5 50.8 50.0 49.8 48.5 – 50.7
83 Guatemala 51.0 51.5 52.5 52.1 52.0 51.4 50.4 49.9 51.3
84 Thailand 52.9 52.6 52.6 52.1 51.3 51.4 51.2 51.1 51.9
85 Tanzania 59.2 58.3 57.6 56.8 56.4 55.4 54.8 – 56.9
86 Peru 60.2 59.9 60.4 59.4 59.3 58.5 57.7 57.0 59.0
87 Panama 64.5 64.1 64.9 65.3 64.3 62.8 61.1 – 63.9
88 Bolivia 67.3 67.1 67.6 67.9 68.0 67.4 65.7 64.4 66.9
Time Average 36.2 35.9 36.0 35.9 35.7 35.2 34.6 34.2
Large shadow economies in some developing countries are only to some

extent an issue of tax burden and regulation, given the simple fact that the
limited local economy means that citizens are often unable to earn a living
wage in a legitimate manner. Working in the shadow economy is often
the only way of achieving a minimal standard of living. It should also be
noted that the average size of the Asian shadow economy is smaller than
the average shadow economy of African and Latin American countries.
Eastern European and Central Asian (mostly former transition) countries

The measurement of the size and trend of the shadow economies in the
transition countries has been undertaken since the late 1980s starting with
the works of Kaufmann and Kaliberda (1996), Johnson, Kaufmann and
Shleifer (1997) and Lack (2000). They all use the physical input (electri-
city) method and come up with larger figures than ours.25 In the works of
Alexeev and Pyle (2003) and Belev (2003) the above mentioned studies are
critically evaluated arguing that the estimated sizes of the unofficial econo-
mies are to a large extent a historical phenomenon and partly determined
by institutional factors.


Table 1.3 Ranking of 98 developing countries according to size of the

shadow economy

Average
1999 2000 2001 2002 2003 2004 2005 2006
1 China 13.0 13.1 13.1 13.1 12.9 12.6 12.2 12.1 12.8
2 Singapore 13.1 13.1 13.4 13.3 13.1 12.8 12.5 12.4 13.0
3 Vietnam 15.8 15.6 15.5 15.3 15.3 15.1 14.7 14.5 15.2
4 Mongolia 18.4 18.4 18.4 18.3 18.1 17.6 17.0 16.8 17.9
5 Bahrain 18.6 18.4 18.3 18.2 17.7 17.4 17.2 – 18.0
6 Saudi Arabia 18.6 18.4 18.7 18.6 17.9 17.5 17.1 17.2 18.0
7 Iran, 18.6 18.9 19.0 18.5 18.2 17.8 17.6 17.4 18.3
Islamic Rep.
8 Oman 19.3 18.9 18.8 18.7 18.6 18.4 17.9 17.5 18.5
9 Jordan 19.5 19.4 19.2 19.0 19.0 18.4 17.6 17.6 18.7
10 Syrian Arab 19.0 19.3 19.2 18.9 19.2 19.4 18.8 18.5 19.0
Republic
11 Chile 19.9 19.8 19.8 20.0 19.6 19.3 18.8 18.5 19.5
12 Indonesia 19.3 19.4 19.5 19.9 19.7 19.6 19.1 19.1 19.5
13 Kuwait 20.0 20.1 20.2 20.2 19.7 19.4 19.2 19.1 19.7
14 Israel 22.3 21.9 22.3 23.0 22.8 22.1 21.4 21.0 22.1
15 India 23.3 23.1 22.9 22.6 22.2 21.8 21.4 21.2 22.3
16 Mauritius 23.3 23.1 22.5 22.5 22.5 22.5 22.4 22.1 22.6
17 Argentina 25.2 25.4 26.1 25.9 25.2 24.6 24.3 24.1 25.1
18 United Arab 26.6 26.4 26.9 27.3 26.4 25.6 25.5 24.2 26.1
Emirates
19 Costa Rica 25.8 26.2 26.7 27.0 26.7 26.4 25.7 25.3 26.2
20 Yemen, Rep. 27.7 27.4 27.5 27.4 27.2 26.9 26.3 26.1 27.1
21 Malta 27.6 27.1 27.9 27.6 27.9 28.1 27.6 27.5 27.7
22 South Africa 28.7 28.4 28.4 28.1 28.3 28.0 27.3 26.8 28.0
23 Cyprus 29.3 28.7 28.9 29.2 29.2 28.5 28.4 28.4 28.8
24 Lao PDR 30.9 30.6 30.1 30.0 29.6 29.3 28.5 28.4 29.7
25 Mexico 30.3 30.1 30.4 30.4 30.2 29.7 29.0 28.5 29.8
26 Namibia 32.0 31.4 31.2 30.5 30.2 29.5 29.2 28.4 30.3
27 Lesotho 31.6 31.3 31.2 30.9 31.0 30.7 30.3 29.5 30.8
28 Malaysia 31.4 31.1 31.8 32.0 31.9 31.5 30.9 30.5 31.4
29 Dominican 32.2 32.1 32.2 32.1 31.9 31.7 30.9 31.3 31.8
Republic
30 Equatorial 33.1 32.8 32.2 32.1 31.8 31.1 31.6 31.2 32.0
Guinea
31 Cameroon 32.9 32.8 32.9 32.7 32.5 32.0 31.5 31.4 32.3
32 Fiji 32.7 33.6 33.6 32.9 32.7 31.7 31.3 31.2 32.5
33 Algeria 34.9 34.1 34.2 34.0 33.2 32.5 31.2 31.1 33.1
34 Guyana 33.3 33.6 33.0 33.4 33.7 33.3 32.7 32.6 33.2



Average
1999 2000 2001 2002 2003 2004 2005 2006
35 Botswana 33.8 33.4 33.6 33.3 33.0 32.9 32.9 32.9 33.2
36 Lebanon 33.6 34.1 34.2 34.0 33.8 33.4 33.0 32.9 33.6
37 Trinidad and 35.2 34.4 34.7 34.4 33.7 33.1 32.4 32.0 33.7
Tobago
38 Ecuador 35.6 34.4 34.4 34.2 34.2 33.4 32.2 32.3 33.8
39 Venezuela, RB 33.6 33.6 33.7 34.7 36.1 34.5 32.7 31.8 33.8
40 Sudan 34.1 – – – – – – – 34.1
41 Kenya 34.3 34.3 34.3 35.2 35.5 34.9 33.5 32.3 34.3
42 Egypt, 34.9 35.1 35.3 35.5 34.8 34.5 33.9 32.8 34.6
Arab Rep.
43 Togo 34.4 35.1 35.4 34.5 34.9 35.0 35.0 34.6 34.9
44 Mauritania 35.5 36.1 36.0 35.8 35.8 35.1 34.4 31.7 35.1
45 Morocco 36.2 36.4 36.0 35.7 35.1 35.1 34.7 33.5 35.4
46 Bangladesh 35.8 35.6 35.3 35.7 35.9 35.7 34.9 34.7 35.5
47 Papua New 35.1 36.1 – – – – – – 35.6
Guinea
48 Cape Verde 36.7 36.1 35.5 35.5 35.8 35.6 35.5 34.8 35.7
49 Jamaica 36.1 36.4 36.4 36.5 35.7 35.4 34.5 34.8 35.7
50 Nepal 36.9 36.8 36.9 36.9 36.7 36.8 36.1 36.0 36.6
51 Pakistan 37.3 36.8 37.5 37.5 36.9 36.2 36.5 35.8 36.8
52 Tunisia 38.5 38.4 38.2 38.4 37.6 36.9 35.7 35.6 37.4
53 Colombia 39.4 39.1 39.0 38.7 38.5 38.0 36.9 36.1 38.2
54 Paraguay 38.1 39.8 39.5 39.6 38.7 37.9 37.8 36.7 38.5
55 Suriname 39.9 39.8 39.5 39.1 38.8 37.9 37.4 36.6 38.6
56 Guinea 39.5 39.6 39.3 39.1 39.2 38.7 38.2 37.6 38.9
57 Ethiopia 40.2 40.3 39.4 39.6 40.4 38.9 37.5 36.4 39.1
58 Burundi 39.1 39.2 39.3 39.4 39.5 39.6 39.7 40.3 39.5
59 Brazil 40.6 39.8 40.0 40.1 39.8 39.3 38.9 38.5 39.6
60 Mozambique 41.1 40.3 40.4 39.8 39.8 39.7 38.9 38.6 39.8
61 Guinea-Bissau 40.3 39.6 39.7 40.1 39.9 39.8 39.9 39.6 39.9
62 Rwanda 40.5 40.3 40.6 39.9 40.7 40.2 39.3 39.1 40.1
63 Madagascar 39.6 39.6 39.7 41.9 42.1 40.5 39.4 39.4 40.3
64 Niger 41.7 41.9 40.9 40.3 39.7 40.7 39.7 38.6 40.4
65 Swaziland 43.5 41.4 41.3 40.9 40.2 40.1 39.3 38.9 40.7
66 Burkina Faso 41.0 41.4 41.6 41.6 40.6 40.4 40.0 39.4 40.8
67 Mali 42.5 42.3 41.0 41.2 41.3 41.3 41.2 40.9 41.5
68 Malawi 40.3 40.3 41.6 42.6 42.6 42.7 41.9 40.7 41.6
69 Ghana 42.0 41.9 41.7 41.8 42.6 42.1 41.5 40.3 41.8
70 Philippines 44.1 43.3 43.0 42.4 41.8 40.9 40.1 39.7 41.9
71 Uganda 44.1 43.1 43.2 43.6 43.2 43.0 42.3 41.8 43.0



Average
1999 2000 2001 2002 2003 2004 2005 2006
72 Côte d’Ivoire 42.2 43.2 43.2 44.0 44.2 44.0 43.5 43.9 43.5
73 Sri Lanka 44.7 44.6 44.3 43.8 42.7 42.9 42.5 43.6 43.7
74 Belize 45.4 43.8 43.7 44.1 43.7 43.3 42.9 42.7 43.7
75 Gambia, The 45.4 45.1 44.5 45.7 44.2 43.1 42.9 41.9 44.1
76 Chad 46.3 46.2 45.7 45.5 44.5 41.1 42.1 42.5 44.2
77 Senegal 45.9 45.1 44.7 45.3 44.4 43.9 42.6 42.7 44.3
78 Nicaragua 45.9 45.2 45.0 45.0 44.6 44.3 43.6 43.5 44.7
79 Sierra Leone 46.3 45.6 44.9 44.2 44.1 44.2 44.3 43.6 44.7
80 Central African – – 44.7 45.3 46.1 46.0 46.9 45.3 45.7
Republic
81 El Salvador 46.2 46.3 46.5 46.3 46.0 45.8 45.0 44.6 45.8
82 Angola 49.7 48.9 48.1 47.3 46.6 45.8 45.0 43.6 46.9
83 Gabon 47.3 48.0 48.1 47.7 46.9 47.0 46.1 46.4 47.2
84 Congo, Rep. 50.0 48.2 48.1 47.9 47.7 47.7 46.4 44.9 47.6
85 Congo, 48.0 – – – – – – – 48.0
Dem. Rep.
86 Zambia 49.1 48.9 48.1 48.1 47.8 48.7 47.6 46.9 48.2
87 Honduras 49.7 49.6 49.8 49.6 49.2 48.7 47.5 46.9 48.9
88 Myanmar 51.6 52.6 51.5 50.7 49.0 49.1 47.8 – 50.3
89 Uruguay 50.7 51.1 51.9 52.6 52.1 49.8 48.3 47.7 50.5
90 Benin 51.4 50.6 50.6 51.1 51.1 50.9 49.8 49.7 50.6
91 Guatemala 51.5 51.5 52.7 52.2 52.1 51.8 50.5 49.9 51.5
92 Thailand 53.0 52.6 52.6 52.1 51.2 51.2 51.0 50.6 51.8
93 Haiti 54.9 55.4 56.4 56.7 56.5 55.9 55.9 56.0 56.0
94 Nigeria 58.0 57.9 57.8 57.6 56.3 55.1 53.8 53.0 56.2
95 Tanzania 59.4 58.3 57.7 57.2 56.8 56.3 54.9 54.2 56.8
96 Peru 60.2 59.9 60.3 59.0 58.8 57.8 57.4 56.4 58.7
97 Panama 64.8 64.1 64.7 65.1 64.4 63.5 61.7 60.0 63.5
98 Bolivia 67.0 67.1 67.7 67.7 67.8 67.1 64.7 63.4 66.6
Time Average 37.0 36.7 36.8 36.8 36.5 36.1 35.5 35.1
In Table 1.4, the size and trend of the shadow economies in 21 Eastern
European and Central Asian (mostly former transition) countries are shown
as a percentage of GDP.26 If we first consider the unweighted average shadow
economy of these 21 Eastern European and Central Asian countries, it was
36.9 per cent in 1999 and decreased to 32.6 per cent in 2007. The three coun-
tries with the smallest shadow economies are the Slovak and Czech Republics,
and Hungary with an average size over the period 1999 to 2007 of 18.1, 18.4
and 24.4 per cent. Croatia, Romania and Albania are in the middle with 32.1,


Table 1.4 Ranking of 21 transition countries according to size of the

shadow economy

Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
1 Slovak 18.9 18.9 18.8 18.6 18.3 18.1 17.6 17.2 16.8 18.1
Republic
2 Czech 19.3 19.1 18.9 18.8 18.7 18.4 17.8 17.3 17.0 18.4
Republic
3 Hungary 25.4 25.1 24.8 24.5 24.4 24.1 24.0 23.7 23.7 24.4
4 Slovenia 27.3 27.1 26.7 26.6 26.4 26.2 25.8 25.3 24.7 26.2
5 Poland 27.7 27.6 27.7 27.7 27.5 27.3 26.9 26.4 26.0 27.2
6 Latvia 30.8 30.5 30.1 29.8 29.4 29.0 28.4 27.7 27.2 29.2
7 Estonia – 32.7 32.4 32.0 31.4 31.1 30.5 29.8 29.5 31.2
8 Turkey 32.7 32.1 32.8 32.4 31.8 31.0 30.0 29.5 29.1 31.3
9 Lithuania 33.8 33.7 33.3 32.8 32.0 31.7 31.0 30.4 29.7 32.0
10 Croatia 33.8 33.4 33.2 32.6 32.1 31.7 31.3 30.8 30.4 32.1
11 Romania 34.3 34.4 33.7 33.5 32.8 32.0 31.7 30.7 30.2 32.6
12 Albania 35.7 35.3 34.9 34.7 34.4 33.9 33.7 33.3 32.9 34.3
13 Bulgaria 37.3 36.9 36.6 36.1 35.6 34.9 34.1 33.5 32.7 35.3
14 Macedonia 39.0 38.2 39.1 38.9 38.4 37.4 36.9 36.0 34.9 37.6
15 Kyrgyz 41.4 41.2 40.8 41.4 40.5 39.8 40.1 39.8 38.8 40.4
Republic
16 Kazakhstan 43.8 43.2 42.5 42.0 41.1 40.6 39.8 38.9 38.4 41.1
17 Tajikistan 43.5 43.2 42.9 42.7 42.1 41.7 41.5 41.2 41.0 42.2
18 Russian 47.0 46.1 45.3 44.5 43.6 43.0 42.4 41.7 40.6 43.8
Federation
19 Moldova 45.6 45.1 44.1 44.5 44.6 44.0 43.4 44.3 – 44.5
20 Ukraine 52.7 52.2 51.4 50.8 49.7 48.8 47.8 47.3 46.8 49.7
21 Georgia 68.3 67.3 67.2 67.2 65.9 65.5 65.1 63.6 62.1 65.8
Time Average 36.9 36.3 36.1 35.8 35.3 34.8 34.3 33.7 32.6
32.6 and 34.3 per cent. The highest shadow economies include Moldova,
Ukraine and Georgia with 44.5, 49.7 and 65.8 per cent, respectively.
High-income OECD countries

The size and trend of the shadow economies of 25 high-income OECD
countries over the period 1999 to 2007 is shown in Table 1.5. We first
analyze the average size of the shadow economies of the 25 high-income
OECD countries. The unweighted average was 17.7 per cent in 1999, and
decreased to 16.6 per cent in 2007. Some high-income OECD countries, like
Portugal, have ups and downs, while others (like Belgium and Australia)


Table 1.5 Ranking of 25 OECD countries according to the size of the

shadow economy

Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
1 Switzerland 8.8 8.6 8.6 8.6 8.8 8.6 8.5 8.3 8.1 8.5
2 United States 8.8 8.7 8.8 8.8 8.7 8.6 8.5 8.4 8.4 8.6
3 Luxembourg 10.0 9.8 9.8 9.8 9.8 9.8 9.7 9.6 9.4 9.7
4 Austria 10.0 9.8 9.7 9.8 9.8 9.8 9.8 9.6 9.5 9.8
5 Japan 11.4 11.2 11.2 11.3 11.2 10.9 10.7 10.4 10.3 11.0
6 New Zealand 13.0 12.8 12.6 12.4 12.2 12.0 12.1 12.1 12.0 12.4
7 United 12.8 12.7 12.6 12.6 12.5 12.4 12.4 12.3 12.2 12.5
Kingdom
8 Netherlands 13.3 13.1 13.1 13.2 13.3 13.2 13.2 13.2 13.0 13.2
9 Australia 14.4 14.3 14.3 14.1 13.9 13.7 13.7 13.7 13.5 14.0
10 France 15.7 15.2 15.0 15.1 15.0 14.9 14.8 14.8 14.7 15.0
11 Iceland 16.0 15.9 15.8 16.0 15.9 15.5 15.1 15.0 15.0 15.6
12 Canada 16.3 16.0 15.9 15.8 15.7 15.6 15.5 15.3 15.3 15.7
13 Ireland 16.1 15.9 15.9 15.9 16.0 15.8 15.6 15.5 15.4 15.8
14 Germany 16.4 16.0 15.9 16.1 16.3 16.1 16.0 15.6 15.3 16.0
15 Finland 18.4 18.1 17.9 17.8 17.7 17.6 17.4 17.1 17.0 17.7
16 Denmark 18.4 18.0 18.0 18.0 18.0 17.8 17.6 17.0 16.9 17.7
17 Norway 19.2 19.1 19.0 19.0 19.0 18.5 18.5 18.2 18.0 18.7
18 Sweden 19.6 19.2 19.1 19.0 18.7 18.5 18.6 18.2 17.9 18.8
19 Belgium 22.7 22.2 22.1 22.0 22.0 21.8 21.8 21.4 21.3 21.9
20 Spain 23.0 22.7 22.4 22.4 22.4 22.5 22.4 22.4 22.2 22.5
21 Portugal 23.0 22.7 22.6 22.7 23.0 23.1 23.3 23.2 23.0 23.0
22 Korea, Rep. 28.3 27.5 27.3 26.9 26.8 26.5 26.3 25.9 25.6 26.8
23 Italy 27.8 27.1 26.7 26.8 27.0 27.0 27.1 26.9 26.8 27.0
24 Greece 28.5 28.7 28.2 28.0 27.4 27.1 26.9 26.4 26.5 27.5
25 Mexico 30.8 30.1 30.3 30.4 30.5 30.1 29.9 29.2 28.8 30.0
Time Average 17.7 17.4 17.3 17.3 17.3 17.1 17.0 16.8 16.6
show a steady decrease. The countries with the smallest shadow economies
include Switzerland, the United States and Luxembourg with an average
size over the period 1999 to 2007 of 8.5, 8.6 and 9.7 per cent, respectively.
The largest shadow economies among these 25 high-income OECD coun-
tries include Mexico with 30.0, Greece with 27.5 and Italy with 27.0 per cent.
The total sample of 151 (120) countries

Finally, we present the calibrated estimates of the shadow economies based
on a broad sample of 151 countries (Table 1.7), and with a larger number
of cause variables, calibrated estimates for 120 countries (Table 1.6).


Table 1.6 Ranking of 120 countries according to the size of the shadow
economy

Average
1999 2000 2001 2002 2003 2004 2005 2006
1 Switzerland 8.7 8.6 8.5 8.6 8.7 8.6 8.4 8.2 8.6
2 United States 8.8 8.7 8.7 8.7 8.6 8.5 8.4 8.3 8.6
3 Austria 10.1 9.8 9.7 9.7 9.7 9.7 9.4 9.3 9.7
4 Luxembourg 10.1 9.8 9.9 9.8 9.9 9.7 9.5 9.4 9.8
5 Japan – – 11.1 11.1 11.0 10.8 10.5 10.3 10.8
6 New Zealand 12.9 12.8 12.6 12.3 12.2 12.1 12.2 12.3 12.4
7 United 12.7 12.7 12.7 12.6 12.5 12.3 12.3 12.3 12.5
Kingdom
8 China 13.2 13.1 13.1 13.1 13.0 12.7 12.4 12.3 12.9
9 Singapore 13.2 13.1 13.4 13.3 13.0 12.6 12.4 12.2 12.9
10 Netherlands 13.3 13.1 13.1 13.2 13.3 13.1 12.9 13.0 13.1
11 Australia 14.3 14.3 14.0 13.7 13.5 13.4 13.4 13.3 13.8
12 France 15.4 15.2 15.1 15.0 14.9 14.7 14.5 14.5 14.9
13 Vietnam 15.7 15.6 15.6 15.4 15.2 15.1 14.6 – 15.3
14 Iceland 16.0 15.9 15.8 16.0 16.0 15.3 14.8 14.8 15.6
15 Canada 16.3 16.0 15.8 15.5 15.5 15.5 15.2 15.2 15.6
16 Germany 16.4 16.0 15.7 15.8 16.0 15.7 15.4 15.0 15.8
17 Ireland 16.0 15.9 16.0 16.0 16.0 15.9 15.4 15.3 15.8
18 Hong Kong, 17.2 16.6 16.5 16.5 16.2 15.9 15.4 – 16.3
China
19 Finland 18.4 18.1 17.8 17.7 17.5 17.4 17.1 16.8 17.6
20 Denmark 18.2 18.0 18.0 17.9 17.9 17.8 17.5 16.9 17.8
21 Mongolia 18.2 18.4 18.3 17.9 17.7 17.3 16.8 – 17.8
22 Bahrain 18.4 18.4 18.3 18.2 17.8 17.4 17.1 – 17.9
23 Iran, 18.8 18.9 18.9 18.4 18.1 17.8 17.8 17.5 18.3
Islamic Rep.
24 Slovak 19.0 18.9 18.8 18.7 18.4 18.0 17.3 17.0 18.3
Republic
25 Saudi Arabia 18.6 18.4 18.9 19.2 18.1 17.6 17.3 – 18.3
26 Oman 19.2 18.9 18.6 18.7 18.7 18.6 18.2 – 18.7
27 Jordan 19.5 19.4 19.3 19.0 18.9 18.4 17.9 17.6 18.7
28 Sweden 19.4 19.2 19.2 19.1 18.8 18.5 18.3 18.0 18.8
29 Czech 19.4 19.1 19.3 19.3 19.2 18.8 18.2 17.7 18.9
Republic
30 Syrian Arab 19.1 19.3 19.0 18.9 19.0 18.8 18.6 – 18.9
Republic
31 Norway 19.1 19.1 19.2 19.2 19.2 18.9 18.6 18.5 19.0
32 Kuwait 19.9 20.0 20.0 20.1 19.3 18.8 18.3 18.1 19.3
33 Indonesia 19.6 19.4 19.4 19.6 19.3 19.3 18.6 – 19.3



Average
1999 2000 2001 2002 2003 2004 2005 2006
34 Chile 19.8 19.8 19.7 19.8 19.6 19.4 19.1 19.1 19.5
35 Israel 22.3 21.9 22.2 22.5 22.3 21.7 21.3 20.9 21.9
36 Belgium 22.5 22.2 22.2 22.3 22.1 21.8 21.4 20.9 21.9
37 India 23.3 23.1 22.9 22.7 22.2 21.9 21.4 21.0 22.3
38 Portugal 22.8 22.7 22.7 22.7 22.7 22.7 22.7 22.3 22.6
39 Spain 23.1 22.7 22.6 22.8 22.8 23.0 22.8 22.9 22.8
40 Mauritius 23.5 23.1 22.9 23.1 22.7 22.6 22.7 22.7 22.9
41 Hungary 25.4 25.1 25.0 24.8 24.5 24.1 23.8 23.6 24.5
42 Costa Rica 26.2 26.2 26.3 26.3 26.0 25.8 25.3 24.8 25.9
43 Argentina 25.1 25.4 26.3 27.5 26.6 25.7 24.8 – 25.9
44 United Arab 26.1 26.4 27.1 27.6 26.4 25.3 25.1 – 26.3
Emirates
45 Slovenia 27.4 27.1 27.0 26.8 26.4 26.2 25.8 25.5 26.5
46 Korea, Rep. 28.1 27.5 27.4 26.9 26.9 26.8 26.4 26.1 27.0
47 Italy 27.7 27.1 27.0 26.9 26.9 27.0 26.9 26.7 27.0
48 Poland 27.9 27.6 27.5 27.5 27.3 27.0 26.3 26.2 27.2
49 Greece 28.1 28.7 28.0 27.9 27.1 26.7 26.1 25.5 27.3
50 Malta 27.5 27.1 27.4 27.2 27.6 27.5 27.2 27.1 27.3
51 Yemen, Rep. 27.8 27.4 27.6 27.6 27.4 27.4 26.8 – 27.4
52 South Africa 28.4 28.4 28.4 28.1 27.8 27.2 26.4 26.1 27.6
53 Latvia 30.9 30.5 30.2 29.9 29.3 29.2 28.3 27.8 29.5
54 Lao PDR 30.8 30.6 30.1 29.8 29.4 29.0 28.4 – 29.7
55 Mexico 30.8 30.1 29.9 30.0 29.7 29.5 29.0 – 29.8
56 Lesotho 31.7 31.3 31.0 30.9 30.9 30.2 29.9 28.7 30.6
57 Namibia 31.8 31.4 31.4 31.0 30.6 29.1 29.0 – 30.6
58 Turkey 32.7 32.1 32.9 32.0 31.2 30.4 29.6 28.7 31.2
59 Malaysia 31.9 31.1 31.7 31.4 31.3 30.9 30.6 – 31.3
60 Estonia – 32.7 32.4 32.1 31.6 31.3 30.0 29.4 31.4
61 Dominican 32.4 32.1 32.8 32.7 32.2 32.2 31.2 31.2 32.1
Republic
62 Fiji 32.7 33.6 33.1 32.1 32.1 31.4 30.9 – 32.3
63 Croatia 33.9 33.4 33.0 32.2 32.1 31.8 31.0 30.6 32.3
64 Lithuania 33.9 33.7 33.2 32.9 32.2 32.0 31.1 30.6 32.4
65 Cameroon 33.6 32.8 32.8 32.5 32.3 31.9 31.5 – 32.5
66 Romania 34.1 34.4 33.7 33.0 32.8 31.5 30.7 29.9 32.5
67 Algeria 34.2 34.1 33.9 33.5 32.3 31.8 30.9 31.0 32.7
68 Botswana 33.6 33.4 33.5 33.5 33.0 32.5 32.4 – 33.1
69 Ecuador 34.6 34.4 34.3 33.7 33.2 32.3 31.1 – 33.4
70 Lebanon 33.7 34.1 34.2 33.8 33.5 32.9 32.7 33.0 33.5
71 Kenya 34.1 34.3 33.8 34.6 34.9 33.8 32.6 31.2 33.7



Average
1999 2000 2001 2002 2003 2004 2005 2006
72 Venezuela, RB 33.4 33.6 33.6 35.2 37.1 35.3 33.7 – 34.6
73 Albania – – – 35.5 35.0 34.7 34.3 33.4 34.6
74 Egypt, 35.1 35.1 35.2 35.8 35.3 34.9 34.2 33.3 34.9
Arab Rep.
75 Jamaica 36.3 36.4 36.2 36.2 34.9 34.5 34.1 33.5 35.3
76 Bulgaria 37.2 36.9 36.4 36.0 35.6 35.1 33.9 33.4 35.6
77 Papua 35.1 36.1 – – – – – – 35.6
New Guinea
78 Morocco 36.7 36.4 35.8 35.6 35.1 34.6 35.4 – 35.7
79 Trinidad and – – 36.3 36.9 36.0 35.7 35.2 34.2 35.7
Tobago
80 Bangladesh 36.2 35.6 35.4 36.2 36.3 36.2 35.5 – 35.9
81 Cape Verde – – – – – – 36.8 35.6 36.2
82 Pakistan 37.2 36.8 37.2 37.1 36.4 35.6 35.5 34.7 36.3
83 Nepal 37.1 36.8 36.7 36.9 36.5 36.6 36.1 35.9 36.6
84 Tunisia 38.5 38.4 38.0 38.3 37.7 37.2 36.5 36.0 37.6
85 Colombia 39.5 39.1 39.0 39.1 38.5 37.5 36.5 35.2 38.1
86 Paraguay 38.6 39.8 39.1 39.8 38.7 38.2 37.9 36.7 38.6
87 Ethiopia 40.2 40.3 39.5 40.1 40.6 38.9 37.6 – 39.6
88 Brazil 40.8 39.8 39.8 39.9 40.1 39.2 38.8 – 39.8
89 Kyrgyz 41.4 41.2 40.9 40.5 40.1 39.7 39.3 39.0 40.3
Republic
90 Mali 42.4 42.3 40.8 40.4 40.4 40.7 40.7 40.6 41.0
91 Ghana 42.3 41.9 41.6 41.5 41.0 40.8 38.8 – 41.1
92 Madagascar 40.3 39.6 39.1 44.6 43.0 41.1 40.0 – 41.1
93 Burkina Faso 41.2 41.4 41.7 41.8 41.0 41.0 40.6 – 41.3
94 Kazakhstan 43.3 43.2 42.7 42.4 41.7 41.0 40.3 39.3 41.7
95 Philippines 44.1 43.3 43.0 42.3 41.7 41.2 40.0 39.4 41.9
96 Malawi 40.8 40.3 42.0 43.7 43.0 42.2 42.1 – 42.0
97 Guinea 43.3 42.9 42.5 42.4 42.0 41.6 41.0 – 42.3
98 Tajikistan 43.7 43.2 42.9 42.3 41.7 41.4 – – 42.5
99 Uganda 44.3 43.1 42.7 43.1 42.5 42.8 42.5 41.4 42.8
100 Russian 46.7 46.1 44.8 43.8 42.8 42.2 41.3 41.0 43.6
Federation
101 Sri Lanka 45.3 44.6 44.8 44.1 43.6 43.6 43.1 43.7 44.1
102 Chad 46.4 46.2 45.5 45.3 44.5 41.2 42.1 – 44.5
103 Côte d’Ivoire 41.7 43.2 44.3 45.5 45.6 45.5 45.6 46.3 44.7
104 Nicaragua 45.8 45.2 44.9 45.2 44.9 44.4 43.8 43.7 44.7
105 Senegal 45.0 45.1 44.1 – – – – – 44.7
106 Sierra Leone 47.0 46.3 45.5 44.8 44.4 44.1 – – 45.3



Average
1999 2000 2001 2002 2003 2004 2005 2006
107 El Salvador 46.7 46.3 46.1 45.9 45.2 45.0 44.4 43.8 45.4
108 Central – – 43.6 44.3 46.5 46.6 46.9 – 45.6
African
Republic
109 Congo, Rep. 49.7 48.2 47.6 46.9 46.6 46.0 44.8 – 47.1
110 Honduras 50.5 49.6 50.0 49.9 49.2 48.5 47.8 – 49.3
111 Zambia 51.3 50.8 50.2 50.4 50.0 50.0 48.9 47.9 49.9
112 Ukraine 52.6 52.2 51.7 51.2 50.6 49.6 48.1 47.3 50.4
113 Benin 51.8 50.7 50.8 50.8 50.5 50.6 49.8 – 50.7
114 Uruguay 50.4 51.1 51.3 53.4 53.3 50.5 48.3 47.7 50.8
115 Thailand 53.2 52.6 52.5 51.5 50.1 49.8 49.3 48.9 51.0
116 Guatemala 51.3 51.5 52.4 51.8 51.3 50.8 50.0 49.1 51.0
117 Tanzania 59.0 58.3 57.7 56.8 56.6 55.7 55.2 – 57.0
118 Peru 60.0 59.9 60.2 58.4 58.1 57.3 57.1 56.3 58.4
119 Georgia 68.0 67.3 66.9 66.6 65.7 65.5 64.5 63.4 66.0
120 Bolivia 67.2 67.1 67.4 67.9 68.1 67.2 63.8 62.6 66.4
Time Average 31.7 31.6 31.4 31.3 31.0 30.5 29.8 28.4
Table 1.6 presents the calibrated estimation of the size of the shadow
economy for 120 countries over the period 1999 to 2006. For these 120
countries, we have additional cause variables. As a consequence, the
results are somewhat different. For the year 1999, when using the 151
country sample, the unweighted average of the shadow economy was
33.3 per cent, and when using the sample with only 120 countries the
same average is 31.7 per cent, which is a rather modest difference.27 This
difference is due to the smaller number of countries in the second sample
because countries with large shadow economies such as Haiti, Zimbabwe
and Azerbaijan are not included in the second sample of 120 countries.
Alternatively, the differences may occur because specification 3 (120 coun-
tries) uses more cause variables and is thus the more specific empirical
model. Given the significance of the additionally included variables and
the confirmatory nature of the MIMIC model, specification 3 is superior
to specification 4. The better empirical model of specification 3 thus, hope-
fully, leads to more precise estimates of the size of the shadow economies
in these 120 countries.
The countries with the smallest shadow economies among the 120 coun-
tries are Switzerland, the United States and Austria with an average value
over the period 1999 to 2006 of 8.6, 8.6 and 9.7 per cent. In the middle, we


Table 1.7 Ranking of 151 countries according to the size of the shadow
economy

Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
1 Switzerland 8.8 8.6 8.6 8.6 8.8 8.6 8.5 8.3 8.1 8.5
2 United States 8.8 8.7 8.8 8.8 8.7 8.6 8.5 8.4 8.4 8.6
3 Luxembourg 10.0 9.8 9.8 9.8 9.8 9.8 9.7 9.6 9.4 9.7
4 Austria 10.0 9.8 9.7 9.8 9.8 9.8 9.8 9.6 9.5 9.7
5 Japan 11.4 11.2 11.2 11.3 11.2 10.9 10.7 10.4 10.3 11.0
6 New Zealand 13.0 12.8 12.6 12.4 12.2 12.0 12.1 12.1 12.0 12.4
7 Macao, China 13.3 13.1 13.0 12.9 12.5 12.1 11.9 11.7 11.1 12.4
8 United 12.8 12.7 12.6 12.6 12.5 12.4 12.4 12.3 12.2 12.5
Kingdom
9 China 13.2 13.1 13.0 12.9 12.8 12.6 12.5 12.2 11.9 12.7
10 Singapore 13.3 13.1 13.3 13.3 13.1 12.8 12.7 12.4 12.2 12.9
11 Netherlands 13.3 13.1 13.1 13.2 13.3 13.2 13.2 13.2 13.0 13.2
12 Australia 14.4 14.3 14.3 14.1 13.9 13.7 13.7 13.7 13.5 14.0
13 France 15.7 15.2 15.0 15.1 15.0 14.9 14.8 14.8 14.7 15.0
14 Vietnam 15.8 15.6 15.5 15.3 15.2 15.1 14.7 14.6 14.4 15.1
15 Iceland 16.0 15.9 15.8 16.0 15.9 15.5 15.1 15.0 15.0 15.6
16 Canada 16.3 16.0 15.9 15.8 15.7 15.6 15.5 15.3 15.3 15.7
17 Ireland 16.1 15.9 15.9 15.9 16.0 15.8 15.6 15.5 15.4 15.8
18 Germany 16.4 16.0 15.9 16.1 16.3 16.1 16.0 15.6 15.3 16.0
19 Hong Kong, 17.0 16.6 16.6 16.6 16.4 15.9 15.5 15.0 14.7 16.0
China
20 Mongolia 18.4 18.4 18.3 18.0 17.7 17.4 17.1 16.7 16.4 17.6
21 Finland 18.4 18.1 17.9 17.8 17.7 17.6 17.4 17.1 17.0 17.7
22 Denmark 18.4 18.0 18.0 18.0 18.0 17.8 17.6 17.0 16.9 17.7
23 Bahrain 18.6 18.4 18.2 18.0 17.8 17.4 17.1 – – 17.9
24 Saudi Arabia 18.7 18.4 18.7 19.2 18.3 17.7 17.4 17.4 16.8 18.1
25 Slovak 18.9 18.9 18.8 18.6 18.3 18.1 17.6 17.2 16.8 18.1
Republic
26 Iran, Islamic 19.1 18.9 19.0 18.7 18.2 17.9 18.1 17.7 17.3 18.3
Rep.
27 Czech 19.3 19.1 18.9 18.8 18.7 18.4 17.8 17.3 17.0 18.4
Republic
28 Oman 19.1 18.9 18.5 18.5 18.4 18.3 18.0 17.6 – 18.4
29 Jordan 19.4 19.4 19.2 18.9 18.7 18.3 18.0 17.5 17.2 18.5
30 Norway 19.2 19.1 19.0 19.0 19.0 18.5 18.5 18.2 18.0 18.7
31 Sweden 19.6 19.2 19.1 19.0 18.7 18.5 18.6 18.2 17.9 18.8
32 Quatar – 19.0 19.3 19.0 19.6 17.4 18.4 – – 18.8
33 Indonesia 19.7 19.4 19.4 19.3 19.1 18.8 18.6 18.3 17.9 18.9



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
34 Syrian Arab 19.3 19.3 19.2 19.1 19.3 19.1 19.0 18.7 18.5 19.0
Republic
35 Chile 19.9 19.8 19.6 19.6 19.4 19.1 18.9 18.7 18.5 19.3
36 Kuwait 20.1 20.1 20.2 20.3 19.3 18.8 18.1 17.9 – 19.3
37 Belgium 22.7 22.2 22.1 22.0 22.0 21.8 21.8 21.4 21.3 21.9
38 Israel 22.7 21.9 22.3 22.7 22.7 22.1 21.8 21.2 20.7 22.0
39 India 23.2 23.1 22.8 22.6 22.3 22.0 21.7 21.2 20.7 22.2
40 Spain 23.0 22.7 22.4 22.4 22.4 22.5 22.4 22.4 22.2 22.5
41 Mauritius 23.3 23.1 22.9 23.0 22.7 22.4 22.4 22.2 21.9 22.7
42 Portugal 23.0 22.7 22.6 22.7 23.0 23.1 23.3 23.2 23.0 23.0
43 Hungary 25.4 25.1 24.8 24.5 24.4 24.1 24.0 23.7 23.7 24.4
44 Taiwan 25.7 25.4 25.7 25.4 25.2 24.7 24.5 24.2 23.9 25.0
45 Argentina 25.2 25.4 26.1 27.6 26.4 25.5 24.7 23.8 23.0 25.3
46 Costa Rica 26.1 26.2 26.4 26.4 26.1 25.9 25.6 25.0 24.0 25.8
47 United Arab 26.3 26.4 27.0 27.4 26.3 25.4 24.8 23.5 – 25.9
Emirates
48 Slovenia 27.3 27.1 26.7 26.6 26.4 26.2 25.8 25.3 24.7 26.2
49 Bahamas, The 26.3 26.2 26.4 26.5 27.0 27.4 26.7 26.2 26.2 26.5
50 Korea, Rep. 28.3 27.5 27.3 26.9 26.8 26.5 26.3 25.9 25.6 26.8
51 Italy 27.8 27.1 26.7 26.8 27.0 27.0 27.1 26.9 26.8 27.0
52 Yemen, Rep. 27.7 27.4 27.3 27.2 27.0 27.0 26.6 26.8 26.8 27.1
53 Poland 27.7 27.6 27.7 27.7 27.5 27.3 26.9 26.4 26.0 27.2
54 Malta 27.4 27.1 27.3 27.3 27.5 27.6 27.3 27.0 26.5 27.2
55 South Africa 28.4 28.4 28.4 28.0 27.8 27.1 26.5 26.0 25.2 27.3
56 Greece 28.5 28.7 28.2 28.0 27.4 27.1 26.9 26.4 26.5 27.5
57 Cyprus 29.2 28.7 28.2 27.8 28.2 28.1 27.7 27.3 26.5 28.0
58 Bhutan 29.6 29.4 29.2 29.1 28.7 28.7 28.3 28.2 27.7 28.7
59 Latvia 30.8 30.5 30.1 29.8 29.4 29.0 28.4 27.7 27.2 29.2
60 Maldives 30.3 30.3 30.0 29.4 29.2 28.9 29.6 29.3 28.6 29.5
61 Lao PDR 30.9 30.6 30.2 30.0 29.8 29.4 28.9 28.4 28.0 29.6
62 Mexico 30.8 30.1 30.3 30.4 30.5 30.1 29.9 29.2 28.8 30.0
63 Namibia 31.4 31.4 31.2 31.3 30.7 29.7 29.6 28.8 28.5 30.3
64 Lesotho 31.7 31.3 31.1 31.0 30.7 30.1 30.2 29.3 28.8 30.5
65 Malaysia 32.2 31.1 31.6 31.5 31.2 30.7 30.4 30.0 29.6 30.9
66 Brunei 31.3 31.1 31.0 30.2 29.9 31.2 31.8 30.8 31.2 30.9
Darussalam
67 Estonia – 32.7 32.4 32.0 31.4 31.1 30.5 29.8 29.5 31.2
68 Turkey 32.7 32.1 32.8 32.4 31.8 31.0 30.0 29.5 29.1 31.3
69 Equatorial 32.7 32.8 32.0 31.5 31.2 30.8 30.5 30.6 30.1 31.4
Guinea



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
70 Dominican 32.4 32.1 32.4 32.1 32.1 32.4 31.7 31.0 30.5 31.9
Republic
71 Cameroon 33.3 32.8 32.4 32.1 31.7 31.6 31.6 31.4 31.4 32.0
72 Lithuania 33.8 33.7 33.3 32.8 32.0 31.7 31.0 30.4 29.7 32.0
73 Croatia 33.8 33.4 33.2 32.6 32.1 31.7 31.3 30.8 30.4 32.1
74 Ecuador 34.2 34.4 33.7 33.3 32.8 31.6 30.8 30.4 30.4 32.4
75 Fiji 32.9 33.6 33.3 32.6 32.5 31.9 31.4 31.0 32.6 32.4
76 Algeria 34.2 34.1 33.8 33.3 32.5 31.7 31.1 31.0 31.2 32.6
77 Romania 34.3 34.4 33.7 33.5 32.8 32.0 31.7 30.7 30.2 32.6
78 Botswana 33.9 33.4 33.2 33.3 33.0 32.8 32.7 32.3 31.9 33.0
79 Lebanon 34.1 34.1 33.7 33.5 33.2 32.4 32.4 32.8 32.0 33.1
80 Kenya 33.7 34.3 34.0 34.8 34.6 33.7 32.7 31.1 29.5 33.2
81 Trinidad and 34.7 34.4 34.3 34.4 33.4 33.1 32.9 31.9 31.5 33.4
Tobago
82 Solomon 31.7 33.4 34.5 34.8 34.7 33.8 33.4 33.2 32.7 33.6
Islands
83 Bosnia & 34.3 34.1 34.0 33.9 33.5 33.6 33.2 32.9 32.8 33.6
Herzegovina
84 Libyan Arab 34.7 35.1 34.5 33.8 34.9 33.9 33.1 32.0 30.9 33.7
Jamahiria
85 Guyana 33.4 33.6 33.3 33.7 33.9 33.4 34.3 33.8 34.0 33.7
86 Venezuela, RB 33.8 33.6 33.5 35.5 36.9 34.9 33.5 32.0 30.9 33.8
87 Albania 35.7 35.3 34.9 34.7 34.4 33.9 33.7 33.3 32.9 34.3
88 Jamaica 36.4 36.4 36.2 36.2 34.4 33.9 34.0 32.9 32.5 34.8
89 Egypt, Arab 35.5 35.1 35.2 35.7 35.4 35.0 34.8 34.1 33.1 34.9
Rep.
90 Morocco 36.5 36.4 35.7 35.5 35.0 34.2 34.9 33.1 33.1 34.9
91 Bangladesh 36.0 35.6 35.5 35.7 35.6 35.5 35.1 34.5 34.1 35.3
92 Bulgaria 37.3 36.9 36.6 36.1 35.6 34.9 34.1 33.5 32.7 35.3
93 Cape Verde 36.5 36.1 35.9 35.9 35.7 35.8 35.4 34.1 33.4 35.4
94 Pakistan 37.0 36.8 37.0 36.8 36.2 35.3 34.9 33.8 33.6 35.7
95 Papua New 35.5 36.1 36.8 37.1 37.1 37.0 37.2 37.1 36.5 36.7
Guinea
96 Nepal 37.2 36.8 36.7 37.1 36.9 36.8 36.7 36.3 36.0 36.7
97 Tunisia 38.7 38.4 37.8 37.8 37.4 36.9 36.7 35.9 35.4 37.2
98 Colombia 39.4 39.1 38.9 38.9 37.9 37.1 36.1 35.1 33.5 37.3
99 Macedonia, 39.0 38.2 39.1 38.9 38.4 37.4 36.9 36.0 34.9 37.6
FYR
100 Suriname 39.7 39.8 39.3 38.9 38.1 36.9 36.5 35.9 35.1 37.8
101 Ethiopia 40.6 40.3 39.5 39.6 40.1 38.6 37.7 36.3 35.1 38.7



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
102 Comoros 39.3 39.6 39.0 37.7 37.6 39.0 38.0 38.4 39.4 38.7
103 Paraguay 38.0 39.8 39.7 40.1 39.1 38.3 38.2 37.4 – 38.8
104 Guinea 39.7 39.6 39.3 38.7 38.8 38.5 38.4 38.9 39.2 39.0
105 Brazil 40.8 39.8 39.9 39.9 39.6 38.6 38.4 37.8 36.6 39.0
106 Burundi 39.1 39.5 39.6 39.4 39.6 39.6 39.7 39.6 39.6 39.5
107 Eritrea 38.1 40.3 39.4 39.4 40.3 40.6 40.5 41.2 41.4 40.1
108 Kyrgyz 41.4 41.2 40.8 41.4 40.5 39.8 40.1 39.8 38.8 40.4
Republic
109 Burkina Faso 41.3 41.4 41.3 41.4 40.3 40.1 39.7 39.7 39.6 40.5
110 Ghana 42.0 41.9 41.8 41.6 41.3 40.9 39.5 38.6 38.3 40.6
111 Mali 42.5 42.3 40.8 40.2 39.9 40.6 40.1 39.9 39.9 40.7
112 Madagascar 40.1 39.6 38.7 44.8 43.4 41.6 40.8 39.8 38.5 40.8
113 Guinea-Bissau 40.4 39.6 39.6 40.7 41.5 41.9 41.7 41.5 41.6 40.9
114 Kazakhstan 43.8 43.2 42.5 42.0 41.1 40.6 39.8 38.9 38.4 41.1
115 Philippines 43.8 43.3 43.0 42.5 42.0 41.6 40.1 39.5 38.3 41.6
116 Malawi 39.9 40.3 42.5 44.4 43.4 42.5 42.6 41.3 39.4 41.8
117 Tajikistan 43.5 43.2 42.9 42.7 42.1 41.7 41.5 41.2 41.0 42.2
118 Uganda 43.5 43.1 42.9 42.9 42.5 42.4 42.2 41.0 40.3 42.3
119 Belize 45.2 43.8 43.3 43.4 42.3 42.0 42.1 41.7 42.0 42.9
120 Chad 45.8 46.2 45.5 45.1 44.2 41.5 41.1 41.7 42.2 43.7
121 Senegal 45.0 45.1 44.5 45.1 44.4 43.2 42.3 42.4 41.7 43.8
122 Russian 47.0 46.1 45.3 44.5 43.6 43.0 42.4 41.7 40.6 43.8
Federation
123 Sri Lanka 45.2 44.6 44.6 44.1 43.8 43.9 43.4 42.9 42.2 43.9
124 Armenia 46.6 46.3 45.4 44.5 43.9 43.6 42.7 42.1 41.1 44.0
125 Liberia 44.2 43.2 43.2 43.1 45.0 45.4 44.9 44.5 44.2 44.2
126 Gambia, The 46.1 45.1 44.7 47.1 45.4 43.8 43.6 42.4 40.9 44.3
127 Nicaragua 45.7 45.2 45.3 45.5 45.0 44.2 43.8 43.5 43.1 44.6
128 Central 42.8 42.6 43.1 44.0 46.9 47.3 46.9 45.9 45.1 45.0
African
Republic
129 El Salvador 46.5 46.3 46.2 45.6 45.2 44.9 44.5 43.8 43.0 45.1
130 Côte d’Ivoire 41.4 43.2 44.3 45.5 46.0 46.1 46.3 46.8 47.0 45.2
131 Sierra Leone 48.6 48.6 47.6 45.4 44.8 44.4 44.3 43.6 42.9 45.6
132 Congo, Rep. 49.5 48.2 47.2 46.8 46.8 46.2 44.7 43.3 44.6 46.4
133 Belarus 48.3 48.1 47.9 47.6 47.0 46.1 45.2 44.2 43.3 46.4
134 Angola 48.8 48.8 48.4 47.4 47.3 47.1 45.0 44.0 42.1 46.6
135 Zambia 49.3 48.9 48.3 48.1 47.5 46.8 46.3 45.0 43.9 47.1
136 Congo, Dem. 47.2 48.0 48.2 48.1 47.1 46.9 46.8 46.8 46.7 47.3
Rep.



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
137 Gabon 46.2 48.0 47.4 47.6 47.5 48.0 47.7 48.0 47.3 47.5
138 Honduras 50.3 49.6 49.7 49.6 48.9 48.3 47.3 46.1 45.1 48.3
139 Cambodia 50.4 50.1 49.6 50.0 49.2 48.8 47.8 46.8 46.0 48.7
140 Ukraine 52.7 52.2 51.4 50.8 49.7 48.8 47.8 47.3 46.8 49.7
141 Benin 51.2 50.2 49.8 49.6 49.3 49.5 49.8 49.6 49.1 49.8
142 Guatemala 51.6 51.5 51.6 51.2 50.7 50.5 50.2 49.0 47.9 50.5
143 Thailand 53.4 52.6 52.4 51.5 50.2 49.6 49.0 48.5 48.2 50.6
144 Uruguay 50.5 51.1 51.7 54.0 53.6 51.1 49.2 48.5 46.1 50.6
145 Haiti 54.8 55.4 56.1 56.5 56.4 57.4 57.1 57.0 57.1 56.4
146 Tanzania 58.6 58.3 57.7 56.9 56.6 56.0 55.4 54.7 53.7 56.4
147 Peru 60.1 59.9 60.2 59.1 58.6 57.9 57.2 55.7 53.7 58.0
148 Azerbaijan 61.0 60.6 60.3 60.0 59.1 58.6 56.7 54.0 52.0 58.0
149 Zimbabwe 59.6 59.4 61.5 62.8 63.7 62.3 62.0 62.3 62.7 61.8
150 Georgia 68.3 67.3 67.2 67.2 65.9 65.5 65.1 63.6 62.1 65.8
151 Bolivia 67.0 67.1 67.6 67.7 67.7 66.9 64.3 62.8 63.5 66.1
Time Average 33.3 33.0 32.9 32.9 32.6 32.2 31.9 31.5 31.3
find Malaysia, Estonia and the Dominican Republic, with average shadow
economies over 1999 to 2006 of 31.3, 31.4, and 32.1 per cent. The three
countries with the highest shadow economy are now Peru, Georgia and
Bolivia with an average value over the period 1999 to 2006 of 58.4, 66.0
and 66.4 per cent, respectively.
Looking at Table 1.7 (151 countries), we see that the unweighted
average of the shadow economy in this sample for the year 1999 is 33.3 per
cent and steadily decreases to 31.3 per cent in 2007. The three countries
with the smallest shadow economies are Switzerland, the United States
and Luxembourg with an average size (over 1999 to 2007) of 8.5, 8.6 and
9.7 per cent of official GDP. In the middle of the distribution, we find the
Fiji Islands, Algeria and Romania, with average sizes of 32.4, 32.6 and
32.6 per cent. The three countries with the largest shadow economies are
Zimbabwe, Georgia and Bolivia with an average size of 61.8, 65.8 and
66.1 per cent. In general, comparing the calibrations of the two samples
(sample with 151 observations and sample with 120 observations), we can
see that the size and trend of the shadow economy are robust for most
of the countries over the period 1999 to 2006/7 with only a few minor
differences.
Having estimated and calculated the size and trend of the shadow
economy according to four different MIMIC model specifications, we


finally compare how much the different estimates vary for each country
given the different models. For this purpose, we calculated the range of
the estimates for each country, that is the difference between the maximum
and the minimum estimate. It turned out that the variation in estimates is
on average relatively low. However, in Guinea (between specifications 3
and 4 as well as between specifications 2 and 4), and Zambia (specifica-
tions 3 and 4 as well as between specifications 2 and 4) the maximum
differences in the range are 3.3 as well as 3.2 per cent, and 2.8 as well as
2.7 per cent, respectively. These rather large differences might be a conse-
quence of the parsimony of specification 4. Except for these exemptions,
all models estimated predict almost the same size of the shadow economy
for each country. Calculating pairwise correlations, we find that the corre-
lation coefficients are extremely high. For example, between specifications
1 and 4 they are for all years above 0.98; meaning that for each country
the predicted sizes of the shadow economy are almost indistinguishable
from each other, regardless of the specification used for prediction. This
allows us to add 11 countries to our maximum sample estimation of
151 countries, which are not included in specification 6, but for which
we have calculated the size of the shadow economy using specification
1 and specification 5 (shown in Appendix 1.4).28 These countries are:
Mauritania, Mozambique, Myanmar, Niger, Nigeria, Panama, Rwanda,
Sudan, Swaziland, Togo (taken from Table 1.3) and Moldova (cali-
brated according to specification 5 presented in Appendix 1.4). Appendix
1.5 presents alphabetically ordered shadow economy estimates for 162
countries around the world.
We turn now to analyze our measurement estimates of the shadow
economy. First, a visual quick check for normality (i.e. a Q–Q plot, Figure
1.2) hints that the measure is normal and that there are no overall outliers
on the top, or on the bottom of the distribution. A formal test for nor-
mality (the Kolmogorov–Smirnov test, not presented) indicates that we
cannot reject the the null hypothesis of normality. Thus, we are confident
that our measure follows a normal distribution.
Second, we analyze the measurement estimates by regions. To do so
we used the regions as defined by the World Bank. The World Bank dis-
tinguishes eight world regions. The mean, median, minimum, maximum
and standard deviation in each region are presented in Table 1.8. The
medians by region are ploted in Figure 1.3 ordered from the highest at
the top to the lowest at the bottom. The regional results are very clear:
Sub-Saharan Africa has the highest estimates of the shadow economy
(with a median of 40.5), followed by Latin America and the Carribean
(38.7), and Europe and Central Asia (35.8). At the bottom of the distri-
bution we find the OECD countries with a median of 16.0. The table also


80
60
Informality
40
20
0
0 20 40 60 80
Inverse normal
Figure 1.2 Q–Q plot of the informality measure
Table 1.8 Average informality (unweighted) by World Bank’s regions
Region Mean median min max sd

EAP East Asia and Pacific 32.3 32.5 12.7 51.0 13.3
ECA Europe and Central Asia 38.5 35.8 18.2 66.7 11.0
LAC Latin America and the 41.2 38.7 19.3 66.1 12.3
Caribbean
MENA Middle East and North Africa 28.0 32.7 18.2 37.2 7.9
OECD High Income OECD 16.8 16.0 8.7 27.9 5.6
OHIE Other High Income 22.8 25.0 12.4 33.4 6.7
SAS South Asia 33.2 35.3 22.2 43.7 6.9
SSA Sub-Saharan Africa 40.8 40.5 22.6 61.8 7.6
World 33.1 33.5 8.7 66.7 12.8
shows that there are big disparities within regions, which is also shown
in Figure 1.3.
Table 1.8 presents on the bottom line the simple unweighted yearly
average which is not the average informality for the world but the average
world’s informality when one weights every country equally. In order
to measure how much of the GDP in the world is really informal, we
weighted by total country GDP. In particular, for every country/year


SSA
LAC
ECA
SAS
MENA
EAP
OHIE
OECD
0 20 40 60 80
Informality
Figure 1.3 Average shadow economy measure by region
Table 1.9 Average informality weighted by total GDP in 2005
Region Mean median min max sd

EAP East Asia and Pacific 17.5 12.7 12.7 51.0 10.7
ECA Europe and Central Asia 36.5 32.8 18.2 66.7 8.6
LAC Latin America and the 34.7 33.7 19.3 66.1 8.0
Caribbean
MENA Middle East and North Africa 27.3 32.7 18.2 37.2 7.8
OECD High Income OECD 13.5 11.0 8.7 27.9 5.6
OHIE Other High Income 20.8 19.5 12.4 33.4 4.8
SAS South Asia 25.1 22.2 22.2 43.7 5.9
SSA Sub-Saharan Africa 38.4 34.1 22.6 61.8 11.3
World 17.2 13.4 8.7 66.7 9.9
we weighted the rate of informality by the total GDP. This gives us the
GDP in current billion US dollars that is informal for each country/year.
Then we added up this amount and divided it by the total GDP of the
sample. The same had also been done for the sub-samples of the eight
world regions the World Bank distinguishes. According to these calcula-
tions, Table 1.9 shows much lower rates of informal GDP for the world
as a whole, with an average of 17.2 per cent. The results with respect to


the countries’ development stage are very impressive too: the averages of
the weighted yearly informality estimates demonstrate that Sub-Saharan
Africa has the largest shadow economies (with an average of 38.4 per
cent) followed by Europe and Central Asia (with an average of 36.5 per
cent). At the bottom of the distribution we find the OECD countries with
and average of 13.5 per cent, which is consistent with the fact that richer
economies have lower informality rates.
Finally, we present the informality measurement country-by-country
in a world map view. Countries shown with darker colours in Figure 1.4
indicate higher levels of informality. Among them: Azerbaijan, Bolivia,
Peru, Panama, Tanzania and Zimbabwe. Countries shown with ligther
colours indicate countries with lower levels of informality. Among them:
Austria, Japan, Luxembourg, Switzerland, the United States and the
United Kingdom.
1.4 SUMMARY AND CONCLUSIONS
There are many obstacles to overcome when measuring the size of the
shadow economy and when analyzing its consequences on the official
economy. But, as this chapter shows, some progress can be made. We
provide estimates of the size of the shadow economies for 162 countries
over the period 1999 to 2006/2007 using the MIMIC procedure for the
econometric estimation and a benchmarking procedure for calibrating the
estimated MIMIC into absolute values of the size of the shadow economy.
Some new knowledge/insights are gained with respect to the size and
trend of the shadow economy of 162 countries,29 leading to three main
conclusions:
● The first conclusion from these results is that for all countries inves-
tigated the shadow economy has reached a remarkably large size with
a weighted (unweighted) average value of 17.2 (33.1) per cent of offi-
cial GDP. However, equally important is the clear negative trend of
the size of the shadow economy over time. The unweighted average
size of the 162 countries decreased from 34.0 per cent of official
GDP in 1999 to 31.0 per cent in 2007; for the 21 transition countries
from 36.9 per cent in 1999 to 32.6 per cent in 2007.
● The second conclusion is that shadow economies are a complex
phenomenon present to a large extent in all type of economies (devel-
oping, transition and highly developed). People engage in shadow
economic activities for a variety of reasons – especially in response
to government actions, most notably, taxation and regulation.


World view of informality
Figure 1.4
49


● The third conclusion is that there are regional disparities in the

level of informality, but obvious regional clusters. At the top level of
informality we find Sub-Saharan Africa, and at the lowest level of
informality we find the OECD countries.
Considering these conclusions, it is obvious that one of the big chal-

lenges for every government is to undertake efficient incentive-orientated
policy measures in order to make work less attractive in the shadow
economy and, thus, to make work in the official economy more attractive.
Successful implementation of such policies may lead to a stabilization, or
even reduction, of the size of the shadow economy. Of course, even after
20 years of intensive research on the shadow economy, its size, causes
and consequences are still heatedly debated in the literature indicating
that further research is necessary to improve our understanding about the
shadow economy.
NOTES
1. Responsibility for the content of this chapter is ours and should not be attributed
to our affiliated institutions. This is a background study for ‘In from the Shadow:
Integrating Europe’s Informal Labor’, a World Bank regional report on the informal
sector in Central, Southern Europe and the Baltic countries (Task number P112988).
We are grateful for suggestions and comments received at the 2010 Annual Meeting
of the Public Choice Society (Monterrey, CA) and also at the 2010 Annual Meeting
of the European Public Choice Society (Izmir, Turkey). A short version of this article,
presenting selected results, has been published in Schneider, Buehn and Montenegro
(2010).
2. The literature on the ‘shadow’, ‘underground’, ‘informal’, ‘second’, ‘cash’ or ‘parallel’
economy is increasing. Various topics – how to measure it, its causes, its effect on the
official economy and so forth – are analyzed. See, for example, survey-type publica-
tions by Frey and Pommerehne (1984), Thomas (1992), Schneider (1994a, 1994b, 1997,
1998a, 2003, 2005, 2007), Loayza (1996), Pozo (1996), Johnson, Kaufmann and Shleifer
(1997), Lippert and Walker (1997), Johnson, Kaufmann and Zoido-Lobatón (1998a,
1998b), Belev (2003), Gerxhani (2004) and Pedersen (2003). For an overall survey of
the global evidence of the size of the shadow economy, see Schneider and Enste (2000,
2002), Alm, Martinez and Schneider (2004), Bajada and Schneider (2005), Kazemier
(2005), Enste and Schneider (2006) and Feld and Schneider (2010).
3. Compare the different opinions of Giles (1999a, b), Tanzi (1999), Thomas (1999) and
Pedersen (2003) and Janisch and Brümmerhoff (2005).
4. This chapter focuses on the size and trend of the shadow economy for countries and
does not show any disaggregated values for specific regions. For the EU regions an
estimation was done by Herwartz, Schneider and Tafenau (2009). Recently some initial
studies were undertaken to measure the size of the shadow economy, as well as the
‘grey’ or ‘shadow’ labour force for urban regions or states (e.g. California). Compare
Marcelli, Pastor and Joassart (1999), Williams and Windebank (1998, 2001a, b), Chen
(2004), Marcelli (2004), Williams (2004a, b, 2005a, b, 2006), Flaming, Haydamack and
Jossart (2005) and Alderslade, Talmage and Freeman (2006) and Brueck, Haisten-
DeNew and Zimmermann (2006).


5. For a literature review see Schneider and Enste (2000) and Feld and Schneider (2010).
6. This definition is used, for example, by Feige (1989, 1994), Schneider (1994b, 2003,
2005, 2007), Feld and Schneider (2010) and Frey and Pommerehne (1984). Do-it-
yourself activities are not included. For estimates of the shadow economy and the do-
it-yourself activities for Germany, see Karmann (1986, 1990) and Buehn, Karmann and
Schneider (2009).
7. This definition is taken from Dell’Anno (2003), Dell’Anno and Schneider (2004) and
Feige (1989); see also Thomas (1999), Fleming, Roman and Farrell (2000).
8. See the excellent discussion of the definition of the shadow economy in Pedersen (2003,
pp. 13–19) and Kazemier (2005) who uses a similar one.
9. Refer to the survey of Andreoni, Erard and Feinstein (1998) and the paper by Kirchler,
Maciejovsky and Schneider (2002), as well as the survey by Feld and Schneider (2009).
10. See Thomas (1992), Lippert and Walker (1997), Schneider (1994a, b, 1997, 1998a, b,
2000, 2003, 2005, 2007), Johnson, Kaufmann, and Zoido-Lobatón (1998a,1998b), Giles
(1999a), Tanzi (1999), Mummert and Schneider (2001), Giles and Tedds (2002) and
Dell’Anno (2003), as well as Feld and Schneider (2009, Chapter 2, this volume), among
others.
11. There is however a body of empirical evidence showing that movements into (infor-
mal) self-employment are procyclical. For example, Taylor (1996) suggests a ‘pull’ of
aspiring entrepreneurs into self-employment when unemployment is low and offers of
salaried employment are abundant. In good times, individuals may choose to become
self-employed knowing that if their venture fails, an offer of formal salaried employ-
ment will not be hard to find. Workers considering self-employment wait for a favour-
able business climate to leave a protected salaried job. Thus, in good economic times
when aggregate demand is high and businesses are more likely to flourish there is always
a wage-employment safety net that lowers the risks of becoming an entrepreneur.
Maloney (1998a, b) presents evidence of procyclical movement into self-employment in
Mexico, Arango and Maloney (2000) find that the share of self-employed in Argentina
increases as economic conditions improve, while Fiess, Maloney and Shankar (2000)
show similar increases in the share of self-employed in Colombia, Brazil and Chile
during periods of expansion.
12. The pioneers of this approach are Weck (1983) and Frey and Weck-Hannemann
(1984), who applied this approach to cross-sectional data from the 24 OECD countries
for various years. Before turning to this approach they developed the concept of ‘soft
modeling’ (Frey, Weck and Pommerehne, 1982; Frey and Weck, 1983a, b), an approach
which has been used to provide a ranking of the relative size of the shadow economy in
different countries.
13. The latest papers dealing extensively with the MIMIC approach, its development
and its weaknesses are from Giles (1999a, 1999b, 1999c), Giles, Tedds and Werkneh
(2002), Dell’Anno (2003) and an excellent study by Giles and Tedds (2002), as well as
Bajada and Schneider (2005), Breusch (2005a, b), Schneider (2005, 2007), Pickhardt
and Sardà Pons (2006), Chatterjee, Chaudhury and Schneider (2006), Buehn, Karmann
and Schneider (2009), and for a detailed discussion of the strengths and weaknesses see
Dell’Anno and Schneider (2009). Buehn and Schneider (2011) study for the first time
the impact of enforcement on the shadow economy for a selected number of countries.
14. Estimation of a SEM with latent variables can be done by means of a computer
program for the analysis of covariance structures, such as LISREL (Linear Structural
Relations). A useful overview of the LISREL software package in an economics journal
is Cziraky (2004). General overviews of the SEM approach are given in, for example,
Hayduk (1987), Bollen (1989), Hoyle (1995), Maruyama (1997), Byrne (1998), Muthen
(2002) and Cziraky (2005).
15. In contrast, in an exploratory factor analysis a model is not specified in advance, that
is beyond the specification of the number of latent variables (factors) and observed
variables the researcher does not specify any structure of the model. This means,
assuming that all factors are correlated, all observed variables are directly influenced


by all factors, and measurement errors are all uncorrelated with each other. In practice,
however, the distinction between a confirmatory and an exploratory factor analysis is
less strong. Facing poorly fitting models, researchers using SEM techniques or a con-
firmatory factor analysis often modify their models in an exploratory way in order to
improve the fit. Thus, most applications fall between the two extreme cases of confirma-
tory (non-specified model structure) and exploratory (ex-ante specified model) factor
analysis.
16. Appendix 1.4 presents three additional specifications for transition and high-income
OECD countries.
17. Due to data limitations, the three different categories of these countries do not add up
to 151 countries. Classifying a country as developing, Eastern European and Central
Asian, or high-income OECD follows the World Bank guidelines, for example, using a
benchmark per capita income of US$9.265 or less for developing countries.
18. Here we have the problem that in many developing and Eastern European and Central
Asian countries, the US dollar (or the Euro) is also a widely used currency, which is not
considered here, because we could not obtain any reliable figures of the amount of US
dollars (Euros) in these countries.
19. x1t equals size of government, x2t and x3t denote the fiscal and business freedom index
and x4t represents GDP per capita. According to the MIMIC approach, all variables are
taken as standardized deviations from mean.
20. Appendix 1.1 discusses the the currency demand approach in detail. Again, the MIMIC
model treats hidden output as a latent variable, and uses several (measurable) causal
and indicator variables. The cash-demand equation is not used as an input to determine
the variation in the hidden economy over time – it is used only to obtain the long-run
average value of hidden output (base value), so that the index for this ratio predicted by
the MIMIC model can be used to calculate a level and the percentage units of the shadow
economy. Overall, this latest combination of the currency demand and MIMIC approach
clearly shows that some progress in the estimation technique of the shadow economy has
been achieved and a number of critical points have been overcome. However, objections
can also be raised against the MIMIC method: (1) instability in the estimated coef-
ficients with respect to sample size changes, (2) instability in the estimated coefficients
with respect to alternative specifications, (3) difficulty of obtaining reliable data on cause
variables other than tax variables, and (4) the reliability of the variables grouping into
‘causes’ and ‘indicators’ in explaining the variability of the shadow economy.
21. Calibration is performed separately for each country. The base values typically origi-
nate from the year 2000. Regarding the developing countries, we sometimes opted for
base values originating from the year 2005 because of data availability. The MIMIC
index has been adjusted to the positive range by adding a positive constant.
22. See also Tanzi (1999), Thomas (1992, 1999), Pedersen (2003), Ahumada et al. (2004),
Breusch (2005a, 2005b) and Janisch and Brümmerhoff (2005), Schneider (2005).
23. For an extensive and excellent literature survey of the research about the shadow
economy in developing countries, see Gerxhani (2004),who stresses thoroughout her
paper that the distinction between developed and developing countries with respect
to the shadow economy is of great importance. Due to space reasons this point is
not further elaborated here; nor are the former results and literature discussed. See
Schneider and Enste (2000) for this.
24. It should be mentioned that mainland China and Vietnam are still communist countries
with partly market economies, so that the figures of these two countries may be biased.
25. Their estimates for the early 1990s are on average 10 to 20 per cent higher than our esti-
mates (1999 to 2007) and up to as twice as large as estimates using the currency demand
and the MIMIC approach for the same time period.
26. The estimates were calibrated using specification 4 except for Moldova. Its size of the
shadow economy was derived using the estimation results of specification 5 presented
in Appendix 1.4.
27. As we have a lot of missing values in this specification for the year 2007, estimates for


the year of 2007 are not shown here because it may be misleading – over a third of the
countries do not have an estimate for the year 2007.
28. The reason for this is that these specifications are based on a paper in which we used a
slightly different set of countries (Schneider and Buehn, 2009).
29. In Appendix 1.1 some critical discussion of these two methods is given; they have well
known weaknesses (compare also Pedersen, 2003, and Feld and Schneider, 2010).
REFERENCES
Ahumada, H., F. Alvaredo, A. Canavese and P. Canavese (2004), The demand for
currency approach and the size of the shadow economy: a critical assessment.
Discussion Paper. Delta Ecole, Normale Superieure, Paris.
Alexeev, M. and W. Pyle (2003), A note on measuring the unofficial economy in
the Former Soviet Republics. Economics of Transition, 11(1), 1–23.
Alderslade, J., J. Talmage and Y. Freeman (2006), Measuring the informal
economy: one neighborhood at a time. Discussion Paper. The Brooking
Institution Metropolitan Policy Program, Washington DC, September 2006.
Alm, J., J. Martinez-Vazquez and F. Schneider (2004), Sizing the problem of the
hard-to-tax. Working Paper. Georgia State University: USA.
Andreoni, J., B. Erard and J. Feinstein (1998), Tax compliance. Journal of
Economic Literature, 36, 818–860.
Bajada, C. and F. Schneider (2005), Size, Causes and Consequences of the
Underground Economy: An International Perspective. Aldershot, UK: Ashgate
Publishing Company.
Belev, B. (2003), The Informal Economy in the EU Accession Countries: Size,
Scope, Trends and Challenges to the Process of EU Enlargement. Sofia: Center
for Study of Democracy.
Bhattacharyya, D.K. (1999), On the economic rationale of estimating the hidden
economy. The Economic Journal, 109(456), 348–359.
Blades, D. (1982), The hidden economy and the national accounts. OECD
(Occasional Studies), Paris, pp. 28–44.
Bollen, K.A. (1989), Structural Equations with Latent Variables. New York: Wiley.
Breusch, T. (2005a), Estimating the underground economy, using MIMIC models.
Working Paper. National University of Australia, Canberra, Australia.
Breusch, T. (2005b), The Canadian underground economy: an examination of
Giles and Tedds. Canadian Tax Journal, 53(2), 367–391.
Brueck, T., J.B. Haisten-DeNew and K.F. Zimmermann (2006), Creating
low-skilled jobs by subsidizing market contracted household work. Applied
Economics, 38(4), 899–911.
Buehn, A., A. Karmann and F. Schneider (2009), Shadow economy and do-it-
yourself activities: the German case. Journal of Institutional and Theoretical
Economics, 164(4), 701–722.
Buehn, A. and F. Schneider (2011), Shadow economies around the world: novel
insights, accepted knowledge, and new estimates, International Tax and Public
Finance, forthcoming.
Byrne, B.M. (1998), Structural Equation Modelling with LISREL, PRELIS and
SIMPLIS: Basic Concepts, Applications and Programming. Mahwah, NJ:
Lawrence Erlbaum Associates.


Cagan, P. (1958), The demand for currency relative to the total money supply.
Journal of Political Economy, 66(3), 302–328.
Chatterjee, S., K. Chaudhury and F. Schneider (2006), The size and development
of the Indian shadow economy and a comparison with other 18 Asian countries:
an empirical investigation. Journal of Development Economics, April 2006.
Chen, M. (2004), Rethinking the informal economy: linkages with the formal
economy and the formal regulatory environment. Paper presented at the EGDI-
WIDR Conference ‘unleashing human potential: linking the informal and
formal sectors, Helsinki, Finland, 2004.
Cziraky, D. (2004), LISREL 8.54: A program for structural equation modelling
with latent variables. Journal of Applied Econometrics, 19, 135–141.
Cziraky, D. (2005), A unifying statistical framework for dynamic structural equa-
tion models with latent variables. Available at: http://stats.lse.ac.uk/ciraki/
framework.pdf.
Dell’Anno, R. (2003), Estimating the shadow economy in Italy: a structural equa-
tion approach. Discussion Paper. Department of Economics and Statistics,
University of Salerno.
Dell’Anno, R. (2007), The shadow economy in Portugal: an analysis with the
MIMIC approach. Journal of Applied Economics, 10, 253–277.
Dell’Anno, R. and F. Schneider (2004). The shadow economy of Italy and
other OECD countries: what do we know? Discussion Paper. Department of
Economics, University of Linz, Linz.
Dell’Anno, R. and F. Schneider (2005), Estimating the underground economy by
using MIMIC models: a response to T. Breusch’s critique. Discussion Paper.
Department of Economics, University of Linz, Linz.
Dell’Anno, R. and F. Schneider (2006), Estimating the underground economy: a
response to T. Breusch’s critique. Discussion Paper. Department of Economics,
Johannes Kepler University Linz, Linz.
Dell’Anno, R. and F. Schneider (2009), A complex approach to estimate the
shadow economy: the structural equation modelling. In M. Faggini and T. Lux
(eds), Coping with the Complexity of Economics. Heidelberg: Springer-Verlag,
pp. 110–130.
Dell’Anno, R. and O.H. Solomon (2008), Shadow economy and unemployment
rate in USA: is there a structural relationship? An empirical analysis. Applied
Economics, 40, 2537–2555.
Deregulation Commission (1991), Opening of markets and competition. Report
presented to the German Federal Government, Bonn.
Enste, D. and F. Schneider (2006), Wie groß ist die Schattenwirtschaft? Des
Rätsels Lösung. Wirtschaftsdienst—Zeitschrift für Wirtschaftspolitik, 86(2),
185–191.
Feige, E.L. (1986), A re-examination of the ‘underground economy’ in the
United States. IMF Staff Papers, 33(4). Supplement to Public Finance/Finances
Publiques, 49, 119–136.
Feige, E.L. (1989), The Underground Economies: Tax Evasion and Information
Distortion. Cambridge: Cambridge University Press.
Feige, E.L. (1994), The underground economy and the currency enigma.
Supplement to Public Finance/Finances Publiques, 49, 119–136.
Feige, E.L. (1996), Overseas holdings of US currency and the underground
economy. In S. Pozo (ed.): Exploring the Underground Economy. Kalamazoo,
Michigan: W.E. Upjohn Institute for Employment Research, pp. 5–62.


Feld, L. and F. Schneider (2009), Survey on the shadow economy and unde-
clared earnings in OECD countries. Invited Paper. German Economic Review,
Department of Economics, University of Linz, Linz.
Fiess, N., W.F. Maloney and R. Shankar (2000), The Informal Sector, Wage
Rigidities, and Real Exchange Rates. Washington, DC: World Bank, Poverty
Reduction and Economic Management Department.
Flaming, D., B. Hayolamak and P. Jossart (2005), Hopeful Workers, Marginal
Jobs: LA’s Off-The-Books Labour Force. Economic Roundtable, Los Angeles,
CA, 2005.
Fleming, M.H., J. Roman and G. Farrel (2000), The shadow economy. Journal of
International Affairs, Spring 2000, 53(2), 64–89.
Frey, B.S. and H. Weck (1983a), Bureaucracy and the shadow economy: a macro-
approach. In H. Hanusch (ed.), Anatomy of Government Deficiencies. Berlin:
Springer, pp. 89–109.
Frey, B.S. and H. Weck (1983b), Estimating the shadow economy: a ‘naive’
approach. Oxford Economic Papers, 35, 23–44.
Frey, B.S. and H. Weck-Hannemann (1984), The hidden economy as an
‘unobserved’ variable. European Economic Review, 26(1), 33–53.
Frey, B.S. and W. Pommerehne (1984), The hidden economy: state and prospect
for measurement. Review of Income and Wealth, 30(1), 1–23.
Frey, B.S., H. Weck and W.W. Pommerehne (1982), Has the shadow economy
grown in Germany? An exploratory study. Weltwirtschaftliches Archiv, 118(4),
499–524.
Friedman, E., S. Johnson, D. Kaufmann and P. Zoido-Lobatón (2000), Dodging
the grabbing hand: the determinants of unofficial activity in 69 countries.
Journal of Public Economics, 76(4), 459–493.
Garcia, G. (1978), The Currency Ratio and the subterranean economy. Financial
Analysts Journal, 69(1): 64–66.
Gerxhani, K. (2004), The informal sector in developed and less-developed
countries: a literature survey. Public Choice, 114(3–4), 295–318.
Giles, D.E.A. (1999a), Measuring the hidden economy: implications for
econometric modelling. The Economic Journal, 109(456), 370–380.
Giles, D.E.A. (1999b), Modelling the hidden economy in the tax-gap in New
Zealand. Empirical Economics, 24(4), 621–640.
Giles, D.E.A. (1999c), The rise and fall of the New Zealand underground
economy: are the reasons symmetric? Applied Economic Letters, 6, 185–189.
Giles, D.E.A. and L.M. Tedds (2002), Taxes and the Canadian underground
economy. Canadian Tax Paper No. 106, Canadian Tax Foundation, Toronto/
Ontario.
Giles, D.E.A., L.M. Tedds and G. Werkneh (2002), The Canadian underground
and measured economies. Applied Economics, 34(4), 2347–2352.
Gutmann, P.M. (1977), The subterranean economy. Financial Analysts Journal,
34(1), 24–27.
Hayduk, L.A. (1987), Structural Equation Modelling with LISREL. Essential and
Advances. London: The Johns Hopkins University Press.
Herwartz, H., F. Schneider and E. Tafenau (2009), One Share Fits All? Regional
Variations in the Extent of Shadow Economy in Europe. Discussion Paper.
Department of Economics, University of Linz and Kiel.
Hoyle, R.H. (ed.) (1995), Structural Equation Modeling: Concepts, Issues, and
Applications. Thousand Oaks, CA: Sage Publications.


Hill, R. and M. Kabir (1996), Tax rates, the tax mix, and the growth of the under-
ground economy in Canada: what can we infer? Canadian Tax Journal/ Revue
Fiscale Canadienne, 44(6), 1552–1583.
Isachsen, A.J. and S. Strom (1985), The size and growth of the hidden economy in
Norway. Review of Income and Wealth, 31(1), 21–38.
Janisch, U. and D. Brümmerhoff (2005), Möglichkeiten und Grenzen der
Schätzung der Schattenwirtschaft: Eine kritische Auseinandersetzung mit den
Schätzergebnissen der Bargeldmethode nach Schneider. Discussion paper.
Universität Rostock.
Johnson, S., D. Kaufmann and A. Shleifer (1997), The unofficial economy in tran-
sition. Brookings Papers on Economic Activity, Fall, Washington DC.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998a), Regulatory discre-
tion and the unofficial economy. The American Economic Review, 88(2),
387–392.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998b), Corruption, public
finances and the unofficial economy. Discussion Paper. The World Bank,
Washington, DC.
Karmann, A. (1986), Monetäre Ansätze zur Erfassung der Schattenwirtschaft: Ein
Vergleich verschiedener Messansätze. Kredit und Kapitel, 19(3), 233–247.
Karmann, A. (1990), Schattenwirtschaft und ihre Ursachen: Eine empirische
Analyse zur Schwarzwirtschaft und Selbstversorgung in der Bundesrepublik
Deutschland. Zeitschrift für Wirtschafts- und Sozialwissenschaften (ZWS),
110(3), 185–206.
Kaufmann, D. and A. Kaliberda (1996), Integrating the unofficial economy into
the dynamics of post socialist economies: a framework of analyses and evidence.
In B. Kaminski (ed.), Economic Transition in Russia and the New States of
Eurasia. London: M.E. Sharpe, pp. 81–120.
Kazemier, B. (2005), The underground economy: a survey of methods and esti-
mates. Discussion Paper. Statistics Netherlands, Voorburg, Netherlands.
Kirchgässner, G. (1983), Size and development of the West German shadow
economy, 1955–1980. Zeitschrift für die gesamte Staatswissenschaft, 139(2),
197–214.
Kirchgässner, G. (1984), Verfahren zur Erfassung des in der Schattenwirtschaft
erarbeiteten Sozialprodukts. Allgemeines Statistisches Archiv, 68(4), 378–405.
Kirchler, E., B. Maciejovsky and F. Schneider (2002), Everyday representations of
tax avoidance, tax evasion and tax flight: do legal differences matter? Economic
Psychology, 8(3), 1–19.
Klovland, J. (1984), Tax evasion and the demand for currency in Norway and
Sweden: is there a hidden relationship? Scandinavian Journal of Economics,
86(4), 423–439.
Lackó, M. (2000), Hidden economy – an unknown quantity: comparative analysis
of hidden economics in transition countries 1989–95. Economics of Transition,
8(1), 117–149.
Lippert, O. and M. Walker (eds) (1997), The Underground Economy: Global
Evidences of its Size and Impact. Vancouver, BC: The Frazer Institute.
Loayza, N.V., (1996), The economics of the informal sector: a simple model and
some empirical evidence from Latin America. Carnegie-Rochester Conference
Series on Public Policy, Elsevier, 45(1), 129–162.
Maloney, W.F. (1998a), Are labor markets in developing countries dualistic?
Policy Research Working Paper No. 1941. World Bank, Washington DC.


Maloney, W.F. (1998b), The structure of the labor market in developing countries:
time series evidence and competing views. Policy Research Working Paper No.
1940. World Bank, Washington DC.
Marcelli, E.A. (2004), Unauthorized Mexican immigration, the labour and
other lower-wage informal employment in California. Regional Studies, 38(1),
1–13.
Marcelli, E.A., M. Pastor Jr. and P.M. Joassart (1999), Estimating the effects
of informal economic activity: evidence from Los Angeles County. Journal of
Economic Issues, 33(3), 579–607.
Maruyama, G.M. (1997), Basic of Structural Equation Modeling. Thousand Oaks,
CA: Sage Publications.
Mummert, A. and F. Schneider (2001), The German shadow economy: parted in a
united Germany? Finanzarchiv, 58(3), 260–285.
Muthen, B.O. (2002), Beyond SEM: general latent variable modeling.
Behaviormetrika, 29, 81–117.
Park, T. (1979), Reconciliation between personal income and taxable income.
Mimeo, Washington DC: Bureau of Economic Analysis, pp. 1947–1977.
Pedersen, S. (2003), The shadow economy in Germany, Great Britain and
Scandinavia: a measurement based on questionnaire service. Study No. 10. The
Rockwool Foundation Research Unit, Copenhagen.
Pickhardt, M. and J. Sardà Pons (2006), Size and scope of the shadow economy in
Germany. Applied Economics, 38(4), 1707–1713.
Pozo, S. (ed.) (1996), Exploring the Underground Economy: Studies of Illegal
and Unreported Activity. Michigan: W.E. Upjohn, Institute for Employment
Research.
Rogoff, K. (1998), Blessing or curse? Foreign and underground demand for Euro
notes. Economic policy: The European Forum, 26, 261–304.
Schneider, F. (1986), Estimating the size of the Danish shadow economy using
the currency demand approach: an attempt. The Scandinavian Journal of
Economics, 88(4), 643–668.
Schneider, F. (1994a), Can the shadow economy be reduced through major tax
reforms? An empirical investigation for Austria. Supplement to Public Finance/
Finances Publiques, 49, 137–152.
Schneider, F. (1994b), Measuring the size and development of the shadow
economy. can the causes be found and the obstacles be overcome? H.
Brandstaetter and W. Güth (eds), Essays on Economic Psychology. Berlin,
Heidelberg: Springer Publishing Company, pp. 193–212.
Schneider, F. (1997), The shadow economies of Western Europe. Journal of the
Institute of Economic Affairs, 17(3), 42–48.
Schneider, F. (1998a), Further empirical results of the size of the shadow economy
of 17 OECD-countries over time. Paper presented at the 54th Congress of the
IIPF, Córdoba, Argentina, Department of Economics, University of Linz, Linz,
Austria.
Schneider, F. (1998b), Stellt das Anwachsen der Schwarzarbeit eine wirtschaft-
spolitische Herausforderung dar? Einige Gedanken aus volkswirtschaftlicher
Sicht. Linz, Mitteilungen des Instituts für angewandte Wirtschaftsforschung
(IAW), I/98, 4–13.
Schneider, F. (2000), The increase of the size of the shadow economy of 18 OECD-
countries: some preliminary explanations. Paper presented at the Annual Public
Choice Meeting, 10–12 March 2000, Charleston, SC.


Schneider, F. (2003), The shadow economy. In C.K. Rowley and

F. Schneider (eds), Encyclopedia of Public Choice. Dordrecht: Kluwer
Academic Publishers.
Schneider, F. (2004), Arbeit im Schatten: Wo Deutschlands Wirtschaft wirklich
wächst. Gabler: Wiesbaden.
Schneider, F. (2005), Shadow economies around the world: what do we really
know? European Journal of Political Economy, 21(3), 598–642.
Schneider, F. (2007), Shadow economies and corruption all over the world: new
estimates for 145 countries. Economics, 2007–9, July 2007.
Schneider, F. and D. Enste (2000), Shadow economies: size, causes, and
consequences. The Journal of Economic Literature, 38(1): 77–114.
Schneider, F. and D. Enste (2002), The Shadow Economy: Theoretical Approaches,
Empirical Studies, and Political Implications. Cambridge, UK: Cambridge
University Press.
Schneider, F. and A. Buehn (2009), Shadow economies and corruption all over the
world: revised estimates for 120 countries. Economics: The Open-Access, Open-
Assessment E-Journal, 1, 2007-9 (Version 2).
Schneider, F., A. Buehn and C.E. Montenegro (2010), New estimates for the
shadow economies all over the world, International Economic Journal, 24,
443–461.
Smith, P. (1994), Assessing the size of the underground economy: the Statistics
Canada Perspectives. Canadian Economic Observer, Catalogue No.: 11-010,
3.16-33, at 3.18.
Tanzi, V. (1980), The underground economy in the United States: estimates and
implications. Banca Nazionale del Lavoro, 135(4), 427–453.
Tanzi, V. (1982) (ed.), The Underground Economy in the United States and Abroad.
Lexington, MA, Lexington Books.
Tanzi, V. (1983), The underground economy in the United States: annual esti-
mates, 1930–1980. IMF-Staff Papers, 30(2), 283–305.
Tanzi, V. (1999), Uses and abuses of estimates of the underground economy. The
Economic Journal, 109(456), 338–340.
Taylor, M.P. (1996), Earnings, independence or unemployment: why become self
employed? Oxford Bulletin of Economics and Statistics, 58(2), 253–265.
Thomas, J.J. (1986), The underground economy in the United States: a comment
on Tanzi. IMF-Staff Papers, 33(4), 782–789.
Thomas, J.J. (1992), Informal Economic Activity. LSE, Handbooks in Economics.
London: Harvester Wheatsheaf.
Thomas, J.J. (1999), Quantifying the black economy: measurement without theory
yet again? The Economic Journal, 109(456), 381–389.
Torgler, B. and F. Schneider (2005), Attitudes towards paying taxes in Austria: an
empirical analysis. Empirica, 204–225.
Weck, H. (1983), Schattenwirtschaft: Eine Möglichkeit zur Einschränkung
der öffentlichen Verwaltung? Eine ökonomische Analyze. Bern-Frankfurt:
Lang.
Williams, C.C. (2004a), Cash-In-Hand Work: The Underground Sector and the
Hidden Economy of Favours. Houndmills, Hampshire: Palgrave Macmillan
Publishing Company.
Williams, C.C. (2004b), Geographical variations in the nature of undeclared work.
Geographisca Analer, 86(B3), 187–200.
Williams, C.C. (2005a), Fostering community engagement and tackling undeclared


work: the case for an evidence-based ‘join-up’. Public Policy Approach, Regional
Studies, 39(8), 1145–1155.
Williams, C.C. (2005b), Small Business and the Informal Economy: Making the
Transition to the Informal Economy: The Evidence Base. London: Small Business
Service.
Williams, C.C. (2006), What is to be done about undeclared work? Evaluating the
policy options. Policy and Politics, 34(1), 91–113.
Williams, C.C. and J. Windebank (1995), Black market work in the European
Community: peripheral work for peripheral localities? International Journal of
Urban and Regional Research, 19(1), 23–39.
Williams, C.C. and J. Windebank (1998), Informal Employment in the Advanced
Economies: Implications for Work and Welfare. London: Routledge.
Williams, C.C. and J. Windebank (2001a), Beyond profit motivated exchange:
some lessons from the study of paid informal work. European Urban and
Regional Studies, 8(1), 49–61.
Williams, C.C. and J. Windebank (2001b), Reconceptualizing paid informal
exchange: some lessons from English cities. Environment and Planning, A 33(1),
121–140.
Zilberfarb, B.-Z. (1986), Estimates of the underground economy in the United
States, 1930–80. IMF-Staff Papers, 33(4), 790–798.


APPENDIX 1.1 THE CURRENCY DEMAND

APPROACH
The currency demand approach, which is also called an ‘indicator’
approach, is a macroeconomic method that uses various economic and
other indicators containing information about the development of the
shadow economy (over time), and leaves some traces of the shadow
economy. This approach was first used by Cagan (1958), who calculated a
correlation of the currency demand and the tax pressure (as one cause of
the shadow economy) for the United States over the period 1919 to 1955.
Twenty years later, Gutmann (1977) used the same approach, but without
any statistical procedures. Cagan’s approach was further developed by
Tanzi (1980, 1983), who econometrically estimated a currency demand
function for the United States for the period 1929 to 1980, in order to
calculate the shadow economy. His approach assumes that shadow (or
hidden) transactions are undertaken in the form of cash payments, so as
to leave no observable traces for the authorities. An increase in the size of
the shadow economy will, therefore, increase the demand for currency. To
isolate the resulting excess demand for currency, an equation for currency
demand is econometrically estimated over time. All conventional possible
factors, such as the development of income, payment habits, interest rates
and so on, are controlled for. Additionally, variables such as the direct and
indirect tax burden, government regulation and the complexity of the tax
system (which are assumed to be the major factors causing people to work
in the shadow economy), are included in the estimation equation. The
basic regression equation for the currency demand, proposed by Tanzi
(1983), is the following:
ln (C/M2) t 5 b0 1 b1 ln (1 1 TW) t 1 b2 ln (WS/Y) t 1 b3 lnRt
1 b4 ln (Y/N) t 1 ut (A1.1)
with b1 > 0, b2 > 0, b3 < 0, b4 > 0

where:
ln denotes natural logarithms, C/M2 is the ratio of cash holdings to
current and deposit accounts, TW is a weighted average tax rate (to
proxy changes in the size of the shadow economy), WS/Y is a proportion
of wages and salaries in national income (to capture changing payment
and money holding patterns), R is the interest paid on savings deposits
(to capture the opportunity cost of holding cash), and Y/N is per capita
income.1
Any excess increase in currency, or the amount unexplained by the


conventional or normal factors (mentioned above) is, then, attributed to

the rising tax burden and the other reasons leading people to work in the
shadow economy. Figures for the size and trend of the shadow economy
can be calculated, in a first step, by comparing the difference between the
development of currency when the direct and indirect tax burden (and
government regulations) are held at their lowest value, and the develop-
ment of currency with the current (much higher) burden of taxation and
government regulations. Assuming in a second step the same velocity
for currency used in the shadow economy as for legal M1 in the official
economy, the size of the shadow can be computed and compared to the
official GDP.
The currency demand approach is one of the most commonly-used
approaches. It has been applied to many OECD countries,2 but has, nev-
ertheless, been criticized on various grounds.3 The most commonly raised
objections to this method are:
1. Not all transactions in the shadow economy are paid in cash. Isachsen
and Strom (1985) used the survey method to find out that in Norway,
in 1980, roughly 80 per cent of all transactions in the hidden sector
were paid in cash. The size of the total shadow economy (including
barter) may thus be even larger than previously estimated.
2. Most studies consider only one particular factor, the tax burden, as a
cause of the shadow economy. But others (such as the impact of regu-
lation, taxpayers’ attitudes toward the state, ‘tax morality’, and so on)
are not considered, because reliable data for most countries are not
available. If, as seems likely, these other factors also have an impact
on the extent of the hidden economy, it might again be higher than
reported in most studies.4
3. As discussed by Garcia (1978), Park (1979) and Feige (1996), increases
in currency demand deposits are due largely to a slowdown in demand
deposits rather than to an increase in currency caused by activities in
the shadow economy, at least in the case of the United States.
4. Blades (1982) and Feige (1986, 1996), criticize Tanzi’s studies on
the grounds that the US dollar is used as an international currency.
Instead, Tanzi should have considered (and controlled) the presence
of US dollars, which are used as an international currency and are
held in cash abroad.5 Moreover, Frey and Pommerehne (1984) and
Thomas (1986, 1992, 1999) claim that Tanzi’s parameter estimates are
not very stable.6
5. Most studies assume the same velocity of money in both types of
economies. As argued by Hill and Kabir (1996) for Canada and by
Klovland (1984) for the Scandinavian countries, there is already


considerable uncertainty about the velocity of money in the official

economy, and the velocity of money in the hidden sector is even more
difficult to estimate. Without knowledge about the velocity of cur-
rency in the shadow economy, one has to accept the assumption of
‘equal’ money velocity in both sectors.
6. Ahumada et al. (2004) show that the currency approach, together
with the assumption of equal income velocity of money in both the
reported and the hidden transaction is only correct if the income elas-
ticity is 1. As this is not the case for most countries, the calculation has
to be corrected.
7. Finally, the assumption of no shadow economy in a base year is open
to criticism. Relaxing this assumption would again imply an upward
adjustment of the size of the shadow economy.


APPENDIX 1.2 VARIABLE DEFINITIONS AND

DATA SOURCES
Business freedom: Subcomponent of the Economic Freedom Index. It
measures the time and efforts of business activity. It ranges from 0 to 100,
where 0 5 least business freedom and 100 5 maximum business freedom.
Source: Heritage Foundation.
Economic freedom: Economic Freedom Index. It ranges from 0 to 100,

where 0 5 least economic freedom and 100 5 maximum economic
freedom.
Fiscal freedom: Subcomponent of the Economic Freedom Index. It meas-

ures the fiscal burden in an economy, that is top tax rates on individual
and corporate income. It ranges from 0 to 100, where 0 5 least fiscal
freedom and 100 5 maximum degree of fiscal freedom.
Currency: M0 over M1. It corresponds to the currency outside the banks

(M0) as a proportion of M1.
Source: International Monetary Fund.
In specification 4 and 5 we use currency over M2 because of higher data
availability.
Source: ECB.
Labour force participation rate: Labour force participation rate, total

(% of total population). Labour force participation rate is the propor-
tion of the population that is economically active: all people who supply
labour for the production of goods and services during a specified
period.
Source: International Labor Organization, Estimates and Projections
of the Economically Active Population database. The data for Taiwan,
China, was obtained from the Taiwan’s Statistical Office website.
GDP per capita (PPP): Corresponds to the GDP per capita based on pur-
chasing power parity (PPP) (constant 2005 international $). GDP PPP is
gross domestic product converted to international dollars using PPP rates.
An international dollar has the same purchasing power over GDP as the
US dollar has in the United States. GDP at purchaser’s prices is the sum
of gross value added by all resident producers in the economy plus any
product taxes and minus any subsidies not included in the value of the


products. It is calculated without making deductions for depreciation of

fabricated assets or for depletion and degradation of natural resources.
Data are in constant 2005 international dollars.
Source: World Bank, International Comparison Program database.
Unemployment rate: Unemployment, total (% of total labour force).

Unemployment refers to the share of the labour force that is without work
but available for and seeking employment. Definitions of labour force and
unemployment differ by country.
Source: International Labor Organization, Key indicators of the labor
market database. Given that this data set contains many missing values,
the source was complemented with data from the PRS Group and also
with data from some national statistical offices’ websites, and also from
the World Bank’s Development Data Platform.
Unemployment rate estimated: In spite of all the efforts to fill in the gaps
many missing values still remained. To fill them up, a structural model of
the determinants of the unemployment rate was estimated. In this model
the dependent variable is the unemployment rate and the predictors are:
● The employment rate of the female population that are 15 years or

older
● The employment rate of the male population that are 15 years of
older
● The female labour force participation rate
● The male labour force participation rate
● The proportion of the population 15–64 that is female
● The proportion of the population 15–64 that is male
● The GDP growth rate of the previous period
● The regression also included country fixed effects
The predictors were selected so that they would be relevant to explain the
unemployment rate, but also so they would be available for most of the
countries in the sample. The model had excellent predictive power. Using
this model we came up with unemployment estimates for some of the
missing unemployment rates.
Size of government: General government final consumption expenditure

(% of GDP). General government final consumption expenditure (for-
merly general government consumption) includes all government current
expenditures for purchases of goods and services (including compensation
of employees). It also includes most expenditure on national defence and


security, but excludes government military expenditures that are part of

government capital formation.
Source: United Nations Statistical Database. Available online at http://
unstats.un.org/unsd/snaama/dnllist.asp. The data for Taiwan, China,
comes from the World Bank’s Development Data Platform.
Share of direct taxes: Direct taxes as a proportion of total overall taxation.

Source: World Bank and Penn World Table (PWT 6.2).
Regulatory quality: Includes measures of the incidence of market-

unfriendly policies such as price controls or inadequate bank supervision,
as well as perceptions of the burdens imposed by excessive regulation in
areas such as foreign trade and business development. The scores of this
index lie between 22.5 and 2.5, with higher scores corresponding to better
outcomes.
Source: Worldwide Governance Indicators: 1996–2009, World Bank.
Available online at: web.worldbank.org.
Government effectiveness: Captures perceptions of the quality of public

services, the quality of the civil service and the degree of its independence
from political pressures, the quality of policy formulation and implemen-
tation and the credibility of the government’s commitment to such poli-
cies. The scores of this index lie between 22.5 and 2.5, with higher scores
corresponding to better outcomes.
Source: Worldwide Governance Indicators: 1996–2009, World Bank.
Available online at: web.worldbank.org.
Inflation rate: Inflation, GDP deflator (annual %). Inflation as measured

by the annual growth rate of the GDP implicit deflator shows the rate of
price change in the economy as a whole. The GDP implicit deflator is the
ratio of GDP in current local currency to GDP in constant local currency.
Source: United Nations Statistical Database. Available online at: http://
unstats.un.org/unsd/snaama/dnllist.asp.
Openness: Corresponds to trade (% of GDP). Trade is the sum of exports

and imports of goods and services measured as a share of gross domestic
product.
Source: United Nations Statistical Database. Available online at: http://
unstats.un.org/unsd/snaama/dnllist.asp.
Total population ages 15 to 64: Corresponds to total population ages

15–64.


Source: World Bank staff estimates from various sources including census
reports, the United Nations Population Division’s World Population
Prospects, national statistical offices, household surveys conducted by
national agencies and Macro International. For Taiwan, China, the data
comes from the National Statistical Office.
Population total: Population, total. Total population is based on the de

facto definition of population, which counts all residents regardless of
legal status or citizenship – except for refugees not permanently settled in
the country of asylum, who are generally considered part of the popula-
tion of their country of origin. The values shown are midyear estimates.
Source: World Bank staff estimates from various sources including census
reports, the United Nations Population Division’s World Population
Prospects, national statistical offices, household surveys conducted by
national agencies and Macro International.
Total labour force: Labour force, total. Total labour force comprises
people aged 15 and older who meet the International Labor Organization
definition of the economically active population: all people who supply
labour for the production of goods and services during a specified period.
It includes both the employed and the unemployed. While national prac-
tices vary in the treatment of such groups as the armed forces and seasonal
or part-time workers, in general the labour force includes the armed
forces, the unemployed and first-time job-seekers, but excludes homemak-
ers and other unpaid caregivers and workers in the informal sector.
Source: International Labor Organization, using World Bank population
estimates.


APPENDIX 1.3 DESCRIPTIVE STATISTICS
Variable Mean Standard Min Max

Deviation
Specification 1
Size of government 14.22 5.57 3.59 44.61
Share of direct taxation 26.79 13.76 2.44 82.40
Fiscal freedom 82.55 9.03 52.91 100.00
Business freedom 44.67 17.75 10.00 94.58
Unemployment rate 11.51 8.04 0.00 39.70
GDP per capita 6806.64 8374.87 319.38 48810.29
Growth rate of GDP per capita 2.31 3.64 –17.61 16.24
Labour force participation rate 66.87 10.26 44.00 92.20
Currency (M0 over M1) 43.74 17.65 1.20 92.99
Specification 2
Fiscal freedom 81.28 9.81 32.56 100.00
GDP per capita 6383.75 8243.83 319.38 51586.21
Currency (M0 over M1) 44.00 16.91 1.20 92.99
Specification 3
Share of direct taxation 29.71 17.20 2.44 92.00
Fiscal freedom 81.48 9.45 50.29 100.00
GDP per capita 10361.04 10986.63 340.18 48810.29
Government effectiveness 0.12 0.90 –1.59 2.64
Growth rate of labour force 0.02 0.02 –0.14 0.11
Currency (M0 over M1) 40.84 18.93 0.02 90.82
Specification 4
GDP per capita 9386.87 11276.40 101.00 66597.70
Government effectiveness –0.09 0.90 –2.51 2.64
Currency (M0 over M1) 42.01 19.62 0.00 97.93


APPENDIX 1.3 (continued)
Variable Mean Standard Min Max

Deviation
Specification 5
Fiscal freedom 80.61 9.47 41.00 96.04
Economic freedom 57.83 8.96 33.71 79.51
Openness 95.97 34.58 29.45 199.68
Inflation 29.22 99.08 –0.92 953.46
Growth rate of labour force 0.00 0.02 –0.14 0.07
Currency (M0 over M1) 48.26 18.06 16.27 90.82
Specification 6
Total tax burden 35.96 7.76 16.57 51.79
Fiscal freedom 70.76 9.03 51.12 88.10
GDP per capita 28412.90 9397.76 7273.22 75597.47
Regulatory quality 1.34 0.41 0.32 2.01
Currency (M0 over M2) 5.37 3.11 0.28 14.98
Specification 7
Total tax burden 37.18 7.15 17.34 51.79
Regulatory quality 1.39 0.37 0.33 2.01
GDP per capita 30988.48 8732.90 11485.83 72783.16
Currency (M0 over M2) 5.19 2.84 0.34 14.87


APPENDIX 1.4 ADDITIONAL SPECIFICATIONS

This appendix presents three additional specifications for 21 Eastern
European and Central Asian (mostly former transition) countries as well
as 25 high-income OECD countries. For the 21 Eastern European and
Central Asian countries, we use as cause variables the size of government
and the fiscal freedom index. For state regulation, we use the economic
freedom index of the Heritage Foundation which ranges from 0 to 100,
where 0 is least economic freedom and 100 maximum economic freedom
(negative sign expected), and for the state of the economy the unemploy-
ment rate, inflation rate and openness (sum of export and imports of
goods and services, as a percentage of GDP). The inflation rate and a
measure for openness are included to take into account the transition
process and periods of high inflation in the late 1990s/early 2000s in the
transition countries. These two variables are measured as follows:
1. Inflation rate: GDP deflator (annual rate in per cent); inflation is

measured by the annual growth rate of the GDP implicit deflator,
which shows the rate of price changes in the economy as a whole
(positive sign expected).
2. Openness: openness corresponds to trade (as a per cent of GDP).
Trade is the sum of exports and imports of goods and services, meas-
ured as a share of gross domestic product (negative sign expected).
As indicator variables, we use the growth rate of GDP per capita, the
growth rate of total labour force and the ratio M0 over M1. For the 25
high-income OECD countries, we use the total tax burden (total tax rev-
enues as a percentage of GDP), the fiscal and business freedom indices, a
regulatory quality index and the unemployment rate as causal variables.
The regulatory quality index is the World Bank’s regulatory quality index
which includes measures of the incidents of market-unfriendly policies,
such as price controls or inadequate bank supervision, as well as percep-
tions of the burdens imposed by excessive regulation in areas such as
foreign trade and business development. The index scores between –2.5
and 12.5 with higher scores corresponding to better outcomes (negative
sign expected). As indicator variables, we use the labour force participa-
tion rate, GDP per capita and a measure for currency defined as:
1. Currency over M2: It corresponds to the currency outside the banks

as a proportion of M2.
2. GDP per capita: GDP per capita is gross domestic product converted
to international dollars using PPP rates, divided by the population.


In specification 5, the MIMIC estimation result for the 21 Eastern

European and Central Asian (mostly former transition) countries over
the period 1994 to 2006 is shown. If we begin with the cause varia-
bles, the size of government and the fiscal freedom variable (both captur-
ing the overall state burden), are highly statistically significant and have
the expected signs. With respect to regulation, the economic freedom
variable has the expected negative sign and is statistically significant.
As these countries experienced periods of high inflation, we include the
inflation rate, which has the expected positive sign and is highly statisti-
cally significant. The variable openness, modelling the transition process,
is also statistically significant. Considering the indicator variables, the
growth rate of the total labour force is statistically significant, as well
as the growth rate of GDP per capita. Also, the test statistics are quite
satisfactory.
In specifications 6 and 7, the estimation results for the 25 high-income
OECD countries are shown. Specification 6 shows the estimation over
the period 1996 to 2006, and specification 7 results over the period
1996 to 2007.7 Considering the results of specification 6 over the period
1996 to 2006, the two variables capturing government burden (total tax
burden and fiscal freedom) are highly statistically significant and have
the expected sign. The unemployment rate has the expected sign and is
statistically significant at the 95 per cent confidence level. The two vari-
ables capturing the regulatory burden, business freedom and regulatory
quality have the expected signs and are highly statistically significant.
Turning to the indicator variables, the labour force participation rate
and currency (ratio of M0 over M2) are both highly statistically sig-
nificant. Also, the test statistics for this equation are quite satisfactory.
Turning to specification 7, where we present the results over the period
1996 to 2007, we use the same set of causal variables but exclude fiscal
and business freedom, which allows us to estimate the model up to the
year 2007. We can see that all causal variables are highly statistically
significant and all have the expected signs. The same is true for the indi-
cator variables.


Table A1.4.1 Additional MIMIC model estimation results
Independent Specification 5 Specification 6 Specification 7

variables 21 Transition Countries 25 High Income OECD 25 High Income OECD
(1994 – 2006) Countries (1996 – 2006) Countries (1996 – 2007)
Causal variables
Size of government 0.18
(3.49)***
Total tax burden 0.05 0.06
(2.05)** (1.78)*
Fiscal freedom −0.08 −0.07
(1.68)* (2.84)***
Business freedom −0.23
(5.93)***
Economic freedom −0.09
(1.91)*
Unemployment rate 0.08 0.05 0.11
(1.84)* (1.89)* (3.16)***
Regulatory quality −0.21 −0.31
(5.45)*** (6.50)***
Openness −0.15
(2.47)**
Inflation rate 0.22
(2.83)***
Indicator variables
Growth rate of GDP −0.76
per capita (4.41)***
GDP per capita −1.52 −1.25
(6.71)*** (8.36)***
Labor force −1.11 −1.03
participation rate (5.45)*** (7.70)***
Growth rate of −0.83
labour force (3.90)***
Currency 1 1 1
Statistical tests
RMSEA (p-value) 0.00 0.00 0.00
(1.00) (0.88) (0.99)
Chi-square (p-value) 17.75 17.74 3.55
(0.91) (0.60) (0.94)
AGFI 0.97 0.95 0.99
Degrees of freedom 27 20 9
Number of 213 145 243
observations
Note: Absolute z-statistics in parentheses. ***, **, * denote significance at the 1, 5 and
10% significance levels. All variables are used as their standard deviations from the mean.
According to the MIMIC models identification rule (see also Section 1.3.1), one indicator
has to be fixed to an a priori value. We have consistently chosen the currency variable.
The degrees of freedom are determined by 0.5(p 1 q)(p 1 q 1 1) – t; with p 5 number of
indicators; q 5 number of causes; t 5 the number for free parameters.


APPENDIX 1.5 LISTING OF 162 COUNTRIES IN

ALPHABETICAL ORDER

Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
1 Albania 35.7 35.3 34.9 34.7 34.4 33.9 33.7 33.3 32.9 34.3
2 Algeria 34.2 34.1 33.8 33.3 32.5 31.7 31.1 31.0 31.2 32.5
3 Angola 48.8 48.8 48.4 47.4 47.3 47.1 45.0 44.0 42.1 46.5
4 Argentina 25.2 25.4 26.1 27.6 26.4 25.5 24.7 23.8 23.0 25.3
5 Armenia 46.6 46.3 45.4 44.5 43.9 43.6 42.7 42.1 41.1 44.0
6 Australia 14.4 14.3 14.3 14.1 13.9 13.7 13.7 13.7 13.5 14.0
7 Austria 10.0 9.8 9.7 9.8 9.8 9.8 9.8 9.6 9.5 9.8
8 Azerbaijan 61.0 60.6 60.3 60.0 59.1 58.6 56.7 54.0 52.0 58.0
9 Bahamas, The 26.3 26.2 26.4 26.5 27.0 27.4 26.7 26.2 26.2 26.5
10 Bahrain 18.6 18.4 18.2 18.0 17.8 17.4 17.1 – – 17.9
11 Bangladesh 36.0 35.6 35.5 35.7 35.6 35.5 35.1 34.5 34.1 35.3
12 Belarus 48.3 48.1 47.9 47.6 47.0 46.1 45.2 44.2 43.3 46.4
13 Belgium 22.7 22.2 22.1 22.0 22.0 21.8 21.8 21.4 21.3 21.9
14 Belize 45.2 43.8 43.3 43.4 42.3 42.0 42.1 41.7 42.0 42.9
15 Benin 51.2 50.2 49.8 49.6 49.3 49.5 49.8 49.6 49.1 49.8
16 Bhutan 29.6 29.4 29.2 29.1 28.7 28.7 28.3 28.2 27.7 28.8
17 Bolivia 67.0 67.1 67.6 67.7 67.7 66.9 64.3 62.8 63.5 66.1
18 Bosnia & 34.3 34.1 34.0 33.9 33.5 33.6 33.2 32.9 32.8 33.6
Herzegovina
19 Botswana 33.9 33.4 33.2 33.3 33.0 32.8 32.7 32.3 31.9 32.9
20 Brazil 40.8 39.8 39.9 39.9 39.6 38.6 38.4 37.8 36.6 39.0
21 Brunei Darussalam 31.3 31.1 31.0 30.2 29.9 31.2 31.8 30.8 31.2 30.9
22 Bulgaria 37.3 36.9 36.6 36.1 35.6 34.9 34.1 33.5 32.7 35.3
23 Burkina Faso 41.3 41.4 41.3 41.4 40.3 40.1 39.7 39.7 39.6 40.5
24 Burundi 39.1 39.5 39.6 39.4 39.6 39.6 39.7 39.6 39.6 39.5
25 Cambodia 50.4 50.1 49.6 50.0 49.2 48.8 47.8 46.8 46.0 48.7
26 Cameroon 33.3 32.8 32.4 32.1 31.7 31.6 31.6 31.4 31.4 32.0
27 Canada 16.3 16.0 15.9 15.8 15.7 15.6 15.5 15.3 15.3 15.7
28 Cape Verde 36.5 36.1 35.9 35.9 35.7 35.8 35.4 34.1 33.4 35.4
29 Central African 42.8 42.6 43.1 44.0 46.9 47.3 46.9 45.9 45.1 45.0
Republic
30 Chad 45.8 46.2 45.5 45.1 44.2 41.5 41.1 41.7 42.2 43.7
31 Chile 19.9 19.8 19.6 19.6 19.4 19.1 18.9 18.7 18.5 19.3
32 China 13.2 13.1 13.0 12.9 12.8 12.6 12.5 12.2 11.9 12.7
33 Colombia 39.4 39.1 38.9 38.9 37.9 37.1 36.1 35.1 33.5 37.3
34 Comoros 39.3 39.6 39.0 37.7 37.6 39.0 38.0 38.4 39.4 38.7
35 Congo, Dem. Rep. 47.2 48.0 48.2 48.1 47.1 46.9 46.8 46.8 46.7 47.3
36 Congo, Rep. 49.5 48.2 47.2 46.8 46.8 46.2 44.7 43.3 44.6 46.4
37 Costa Rica 26.1 26.2 26.4 26.4 26.1 25.9 25.6 25.0 24.0 25.7



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
38 Côte d’Ivoire 41.4 43.2 44.3 45.5 46.0 46.1 46.3 46.8 47.0 45.2
39 Croatia 33.8 33.4 33.2 32.6 32.1 31.7 31.3 30.8 30.4 32.1
40 Cyprus 29.2 28.7 28.2 27.8 28.2 28.1 27.7 27.3 26.5 28.0
41 Czech Republic 19.3 19.1 18.9 18.8 18.7 18.4 17.8 17.3 17.0 18.4
42 Denmark 18.4 18.0 18.0 18.0 18.0 17.8 17.6 17.0 16.9 17.7
43 Dominican 32.4 32.1 32.4 32.1 32.1 32.4 31.7 31.0 30.5 31.9
Republic
44 Ecuador 34.2 34.4 33.7 33.3 32.8 31.6 30.8 30.4 30.4 32.4
45 Egypt, Arab Rep. 35.5 35.1 35.2 35.7 35.4 35.0 34.8 34.1 33.1 34.9
46 El Salvador 46.5 46.3 46.2 45.6 45.2 44.9 44.5 43.8 43.0 45.1
47 Equatorial Guinea 32.7 32.8 32.0 31.5 31.2 30.8 30.5 30.6 30.1 31.4
48 Eritrea 38.1 40.3 39.4 39.4 40.3 40.6 40.5 41.2 41.4 40.1
49 Estonia – 32.7 32.4 32.0 31.4 31.1 30.5 29.8 29.5 31.2
50 Ethiopia 40.6 40.3 39.5 39.6 40.1 38.6 37.7 36.3 35.1 38.6
51 Fiji 32.9 33.6 33.3 32.6 32.5 31.9 31.4 31.0 32.6 32.4
52 Finland 18.4 18.1 17.9 17.8 17.7 17.6 17.4 17.1 17.0 17.7
53 France 15.7 15.2 15.0 15.1 15.0 14.9 14.8 14.8 14.7 15.0
54 Gabon 46.2 48.0 47.4 47.6 47.5 48.0 47.7 48.0 47.3 47.5
55 Gambia, The 46.1 45.1 44.7 47.1 45.4 43.8 43.6 42.4 40.9 44.3
56 Georgia 68.3 67.3 67.2 67.2 65.9 65.5 65.1 63.6 62.1 65.8
57 Germany 16.4 16.0 15.9 16.1 16.3 16.1 16.0 15.6 15.3 16.0
58 Ghana 42.0 41.9 41.8 41.6 41.3 40.9 39.5 38.6 38.3 40.7
59 Greece 28.5 28.7 28.2 28.0 27.4 27.1 26.9 26.4 26.5 27.5
60 Guatemala 51.6 51.5 51.6 51.2 50.7 50.5 50.2 49.0 47.9 50.5
61 Guinea 39.7 39.6 39.3 38.7 38.8 38.5 38.4 38.9 39.2 39.0
62 Guinea-Bissau 40.4 39.6 39.6 40.7 41.5 41.9 41.7 41.5 41.6 40.9
63 Guyana 33.4 33.6 33.3 33.7 33.9 33.4 34.3 33.8 34.0 33.7
64 Haiti 54.8 55.4 56.1 56.5 56.4 57.4 57.1 57.0 57.1 56.4
65 Honduras 50.3 49.6 49.7 49.6 48.9 48.3 47.3 46.1 45.1 48.3
66 Hong Kong, China 17.0 16.6 16.6 16.6 16.4 15.9 15.5 15.0 14.7 16.0
67 Hungary 25.4 25.1 24.8 24.5 24.4 24.1 24.0 23.7 23.7 24.4
68 Iceland 16.0 15.9 15.8 16.0 15.9 15.5 15.1 15.0 15.0 15.6
69 India 23.2 23.1 22.8 22.6 22.3 22.0 21.7 21.2 20.7 22.2
70 Indonesia 19.7 19.4 19.4 19.3 19.1 18.8 18.6 18.3 17.9 18.9
71 Iran, Islamic Rep. 19.1 18.9 19.0 18.7 18.2 17.9 18.1 17.7 17.3 18.3
72 Ireland 16.1 15.9 15.9 15.9 16.0 15.8 15.6 15.5 15.4 15.8
73 Israel 22.7 21.9 22.3 22.7 22.7 22.1 21.8 21.2 20.7 22.0
74 Italy 27.8 27.1 26.7 26.8 27.0 27.0 27.1 26.9 26.8 27.0
75 Jamaica 36.4 36.4 36.2 36.2 34.4 33.9 34.0 32.9 32.5 34.8
76 Japan 11.4 11.2 11.2 11.3 11.2 10.9 10.7 10.4 10.3 11.0



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
77 Jordan 19.4 19.4 19.2 18.9 18.7 18.3 18.0 17.5 17.2 18.5
78 Kazakhstan 43.8 43.2 42.5 42.0 41.1 40.6 39.8 38.9 38.4 41.1
79 Kenya 33.7 34.3 34.0 34.8 34.6 33.7 32.7 31.1 29.5 33.2
80 Korea, Rep. 28.3 27.5 27.3 26.9 26.8 26.5 26.3 25.9 25.6 26.8
81 Kuwait 20.1 20.1 20.2 20.3 19.3 18.8 18.1 17.9 – 19.4
82 Kyrgyz Republic 41.4 41.2 40.8 41.4 40.5 39.8 40.1 39.8 38.8 40.4
83 Lao PDR 30.9 30.6 30.2 30.0 29.8 29.4 28.9 28.4 28.0 29.6
84 Latvia 30.8 30.5 30.1 29.8 29.4 29.0 28.4 27.7 27.2 29.2
85 Lebanon 34.1 34.1 33.7 33.5 33.2 32.4 32.4 32.8 32.0 33.1
86 Lesotho 31.7 31.3 31.1 31.0 30.7 30.1 30.2 29.3 28.8 30.5
87 Liberia 44.2 43.2 43.2 43.1 45.0 45.4 44.9 44.5 44.2 44.2
88 Libyan Arab 34.7 35.1 34.5 33.8 34.9 33.9 33.1 32.0 30.9 33.7
Jamahiria
89 Lithuania 33.8 33.7 33.3 32.8 32.0 31.7 31.0 30.4 29.7 32.0
90 Luxembourg 10.0 9.8 9.8 9.8 9.8 9.8 9.7 9.6 9.4 9.7
91 Macao, China 13.3 13.1 13.0 12.9 12.5 12.1 11.9 11.7 11.1 12.4
92 Macedonia, FYR 39.0 38.2 39.1 38.9 38.4 37.4 36.9 36.0 34.9 37.6
93 Madagascar 40.1 39.6 38.7 44.8 43.4 41.6 40.8 39.8 38.5 40.8
94 Malawi 39.9 40.3 42.5 44.4 43.4 42.5 42.6 41.3 39.4 41.8
95 Malaysia 32.2 31.1 31.6 31.5 31.2 30.7 30.4 30.0 29.6 30.9
96 Maldives 30.3 30.3 30.0 29.4 29.2 28.9 29.6 29.3 28.6 29.5
97 Mali 42.5 42.3 40.8 40.2 39.9 40.6 40.1 39.9 39.9 40.7
98 Malta 27.4 27.1 27.3 27.3 27.5 27.6 27.3 27.0 26.5 27.2
99 Mauritania 35.5 36.1 36.0 35.8 35.8 35.1 34.4 31.7 – 35.1
100 Mauritius 23.3 23.1 22.9 23.0 22.7 22.4 22.4 22.2 21.9 22.7
101 Mexico 30.8 30.1 30.3 30.4 30.5 30.1 29.9 29.2 28.8 30.0
102 Moldova 45.6 45.1 44.1 44.5 44.6 44.0 43.4 44.3 – 44.5
103 Mongolia 18.4 18.4 18.3 18.0 17.7 17.4 17.1 16.7 16.4 17.6
104 Morocco 36.5 36.4 35.7 35.5 35.0 34.2 34.9 33.1 33.1 34.9
105 Mozambique 41.1 40.3 40.4 39.8 39.8 39.7 38.9 38.6 – 39.8
106 Myanmar 51.6 52.6 51.5 50.7 49.0 49.1 47.8 – – 50.3
107 Namibia 31.4 31.4 31.2 31.3 30.7 29.7 29.6 28.8 28.5 30.3
108 Nepal 37.2 36.8 36.7 37.1 36.9 36.8 36.7 36.3 36.0 36.7
109 Netherlands 13.3 13.1 13.1 13.2 13.3 13.2 13.2 13.2 13.0 13.2
110 New Zealand 13.0 12.8 12.6 12.4 12.2 12.0 12.1 12.1 12.0 12.4
111 Nicaragua 45.7 45.2 45.3 45.5 45.0 44.2 43.8 43.5 43.1 44.6
112 Niger 41.7 41.9 40.9 40.3 39.7 40.7 39.7 38.6 – 40.4
113 Nigeria 58.0 57.9 57.8 57.6 56.3 55.1 53.8 53.0 – 56.2
114 Norway 19.2 19.1 19.0 19.0 19.0 18.5 18.5 18.2 18.0 18.7
115 Oman 19.1 18.9 18.5 18.5 18.4 18.3 18.0 17.6 – 18.4



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
116 Pakistan 37.0 36.8 37.0 36.8 36.2 35.3 34.9 33.8 33.6 35.7
117 Panama 64.8 64.1 64.7 65.1 64.4 63.5 61.7 60.0 – 63.5
118 Papua New Guinea 35.5 36.1 36.8 37.1 37.1 37.0 37.2 37.1 36.5 36.7
119 Paraguay 38.0 39.8 39.7 40.1 39.1 38.3 38.2 37.4 – 38.8
120 Peru 60.1 59.9 60.2 59.1 58.6 57.9 57.2 55.7 53.7 58.0
121 Philippines 43.8 43.3 43.0 42.5 42.0 41.6 40.1 39.5 38.3 41.6
122 Poland 27.7 27.6 27.7 27.7 27.5 27.3 26.9 26.4 26.0 27.2
123 Portugal 23.0 22.7 22.6 22.7 23.0 23.1 23.3 23.2 23.0 23.0
124 Quatar – 19.0 19.3 19.0 19.6 17.4 18.4 0.0 0.0 14.1
125 Romania 34.3 34.4 33.7 33.5 32.8 32.0 31.7 30.7 30.2 32.6
126 Russian Federation 47.0 46.1 45.3 44.5 43.6 43.0 42.4 41.7 40.6 43.8
127 Rwanda 40.5 40.3 40.6 39.9 40.7 40.2 39.3 39.1 – 40.1
128 Saudi Arabia 18.7 18.4 18.7 19.2 18.3 17.7 17.4 17.4 16.8 18.1
129 Senegal 45.0 45.1 44.5 45.1 44.4 43.2 42.3 42.4 41.7 43.7
130 Sierra Leone 48.6 48.6 47.6 45.4 44.8 44.4 44.3 43.6 42.9 45.6
131 Singapore 13.3 13.1 13.3 13.3 13.1 12.8 12.7 12.4 12.2 12.9
132 Slovak Republic 18.9 18.9 18.8 18.6 18.3 18.1 17.6 17.2 16.8 18.1
133 Slovenia 27.3 27.1 26.7 26.6 26.4 26.2 25.8 25.3 24.7 26.2
134 Solomon Islands 31.7 33.4 34.5 34.8 34.7 33.8 33.4 33.2 32.7 33.6
135 South Africa 28.4 28.4 28.4 28.0 27.8 27.1 26.5 26.0 25.2 27.3
136 Spain 23.0 22.7 22.4 22.4 22.4 22.5 22.4 22.4 22.2 22.5
137 Sri Lanka 45.2 44.6 44.6 44.1 43.8 43.9 43.4 42.9 42.2 43.9
138 Sudan 34.1 – – – – – – – – 34.1
139 Suriname 39.7 39.8 39.3 38.9 38.1 36.9 36.5 35.9 35.1 37.8
140 Swaziland 43.5 41.4 41.3 40.9 40.2 40.1 39.3 38.9 – 40.7
141 Sweden 19.6 19.2 19.1 19.0 18.7 18.5 18.6 18.2 17.9 18.8
142 Switzerland 8.8 8.6 8.6 8.6 8.8 8.6 8.5 8.3 8.1 8.5
143 Syrian Arab 19.3 19.3 19.2 19.1 19.3 19.1 19.0 18.7 18.5 19.1
Republic
144 Taiwan 25.7 25.4 25.7 25.4 25.2 24.7 24.5 24.2 23.9 25.0
145 Tajikistan 43.5 43.2 42.9 42.7 42.1 41.7 41.5 41.2 41.0 42.2
146 Tanzania 58.6 58.3 57.7 56.9 56.6 56.0 55.4 54.7 53.7 56.4
147 Thailand 53.4 52.6 52.4 51.5 50.2 49.6 49.0 48.5 48.2 50.6
148 Togo 34.4 35.1 35.4 34.5 34.9 35.0 35.0 34.6 – 34.9
149 Trinidad and 34.7 34.4 34.3 34.4 33.4 33.1 32.9 31.9 31.5 33.4
Tobago
150 Tunisia 38.7 38.4 37.8 37.8 37.4 36.9 36.7 35.9 35.4 37.2
151 Turkey 32.7 32.1 32.8 32.4 31.8 31.0 30.0 29.5 29.1 31.3
152 Uganda 43.5 43.1 42.9 42.9 42.5 42.4 42.2 41.0 40.3 42.3
153 Ukraine 52.7 52.2 51.4 50.8 49.7 48.8 47.8 47.3 46.8 49.7



Average
1999 2000 2001 2002 2003 2004 2005 2006 2007
154 United Arab 26.3 26.4 27.0 27.4 26.3 25.4 24.8 23.5 – 25.9
Emirates
155 United Kingdom 12.8 12.7 12.6 12.6 12.5 12.4 12.4 12.3 12.2 12.5
156 United States 8.8 8.7 8.8 8.8 8.7 8.6 8.5 8.4 8.4 8.6
157 Uruguay 50.5 51.1 51.7 54.0 53.6 51.1 49.2 48.5 46.1 50.6
158 Venezuela, RB 33.8 33.6 33.5 35.5 36.9 34.9 33.5 32.0 30.9 33.8
159 Vietnam 15.8 15.6 15.5 15.3 15.2 15.1 14.7 14.6 14.4 15.1
160 Yemen, Rep. 27.7 27.4 27.3 27.2 27.0 27.0 26.6 26.8 26.8 27.1
161 Zambia 49.3 48.9 48.3 48.1 47.5 46.8 46.3 45.0 43.9 47.1
162 Zimbabwe 59.6 59.4 61.5 62.8 63.7 62.3 62.0 62.3 62.7 61.8
Time Average 34.0 33.7 33.6 33.6 33.3 32.9 32.5 31.9 31.0
NOTES
1. The estimation of such a currency demand equation has been criticized by Thomas
(1999), but part of this criticism has been considered by the work of Giles (1999a, b) and
Bhattacharyya (1999), who both use the latest econometric techniques.
2. See Karmann (1986 and 1990), Schneider (1997, 1998a, 2005), Johnson, Kaufmann and
Zoido-Lobatón (1998a) and Williams and Windebank (1995).
3. See Feige (1986); Thomas (1992, 1999); Pozo (1996); Pedersen (2003) and Ahumada et al.
(2004); Janisch and Brümmerhof (2005); and Breusch (2005a, b).
4. One (weak) justification for the use of only the tax variable is that this variable has by
far the strongest impact on the size of the shadow economy in the studies known to the
authors. The only exception is the study by Frey and Weck-Hannemann (1984) where
the variable ‘tax immorality’ has a quantitatively larger and statistically stronger influ-
ence than the direct tax share in the model approach. In the study of Pommerehne and
Schneider (1985) for the United States data, tax immorality and minimum wage rates
are – besides various tax measures – available, and the tax variable has a dominating
influence and contributes roughly 60–70 per cent of the size of the shadow economy. See
also Zilberfarb (1986).
5. In another study by Tanzi (1982, esp. pp. 110–113), he explicitly deals with this criticism.
A very careful investigation of the amount of US dollars used abroad and in the shadow
economy and to ‘classical’ crime activities has been undertaken by Rogoff (1998), who
concludes that large denomination bills are the major driving force for the growth of the
shadow economy and classical crime activities are due largely to reduced transactions
costs.
6. However, in studies for European countries, Kirchgässner (1983, 1984) and Schneider
(1986) reach the conclusion that the estimation results for Germany, Denmark, Norway
and Sweden are quite robust when using the currency demand method. Hill and Kabir
(1996) find for Canada that the rise of the shadow economy varies with respect to the
tax variable used; they conclude ‘when the theoretically best tax rates are selected and a


range of plausible velocity values is used, this method estimates underground economic
growth between 1964 and 1995 at between 3 and 11 percent of GDP’ (Hill and Kabir
[1996, p. 1553]).
7. A number of variables is not available for 2007, hence we have two different sets of cause
variables.


2 Survey on the shadow economy and
undeclared work in OECD countries*
Lars P. Feld and Friedrich Schneider
2.1 INTRODUCTION
Fighting tax evasion and the shadow economy has been an important
policy goal in OECD countries during recent decades. One notable
example is Germany. When the so-called ‘Black Activities’ Act’ or more
exactly the ‘Law to intensify the fight against black activities and accom-
panying tax evasion’ (SchwarzArbG, Bundesrats-Drucksache 155/04a)
was passed in 2004, this was but the preliminary peak of the German gov-
ernment’s efforts to deter tax non-compliance. It was followed by a raid
by German investigators against dishonest taxpayers in Liechtenstein in
2008 (the Zumwinkel Affair) and the most recent pressure on Switzerland
to extend administrative and legal cooperation to tax evasion (in addition
to tax fraud).
In all these cases, deterrence appears to be the policy instrument of
choice. Regarding the shadow economy, for example, the German ‘Black
Activities’ Act’ establishes a uniform law for illegal activities, which were
part of the criminal tax code, of social security legislation or immigra-
tion laws before, and thus facilitates deterrence. Moreover, it increases
punishment by focusing on employers. Misdemeanors for offences against
declaration duties are defined and fined up to €300 000. A fine for resist-
ance against or neglect of audit duties up to €50 000 is enacted and illegal
employment of foreigners can be sentenced by penalty or prison from one
to up to six years depending on the severity of the offence and circum-
stances. Finally, the ‘Black Activities’ Act’ aims at an increase in the prob-
ability of detection by extending the investigation force, that is customs
agencies are authorized to support tax and social security agencies.
In addition, German policy against the shadow economy comprises
positive measures in order to give people incentives to work in the official
economy. This includes some reduction of the tax burden for low income
people who presumably work more heavily in the underground economy.
The mini job legislation, for example, requires employers of people who
78


Survey on the shadow economy in OECD countries 79
earn up to € 400 per month to pay social security contributions and taxes
of only 30.1 per cent without the usual 50 per cent split between employer
and employee. There are further reductions of these contributions in the
case of private households who can also deduct some of their expenses for
housekeeping from their tax payments. Further tax deductions have been
introduced for craftsmen expenses allowing the deduction of 20 per cent of
expenses up to €1200 which implies a tax bonus of up to €6000 of labour
costs (including VAT).
In Germany, these policy measures have been accompanied by a media
campaign by the federal finance ministry which aims at a change of public
opinion against the shadow economy. People should become aware that
working in the underground economy is not a peccadillo only because it
appears to be widespread. Similar legal measures and media campaigns
have been conducted by the Swiss federal government in connection with
the Swiss ‘Black Activities’ Act’ of 2008, but are widespread also in most
EU countries (Williams, Horlings and Renooy, 2008).
To date, the success of these recent policies (in Germany and Switzerland)
has, however, not been evaluated properly. The idea of such an evaluation
study has not been pursued by the German Federal Finance Ministry
further.1 In this chapter, we are not able to show results of such an evalua-
tion as the data necessary for that are not available. However, we take stock
of the research conducted more recently in order to shed some light on the
debate on success and failure of the policies to fight the shadow economy.
The recent research more heavily relies on questionnaire and survey designs
than that previously done and thus allows for a closer look on the influence
of deterrence, tax policies, regulatory policies or tax morale on the prob-
ability of individuals working underground. Direct estimates of the shadow
economy are put into a wider context by comparison with estimates from
indirect approaches (like the MIMIC approach). Finally, the interactions
between the shadow and the official economies are discussed. The overall
picture we draw is rather differentiated with respect to policies, but also to
the assessment as to how detrimental the shadow economy actually is.
In this survey, we are mainly concerned with the shadow economy,
black activities, the underground economy or undeclared work. Tax
evasion is not considered in order to keep the subject of this chapter
tractable and because too many additional aspects would be involved.2
Also, tax morale or experimental studies on tax compliance are beyond
the scope of this chapter.3 This survey is organized as follows: Section 2.2
presents theoretical considerations from the definition and measurement
of the shadow economy to the main factors determining its size. In Section
2.3, results from survey studies are discussed. They are compared with the
main evidence from the MIMIC approach in Section 2.4. The interaction


between shadow and official economies is focused on in Section 2.5.

Section 2.6 concludes.
2.2 SOME THEORETICAL CONSIDERATIONS

ABOUT THE SHADOW ECONOMY
2.2.1 Defining the Shadow Economy
Most authors trying to measure the shadow economy still face the dif-
ficulty of definition.4 According to one commonly used definition it
comprises all currently unregistered economic activities that contribute
to the officially calculated (or observed) Gross National Product.5 Smith
(1994, p. 18) defines it as ‘market-based production of goods and services,
whether legal or illegal, that escape detection in the official estimates of
GDP’. Put differently, one of the broadest definitions is: ‘those economic
activities and the income derived from them that circumvent or other-
wise avoid government regulation, taxation or observation’.6 As these
definitions still leave room for interpretation, Table 2.1 may provide a
better feeling as to what could be a reasonable consensus definition of the
underground (or shadow) economy.
From Table 2.1, it is obvious that a broad definition of the shadow
economy includes unreported income from the production of legal goods
and services, either from monetary or barter transactions – and so includes
all productive economic activities that would generally be taxable were
they reported to the state (tax) authorities. In this chapter, the follow-
ing more narrow definition of the shadow economy is used.7 The shadow
economy includes all market-based legal production of goods and services
that are deliberately concealed from public authorities for the following
reasons:

minimum wages, maximum working hours and safety standards, and
4. to avoid complying with certain administrative obligations, such as
Thus, we will not deal with typically illegal underground economic activi-
ties that fit the characteristics of classical crimes like burglary, robbery,
drug dealing and so forth. We also exclude the informal household
economy which consists of all household services and production.


Table 2.1 A taxonomy of types of underground economic activitiesa
Type of Activity Monetary Transactions Non Monetary

Transactions
ILLEGAL Trade with stolen goods; drug Barter of drugs, stolen
ACTIVITIES dealing and manufacturing; goods, smuggling etc.
prostitution; gambling; Produce or growing
smuggling; fraud; etc. drugs for own use.
Theft for own use.
Tax Evasion Tax Tax Tax
Avoidance Evasion Avoidance
LEGAL Unreported Employee Barter All do-it-
ACTIVITIES income from discounts, of legal yourself
self-employment; fringe services work and
wages, salaries benefits and goods neighbour
and assets from help
unreported work
related to legal
services and
goods
Notes: a Structure of the table is taken from Lippert and Walker (1997, p. 5) with
additional remarks.
2.2.2 Measuring the Shadow Economy
The definition of the shadow economy plays an important role in assessing

its size. By having a clear definition, a number of ambiguities and contro-
versies can be avoided. In general, there are two types of underground
economic activity: illicit employment and the production of goods and
services consumed within the household.8 The following analysis focuses
on the former type and excludes illegal activities such as drug production,
crime and human trafficking. The latter type includes the production of
goods and services, consumed within the household, or childcare and is
not part of this analysis either. Thus, it only focuses on productive eco-
nomic activities that would normally be included in the national accounts
but which remain underground due to tax or regulatory burdens.9
Although such legal activities contribute to the country’s value added,
they are not captured in the national accounts because they are produced
in illicit ways (e.g. by people without proper qualifications or without a
master craftsman’s certificate). From the economic and social perspective,
soft forms of illicit employment, such as moonlighting (e.g. construction


work in private homes) and its contribution to aggregate value added can
be assessed rather positively.
Although the issue of the shadow economy has been investigated for
a long time, the discussion regarding the ‘appropriate’ methodology to
assess its scope has not yet come to an end.10 There are three methods of
assessment:
(1) Direct procedures at a micro level that aim at determining the size of
the shadow economy at one particular point in time. An example is
the survey method;
(2) Indirect procedures that make use of macroeconomic indicators in
order to proxy the development of the shadow economy over time;
(3) Statistical models that use statistical tools to estimate the shadow
economy as an ‘unobserved’ variable.
The estimation of the shadow economy of highly developed OECD coun-

tries (with a stronghold on Austria and Germany) is firstly based on a com-
bination of the MIMIC procedure and the currency demand method, that
is a combination of methods (2) and (3).11 The MIMIC procedure assumes
that the shadow economy remains an unobserved phenomenon (latent
variable) which can be estimated using quantitatively measurable causes of
illicit employment, for example, tax burden and regulation intensity, and
indicators reflecting illicit activities, for example, currency demand, official
GDP and official working time. A disadvantage of the MIMIC procedure
is the fact that it produces only relative estimates of the size and the devel-
opment of the shadow economy. Thus, the currency demand method12 is
used to calibrate the relative into absolute estimates by using two or three
absolute values of the absolute size of the shadow economy.
Secondly, the size of the shadow economy is estimated by using survey
methods. In order to minimize the number of respondents dishonestly reply-
ing or totally declining answers to the sensitive questions, structured inter-
views are undertaken (usually face-to-face) in which the respondents are
slowly getting accustomed to the main purpose of the survey. As done in the
contingent valuation method (CVM) in environmental economics (Kopp,
Pommerehne and Schwarz, 1997), the first part of the questionnaire aims at
shaping respondents’ perception as to the issue at hand. In the second part,
questions about respondents’ activities in the shadow economy are asked,
and the third part contains the usual socio-demographic questions.
In the survey studies undertaken by Feld and Larsen (2005, 2008, 2011),
the interviewers led up to the sensitive questions by asking respondents
about their do-it-yourself activities at home, and then continued to the
part about the shadow economy in the following way:


The next questions are about what are popularly called ‘black activities’.
There is considerable evidence to show that a large part of the population
accept ‘black activities’ and ‘black transactions’ – i.e. activities which circum-
vent the Inland Revenue, where all parties benefit because they do not pay tax
or VAT, etc. This can involve ‘black activities’ which you pay for in cash, but
can also include reciprocal favors between friends, acquaintances and family
members.
Have you carried out activities of this kind during the past year? 13
In addition to the studies by Merz and Wolff (1993), Feld and Larsen
(2005, 2008, 2011) and Enste and Schneider (2006) for Germany, the
survey method has been applied in the Nordic countries and Great Britain
(Isachsen and Strøm, 1985; Pedersen, 2003) as well as in the Netherlands
(van Eck and Kazemier, 1988; Kazemier, 2006). While the question-
naires underlying these studies are broadly comparable in design, recent
attempts by the European Union to provide survey results for all EU
member states runs into difficulties regarding comparability (Renooy et
al., 2004; European Commission, 2007): the wording of the questionnaires
becomes more and more cumbersome depending on the culture of differ-
ent countries with respect to the underground economy.
These two sets of approaches are most broadly used in the literature.
Although each has its drawbacks, and although biases in the estimates
of the shadow economy almost certainly prevail, no better data are cur-
rently available. In tax compliance research, the most interesting data
stem from actual tax audits by the US Internal Revenue Service (IRS).
In the Taxpayer Compliance Measurement Program (TCMP), actual
compliance behaviour of taxpayers is observed and is used for empirical
analysis (Andreoni, Erard and Feinstein, 1998). The approach of the IRS
is broader in a certain sense as tax evasion from all sources of income
is considered, while the two methods discussed before aim at capturing
the shadow economy or undeclared work and thus mainly measure tax
evasion from labour income. Even the data obtained from the TCMP is
biased, however, because the actually detected tax non-compliance could
only be the tip of the iceberg. Although the perfect data on tax non-
compliance does therefore not exist, the imperfect data in this area can still
provide interesting insights also regarding the size, the development and
the determinants of the shadow economy.
2.2.3 The Main Causes Determining the Shadow Economy
A useful starting point for a theoretical discussion of tax non-compliance

is the paper by Allingham and Sandmo (1972) on income tax evasion.
While the shadow economy and tax evasion are not congruent, activities


in the shadow economy in most cases imply the evasion of direct or indi-
rect taxes, such that the factors affecting tax evasion will most certainly
also affect the shadow economy. According to Allingham and Sandmo
(1972), tax compliance depends on its expected costs and benefits. The
benefits of tax non-compliance result from the individual marginal tax
rate and the true individual income. In the case of the shadow economy
the individual marginal tax rate is obtained by calculating the overall mar-
ginal tax burden from indirect and direct taxes including social security
contributions. The individual income generated in the shadow economy is
usually categorized as labour income and less probably as capital income.
The expected costs of non-compliance derive from deterrence enacted
by the state. Tax non-compliance thus depends on the state’s auditing
activities raising the probability of detection and the fines individuals face
when they are caught. As individual morality also plays a role for compli-
ance, additional costs could pertain beyond pure punishment by the tax
administration in the form of psychic costs like shame or regret, but also
additional pecuniary costs if, for example, reputation loss results.
Kanniainen, Pääkönen and Schneider (2004) incorporate many of
these insights in their model of the shadow economy by also considering
labour supply decisions. They hypothesize that tax hikes unambiguously
increase the shadow economy, while the effect of public goods financed by
those taxes depends on the ability to access public goods. Morality is also
included in this analysis. But the costs for individual non-compliers result-
ing from moral norms appear to be mainly captured by state punishment
although self-esteem also plays a role.
A shortcoming of these analyses is the neglected endogeneity of tax
morale and good governance. In contrast, Feld and Frey (2007) argue
that tax compliance is the result of a complicated interaction between
tax morale and deterrence measures. While it must be clear to taxpayers
what the rules of the game are and as deterrence measures serve as signals
for the tax morale a society wants to elicit (Posner 2000a, b), deterrence
could also crowd out the intrinsic motivation to pay taxes. Moreover,
tax morale is not only increased if taxpayers perceive the public goods
received in exchange for their tax payments worth it. It also increases if
political decisions for public activities are perceived to follow fair proce-
dures or if the treatment of taxpayers by the tax authorities is perceived to
be friendly and fair. Tax morale is thus not exogenously given, but is influ-
enced by deterrence, the quality of state institutions and the constitutional
differences among states.
Although this leaves us with a rich set of variables that might influence
the size of the shadow economy, it is only the starting point. As labour
supply decisions are involved, labour and product market regulations are


additionally important. Recent theoretical approaches thus suggest fol-

lowing a differentiated policy to contain the shadow economy’s expansion.
This cautionary note becomes even stronger when the following empirical
evidence is considered:
Deterrence
Although the traditional economic theory of tax non-compliance derives
unambiguous predictions as to their impact only for deterrence measures
and despite the strong focus on deterrence in policies fighting the shadow
economy, there is surprisingly little known about the effects of deterrence
from empirical studies. In their survey on tax compliance, Andreoni,
Erard and Feinstein (1998) report that deterrence matters for tax evasion,
but that the reported effects are rather small. Blackwell (2010) finds
strong deterrence effects of fines and audits in experimental tax evasion.
Regarding the shadow economy, there is, however, little evidence.
This is due to the fact that data on the legal background and the fre-
quency of audits are not available on an international basis. They would
also be difficult to collect even for the OECD member countries. A recent
study by Feld, Schmidt and Schneider (2007) demonstrates this for the
case of Germany. The legal background is quite complicated differentiat-
ing fines and punishment according to the severity of the offence, to true
income of the non-complier, but also regionally given different directives
on sentences by the courts in different Länder (states). Moreover, the
tax authorities at the state level do not reveal how intensively auditing is
taking place. With the available data on fines and audits, Feld, Schmidt
and Schneider (2007) conduct a time series analysis using the estimates of
the shadow economy obtained by the MIMIC approach. According to
their results, deterrence does not have a consistent effect on the German
shadow economy. Conducting Granger causality tests, the direction of
causation (in the sense of precedence) is ambiguous leaving room for an
impact of the shadow economy on deterrence instead of deterrence on the
shadow economy.
Feld and Larsen (2005, 2008, 2011) follow a different approach by using
individual survey data for Germany. First replicating Pedersen (2003),
who reports a negative impact of the subjectively perceived risk of detec-
tion by state audits on the probability of working in the shadows for the
year 2001, they then extend it by adding subjectively perceived measures
of fines and punishment. Fines and punishment do not exert a negative
influence on the shadow economy in any of the annual waves of surveys,
or in the pooled regressions for the years 2004–2007 (about 8000 observa-
tions overall). The subjectively perceived risk of detection has a robust
and significant negative impact in individual years only for women. In the


pooled sample for 2004–2007, which minimizes sampling problems, the

probability of detection has a significantly negative effect on the probabil-
ity of working in the shadow economy also for men (keeping the one for
women) and is robust across different specifications.14
Pedersen (2003) reports negative effects of the subjectively perceived
risk of detection on the probability of conducting undeclared work in the
shadows for men in Denmark in 2001 (marginally significant), for men in
Norway in 1998/2002 (highly significant),15 men and women in Sweden in
1998 (highly significant in the first and marginally significant in the second
case), and no significant effect for Great Britain in 2000. Moreover, van Eck
and Kazemier (1988) report a significant negative effect of a high perceived
probability of detection on participation in the hidden labour market for
the Netherlands in 1982/1983. None of these studies include perceived fines
and punishments as explanatory variables. The large scale survey study
on Germany by Feld and Larsen (2005, 2011) thus appears to be the most
careful analysis of deterrence effects on undeclared work up to date.
Overall, this is far from convincing evidence on the proper working of
deterrence as it is always the combination of audits and fines that matters
according to theoretical analysis, but also to pure plausibility argu-
ments. The reasons for the unconvincing evidence of deterrence effects
are discussed in the tax compliance literature by Andreoni, Erard and
Feinstein (1998), Feld and Frey (2007) or Kirchler (2007). They range
from interactions between tax morale and deterrence, thus the possibility
that deterrence crowds out tax morale, to more mundane arguments like
misperceptions of taxpayers. Likewise, these reasons could be important
for the evidence on the deterrence effects on work in the shadow economy.
As the latter mainly stems from survey studies, the insignificant findings
for fines and punishment may also result from shortcomings in the survey
design.
Tax and social security contribution burdens

In contrast to deterrence, almost all studies ascertain that the tax and social
security contribution burdens are among the main causes for the existence
of the shadow economy.16 Since taxes affect labour–leisure choices and
stimulate labour supply in the shadow economy, the distortion of the
overall tax burden is a major concern. The bigger the difference between
the total labour cost in the official economy and after-tax earnings (from
work), the greater is the incentive to reduce the tax wedge and work in the
shadow economy. Since the tax wedge depends on the level and increase
of the social security burden/payments and the overall tax burden, they
are key features of the existence and the increase of the shadow economy.
But even tax reforms with major tax rate deductions will not necessarily


lead to a substantial decrease of the shadow economy.17 Such reforms

are able to stabilize the size of the shadow economy and avoid a further
increase. Social networks and personal relationships, high profit from
irregular activities and associated investments in real and human capital
prevent people from transferring to the official economy. For Canada,
Spiro (1993) found similar reactions of people facing an increase in indirect
taxes (VAT, GST). This fact makes it even more difficult for politicians to
carry out major reforms because they may not gain a lot from them.
Empirical results of the influence of the tax burden on the shadow
economy is provided by Schneider (1994a, 2000, 2004, 2005) and Johnson,
Kaufmann and Zoido-Lobatón (1998a, b); they report a statistically
significant influence of taxation on the shadow economy. For Austria,
the driving force of the shadow economy is the direct tax burden (includ-
ing social security payments); this biggest influence is followed by the
intensity of regulation and complexity of the tax system. Schneider (1986)
obtains a similar result for Scandinavian countries (Denmark, Norway
and Sweden). In all three countries various tax variables: average direct
tax rate, average total tax rate (indirect and direct tax rate) and marginal
tax rates have the expected positive effect (on currency demand) and are
highly statistically significant. These findings are also supported by studies
of Kirchgässner (1983) for Germany and by Klovland (1984) for Norway
and Sweden. In the survey studies by Feld and Larsen (2005, 2011), per-
ceived tax rates do not have a robust and significant effect on the prob-
ability of engaging in undeclared work. Contrary to deterrence, a closer
look at the data reveals much more clearly that respondents had huge
difficulties of properly assessing their tax burden.
Increased intensity of regulations, for example, labour market regula-
tions, trade barriers and labour restrictions for immigrants, is another
important factor which reduces the freedom (of choice) for individu-
als engaged in the official economy. Johnson, Kaufmann and Zoido-
Lobatón (1998b) find significant empirical evidence of the influence
of (labour) regulations on the shadow economy; and the impact is
clearly described and theoretically derived in other studies, for example,
for Germany (Deregulierungskommission/ Deregulation Commission
1991).18 Regulations lead to a substantial increase in labour costs in the
official economy. But since most of these costs can be shifted to employees,
regulations provide for another incentive to work in the shadow economy
where they can be avoided. Johnson, Kaufmann and Shleifer (1997) report
empirical evidence supporting their model which predicts that countries
with higher general regulation of their economies tend to have a higher


share of the unofficial economy in total GDP. They conclude that it is the
enforcement of regulation which is the key factor for the burden levied on
firms and individuals, and not the overall extent of regulation – mostly
not enforced – which drives firms into the shadow economy. Friedman
et al. (2000) arrive at a similar conclusion. In their study, every available
measure of regulation is significantly correlated with the share of the unof-
ficial economy and the estimated sign of the relationship is unambiguous:
more regulation is correlated with a larger shadow economy.
These findings demonstrate that governments should put more empha-
sis on improving enforcement of laws and regulations, rather than increas-
ing their number. Some governments, however, prefer this policy option
(more regulations and laws), when trying to reduce the shadow economy,
mostly because it leads to an increase in power for the bureaucrats and to
a higher rate of employment in the public sector.
Public sector services

An increase of the shadow economy can lead to reduced state revenues
which in turn reduce the quality and quantity of publicly provided goods
and services. Ultimately, this can lead to an increase in the tax rates for
firms and individuals in the official sector, quite often combined with
a deterioration in the quality of the public goods (such as the public
infrastructure) and of the administration, with the consequence of even
stronger incentives to participate in the shadow economy. Johnson,
Kaufmann and Zoido-Lobatón (1998a, b) present a simple model of this
relationship. According to their findings smaller shadow economies occur
in countries with higher tax revenues achieved by lower tax rates, fewer
laws and regulations and less bribery facing enterprises. Countries with
a better rule of law, which is financed by tax revenues, also have smaller
shadow economies. Transition countries have higher levels of regulation
leading to a significantly higher incidence of bribery, higher effective taxes
on official activities and a large discretionary framework of regulations
and consequently a higher shadow economy. Their overall conclusion is
that
wealthier countries of the OECD, as well as some in Eastern Europe, find them-
selves in the ‘good equilibrium’ of relatively low tax and regulatory burden,
sizeable revenue mobilization, good rule of law and corruption control, and
a [relatively] small unofficial economy. By contrast, a number of countries in
Latin American and the former Soviet Union exhibit characteristics consistent
with a ‘bad equilibrium’: tax and regulatory discretion and burden on the firm
is high, the rule of law is weak, and there is a high incidence of bribery and a rel-
atively high share of activities in the unofficial economy. (Johnson, Kaufmann
and Zoido-Lobatón, 1998a, p. I).


Other public institutions

Recently, various authors have considered the quality of public institu-
tions as another key factor of the development of the informal sector.
These authors argue19 that the efficient and discretionary application of
tax systems and regulations by government may play a crucial role in
the decision of conducting undeclared work, even more important than
the actual burden of taxes and regulations. In particular, corruption of
bureaucracy and government officials seems to be associated with larger
unofficial activity, while a good rule of law by securing property rights and
contract enforceability, increases the benefits of being formal.
Hence, it is important to analyze theoretically and empirically the effect
of political institutions like the federal political system on the shadow
economy. If the development of the informal sector is considered as a
consequence of the failure of political institutions in promoting an effi-
cient market economy, since entrepreneurs go underground when there is
an inefficient public goods provision, then the effect of institutions of the
individual’s incentive to operate unofficially can be assessed. In a federal
system, competition among jurisdictions and the mobility of individuals
act as constraints on politicians because ‘choices’ will be induced that
provide incentives to adopt policies which are closer to a majority of
voters’ preferences. Frequently, the efficient policies are characterized by
a certain level of taxation, mostly spent in productive public services. In
fact, the production in the formal sector benefits from a higher provision
of the productive public services and is negatively affected by taxation,
while the shadow economy reacts in the opposite way. As fiscal policy gets
closer to a majority of voters’ preferences in federal systems, the size of
the informal sector goes down. This leads to the hypothesis that the size of
the shadow economy should be lower in a federal system than in a unitary
state, ceteris paribus.
Tax morale
In addition to the incentives effects discussed before, the efficiency of the
public sector has an indirect effect on the size of the shadow economy
because it affects tax morale. As Feld and Frey (2007) argue, tax com-
pliance is driven by a psychological tax contract that entails rights and
obligations from taxpayers and citizens on the one hand, but also from the
state and its tax authorities on the other hand. Taxpayers are more heavily
inclined to pay their taxes honestly if they get valuable public services in
exchange. However, taxpayers are honest even in cases when the benefit
principle of taxation does not hold, that is for redistributive policies, if the
political decisions underlying such policies follow fair procedures. Finally,
the treatment of taxpayers by the tax authority plays a role. If taxpayers


are treated like partners in a (tax) contract instead of subordinates in a

hierarchical relationship, taxpayers will stick to their obligations of the
psychological tax contract more easily. In addition to the empirical evi-
dence on these arguments reported by Feld and Frey (2007), Kirchler
(2007) presents a comprehensive discussion of the influence of such factors
on tax compliance.
Regarding the impact of tax morale on the shadow economy, there
is scarce and only recent evidence. Using data on the shadow economy
obtained by the MIMIC approach, Torgler and Schneider (2009) report
the most convincing evidence for a negative effect of tax morale. They
particularly address causality issues and establish a causal negative rela-
tion from tax morale to the size of the shadow economy. This effect is
also robust to the inclusion of additional explanatory factors and speci-
fications. These findings are in line with earlier preliminary evidence by
Körner et al. (2006). Using survey data, Feld and Larsen (2005, 2011) like-
wise report a robust negative effect of tax morale in particular and social
norms in general on the probability of respondents to conduct undeclared
work. Interestingly, the estimated effects of social norms are quantita-
tively more important than the estimated deterrence effects. Van Eck and
Kazemier (1988) also report a marginally significant effect of tax morale
on the participation in the hidden labour market.
Summary of the main causes of the shadow economy

In Table 2.2, an overview of a number of empirical studies summarizes the
various factors influencing the shadow economy. The overview is based
on the studies in which the size of the shadow economy is measured by
the MIMIC approach. As there is no evidence on deterrence using this
approach – at least with respect to the broad panel data base on which this
table draws – the most central policy variable does not show up. This is
an obvious shortcoming of the studies, but one that cannot be coped with
easily due to the lack of internationally comparable deterrence data. In
Table 2.2, two columns are presented, showing the various factors influ-
encing the shadow economy with and without the independent variable,
‘tax morale’. This table clearly demonstrates that the increase of tax and
social security contribution burdens is by far the most important single
contributor to the increase of the shadow economy. This factor explains
some 35–38 per cent or 45–52 per cent of the variance of the shadow
economy with and without including the variable ‘tax morale’. The vari-
able tax morale accounts for some 22–25 per cent of the variance of the
shadow economy,20 there is a third factor, ‘quality of state institutions’,
accounting for 10–12 per cent and a fourth factor, ‘intensity of state regu-
lation’ (mostly for the labour market) for 7–9 per cent. In general, Table


Table 2.2 Main causes of the increase of the shadow economy
Factors influencing the shadow economy Influence on the shadow

economy (in %)
(a) (b)
(1) Increase of the Tax and Social Security 35–38 45–52
Contribution Burdens
(2) Quality of State Institutions 10–12 12–17
(3) Transfers 5–7 7–9
(4) Specific Labour Market Regulations 7–9 7–9
(5) Public Sector Services 5–7 7–9
(6) Tax Morale 22–25 –
Influence of all Factors 84–98 78–96
Notes:
(a) Average values of 12 studies.
(b) Average values of empirical results of 22 studies.
Source: Schneider (2005)
2.2 shows that the independent variables tax and social security burden,
followed by variables tax morale and intensity of state regulations are the
three major driving forces of the shadow economy.
The few studies based on survey data do not allow for a similar table
due to the low number of estimates, but also due to some still open ques-
tions on sampling, randomization and questionnaire design. For example,
the problem as to how an accessible formulation of questions regarding
marginal tax burdens could look like is still not resolved. Aside from
socio-demographic factors, it is safe to conclude, however, that social
norms and the individually perceived probability of detection are the two
most important influential factors to explain participation in the shadow
economy. Comparing these two, the evidence leans towards a relatively
stronger influence of social norms.
2.3 SURVEYS ON THE SHADOW ECONOMY
2.3.1 Germany
The discussion in the previous section highlights that the perception of

taxpayers on the shadow economy and their (moral) reaction to this
phenomenon is an important factor: under what circumstances do people


Table 2.3 Work in the shadow economy: survey results for 2007
(1) Do you work regularly in the shadow economy? Values in per cent
No 77.3
Yes 20.7
(25% male, 16% female)
No answer 2
(2) Do you regularly demand shadow economy Values in per cent
activities?
No 69.2
Yes 30.8
(35.4% male, 26.5% female)
Notes: Representative questionnaire, Germany, January 2007.
Source: IW Köln, Germany
decide to work in the shadow economy? There are a number of empiri-

cal studies which investigate the tax morale of people and their attitudes
towards the shadow economy.21 In this section, some results for Germany
are shown which demonstrate that people do not necessarily have a
bad (moral) sentiment (of guilt or shame) when working in the shadow
economy. The figures in Table 2.3 indicate for the year 2007 to what extent
people regularly work in the shadow economy. A total of 20.7 per cent of
German respondents admit to working in the shadows, and 30.8 per cent
of respondents regularly demand shadow economy activities.
In Table 2.4 the reasons are shown as to why shadow economy activi-
ties are demanded. The most important result is that it is possible to save
money – that is shadow economy activities are much cheaper than the offi-
cial ones. The second most important reason is that tax and social security
burdens are too high (73 per cent of the respondents) and the third reason
is that due to the much higher labour costs in the official economy these
regular services would not be demanded. In particular, the third answer
is interesting, because it indicates that only 22 per cent of the demand of
the shadow economy have substitutive character (that is they would be
demanded in the official economy if there were no shadow economy) and
30 per cent of the respondents respond that they would do it themselves.
From this survey result the conclusion emerges that about 48 per cent of
these activities would not take place if there were no shadow economy.
In Table 2.5, examples of hourly wage rates of shadow economy activities
in Germany are shown. What is surprising here is the huge range of wage
rates in the shadow economy, for example the varying ‘price’ for an hour


Table 2.4 Reasons for shadow economy activities: survey results for
Germany, January 2007
Reasons why shadow economy activities are demanded Values in

per cent
(1) One saves money – or they are much cheaper than the 90
official ones
(2) The tax and social security burden is much too high 73
(3) Due to the high labour costs in the official economy one would 68
not demand these activities (extreme assumption: no shadow
economy – 22% demand in the official economy; 30% do-it-
themselves; and 48% no demand at all!)
(4) The firms offer them themselves 52
(5) It’s so easy to get quick and reliable workers 31
Notes: Representative questionnaire, Germany, January 2007.
Source: IW Köln
Table 2.5 Hourly wage rates of shadow economy activities: survey results
for Germany, 2004
Activity/Type of Worker Town/ Wage rate in the Wage rate in the

Area shadow economy official economy
(in €) (in €)
Painters Berlin 10–17
München 9–15 42
Rhein/Ruhr 10–12
Mechanics Hamburg 13–23
Berlin 15–19 58
München 15–23
Cost of moving household Berlin 300–380
furniture and other München 400–450 1.800
goods (distance 300 km) Rhein/Ruhr 350–420
Notes: Representative questionnaire, May 2003.
of shadow market work by a painter ranges from €9 to €17. Table 2.5 dem-

onstrates also the large difference (a multiplicative factor between 4 and 5)
between the wage rates in the shadow economy and in the official one.
The survey results by Feld and Larsen (2005, 2008) reveal smaller


Table 2.6 Participation in black activities in Germany, 2001–2006
Carried out black activities within the last 12 months

Participation rates Hours: minutes per week
2001 10.4 8 : 14
2004 8.8 7 : 30
2005 11.1 6 : 40
2006 7.2 7 : 16
Notes: 18–74-year olds.
Source: Feld and Larsen (2008)
participation rates in the unofficial economy for the years 2001, 2004, 2005
and 2007 than those reported by the IW Köln for 2007. This is probably
the result from the more conservative approach in the face-to-face inter-
views as described above. These surveys ask who has carried out black
activities during the last 12 months. The results are reported in Table 2.6. In
2001, 10.4 per cent of the respondents answered that yes, they had carried
out black activities during the last 12 months. This share decreased to 8.8
per cent in 2004 only to increase to 11.1 per cent the year after and then
decrease again to 7.2 per cent in 2006. It thus looks as if the participation
rate varies around an average which would be almost exactly one in ten
which is half the figure of the IW Köln.
Regarding the socio-demographic distribution of the participation rates
interesting differences can be observed. As Table 2.7 shows, working in
the shadow economy is more widespread among men than among women,
among the young than among the old, among skilled workers and self-
employed than among other occupations. The variation in the figures
across time also reflects the sampling problems mentioned before.
Table 2.8 contains the average black hourly wages for the period 2001
to 2006. These average wages are in line with those reported in Table 2.5.
But they give a more comprehensive assessment as the amounts are calcu-
lated for overall Germany and all occupations. Given the small number of
observations in the quid pro quo case the higher volatility does not play
an important role for the overall wages. Interestingly, the black hourly
wages remain relatively constant across time and they underline the strong
differences to hourly wages in the regular economy as indicated for some
examples in Table 2.5.
Finally, in Table 2.9, a comparison between the size of the German
shadow economy, using the survey and the MIMIC method, is undertaken


Table 2.7 Proportion that have carried out black activities within the last
12 months by gender, age and occupation
Carried out black activities (%)

2001 2004 2005 2006
Men 14.5 13.4 13.9 9.0
Women 6.5 4.5 8.5 5.3
18–19-year-olds 16.6 24.3 13.9 8.8
20–29-year-olds 19.1 13.4 21.0 11.1
30–39-year-olds 13.2 12.2 13.3 11.2
40–49-year-olds 10.0 10.3 9.4 4.8
50–59-year-olds 7.4 5.1 8.5 7.6
60–69-year-olds 5.6 2.6 8.2 4.3
70–74-year-olds 1.0 3.0 2.9 1.3
Self-employed/ assisting spouse 12.1 1.7 9.9 16.2
Salaried employees 7.1 8.9 10.2 5.0
Skilled workers 19.2 16.6 13.8 13.7
Unskilled workers 8.2 8.9 14.0 6.9
Unemployed 20.7 17.1 19.1 7.0
Pensioners 4.2 3.6 6.3 3.2
Students 27.3 14.6 15.3 8.2
Other 8.7 6.1 10.8 6.3
Total 10.4 8.8 11.1 7.2
No. of persons 5,686 2,143 2,144 1,083
Note: 18–74-year-olds.
with the data by IW Köln. Also an attempt is made to explain the often
observed, large differences between the survey method and the MIMIC
and/or currency demand approach. According to the latter the size of the
German shadow economy in 2006 is 15 per cent of ‘official’ GDP. Using
the survey method, values between 5 and 6 per cent are obtained. Hence,
there is quite a large difference. Using the figures by Feld and Larsen
(2005, 2008), the implied GDP of undeclared work is even smaller at
between 3 and 4 per cent (see Table 2.12).
The first difference between both approaches that explains those dif-
ferences originates from the survey method: usually not the total overall
value added is recorded, but only the value added of undeclared work. If
material is added, another 3–4 per cent comes up. Moreover, other illegal
activities (prostitution and totally illegally working firms in the construc-
tion sector) must be considered such that another 4–5 per cent of the size


Table 2.8 Average black hourly wages by form of payment
2001 2004 2005 2006

Euroa
Cash 9.5 10.2 9.0 9.2
Quid pro quob 11.0 10.4 16.9 13.4
Cash and quid pro quob 10.4 10.5 10.5 10.5
Total 10.3 10.4 13.3 11.8
Notes: 18–74-year-olds who have carried out black activities within the last 12 months.
a
2001: DM converted to Euro by using the average synthetic exchange rate.
b
Quid pro quo: hypothetical wages.
Table 2.9 A comparison of the size of the German shadow economy using
the survey and the MIMIC method, year 2006
Various kinds of shadow Shadow Shadow % share of the

economy activities/values Economy in % Economy overall shadow
of official GDP in bill. Euro economy
(1) Survey method: Shadow 5.0–6.0 117–140 33–40
economy activities (based
on ‘black’ hours worked)
(2) Material (used) 3.0–4.0 70–90 20–25
(3) Illegal activities (goods and 4.0–5.0 90–117 25–33
services)
(4) Already in the official GDP 1.0–2.0 23–45 7–13
included illegal activities
Sum (1) to (4) 13.0–17.0 300–392 85–111
Overall (total) shadow economy 15.0 340 100
(estimated by the MIMIC and
calibrated by the currency demand
procedure)
Source: Enste and Schneider (2006) and own calculation
of these activities measured in per cent of official GDP is gained. Finally,

the statistical offices when calculating the official national accounts (also in
Germany) add (or include) some shadow economy activities in the ‘official’
GDP. Thus another 1–2 per cent of black activities from official GDP are
obtained which sums up to about 15 per cent. If these different kinds of


shadow activities in per cent of overall shadow economy activities are calcu-
lated, undeclared work has the biggest share of between 33 and 40 per cent,
followed by illegal activities in the shadow economy with between 25 and
35 per cent. The MIMIC cum currency demand approach and the survey
approach can thus be reconciled with each other and do not contradict each
other.
2.3.2 Austria
In Austria, in November 2002, Schneider undertook a representative

survey of the population to achieve two goals. First, to get some informa-
tion about the reaction of the Austrian public on the shadow economy;
and second, to estimate the size of the shadow economy in the construc-
tion and craftsman sector (including repairing) considering three groups:
1. A representative sample of the Austrian population between 16 and 65

years old,
2. 55 self-declared shadow economy workers in the construction and
craftsmen sector, and
3. 320 managers (owners) of construction and craftsmen firms.
The following results were obtained:
1. Among the Austrian population (potential labour force) are 918 000
Austrians who supplied shadow economy activities in the construction
and craftsmen sector. Their average hourly earnings in the shadow
economy varied between €15.30 and €15.60, and the average yearly
income from shadow economy activities varied between €1117.00
and €1142.00. This means that 73 hours per year were worked in the
shadow economy.
2. Among the 55 self-declared shadow economy workers a wage rate
of €11.50 per hour and annual earnings in the shadow economy of
€2480.00 were reported using the fact that these groups worked 245
hours per year in the shadow economy.
3. Managers (owners) of construction and craftsmanship firms report a
wage rate for shadow economy workers of €17 per hour and average
earnings per year of €4590.00, assuming that 270 hours per year were
used for shadow economy activities by their employees/workers. A
total of 21 per cent of the managers questioned also stated that more
than 50 per cent of their employees work in the shadow economy,
41 per cent indicated a figure of less than 50 per cent that was in the
shadow economy, and 34 per cent reported that no-one in the firm


works in the shadow economy. To summarize, 62 per cent of the

managers acknowledge that a large percentage of their employees
work in the shadow economy. Further results are that 7 per cent of
the managers think that their employees work between 0 and 2 hours
per week in the shadow economy; 29 per cent assume that they work
between 6 and 10 hours, 28 per cent between 3 and 5 hours and 14 per
cent think that their employees work more then 10 hours per week in
the shadow economy; 22 per cent of all managers have no knowledge
of this fact. In principle, 39 per cent of managers are not in favour of
(do not support) moonlighting by their workers and 61 per cent are in
favour (do support) – an amazingly high percentage!
Finally in Table 2.10 the aggregate values of the size of the shadow
economy in the construction and craftsmen sector in the year 2002 are
presented, based on questionnaire findings. Table 2.10 clearly demon-
strates that the size of the shadow economy in the construction and
craftsmen sector varies considerably from a total value of €2.6 billion
up to €4.2 billion. These differences originate from different hourly wage
rates, ranging from €11.50 to €17 and from the different amount of hours
worked per year in the shadow economy ranging from 245 to 270. Hence
the survey method ‘covers’ between 31.2 per cent and 50.9 per cent of the
value obtained by a macro (MIMIC) approach. These results still leave a
considerable leeway, but the rather large differences may be explained by
the following facts:
1. Table 2.10 contains earnings and not the value added of the shadow
economy.
2. Shadow economy demanders are overwhelmingly households, the
whole area of the shadow economy activities between firms (which are
especially a problem in the construction and craftsmen sectors) are
not considered.
3. All foreign shadow economy activities achieved by foreigners (illegal
immigrants) are not considered.
4. The amount earned in the shadow economy (hourly wage rates and
hours worked per year) varies considerably.
2.3.3 Summary
The results for Germany and Austria, briefly discussed according to recent
surveys, show that the readiness to undertake illicit employment as well as
its acceptance is high in both countries. More than one half of the popula-
tion would demand goods or services produced in the shadow economy if


Table 2.10 Size of the supplied shadow economy in the construction and
craftsmen sector, Austria 2002, based on the questionnaire
findings
Variable/Indicator Questioned people

results from results from results from results from
declared managers of declared managers of
moon- construction moon- construction
lightersa (1) and lightersa (3) and crafts-
craftsmen men firmsa (4)
firmsa (2)
∅ amount of hours 245 245 270 270
worked in the
shadow economy per
year per workera
∅ hourly shadow €11.5 €17 €11.5 €17
economy wage rate
∅ average yearly €2,814 €4,165 €3,105 €4,590
earning
∅ aggregated yearly 225.1 225.1 248.1 248.1
amount of hours million million million million
worked in the
shadow economya
Total earnings of the €2,588.65 €3,826.7 €2,853.15 €4,217.7
shadow economy in million million million million
the year 2002
Total shadow economy 31.2 46.1 34.4 50.9
earnings in % of the
value added of the
shadow economy in
the construction and
craftsmanship sector
(including repairing);
absolute value €8,284
billion in 2002
Notes:
a
As the amount of hours worked varied considerably between the lower bound (245 hours
per year) and the upper bound (270 hours per year), both values have been used for both
groups.
b
Basis of the calculation: 918,864 shadow economy workers in the construction and
craftsmen sector.
Source: Own calculations


given such an opportunity. In other words, if asked whether he/she ‘needs

a receipt/bill?’, every second person would answer ‘no’, saving at least the
value added tax. Around one third of the population is illicitly employed
and, as a result, avoids paying high taxes and other contributions and
escapes the rigidity of regulations.22
2.4 ESTIMATION AND SIZE OF THE SHADOW

ECONOMIES IN OECD COUNTRIES
2.4.1 Econometric Estimation
Following the theoretical considerations in Section 2.2, we develop seven

hypotheses below, all ceteris paribus, which will be empirically tested sub-
sequently using the MIMIC approach:
1. An increase in direct and indirect taxation increases the shadow

economy;
2. An increase in social security contributions increases the shadow
economy;
3. The more the country is regulated, the greater the incentives to work
in the shadow economy;
4. The lower the quality of state institutions, the higher the incentives to
work in the shadow economy;
5. The lower the tax morale, the higher the incentives to work in the
shadow economy;
6. The higher unemployment, the more people engage in shadow
economy activities;
7. The lower GDP per capita in a country, the higher is the incentive to
work in the shadow economy.
As the sample consists of 21 highly developed OECD countries between

1990 and 2005 (pooled cross section time series data), the effect of deter-
rence cannot be empirically tested. The size of fines and punishment and
the probability of detection are only available for one or two countries
across time. The following estimation results thus rather correspond to
the factors reported in Table 2.2 which are gained from an overview of
existing studies.
In Table 2.11, the econometric results using the MIMIC approach
(latent estimation approach) are presented for these 21 OECD-countries
for which we have eight data points of the years 1990/91, 1994/95, 1997/98,
1999/2000, 2001/02, 2002/03, 2003/04 and 2004/05. Aside from the usual


Table 2.11 MIMIC estimation of the shadow economy of 21 highly

developed OECD countries, 1990/91, 1994/95, 1997/98,
1999/2000, 2001/02, 2002/03, 2003/04 and 2004/05
Cause Variables Estimated Coefficients

Share of direct taxation (in % of GDP) l1 5 0.384**
(3.06)
Share of indirect taxation (in % of GDP) l2 5 0.196(*)
(1.84)
Share of social security contributions l3 5 0.506**
(in % of GDP) (3.86)
Burden of state regulation (index of l4 5 0.213(*)
labor market regulation, Heritage (1.96)
Foundation, score 1 least reguled,
score 5 most reguled)
Quality of state institutions l5 5 –0.307*
(rule of law, World Bank, (–2.61)
score –3 worst and 13 best case)
Tax morale (WVS and EVS, Index, l6 5 –0.582**
Scale tax cheating always justified 51, (–3.66)
never justified 510)
Unemployment rate (%) l7 5 0.324**
(2.61)
GDP per capita (in US-$) l8 5 –0.106**
(–3.04)
Indicator Variables Estimated Coefficients
Employment rate l9 5 –0.626**
(in % of population 18–64) (–2.72)
Average working time (per week) l10 5 –1.00 (Residuum)
Annual growth rate of GDP l11 5 –0.274**
(adjusted for the mean of all 22 OECD (–3.03)
countries)
Change of local currency per capita l12 5 0.312**
(3.74)
Test-statistics RMSEa 5 0.0016* (p-value 5 0.903)
Chi-squareb 5 26.43 (p-value 5
0.906)
TMCVc 5 0.049
AGFId 5 0.763
N 5 168
D.F.e 5 67


Notes: t-statistics are in parentheses. (*), *, ** denote significance at the 10 per cent, 5 per
cent, or 1 per cent significance level.
a
Steiger’s Root Mean Square Error of Approximation (RMSEA) for test of close fit;
RMSEA < 0.05; the RMSEA-value varies between 0.0 and 1.0.
b
If the structural equation model is asymptotically correct, then the matrix S (sample
covariance matrix) will be equal to S (q) (model implied covariance matrix). This test
has a statistical validity with a large sample (N ≥ 100) and multinomial distributions;
both are given for all three equations using a test of multinomial distributions.
c
Test of Multivariate Normality for Continuous Variables (TMNCV); p-values of
skewness and kurtosis.
d
Test of Adjusted Goodness of Fit Index (AGFI), varying between 0 and 1; 1 5 perfect
fit.
e
The degrees of freedom are determined by 0.5 (p 1 q) (p 1 q 1 1) – t; with p 5 number
of indicators; q 5 number of causes; t 5 the number for free parameters.
cause variables like direct and indirect taxation, social security contribu-
tions and state regulation, we have added two further causal factors, tax
morale and the quality of state institutions. In addition to the employ-
ment rate, the annual growth rate of GDP and the change of currency
per capita, we use the average working time (per week) as an additional
indicator variable.23 The estimated coefficients of all eight cause vari-
ables are statistically significant and have the theoretically expected signs.
The tax and social security burden variables are quantitatively the most
important ones, followed by the tax morale variable, which has the single
biggest influence. Also, the independent variable quality of state institu-
tions is statistically significant and quite important to determine whether
one is engaged in shadow economy activities or not. The development of
the official economy measured by unemployment and GDP per capita has
a quantitatively important influence on the shadow economy. Turning
to the indicator variables they all have a statistically significant influence
and the estimated coefficients have the theoretically expected signs. The
quantitatively most important independent variables are the employment
rate and the change of currency per capita.24 Summarizing, the economet-
ric results demonstrate that in these OECD countries the social security
contributions and the share of direct taxation have the biggest influence,
followed by tax morale and the quality of state institutions.
2.4.2 The Development and Size of the Shadow Economy in German-

Speaking and Other OECD Countries
Existing estimates of the German shadow economy (measured in per-

centage of official GDP) are shown in Table 2.12 (see also Feld, Schmidt


Table 2.12 The size of the shadow economy in Germany according to different methods (in percentage of official GDP)
Method Shadow economy in Germany (in percentage of official GDP) in: Source
1970 1975 1980 1985 1990 1995 2000 2005
M2678 - SCHNEIDER TEXT.indd 103

a
Survey – 3.6 – – – – – – IfD Allensbach (1975)
– – – – – – 4.1b 3.6b Feld and Larsen (2005,
2008)
Discrepancy between 11.0 10.2 13.4 – – – – – Lippert and Walker
expenditure and income (1997)
Discrepancy between 23.0 38.5 34.0 – – – – – Langfeldt (1984a, b)
official and actual
employment
Physical input method – – – 14.5 14.6 – – – Feld and Larsen (2005)
103
Transactions approach 17.2 22.3 29.3 31.4 – – – –
Currency demand approach 3.1 6.0 10.3 – – – – – Kirchgässner (1983)
12.1 11.8 12.6 – – – – – Langfeldt (1984a, b)
4.5 7.8 9.2 11.3 11.8 12.5 14.7 – Schneider and Enste
(2000)
Latent ((DY)MIMIC) 5.8 6.1 8.2 – – – – – Frey and Weck (1984)
approach – – 9.4 10.1 11.4 15.1 16.3 – Pickhardt and Sarda Pons
(2006)
4.2 5.8 10.8 11.2 12.2 13.9 16.0 15.4 Schneider (2005, 2006)
Soft modelling – 8.3 – – – – – – Weck (1983)
Notes:
a
1974.

b
2001 and 2005; calculated using wages in the official economy.
27/07/2011 16:16
and Schneider, 2007). The oldest estimate uses the survey method of the
Institute for Demoscopy (IfD) in Allensbach, Germany, and shows that
the shadow economy was 3.6 per cent of official GDP in 1974. In a much
later study, Feld and Larsen (2005, 2008) undertook an extensive research
project using the survey method to estimate shadow economic activi-
ties in the years 2001 to 2006.25 Using the officially paid wage rate, they
concluded that these activities reached 4.1 per cent in 2001, 3.1 per cent
in 2004, 3.6 per cent in 2005 and 2.5 per cent in 2006.26 Using the (much
lower) shadow economy wage rate these estimates shrink however to 1.3
per cent in 2001 and 1.0 per cent in 2004, respectively. If we look at the
discrepancy method, for which we have estimates from 1970 to 1980, the
German shadow economy is much larger: using the discrepancy between
expenditure and income, we get approximately 11 per cent for the 1970s,
and using the discrepancy between official and actual employment,
roughly 30 per cent. The physical input methods from which estimates for
the 1980s are available deliver values of around 15 per cent for the second
half of that decade. The (monetary) transaction approach developed by
Feige (1989) places the shadow economy at 30 per cent between 1980 and
1985. Yet another monetary approach, the currency demand approach –
the first person to undertake an estimation for Germany was Kirchgässner
(1983, 1984) – provides values of 3.1 per cent (1970) and 10.3 per cent
(1980). Kirchgässner’s values are quite similar to the ones obtained by
Schneider and Enste (2000, 2002), who also used a currency demand
approach to value the size of the shadow economy at 4.5 per cent in 1970
and 14.7 per cent in 2000. Finally, if we look at latent MIMIC estimation
procedures, the first ones being conducted by Frey and Weck-Hannemann
(1984), and later, Schneider and others followed for Germany, again, the
estimations for the 1970s are quite similar. Furthermore, Schneider’s esti-
mates using a MIMIC approach (Schneider 2005, 2009) are close to those
of the currency demand approach.
Thus, we can see that different estimation procedures produce different
results. It is safe to say that the figures produced by the transaction and
the discrepancy approaches are rather unrealistically large: the size of the
shadow economy at almost one third of official GDP in the mid-1980s
is most likely an overestimate. The figures obtained using the currency
demand and hidden variable (latent) approaches, on the other hand, are
relatively close together and much lower than those produced by other
methods (the discrepancy or transaction approaches). This similarity is
not surprising given the fact that the estimates of the shadow economy
using the latent (MIMIC) approach were measured by taking point
estimates from the currency demand approach. The estimates from the
MIMIC approach can be regarded as the upper bound of the size of the


shadow economy. For the reasons outlined in Section 2.3, the estimates
obtained from the survey approach provide for its lower bound. It should
be noted that the ‘true’ size of the shadow economy does not necessarily
lie between both bounds, nor is it precluded that it is closer to the upper
than the lower bound. But both benchmarks help us to understand the
phenomenon pretty well.
2.4.3 Size and Development of the Shadow Economy in 21 OECD

Countries
In order to calculate the size and development of the shadow economies

of the 21 OECD countries, we have to overcome the disadvantage of the
MIMIC approach, which is, that only relative sizes of the shadow economy
are obtained such that another approach to calculate absolute figures must
be used. For the calculation of the absolute sizes of the shadow economies
from these MIMIC estimation results, we take the already available esti-
mates from the currency demand approach for Austria, Germany, Italy
and the United States (from studies of Dell’Anno and Schneider, 2003;
Bajada and Schneider, 2005; and Schneider and Enste, 2002). As we have
values of the shadow economy (in percentage of GDP) for various years
for the above mentioned countries, we can use them in a benchmark pro-
cedure to transform the index of the shadow economy from the MIMIC
estimations into cardinal values.27
Table 2.13 presents the findings for 21 OECD countries until 2007.
They clearly reveal that since the end of 1990s the size of the shadow
economy in most OECD countries continued to decrease. The unweighted
average for all countries in 1999/2000 was 16.8 per cent and dropped to
13.9 per cent in 2007. This means that since 1997/98 – the year in which
the shadow economy was the biggest in most OECD countries – it has
continuously shrunk. Only in Germany, Austria and Switzerland did
the growing trend last longer and was reversed two or three years ago.
The reduction of the share of the shadow economy from GDP between
1997/98 and 2007 is most pronounced in Italy (−5.0 per cent) and in
Sweden (−4.0 per cent). The German shadow economy ranges in the
middle of the ranking, whereas Austria and Switzerland are located at
the lower end. With 20 per cent to 26 per cent, South European countries
exhibit the biggest shadow economies measured as a share from official
GDP. They are followed by Scandinavian countries whose shadow econo-
mies’ shares in GDP range between 15 and 16 per cent. One reason for
the differences in the size of the shadow economy between these OECD
countries includes, among others, that for example there are fewer regu-
lations in the OECD country the United States compared to the OECD


Table 2.13 The size of the shadow economy in 21 OECD countries between 1989/90 and 2007 estimated using the
money demand and MIMIC methods (in % of official GDP)
OECD-countries Shadow Economy
106
Average Average Average Average Average 2003 2004 2005 2006 2007
1989/90 1994/95 1997/98 1999/00 2001/02
1. Australia 10.1 13.5 14.0 14.3 14.1 13.7 13.2 12.6 11.4 10.7
2. Belgium 19.3 21.5 22.5 22.2 22.0 21.4 20.7 20.1 19.2 18.3
3. Canada 12.8 14.8 16.2 16.0 15.8 15.3 15.1 14.3 13.2 12.6
4. Denmark 10.8 17.8 18.3 18.0 17.9 17.4 17.1 16.5 15.4 14.8
5. Germany 11.8 13.5 14.9 16.0 16.3 17.1 16.1 15.4 14.9 14.6
6. Finland 13.4 18.2 18.9 18.1 18.0 17.6 17.2 16.6 15.3 14.5
7. France 9.0 14.5 14.9 15.2 15.0 14.7 14.3 13.8 12.4 11.8
8. Greece 22.6 28.6 29.0 28.7 28.5 28.2 28.1 27.6 26.2 25.1
9. Great Britain 9.6 12.5 13.0 12.7 12.5 12.2 12.3 12.0 11.1 10.6
10. Ireland 11.0 15.4 16.2 15.9 15.7 15.4 15.2 14.8 13.4 12.7

11. Italy 22.8 26.0 27.3 27.1 27.0 26.1 25.2 24.4 23.2 22.3
12. Japan 8.8 10.6 11.1 11.2 11.1 11.0 10.7 10.3 9.4 9.0
27/07/2011 16:16
14. New Zealand 9.2 11.3 11.9 12.8 12.6 12.3 12.2 11.7 10.4 9.8
15. Norway 14.8 18.2 19.6 19.1 19.0 18.6 18.2 17.6 16.1 15.4
16. Austria 6.9 8.6 9.0 9.8 10.6 10.8 11.0 10.3 9.7 9.4
17. Portugal 15.9 22.1 23.1 22.7 22.5 22.2 21.7 21.2 20.1 19.2

18. Sweden 15.8 19.5 19.9 19.2 19.1 18.6 18.1 17.5 16.2 15.6
19. Switzerland 6.7 7.8 8.1 8.6 9.4 9.5 9.4 9.0 8.5 8.2
20. Spain 16.1 22.4 23.1 22.7 22.5 22.2 21.9 21.3 20.2 19.3
21. USA 6.7 8.8 8.9 8.7 8.7 8.5 8.4 8.2 7.5 7.2
Unweighted average for 12.7 16.2 16.8 16.8 16.7 16.5 16.1 15.6 14.5 13.9
21 OECD countries
Source: Own calculations
107

27/07/2011 16:16
country Germany where the individual’s freedom is limited in many areas

by far-reaching state interventions. Another reason is the large differences
in the direct and indirect tax burden with the lowest in the United States
and Switzerland in this sample.
2.5 INTERACTION OF SHADOW AND OFFICIAL

ECONOMY
2.5.1 Shadow Economy Labour Market and Productivity
Having examined the size, rise and fall of the shadow economy in
terms of value added over time, the analysis now focuses on the ‘shadow
labour market’, as within the official labour market there is a par-
ticularly tight relationship and ‘social network’ between people who are
active in the shadow economy.28 Moreover, by definition every activity
in the shadow economy involves a ‘shadow labour market’ to some
extent:29 Hence, the ‘shadow labour market’ includes all cases, where
the employees or the employers, or both, occupy a ‘shadow economy
position’.
Why do people work in the shadow economy? In the official labour
market, the costs firms (and individuals) have to pay when ‘officially’
hiring someone are increased tremendously by the burden of tax and social
contributions on wages, as well as by the legal administrative regulation
to control economic activity.30 In various OECD countries, these costs are
greater than the wage effectively earned by the worker – providing a strong
incentive to work in the shadow economy.
More detailed theoretical information on the labour supply decision
in the underground economy is given by Lemieux, Fortin and Fréchette
(1994) who use micro data from a survey conducted in Quebec City
(Canada). In particular, their study provides some economic insights
regarding the size of the distortion caused by income taxation and the
welfare system. The results of this study suggest that hours worked in the
shadow economy are quite responsive to changes in the net wage in the
regular (official) sector. Their empirical results attribute this to a (mis-)
allocation of work from the official to the informal sector, where it is not
taxed. In this case, the substitution between labour market activities in
the two sectors is quite high. These empirical findings indicate that ‘par-
ticipation rates and hours worked in the underground sector also tend to
be inversely related to the number of hours worked in the regular sector’
(Lemieux, Fortin and Fréchette, 1994, p. 235). These findings demonstrate
a large negative elasticity of hours worked in the shadow economy with


respect both to the wage rate in the regular sector as well as to a high
mobility between the sectors.
Illicit work can take many shapes. The underground use of labour may
consist of a second job after (or even during) regular working hours. A
second form is shadow economy work by individuals who do not par-
ticipate in the official labour market. A third component is the employ-
ment of people (for example, clandestine or illegal immigrants), who are
not allowed to work in the official economy. Empirical research on the
shadow economy labour market is even more difficult than that on the
shadow economy on the value added, since one has very little knowledge
about how many hours an average ‘shadow economy worker’ is actu-
ally working (from full time to a few hours, only); hence, it is not easy to
provide empirical facts.31
Kucera and Roncolato (2008, p. 321) also deal with informal employ-
ment. They address two issues of crucial importance to labor market
policy:
1. The intensive labour market regulations as one (major) cause of infor-

mal employment, and
2. the so-called ‘voluntary’ informal employment.
Kucera and Roncolato give a theoretical overview on both issues and

also a survey of a number of empirical studies, in which mainly the effect
of official labour market regulations on informal employment is analyzed,
where they find a significant and quantitatively important influence.
The latest OECD study (2009) concludes that informal employment is
the norm, not the exception, in many parts of the world. More than half
of all jobs in the non-agricultural sectors of developing countries – over
900 million workers – can be considered informal. If agricultural workers
in developing countries are included, the estimates size to roughly 2000
million people. In some regions, including Sub-Saharan Africa and South
Asia, over 80 per cent of non-agricultural jobs are informal. Most infor-
mal workers in the developing world are self-employed and work inde-
pendently, or own and manage very small enterprises. According to the
OECD study (2009), informal employment is a result of both, people being
excluded from official jobs and people voluntarily opting out of formal
structures, for example, in many middle income countries incentives drive
individuals and businesses out of the formal sector.
In Table 2.14, the estimates for the shadow economy labour force in
highly developed OECD countries (Austria, Denmark, France, Germany,
Italy, Spain and Sweden) are shown.32 In Austria the shadow economy
labour force arrived at 500 000 to 750 000 or 16 per cent of the official


Table 2.14 Estimates of the size of the ‘shadow economy labour force’ and of the official and shadow economy
productivity in some OECD countries 1974–1998
Countries Year Official Total Economy Size of the Shadow Shadow Shadow Sources of
GDP per (Shadow Economy (in % Economy Economy Shadow
capita in Economy plus of official GDP) Labour Force Participants Economy
110
US$a official GDP Currency in 1000 in % of official Labour
per capita in Demand peoplec Labour Force
US$) Approachb Forced
Austria 90–91 20,636 25,382 5.47 300–380 9.6 Schneider (1998a, b)
97–98 25,874 29,630 8.93 500–750 16.0 and own calculations
Denmark 1980 13,233 18,658 8.6 250 8.3 Mogensen, et. al.
1986 18,496 26,356 9.8 390 13.0 (1995)
1991 25,946 36,558 11.2 410 14.3 and own calculations
1994 34,441 48,562 17.6 420 15.4
France 1975–82 12,539 17,542 6.9 800–1,500 3.0–6.0 De Grazia (1983) and
1997–98 24,363 34,379 14.9 1,400–3,200 6.0–12.0 own calculations
Germany 1974–82 11,940 17,911 10.6 3,000–4,000 8.0–12.0 De Grazia (1983), F.

1997–98 26,080 39,634 14.7 7,000–9,000 19.0–23.0 Schneider (1998a, b)
and own calculations
27/07/2011 16:16
Italy 1979 8,040 11,736 16.7 4,000–7,000 20.0–35.0 Gaetani-d’Aragona
1997–98 20,361 29,425 27.3 6,600–11,400 30.0–48.0 (1979) and own
calculations
Spain 1979–80 5,640 7,868 19.0 1,250–3,500 9.6–26.5 Ruesga (1984) and

1997–98 13,791 19,927 23.1 1,500–4,200 11.5–32.3 own calculations
Sweden 1978 15,107 21,981 13.0 750 13.0–14.0 De Grazia (1983) and
1997–98 25,685 37,331 19.8 1,150 19.8 own calculations
European 1978 9,930 14,458 14.5 15,000 – De Grazia (1983) and
Union 1997–98 22,179 32,226 19.6 30,000 own calculations
OECD 1978 9,576 14,162 15.0 26,000 – De Grazia (1983) and
(Europe) own calculations
1997–98 22,880 33,176 20.2 48,000
Notes:
111
c
Estimated full-time jobs, including unregistered workers, illegal immigrants, and second jobs.
d
In per cent of the population aged 20–69, survey method.
Source:
a
OECD, Paris, various years
b
Own calculations from Schneider (2000, 2001)

27/07/2011 16:16
labour force (mean value) in the years 1997–1998. In Denmark, the

development of the 1980s and 1990s shows that the part of the Danish
population engaged in the shadow economy ranged from 8.3 per cent of
the total labour force (in 1980) to 15.4 per cent in 1994 – quite a remark-
able increase of the shadow economy labour force; it almost doubled over
15 years. In France (in the years 1997–1998) the shadow economy labour
force reached a size of between 6 and 12 per cent of the official labour
force or between 1.6 and 3.2 million in absolute figures. In Germany, this
figure rose from 8 to 12 per cent in 1974 to 19 per cent and to 23 per cent
(8 million) in the year 1997–1998. For France and Germany this is again a
very strong increase in the shadow economy labour force. In other coun-
tries, the amount of the shadow economy labour force is quite large, too:
in Italy 30–48 per cent (1997–1998), Spain 11.5–32 per cent (1997–1998)
and Sweden 19.8 per cent (1997–1998). In the European Union about
30 million people are engaged in shadow economy activities in the years
1997–1998 and in all European OECD countries 48 million work illicitly.
These figures demonstrate that the shadow economy labor market is lively
and may provide an explanation, why, for example, in Germany, one can
observe such a high and persistent unemployment.
Additionally, Table 2.14 contains a preliminary calculation of the total
GDP per capita (including the official and the shadow economy GDP per
capita) in US$. In all countries investigated, total GDP per capita is much
higher – on average in all countries around 40 per cent. This clearly shows
that the productivity in the shadow economy is roughly as high as in the
official economy – a clear indication that the work effort (that is the incen-
tive to work effectively) is as strong in the shadow economy as in the offi-
cial one. In general, these results demonstrate that the shadow economy
labour force has reached a remarkable size in developing countries as well
as in highly developed OECD countries, even though the calculation still
might have many errors.
2.5.2 Shadow Economy and Aggregate Efficiency
Most studies of economic development rest on official output figures.

However, in so doing they neglect a sizeable part of economic activity,
which takes place in the informal sector, and therefore goes unrecorded
in official statistics. Nevertheless, as Tanzi (1999) remarks, though some
of those activities may be illegal, others are legal and socially valuable.
They should therefore be taken into account when measuring a country’s
output. Recently, Meon, Schneider and Weill (2010) analyzed the impact
of adding the shadow economy to official output figures on estimated
production functions and technical efficiency across up to 97 countries.


Including the shadow economy hardly affects the ranking of countries in

terms of efficiency. However, it results in an increase of observed efficiency
scores. Adding the shadow economy to official output figures thus allows
for a more precise estimate of countries’ outputs. Those results are impor-
tant in several aspects. First, they show that estimates of the production
function based on total output differ from those based on official output
figures. Second, they therefore imply that ignoring the shadow economy
leads to mistakes in measured efficiency. Finally, their results provide
guidance to the empirical literature on economic output and productivity
at large. Given that official output figures overlook a sizeable share of total
activity, future research on the determinants and effects of a country’s
production should clearly start with a reflection as to which definition of
output, official or total, is relevant to the question at hand. Their results
suggest that the answer to this question need not always be official output.
2.5.3 Shadow Economy and Unemployment
Although there has been considerable discussion on the size of the shadow
economy, comparatively little attention has been given to the relationship
between unemployment and working in the shadow economy. As Tanzi
(1999) points out, ‘the current literature does not cast much light on these
relationships even though the existence of large underground activities
would imply that one should look more deeply at what is happening in the
labour market’ (p. 347). The objective of a paper by Bajada and Schneider
(2009) is to examine the extent of participation in the shadow economy
by the unemployed. Their paper has investigated the relationship between
the unemployment rate and the shadow economy. Previous literature on
this topic has suggested that the relationship between these two variables
is ambiguous, predominantly because a heterogeneous group of people
working in the shadow economy exists and there are also various cyclical
forces at work, such that they produce a net effect that is weakly correlated
with unemployment. In this paper, they have provided a suggestion for
disentangling these cyclical effects, so as to study the component of the
shadow economy that is influenced directly by those who are unemployed.
They refer to this effect as the ‘substitution effect’ which typically increases
during declining periods of legitimate economic activity (and increasing
unemployment). Equipped with this approach for measuring the ‘substitu-
tion effect’, they discover that a relationship exists between changes in the
unemployment rate and shadow economy activity.
By examining the growth cycle characteristics of the ‘substitution
effect’ component of the shadow economy Bajada and Schneider (2009)
determine that the growth cycles are symmetric (in terms of steepness and


deepness) and that changes in the unemployment rate, whether positive or

negative, had similar impacts on changes in the substitution effect compo-
nent. They suggest that the shadow economy is a source of financial support
during periods of unemployment for those genuinely wanting to participate
in the legitimate economy. Although this does not exclude the possibil-
ity that long-term unemployed may also be participating in the shadow
economy, it would appear that short-term fluctuations in unemployment
directly contribute to short-term fluctuations in the shadow economy.
When Bajada and Schneider consider the various unemployment
support programmes across 12 OECD countries, there appears to be no
real systematic relationship between the generosity of the social security
systems and the nature of short-term shadow economic activity by the
unemployed. Even the various replacement rates across the OECD coun-
tries appear to have little consequence on the rate at which the unemployed
take on and cut back shadow economy activity. There is, however, some
evidence to suggest that extended duration spell in unemployment lasts
anywhere between less than three months to approximately nine months.
On the whole Bajada and Schneider argue that dealing with unemploy-
ment participation in the shadow economy as a way of correcting the
inequity it generates, is best handled by more stringent monitoring of those
receiving unemployment benefits rather than reducing replacement rates as
a way of encouraging re-integration into the work force. A strategy of reduc-
ing replacement rates would not only fail to maintain adequate support for
those experiencing financial hardship during periods of unemployment, it
is likely to have little impact on reducing participation by the unemployed
who are willing and able to engage in shadow economy activity.
2.5.4 Shadow Economy and Do-it-yourself Activities
Bühn, Karmann and Schneider (2009) use a MIMIC model to consistently

disentangle the size and development of the shadow economy and of do-
it-yourself (DIY) activities in Germany the first time. They report a statis-
tically highly significant impact of regulation as well as tax burdens and
social security contributions on the shadow economy. For DIY activities,
they observe a statistically highly significant positive influence of unem-
ployment. In general, the estimated models show satisfactory statistical
properties. According to their calculations the German shadow economy
increased from 2 per cent in 1970 to 17 per cent in 2005. DIY activities
amounted to 4 per cent of official GDP in 1970, slightly increased to 4.49
per cent in 1995, and remained relatively constant until 2005. Taking both
sectors together, they find that in Germany the hidden economy and DIY
activities reached a remarkable size of more than 20 per cent of official


GDP in 2005. While the shadow economy is primarily driven by political

and economic factors like taxation and regulation, DIY activities exhibit
a pattern of responding to slowly and steadily changing variables like
unemployment and social preferences. Their results suggest that shadow
economies are contingent on governmental behaviour while DIY activities
are driven by individual constraints, self-help and mutual aid. Because of
their significant amount and specific dynamics, a comprehensive analysis
of the hidden economy must consider DIY activities.
What type of policy conclusions did they draw from these results? If
the shadow economy and/or DIY activities should be reduced, fewer
regulations, lower taxes and social security contributions might be the two
most efficient means of shifting more activity into the official economy.
Reducing both the intensity of regulation and the amount of contributions
to the social security system in Germany might also result in a lower level
of unemployment. This would reduce individuals’ incentives to engage in
DIY activities. Though their results should be regarded as first steps in
measuring the size of the overall hidden economy, they have demonstrated
that both shadow economic and DIY activities are important and should
be taken into account when seeking to stimulate the official economy
through policy measures.
2.6 CONCLUSIONS
In this survey on the most recent developments in research on the shadow

economy and undeclared work, we start from the observation that in most
OECD countries the policy instrument of choice to prevent people from
working in the shadows has been deterrence. While deterrence policy is
well-founded from a theoretical point of view, the empirical evidence on
its success is weak. There is almost no study using the MIMIC approach
to estimate the size of the shadow economy that tests on the impact of
deterrence empirically. The only study for Germany does not find any
significant effect. In survey studies on undeclared work, the perceived
probability of being detected has a consistent robust significantly negative
effect, but perceived fines and punishment do not. Compared to the impact
of tax morale, deterrence is quantitatively less important. The studies
based on the MIMIC approach also report strong effects of tax morale,
but underline the higher importance of tax policies and state regulation to
increase the shadow economy.
Also, the pure inter-temporal development of the shadow economy, in
particular, its most recent decline cannot be interpreted as a success of
deterrence policies. The discussion of the recent literature underlines that


economic opportunities for employees, the overall situation of the labour

market, not least unemployment are crucial for an understanding of the
dynamics of the shadow economy. Individuals look for ways to improve
their economic situation and thus contribute productively to aggregate
income of a country. This holds regardless of their being active in the
official or the unofficial economy. A strong emphasis on deterrence may
thus backfire.
NOTES
* A shorter version of this chapter has been published as Feld and Schneider (2010).
1. Boockmann and Rincke (2005) outlined how such an evaluation could be done.
2. See Andreoni, Erard and Feinstein (1998) for the authoritative survey, Feld and Frey
(2007) or Kirchler (2007) for broader interdisciplinary approaches, or the papers
by Kirchler, Maciejovsky and Schneider (2003), Kirchler, Hoelzl and Wahl (2007),
Kastlunger et al. (2009).
3. The authoritative scientific work on tax morale is by Torgler (2007). See also Torgler
(2002) for a survey on experimental studies and Blackwell (2010) for a meta-analysis.
4. Our chapter focuses on the size and development of the shadow economy for uniform
countries and not for specific regions. Recently first studies have been undertaken to
measure the size of the shadow economy as well as the ‘grey’ or ‘shadow’ labour force for
urban regions or states (e.g. California). See Williams and Windebank (1998, 2001a, b),
Marcelli, Pastor and Joassart (1999), Chen (2004), Marcelli (2004), Flaming, Hayolamak
and Jossart (2005), Alderslade, Talmage and Freeman (2006), Brück, Haisten-DeNew
and Zimmermann (2006). Herwartz, Schneider and Tafenau (2009) estimate the size of
the shadow economy of 234 EU-NUTS regions for the year 2004 for the first time dem-
onstrating a considerable regional variation in the size of the shadow economy.
5. This definition is used, by Frey and Pommerehne (1984), Feige (1989, 1994) and
Schneider (1994c, 2003, 2005). Do-it-yourself activities are not included. For estimates
of the shadow economy and the do-it-yourself activities for Germany see Karmann
(1986, 1990) or Bühn, Karmann and Schneider (2009).
6. This definition is taken from Del’Anno (2003), Dell’Anno and Schneider (2003) and
Feige (1989); see also Thomas (1999), Fleming, Roman and Farrell (2000) or Feld and
Larsen (2005, p. 25).
7. See also the excellent discussion of the definition of the shadow economy in Pedersen
(2003, pp. 13–19) and Kazemier (2005a) who uses a similar one.
8. For a broader discussion of the definition issue see Thomas (1992), Schneider and Enste
(2002, 2006), Schneider, Volkert and Caspar (2002) and Kazemier (2005a, b).
9. With this definition the problem of having classical crime activities included could be
avoided, because neither the MIMIC procedure nor the currency demand approach
capture these activities: for example, drug dealing is independent of increasing taxes,
especially as the included causal variables are not linked (or causal) to classical crime
activities. See Thomas (1992), Kazemir (2005a, b) and Schneider (2005).
10. For the strengths and weaknesses of the various methods see Schneider (1986, 2001,
2003, 2005, 2006), Feige (1989), Thomas (1992, 1999), Bhattacharyya (1999), Dixon
(1999), Giles (1999a, b, c), Tanzi (1999), Schneider and Enste (2000a, b, 2002, 2006),
Breusch (2005a, b), Feld and Larsen (2005), Dell’Anno and Schneider (2009). See also
the discussion in the appendix.
11. These methods are presented in detail in Schneider (1994a, b, c, 2005) and Schneider
and Enste (2000b, 2002, 2006). Furthermore, these studies discuss advantages and
disadvantages of the MIMIC – and the money demand methods as well as other


estimation methods for assessing the size of illicit employment; for a detailed discussion
see also the appendix or Feld and Larsen (2005).
12. This indirect approach is based on the assumption that cash is used to make transac-
tions within the shadow economy. By using this method one econometrically estimates
a currency demand function including independent variables like tax burden, regulation
and so forth which ‘drive’ the shadow economy. This equation is used to make simula-
tions of the amount of money that would be necessary to generate the official GDP.
This amount is then compared with the actual money demand and the difference is
treated as an indicator for the development of the shadow economy. On this basis the
calculated difference is multiplied by the velocity of money and one gets a value added
figure for the shadow economy. See endnote 10 for references discussing this method
critically.
13. See Feld and Larsen (2005, p. 44).
14. An earlier study by Merz and Wolff (1993) does not analyze the impact of deterrence on
undeclared work.
15. The earlier study by Isachsen and Strøm (1985) for Norway does also not properly
analyze the impact of deterrence on undeclared work.
16. See Thomas (1992), Lippert and Walker (1997), Schneider (1994a, b, c, 1997, 1998a,
b, 1999, 2000, 2003, 2005, 2009), Johnson, Kaufmann, and Zoido-Lobatón (1998a, b),
Giles (1999a), Tanzi (1999), Mummert and Schneider (2001), Giles and Tedds (2002)
and Dell’Anno (2003) as more recent ones.
17. See Schneider (1994a, b, c) for the effects of a major tax reform in Austria on the
shadow economy. He shows that a major reduction in the direct tax burden did not
decrease the size of the shadow economy because legal tax avoidance was abolished and
other factors, like regulations, were not changed; hence for a considerable part of the
taxpayers the actual tax and regulation burden remained unchanged.
18. The importance of regulation on the official and unofficial (shadow) economy has
been more recently investigated by Loayza, Oviedo and Servén (2005a, b). Kucera and
Roncolato (2008) extensively analyze the impact of labour market regulation on the
shadow economy.
19. See Johnson, Kaufman and Zoido-Lobatón (1998a, b), Friedman et al. (2000), Dreher,
Kotsogiannis and Macorriston (2007, 2009), Dreher and Schneider (2010).
20. The importance of this variable with respect to theory and empirical relevance is also
shown in Frey (1997), Feld and Frey (2002a, 2002b, 2007) and Torgler and Schneider
(2009).
21. See Feld and Frey (2002a, b, 2007), Torgler (2002), Feld and Larsen (2005), Halla and
Schneider (2005) and Torgler and Schneider (2005, 2009).
22. See also Lamnek, Olbrich and Schäfer (2000).
23. Using this indicator variable the problem might arise that this variable is influenced by
state regulation, so that it is not exogenous; hence the estimation may be biased.
24. The variable currency per capita or annual change of currency per capita is heavily
influenced by banking innovations; hence this variable is pretty unstable with respect to
the length of the estimation period. Similar problems have already been mentioned by
Giles (1999a) and Giles and Tedds (2002).
25. In our chapter there is no extensive discussion about the various methods to estimate
the size and development of the shadow economy; also we do not discuss the strengths
and weaknesses of each method. See Giles (1999a, b, c), Schneider and Enste (2000),
Pedersen (2003), Feld and Larsen (2005, 2008, 2011) and Schneider (2005).
26. Due to the extraordinarily low rate of participation based on a relatively small sample,
the results for 2006 must be interpreted with extra great care. The results for 2006
should be regarded as tentative and, at the most, as an indication that black activities
do not appear to have increased from 2005 to 2006.
27. This procedure is described in great detail in the papers by Dell’Anno and Schneider
(2003, 2009), see also the appendix where the procedure is briefly described and the
advantages and disadvantages are shown.


28. Pioneering work in this area has been done by Frey (1972, 1975, 1978, 1980), Cappiello
(1986), Lubell (1991), Pozo (1996), Bartlett (1998) and Tanzi (1999).
29. Compare also the latest OECD report (2009).
30. This is especially true in Europe (e.g. in Germany and Austria), where the total tax and
social security burden adds up to 100% on top of the wage effectively earned; see also
Section 2.2.3.
31. For developing countries some literature about the shadow labour market exists
(Dallago, 1990; Chickering and Salahdine, 1991; Loayza, 1996; Pozo, 1996 and OECD
2009).
32. Shadow economy labour force consists of estimated full-time ‘black’ jobs, including
unregistered workers, illegal immigrants and second ‘black’ jobs.
REFERENCES
Alderslade, J., J. Talmage and Y. Freeman (2006), Measuring the informal

economy: one neighborhood at a time. Discussion Paper. The Brooking
Institution Metropolitan Policy Program, Washington DC, September
2006.
Allingham, M.G. and A. Sandmo (1972), Income tax evasion: a theoretical analy-
sis. Journal of Public Economics, 1, 323–338.
Bajada, C. and F. Schneider (2005), The shadow economies of the Asia-Pacific.
Pacific Economic Review, 10, 379–401.
Bajada, C. and F. Schneider (2009), Unemployment and the shadow economy in
the OECD. Revue Economique, 60, 1033–1067.
Bartlett, B. (1998), Corruption, the underground economy, and taxation.
Unpublished Manuscript. National Center for Policy Analysis, Washington
DC.
Bentler, P.M. and C.P. Chou (1987), Practical issues in structural modeling.
Sociological Methods and Research, 16, 78–117.
economy. Economic Journal, 109, 348–359.
Blackwell C. (2010), A meta-analysis of incentive effects in tax compliance
experiments. In J. Alm, J. Martinez-Vazquez and B. Torgler (eds), Developing
Alternative Frameworks Explaining Tax Compliance. London: Routledge.
Bollen, K.A. and J.S. Long (1993), Testing Structural Equation Models. Beverly
Hills, CA: Sage.
Boockmann, B. and J. Rincke (2005), Wirksamkeit der Bekämpfung der
Schwarzarbeit durch die ‘Finanzkontrolle Schwarzarbeit’: Empirische
Untersuchung der Entwicklung der Schwarzarbeit und des hierauf bezogenen
Bewusstseins- und Wertewandels – Machbarkeitsstudie. Final Report to the
Federal Finance Ministry, ZEW Mannheim.
Breusch, T. (2005a), Estimating the underground economy using MIMIC models.
Working Paper, Canberra, Australia. Available at http://econwpa.wustl.edu/
eps/em/papers/0507/0507003.pdf.
Breusch, T. (2005b), The Canadian underground economy: an examination of
Giles and Tedds. Canadian Tax Journal, 53, 367–391.
Brück, T., J.B. Haisten-DeNew and K.F. Zimmermann (2006), Creating


low-skilled jobs by subsidizing market contracted household work. Applied

Economics, 38, 899–911.
Bühn, A., A. Karmann and F. Schneider (2009), Shadow economy and do-it-
yourself activities: the German case. Journal of Institutional Economics, 164/5,
701–722.
Cappiello, M.A. (1986), Proposita di bibliografia ragionata sull’economia som-
mersa nell’industria (Italia 1970–82). In A. Bagnasco (ed.), L’altra metà
dell’economia, La ricerca internazionale sull’economia informale, Liguori,
Naples, pp. 307–345.
Cassar, A. (2001), An index of the underground economy in Malta. Bank of
Valletta Review, 23, 44–62.
economy and the formal regulatory environment. Paper presented at the EGDI-
WIDR Conference ‘Unleashing Human Potential: Linking the Informal and
Formal Sectors’, Helsinki, Finland.
Chickering, L.A. and M. Salahdine (eds) (1991), The Silent Revolution: The
Informal Sector in Five Asian and Near Eastern Countries. San Francisco, CA:
International Center for Economic Growth Publication (ICS Press).
Dallago, B. (1990), The Irregular Economy: The ‘Underground Economy’ and the
‘Black Labour Market’. Dartmouth, UK: Publishing Company.
De Grazia, R. (1983), Le travail clandestin: Situation dans les pays industrialisés à
économie de marché. Genève: Bit.
Dell’Anno R. (2003). Estimating the shadow economy in Italy: a structural equa-
tion approach. Working Paper 2003-7. Department of Economics, University
of Aarhus.
Dell’Anno, R. and F. Schneider (2003), The shadow economy of Italy and other
OECD countries: what do we know?, Journal of Public Finance and Public
Choice, 21, 223–245.
Dell’Anno, R. and F. Schneider (2009), A complex approach to estimate shadow
economy: the structural equation modeling. In M. Faggnini and T. Looks (eds),
Coping with the Complexity of Economics. Berlin: Springer, pp. 110–130.
Dell’Anno, R and O.H. Solomon (2008), Shadow economy and unemployment
rate in USA: is there a structural relationship? An empirical analysis, Applied
Economics, 40, 2537–2555.
Deregulierungskommission (Deregulation Commission) (1991), Marktöffnung und
Wettbewerb. Stuttgart: C.E. Poeschel.
Dixon, H. (1999), Controversy, on the hidden economy, editorial introduction.
Economic Journal, 456, 335–337.
Dreher, A. and F. Schneider (2010), Corruption and the shadow economy: an
empirical analysis. Public Choice, 144(2), 215–238.
Dreher, A., C. Kotsogiannis and S. McCorriston (2007), Corruption around the
world: evidence from a structural model. Journal of Comparative Economics, 35,
443–466.
Dreher, A., C. Kotsogiannis and S. McCorriston (2009), How do institutions
affect corruption and the shadow economy? International Tax and Public
Finance, 16(6), 773–796.
Eck, R. van and B. Kazemier (1988), Features of the hidden economy in the
Netherlands, Review of Income and Wealth, 34, 251–273.
Enste, D. and F. Schneider (2006), Umfang und Entwicklung der Schattenwirtschaft
in 145 Ländern. In F. Schneider and D. Enste (eds), Jahrbuch Schattenwirtschaft


2006/07. Zum Spannungsfeld von Poltik und Ökonomie. Berlin: LIT Verlag,
pp. 55–80.
European Commission (2007), Undeclared work in the European Union.
Special Eurobarometer Report, Brussels. Available at http://ec.europa.eu/
public_opinion/archives/ebs/ebs_284_en.pdf.
Feige, E.L. (ed.) (1989), The Underground Economies. Tax Evasion and Information
Distortion, Cambridge, UK: Cambridge University Press.
Supplement to Public Finance/ Finances Publiques, 49, 119–136.
Feld, L.P. and B.S. Frey (2002a), The tax authority and the taxpayer: an explora-
tory analysis. Unpublished Manuscript, University of Zurich
Feld, L.P. and B.S. Frey (2002b), Trust breeds trust: how taxpayers are treated.
Economics of Governments, 3, 87–99.
Feld, L.P. and B.S. Frey (2007), Tax compliance as the result of a psychological
tax contract: the role of incentives and responsive regulation. Law and Policy,
29, 102–120.
Feld, L.P. and C. Larsen (2005), Black Activities in Germany in 2001 and 2004: A
Comparison Based on Survey Data, Study No. 12. The Rockwool Foundation
Research Unit, Copenhagen.
Feld, L.P. and C. Larsen (2008), ‘Black’ activities low in Germany in 2006. News
from the Rockwool Foundation Research Unit, March 2008, 1–12.
Feld, L.P. and C. Larsen (2011), Undeclared Work, Deterrence and Social Norms:
The Case of Germany. Berlin: Springer.
Feld, L.P., A. Schmidt and F. Schneider (2007), Tax evasion, black activities and
deterrence in Germany: an institutional and empirical perspective. Mimeo,
University of Heidelberg.
Feld, L.P. and F. Schneider (2010), Survey on the shadow economy and unde-
clared earnings in OECD countries. German Economic Review, 11, 109–
149.
Flaming, D., B. Hayolamak and P. Jossart (2005), Hopeful workers, marginal
jobs: LA’s off-the-books labour force. Economic Roundtable, Los Angeles, CA.
Fleming, M.H., J. Roman and G. Farrel (2000), The shadow economy. Journal of
International Affairs, Spring 2000, 53/2, 64–89.
Frey, B.S. (1997), Not Just for the Money: An Economic Theory of Personal
Motivation. Cheltenham, UK: Edward Elgar.
Frey, B.S. and W.W. Pommerehne (1984), The hidden economy: state and pros-
pect for measurement. Review of Income and Wealth, 30, 1–23.
Frey, B.S. and H. Weck-Hannemann (1984), The hidden economy as an ‘unob-
served’ variable. European Economic Review, 26, 33–53.
Frey, L. (1972), Il lavoro a domicilio in Lombardia, Giunta Regionale Lombarda.
Assessorato al Lavoro, Milan.
Frey, L. (1975), Il potenziale di lavoro in Italia. Documenti ISVET, no. 50.
Frey, L. (1978), Il lavoro nero nel 1977 in Italia. Tendenze della occupazione,
no. 6.
Frey, L. (1980), Introduzione all’analisi economica del lavoro minorile. Economia
del Lavoro, 1–2, 5–16.
the grabbing hand: the determinants of unofficial activity in 69 countries,
Journal of Public Economics, 76, 459–493.
Gaetani-D’Aragona, G. (1979), I sommersi. Nord e Sud, 7, 26–46.


Garson, D.G. (2005), Statnotes: An Online Textbook. Available at http://www2.

chass.ncsu.edu/garson/pa765/structur.htm.
Giles, D.E.A. (1999a), Measuring the hidden economy: implications for economet-
ric modelling, Economic Journal, 109, 370–380.
Giles, D.E.A. (1999b), Modelling the hidden economy in the tax-gap in New
Zealand. Empirical Economics, 24, 621–640.
Giles, D.E.A. (1999c), The rise and fall of the New Zealand underground economy:
are the reasons symmetric? Applied Economic Letters, 6, 185–189.
economy. Canadian Tax Paper No. 106, Canadian Tax Foundation, Toronto/
Ontario.
Halla, M. and F. Schneider (2005), Taxes and benefits: two distinct options to
cheat on the state? Discussion Paper. Institute of Economics, University of Linz,
Linz.
Hayduk, L.A. (1987), Structural Equation Modelling with LISREL. Essential and
Advances. London: The Johns Hopkins University Press.
Helberger, C. and H. Knepel (1988), How big is the shadow economy? A re-
analysis of the unobserved-variable approach of B.S. Frey and H. Weck-
Hannemann. European Economic Review, 32, 965–976.
Herwartz, H., F. Schneider and E. Tafenau (2009), One share fits it all? Regional
variation in the shadow economy in the EU regions. Discussion paper.
Universities of Linz and Kiel.
Hill, R. (2002), The underground economy in Canada: boom or bust? Canadian
Tax Journal, 50, 1641–1654.
Institut für Demoskopie (IfD) Allensbach (1975), Freizeitarbeit 1974. Studie
im Auftrag der Kommission für wirtschaftlichen und sozialen Wandel,
Allensbach.
Isachsen, A.J. and S. Strøm (1985), The size and growth of the hidden economy in
Norway. Review of Income and Wealth, 31, 21–38.
sition. Brookings Papers on Economic Activity, 2, 159–221.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998a), Regulatory discre-
tion and the unofficial economy, American Economic Review, Papers and
Proceedings, 88(2), 387–392.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998b), Corruption, public
finances and the unofficial economy. World Bank Policy Research Working
Paper Series No. 2169, The World Bank, Washington DC.
Jöreskog, K.G. and D. Sörbom (1993), New Features in LISREL 8. Chicago:
Scientific Software.
Kanniainen, V., J. Pääkönen and F. Schneider (2004), Fiscal and ethical deter-
minants of shadow economy: theory and evidence. Mimeo, Johannes-Kepler
University Linz.
Karmann, A. (1986), Monetäre Ansätze zur Erfassung der Schattenwirtschaft: Ein
Vergleich verschiedener Messansätze. Kredit und Kapitel, 19, 233–247.
Karmann, A. (1990), Schattenwirtschaft und ihre Ursachen: Eine empirische Analyse
zur Schwarzwirtschaft und Selbstversorgung in der Bundesrepublik Deutschland.
Zeitschrift für Wirtschafts- und Sozialwissenschaften (ZWS), 110, 185–206.
Kastlunger, B., E. Kirchler, L. Mittore and J. Pitters (2009), Sequences of audits,
tax compliance, and taxpaying strategies. Journal of Economic Psychology, 30,
405–418.


Kazemier, B. (2005a), The underground economy: a survey of methods and esti-

mates. Discussion Paper. Statistics Netherlands, Voorburg, Netherlands.
Kazemier, B. (2005b), Monitoring the underground labour market: what surveys
can do. Discussion Paper. Statistics Netherlands, Voorburg, Netherlands.
Kazemier, B. (2006), Monitoring the underground economy: a survey of methods
and estimates. In F. Schneider and D. Enste (eds), Jahrbuch Schattenwirtschaft
2006/07. Zum Spannungsfeld von Politik und Ökonomie. Berlin: LIT Verlag,
pp. 11–53.
economy, 1955–1980. Zeitschrift für die gesamte Staatswissenschaft (ZgS)/
Journal of Institutional and Theoretical Economics (JITE), 139, 197–214.
erarbeiteten Sozialprodukts. Allgemeines Statistisches Archiv, 68, 378–405.
Kirchler, E. (2007), The Economic Psychology of Tax Behaviour, Cambridge, UK:
Cambridge University Press.
Kirchler, E., E. Hoelzl and I. Wahl (2007), Enforced versus voluntary tax com-
pliance: the ‘slippery Slope’ Framework. Journal of Economic Psychology, 29,
210–225.
Kirchler, E., B. Maciejovsky and F. Schneider (2003), Everyday representations of
tax avoidance, tax evasion and tax flight: do legal differences matter? Journal of
Economic Psychology, 24, 535–553.
Klovland, J.T. (1984), Tax evasion and the demand for currency in Norway and
Sweden. Is there a hidden relationship? Scandinavian Journal of Economics, 86,
423–439.
Kopp, R.J., W.W. Pommerehne and N. Schwarz (1997), Determining the Value of
Non-Marketed Goods: Economic, Psychological, and Policy Relevant Aspects of
Contingent Valuation Methods. Boston: Kluwer.
Körner, M., H. Strotmann, L.P. Feld and F. Schneider (2006), Steuermoral – Das
Spannungsfeld von Freiwilligkeit der Steuerhinterzahlung und Regelverstoß durch
Steuerhinterziehung. IAW Forschungsbericht Nr. 64, Tübingen.
Kucera, D. and L. Roncolato (2008), Informal employment: two contested policy
issues. International Labour Review, 147, 321–348.
Lamnek, S., G. Olbrich and W. Schäfer (2000), Forschungsprojekt ‘Devianz im
Sozialstaat’. Eichstätt: gefördert von der Volkswagen-Stiftung.
Langfeldt, E. (1984a), Die Schattenwirtschaft in der Bundesrepublik Deutschland.
Tübingen: Kieler Studien 191, Mohr Siebeck.
Langfeldt, E. (1984b), The unobserved economy in the Federal Republic of
Germany. In E.L. Feige (ed.), The Unobserved Economy. Cambridge, UK:
Cambridge University Press, pp. 236–260.
Lemieux, T., B. Fortin and P. Fréchette (1994), The effect of taxes on labor supply
in the underground economy. American Economic Review, 84, 231–254.
Lippert, O. and M. Walker (eds) (1997), The Underground Economy: Global
Evidences of its Size and Impact. Vancouver, BC: The Frazer Insitute.
Loayza, N.V. (1996), The economics of the informal sector: a simple model and
Series on Public Policy, 45, 129–162.
Loayza, N.V., A.M. Oviedo and L. Servén (2005a), The impact of regulation on
growth and informality: cross country evidence. World Bank, Policy Research
Paper WPS3623, Washington DC.


Loayza, N.V., A.M. Oviedo and L. Servén (2005b), Regulation and firm dynamics.
Discussion Paper. World Bank, Washington DC.
Lubell, H. (1991), The Informal Sector in the 1980’s and 1990’s. Paris: OECD.
Marcelli, E.A. (2004), Unauthorized Mexican immigration, the labour and other
lower-wage informal employment in California. Regional Studies, 38, 1–13.
Marcelli, E.A., M. Pastor Jr. and P.M. Joassart (1999), Estimating the effects
of informal economic activity: evidence from Los Angeles County. Journal of
Economic Issues, 33, 579–607.
Meon, P.-G., F. Schneider and L. Weill (2010), Does taking the shadow economy
into account matter when measuring aggregate efficiency? Applied Economics,
42/2, 160–176.
Merz, J. and K.G. Wolff (1993), The shadow economy: illicit work and household
production – a microanalysis of West Germany. Review of Income and Wealth,
39, 177–194.
Mogensen, G.V., H.K. Kvist, E. Körmendi and S. Pedersen (1995), The Shadow
Economy in Denmark 1994: Measurement and Results. Study No. 3, The
United Germany? Finanzarchiv, 58, 260–285.
OECD (2009), Is Informal Normal? Towards More and Better Jobs. Paris: OECD.
Pedersen, S. (2003), The Shadow Economy in Germany, Great Britain and
Scandinavia: A Measurement Based on Questionnaire Service. Study No. 10, The
Pickhardt, M. and J. Sardà Pons (2006), Size and scope of the underground
economy in Germany. Applied Economics, 38, 1707–1713.
Posner, E.A. (2000a), Law and Social Norms. Cambridge, MA: Harvard University
Press.
Posner, E.A. (2000b), Law and social norms: the case of tax compliance. Virginia
Law Review, 86, 1781–1820.
Pozo, S. (ed.) (1996), Exploring the Underground Economy: Studies of Illegal and
Unreported Activity. Michigan: W.E. Upjohn, Institute for Employment Research.
Renooy, P., S. Ivarsson, O. van der Wusten-Gritsai and E. Meijer (2004),
Undeclared work in an enlarged union – an analysis of undeclared work: an
in-depth study of specific items. Brussels: European Commission. Available at
http://ec.europa.eu/social/main.jsp?catId5647&langId5en.
Ruesga, B.S.M. (1984), Economia oculta y mercado de trabajo: aproximación al
caso español. Información commercial espaniola, 55–61.
Schneider, F. (1986), Estimating the size of the Danish shadow economy using the
currency demand approach: an attempt. Scandinavian Journal of Economics, 88,
643–668.
Schneider, F. (1994b), Determinanten der Steuerhinterziehung der Schwarzarbeit
im internationalen Vergleich. In C. Smekal and E. Theurl (eds), Stand und
Entwicklung der Finanzpsychologie. Baden-Baden: Nomos, pp. 247–288.
Schneider, F. (1994c), Measuring the size and development of the shadow economy.
Can the causes be found and the obstacles be overcome? In H. Brandstaetter and
W. Güth (eds), Essays on Economic Psychology. Berlin: Springer, pp. 193–212.


Schneider, F. (1997), The shadow economies of Western Europe, Economic

Affairs, 17, 42–48.
Schneider, F. (1998a), Further empirical results of the size of the shadow economy
of 17 OECD-countries over time. Paper, the 54th Congress of the IIPF Cordoba,
Argentina. Department of Economics, University of Linz, Linz, Austria.
Schneider, F. (1998b), Stellt das Anwachsen der Schwarzarbeit eine wirtschaft-
spolitische Herausforderung dar? Einige Gedanken aus volkswirtschaftlicher
Sicht. Mitteilungen des Instituts für angewandte Wirtschaftsforschung (IAW),
I/98, 4–13.
Schneider, F. (1999), Ist Schwarzarbeit ein Volkssport geworden? Ein internation-
aler Vergleich des Ausmaßes der Schwarzarbeit von 1970–97. In S. Lamnek and
J. Luedtke (eds), Der Sozialstaat zwischen Markt und Hedeonismus. Opladen:
Leske und Budrich, pp. 126–161.
Schneider, F. (2000), The increase of the size of the shadow economy of 18 OECD-
countries: some preliminary explanations. Paper presented at the Annual Public
Choice Meeting, 10–12 March 2000, Charleston, SC.
Schneider, F. (2001), Die Schattenwirtschaft – Tatbestand, Ursachen,
Auswirkungen. In A. Rauscher (ed.), Die Arbeitswelt im Wandel. Köln: J.P.
Bachem, pp. 127–143.
Schneider, F. (2003), Shadow economy. In C.K. Rowley and F. Schneider (eds),
Encyclopedia of Public Choice Vol. II. Dordrecht: Kluwer Academic Publishers,
pp. 286–296.
Schneider, F. (2004), Einige Volkswirtschaftliche Überlegungen über die
Interaktionen zwischen der Schattenwirtschaft und der offiziellen Wirtschaft.
In K. Bizer, A. Falk and J. Lange (eds), Am Staat vorbei: Transparenz, Fairness
und Partizipation kontra Steuerhinterziehung. Berlin: Duncker & Humblot,
pp. 29–45.
know? European Journal of Political Economy, 21, 598–642.
Schneider, F. (2006), Shadow economies and corruption all over the world:
what do we really know? Discussion paper. Universität Linz: Institut für
Volkswirtschaftslehre.
Schneider, F. (2009), Size and development of the shadow economy in Germany,
Austria and other OECD countries: some preliminary findings. Revue
Economique, 60, 1079–1116.
Schneider, F. and D. Enste (2000a), Schattenwirtschaft und Schwarzarbeit –
Umfang, Ursachen, Wirkungen und wirtschaftspolitische Empfehlungen.
München: Oldenbourg.
Schneider, F. and D. Enste (2000b), Shadow economies: size, causes and conse-
quences. Journal of Economic Literature, 38, 73–110.
University Press.
Schneider, F. and D. Enste (eds) (2006), Jahrbuch Schattenwirtschaft 2006/07. Zum
Spannungsfeld von Politik und Ökonomie. Berlin: LIT Verlag.
Schneider, F., J. Volkert and S. Caspar (2002), Schattenwirtschaft und
Schwarzarbeit: Beliebt bei – Problem für alle: Eine Analyse der schattenwirt-
schaftlichen Aktivitäten in Deutschland (am Beispiel Baden-Württemberg) und
mögliche politische Konsequenzen. Baden-Baden: Nomos.
Smith, P. (1994), Assessing the size of the underground economy: the Canadian


statistical perspectives. Canadian Economic Observer, Catalogue No. 11-010,

3.16-33, at 3.18.
Smith, R.S. (2002), The underground economy: guidance for policy makers?
Canadian Tax Journal, 50, 1655–1661.
Spiro, P.S. (1993), Evidence of a post-GST increase in the underground economy.
Canadian Tax Journal, 41, 247–258.
Stapleton, D.C. (1978), Analyzing political participation data with a MIMIC
model. Sociological Methodology, 15, 52–74.
Stevens, J. (1996), Applied Multivariate Statistics for the Social Sciences, 3rd edn.
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Tanzi, V. (1999), Uses and abuses of estimates of the underground economy.
Thomas, J.J. (1992), Informal Economic Activity. New York: Harvester
Wheatsheaf.
Thomas, J.J. (1999), Quantifying the black economy: ‘measurement without
theory’ yet again? Economic Journal, 109, 381–389.
Torgler, B. (2002), Speaking to theorists and searching for facts: tax morale and
tax compliance in experiments. Journal of Economic Surveys, 16, 657–683.
Torgler, B. (2007), Tax Compliance and Tax Morale: A Theoretical and Empirical
Analysis. Cheltenham, UK: Edward Elgar.
Torgler, B. and F. Schneider (2005), Attitudes towards paying taxes in Austria: an
empirical analysis. Empirica, 32, 231–250.
Torgler, B. and F. Schneider (2009), The impact of tax morale and institutional
quality on the shadow economy. Journal of Economic Psychology, 30, 228–245.
Weck, H. (1983), Schattenwirtschaft: Eine Möglichkeit zur Einschränkung der
öffentlichen Verwaltung? Eine ökonomische Analyse. Bern: Peter Lang.
Economies: Implication for Work and Welfare. London: Routledge.
Williams, C.C. and J. Windebank (2001a), Beyond profit motivated exchange:
some lessons from the study of paid informal work. European Urban and
Regional Studies, 8, 49–61.
Williams, C.C. and J. Windebank (2001b), Reconceptualizing paid informal
exchange: some lessons from English cities. Environment and Planning A, 33,
121–140.
Williams, C.C., E. Horlings and P. Renooy (2008), Tackling Undeclared Work
in the European Union. Research Report. European Foundation for the
Improvement of Living and Working Conditions, Dublin.


APPENDIX 2.1 THE MIMIC ESTIMATION

PROCEDURE, ITS ADVANTAGES
AND DISADVANTAGES1
A2.1.1 MIMIC Model
How does the MIMIC procedure work? Using the standard LISREL
notation of Jöreskog and Sörbom (1993), Equation A2.1.1 is a meas-
urement equation where ht (unobserved or latent) variable determines
yr 5 (y1, y2, . . ., yd) r column vector of indicators subject to a random error
term et. ht is an unobserved or latent and is a scalar. Following Dell’Anno
and Solomon (2008), L is a (d 3 1) column vector of parameters that
relates yt to ht:
yt 5 lht 1 et. (A2.1.1)
Equation A2.1.2 is a structural equation which shows that the unobserved

variable ht is determined by xt set of exogenous causes (x1, x2, . . ., xc) and
zt a structural disturbance error term. g is a (1 3 c) vector of structural
parameters:
ht 5 gxt 1 zt. (A2.1.2)
Without loss of generality, all variables are taken to have zero means.
In Equations (A2.1.1) and (A2.1.2) it is assumed that: the elements of zt
and et are normal, independent and identically distributed;2 the variance
of the structural disturbance term zt is Y and the covariance matrix of
the measurement errors is a diagonal covariance matrix Qe.3 Substituting
Equations A2.1.1 and A2.1.2 yields a reduced form equation which
expresses a relationship between the observed variables xt and yt. This is
shown in Equation A2.1.3:
yt 5 Prxt 1 zt, (A2.1.3)
where: P 5 lgr is a c 3 d reduced form coefficients matrix and has rank

one expressed in terms of c and d elements of l and g; zt 5 lzt 1 et is a
reduced form disturbance vector; z has an d 3 d reduced form covariance
matrix (w) given by:
w 5 l␾lr 1 Qe (A2.1.4)


where: ␾ 5 var (l) and Qe 5 the reduced-form covariance matrix of the

measurement errors.
For the MIMIC nomenclature, the equations system with the relations
between the latent variable ht(shadow economy) and the causes xt is called
the ‘structural model’ (Eq. A2.1.2); the links among indicators yt and
underground economy is called the ‘measurement model’ (Eq. A2.1.3).
A2.1.2 Application of the MIMIC Procedure
In the first step, the researcher has to translate his/her theory into a
structural model. In the second step, it is necessary to fix one coefficient
to the value one in order to give the latent variable an interpretable
scale. In the third step the estimation method has to be chosen. The
Maximum Likelihood Estimation (MLE procedure) is the most widely
used in a MIMIC model. It assumes multivariate normal data and a
reasonable sample size.4 If the data are continuous but not normally
distributed, an alternative method is an asymptotically distribution free
estimation procedure, which in LISREL is known as WLS (weighted
least squares).
All goodness-of-fit measures are a function of sample size and degrees
of freedom. Most of these take into account not only the fit of the model
but also the model complexity. On the one side, if we have a very large
sample, the statistical test will almost certainly be significant with respect
to the degrees of freedom. On the other side, if we have small samples the
model is very likely to be accepted even if the fit is poor. This is particularly
important in the analysis of shadow economy, since usually both the data
availability is poor and the model complexity is high. When the model fit
is not adequate, it has become common practice to modify the model, by
deleting non-significant parameters in order to improve the fit and select
the most suitable model specification.
A2.1.3 Advantages and Disadvantages of the MIMIC Estimates of the

Shadow Economy
Advantages
It is widely accepted by most scholars, who estimate the size and develop-
ment of the shadow economy, that such an empirical exercise is a ‘mine-
field’ regardless of which method is used. In evaluating the estimations of
the shadow economy, we should keep in mind that Schneider (1997) and
Schneider and Enste (2000a) warned that there is no best or commonly
accepted method. Each approach has its strengths and weaknesses and can
provide specific insights and results. Although from the first use the MIMIC


approach has been ‘accompanied’ by strong criticisms,5 in the last 10 years

it has been increasingly used for estimation of the shadow economy.6
The MIMIC approach offers several advantages in comparison with the
other statistical methods to estimate shadow economy. According to Giles
and Tedds (2002), MIMIC provides a wider approach than most other
competing methods, since it allows one to take multiple indicator vari-
ables and multiple causal variables into consideration at the same time.
Moreover, it is quite flexible, allowing one to vary the choice of causal and
indicator variables according to the particular features of the economy
under study, the period in question, and the availability of data. Again,
following Giles and Tedds (2002), the MIMIC model leads to a formal
estimation and to testing procedures, such as those based on the method
of MLE. These procedures are well known and are generally ‘optimal’, if
the sample is sufficiently large.
A further advantage of the MIMIC approach has been stressed by
Schneider and Enste (2000a, b), who emphasize that the MIMIC approach
leads to some progress in the estimation techniques of underground
economy, because this methodology allows a wide flexibility in its applica-
tion, therefore it is potentially superior over all other estimation methods.
Compared with other methods, Cassar (2001) argues that MIMIC does
not need restrictive assumptions to operate (with the exception of the
calibrating process). Also, Thomas (1992) argues that the only real con-
straint of this approach is not in its conceptual structure but the choice of
variables.
Disadvantages
Besides advantages there are, of course, disadvantages, and the following
six ones are shown:
1. When estimating the shadow economy using the MIMIC model

approach the most common objection concerns the meaning of the
latent variable (Helberger and Knepel, 1988; Giles and Tedds, 2002;
Smith, 2002; Dell’Anno, 2003), because the MIMIC approach is
largely a confirmatory rather than exploratory technique. This means
a researcher is more likely to determine whether a certain model
is valid, rather than to ‘find’ a suitable model. Therefore, it is pos-
sible that the theoretical construct ‘shadow economy’ could include
other potential definitions (traditional crime activities, DIY and so
forth). This criticism, which is probably the most common in litera-
ture, remains difficult to overcome as it goes back to the theoretical
assumptions behind the choice of variables and empirical limitation
on the availability of data.


2. Another objection is expressed by Helberger and Knepel (1988). They

argue that the MIMIC estimations lead to unstable estimated coef-
ficients with respect to changes in the sample size and with respect to
alternative model specifications. As Dell’Anno (2003) shows, instabil-
ity disappears asymptotically as the sample size grows large and if
data are stationary and normally distributed.
Dell’Anno (2003) points out additional objections: (a) to calculate
the confidence intervals associated with estimates of the latent vari-
able, (b) to test the hypothesis of independence between structural and
measurement errors, (c) to identify exhaustively the properties of the
residuals, and (d) to apply the SEM approach to small sample sizes
and time series analysis. We believe that these cited weaknesses are the
main limitations of this approach.
3. A further criticism is pointed out by Dell’Anno (2003). When using
the MIMIC approach he finds that there is a frequent possibility that
an indefinite covariance matrix is encountered in the estimation proce-
dure. According to Bollen and Long (1993), this problem arises when
the data contain too little information, like small sample size, too few
indicator variables, small factor loadings, missing values and so forth.
4. Another criticism of the reliability of the MIMIC estimates of shadow
economy is related to the benchmark method (Breusch 2005a, b). This
criticism has its origin in the complications researchers face, when he
or she wants to convert the index of shadow economy (estimated by
the MIMIC model) into cardinal values. This is not an easy task, as
the latent variable and its unit measure are not observed. The model
just provides a set of estimated coefficients from which one can calcu-
late an index which shows the dynamics of the unobserved factor.
Such a calibration – regardless which methodology is used –
requires experimentation, and a comparison of the calibrated values
in a wide academic debate, although at this stage of research on the
MIMIC approach, it is not clear which benchmark method is the best
or most reliable one. The way in which to proceed here is still prob-
lematical and unexplored, hence every suggestion about this aspect of
technique is welcome.7
5. The last criticism of MIMIC estimates refers to the methodological
side and, in accordance with the aims of this chapter, may be the most
relevant. Breusch (2005b) argues that the statistical properties/nature
of the MIMIC approach are unsuitable for economic questions/prob-
lems because this approach was designed for psychometric application
and for ‘measuring intelligence seems far removed from estimating
the underground economy’. Dealing with this critique, the main
problem of the MIMIC approach lies in the strong difference between


economic and psychological variables. Although, we agree that it

is (still) problematical to apply this methodology to an economic
dataset and to specify a macroeconomic model through the MIMIC
framework, it does not mean we should abandon this approach. On
the contrary, following an interdisciplinary approach to economics,
the marked criticism should be considered as an incentive for further
(economic) research in this field rather than to suggest not using
this method because of the difficulties in the implementation of the
MIMIC method.
NOTES
1. This appendix closely follows Dell’Anno and Schneider (2009, pp. 110–130).
2. The assumption on independence between structural disturbance zt, and measurement
errors the measurement errors et could be considered as too restrictive, when mainly
using economic dataset and, consequently, espoused to question the validity of this
approach. Hayduk (1987, p. 193) explains it ‘is purely a matter of arbitrary conven-
tion’ for SEM analysis. Dell’Anno (2003) in the context MIMIC model presents a re-
parameterization of the MIMIC able to test the assumption on independence between
structural disturbance zt, and measurement errors et.
3. In the standard MIMIC model the measurement errors are assumed to be independent
of each other, but this restriction could be relaxed (Stapleton, 1978, p. 53).
4. There are several rules of thumb about the sample size in the literature (Garson, 2005):
the sample size should be at least 50 observations or have more than eight times the
observations than the number of independent variables in the model. Another one, based
on Stevens (1996), is to have at least 15 observations per measured variable or indicator.
Bentler and Chou (1987) recommend at least five observations per parameter estimate
(including error terms as well as path coefficients). If possible, the researcher should go
beyond these minimum sample size recommendations particularly when the data are not
normally distributed or are incomplete.
5. Compare, for example, the criticism by Helberger and Knepel (1988) with respect to the
pioneering work of Frey and Weck-Hannemann (1984).
6. Compare the studies cited in Section 2.4.
7. In Dell’Anno and Schneider (2009, pp. 110–130), four different benchmarking strategies
are discussed.


3 The size and development of
the shadow economy in India: a
first attempt at a public choice
explanation
Kausik Chaudhuri and Friedrich Schneider
3.1 INTRODUCTION
The gap between the observable and the actual has for long been the
anathema of social scientists. Over the years this has led to the con-
ceptualization of the ‘hidden economy’. Black, shadow, underground,
unobserved, unofficial, subterranean, unrecorded, informal, irregular,
second, twilight, parallel – synonyms used for the ‘hidden economy’ seem
to highlight the fact that this concept essentially captures the activities
beyond measurement by fiscal or economic factors.1 The hidden economy
basically consists of legal and illegal activities outside the reach of the gov-
ernment.2 Studies show that underground activities have been on the rise
since the 1970s when the presence of government activity became stronger
in economies around the world. With an increase in tax rates to finance
larger public spending programmes, the desire to escape taxes and regula-
tory restrictions also gained in prominence (Tanzi and Schuknecht, 1997).
The increasing size of the underground economy has received significant
media attention in recent past. Such media attention brought the nexus of
the black economy into the public glare creating a consciousness about the
gravity of the phenomenon all over the world.
In this chapter, we deal with two questions: (1) we provide a detailed
estimate of the size of the hidden economy in India using a Multiple
Indicator Multiple Cause (MIMIC) model and (2) we provide a first
attempt to demonstrate the role of the print media, elections and types of
government in determining the growth of the size of the hidden economy
using data from India. Our approach thus demonstrates the importance of
policy actions in increasing government responsiveness to curb the size of
the hidden economy.
131


Now, how do we define the hidden economy? Experts have real-

ized that arriving at a solitary definition of the underground or hidden
economy is a mammoth, if not an irrational and inhibiting, exercise.
In this chapter, we adopt a relatively broad definition. The shadow
or hidden economy deals with the portion of the income earned from
legal and illegal activities that cannot be accounted for by the stand-
ard measurement procedures used in compilation of national income
accounts. Hence, a commonly used working definition is: all currently
unregistered economic activities, which contribute to the officially cal-
culated (or observed) Gross National Product.3 Our definition is closely
related to the existing work on this subject. Tanzi (1999) suggests that
the shadow economy crops up because of the presence of activities that
are difficult to measure. Schneider (1986) sums up this point by defin-
ing the underground economy as ‘all economic activities that contribute
to the value added and should be included in national income in terms
of national accounting conventions but are presently not registered by
national measurement agencies’. Bhattacharyya (1999) describes the
hidden economy as reflected by the unrecorded national income ‘calcu-
lated as the difference between the potential national income for the given
currency in circulation and the recorded national income’. Bagachwa and
Naho (1995) consider it as a combination of informal (small-scale pro-
duction and distribution units), parallel (illegal production of legal activi-
ties) and black market activities (production and distribution of market
and non-market goods forbidden by the government). Acharya (1985),
in the Indian context, refers to the black economy as ‘the aggregate of
incomes which are taxable but are not reported to the tax authorities’
and also ‘the extent to which estimates of national income and output
are biased downwards because of deliberate, false reporting of incomes,
output and transactions for reasons of tax evasion, flouting of other eco-
nomic controls and related motives’. In general, a precise definition seems
quite difficult, if not impossible as ‘the shadow economy develops all the
time according to the “principle of running water”: it adjusts to changes
in taxes, to sanctions from the tax authorities and to general moral atti-
tudes, etc.’ (Mogensen et al., 1995, p. 5).
The chapter is organized in the following manner. In Section 3.2, we
discuss the methodology and in Section 3.3, we document the size of the
hidden economy estimates obtained for India. Section 3.4 highlights the
potential role of civic and political institutions to curb the size of the
hidden economy. In Section 3.5, we discuss the empirical results focusing
on the role of the print media and political variables to explain the growth
in the size of the hidden economy for the Indian context. Section 3.6 con-
cludes with policy recommendations.


The shadow economy in India 133
3.2 MIMIC MODEL AND THE HIDDEN ECONOMY

In this chapter, we attempt to estimate the hidden economy by multiple indi-
cator multiple cause approach (MIMIC). Frey and Weck-Hanneman (1984)
were the first to employ this methodology for the estimation of the hidden
economy of a cross-section of 17 OECD countries for the period of 1960–78.
They borrowed from the statistical theory of unobserved variables devel-
oped by Zellner (1970), Goldberger (1972) and Jöreskog and Goldberger
(1975). Frey and Weck-Hanneman (1984) arrived at a ranking of OECD
countries on the basis of the size of their underground economies. In the late
1970s, the Scandinavian and Benelux countries were seen to have very large
hidden economies followed by the United States in the middle and then by
Switzerland and Japan, which show very low underground activity for that
period. Aigner, Schneider and Ghosh (1988) use a variant of the MIMIC
approach, namely the DYMIMIC (the dynamic multiple indicators mul-
tiple causes) approach to assess the size of the US hidden economy for the
time period 1939–82. The results of this study have found a peak in the US
hidden economy size around 1943–44 and a trough in 1967–68.
In recent years, a lot of work has been done using the unobserved or
latent variable approach, particularly in the context of New Zealand and
Canada. Giles (1999a, 1999b), Giles and Caragata (2001) and Giles and
Tedds (2002) have used the time series data for the New Zealand and
the Canadian economy to arrive at hidden economy estimates using the
MIMIC approach.
Our chapter also uses a factor-analytic MIMIC4 variable approach.
The MIMIC model is actually a variant of the LISREL (linear independ-
ent structural relationships) models of Jöreskog and Sorbom (1993a, b)
and others. These kinds of models can only yield a time series index for
the latent variables expressed as percentage of the official GDP/labour
force. This ordinal index has to be converted into a cardinal series of
values of hidden economy sizes by scaling up the ordinal values to some
cardinal value that has been obtained in the past through other methods of
estimation like the electricity or the currency demand approach.
The MIMIC model equations can be stated as:
y 5 lh 1 e (3.1)
h 5 g9x 1 z (3.2)
where, y is a column vector of ‘p’ indicators of the latent variable, h, and

x is a column vector of the ‘q’ ‘causes’ of h. In other words, Equation (3.1)


is the measurement model for h and Equation (3.2) is the structural equa-
tion for the latent variable, h. e is a (p × 1) measurement error while z is
the scalar structural error. It is assumed that z and all the elements of e are
mutually uncorrelated, with Var (z) 5 y, and Cov (e) 5 Qe. Substituting
Equation (3.2) into (3.1), the MIMIC model can be conceived as a
p-equation multivariate regression model that takes the standard form,
y 5 Px 1 u, (3.3)
where P 5 lgr, u 5 lz 1 u, and Cov ? (z) 5 llry 1 Qe.

The p-equation model in Equation (3.3) seems to have a regressor
matrix of rank equal to one and an error covariance matrix that is simi-
larly constrained. We estimate the model using the maximum likelihood
estimation procedure.
3.3 MIMIC MODEL AND THE SIZE OF THE INDIAN

HIDDEN ECONOMY
Our study is different to the earlier underground economy studies con-

ducted in the Indian context by Gupta and Mehta (1981), Chopra (1982),
Acharya (1985), Bhattacharyya and Ghose (1998) and Bhattacharyya
(1999) in the following ways. Acharya (1985), Bhattacharyya and Ghose
(1998) and Bhattacharyya (1999) used traditional cash demand estimation
methodology, which has been criticized in the literature for its focus on
just one facet of the hidden economy. Gupta and Mehta (1981) have used
a physical input approach whereas Chopra’s method is in close line with
the one suggested by Kaldor (1956).5
In our chapter, we use the MIMIC model. To the best of our knowl-
edge, this is the first attempt to provide the size of the hidden economy
in India over a long period of time using the MIMIC model. We use
different combinations of the causes and indicators to arrive at different
hidden economy estimates. For the indicators, we use the growth rate of
real GDP of the economy and the cash to M3 ratio of the economy or
the level of cash circulation in real terms in the economy.6 The ‘causal’
variables that we include in our model allow for unemployment and
income effects; the tax-bracket ‘creep’ effect of inflation; and measures
of overall ‘tax burden’ and the nature of the tax-mix (indirect and direct
taxes). Specifically, we select different combinations from the set of vari-
ables including inflation, the number of persons working in the public
sector and the ratio of indirect taxes, direct taxes and corporate taxes
over GDP in the four different MIMIC specifications. After obtaining


Table 3.1 Variables used in MIMIC model
Names of the Description of the variables used

variables used
INDICATOR VARIABLES
grgdp Growth rate of Real GDP
cashm3 Currency to M3 ratio
dcashm3 First difference of the currency to M3 ratio
rcurr Level of real currency in the economy
drcurr First difference of level of real currency in the economy
CAUSAL VARIABLES
inflation Natural logarithm of the ratio of current year’s CPI to previous
year’s CPI
corgdp Ratio of total corporate taxes to nominal GDP
dirtgdp Ratio of total direct taxes to nominal GDP
indigdp Ratio of total indirect taxes to nominal GDP
dindgdp First difference of the ratio of total indirect taxes to nominal
GDP
npubemp Number of persons employed in the public sector (comprising
the central government 1 state government 1 quasi central
and state government 1 local bodies)
dnpubemp First difference of the number of persons employed in the
public sector
the ordinal series, cardinal estimates are obtained by calibration to some

other hidden economy estimate obtained from other studies. In this
context, we have used Bhattacharyya’s (1999) hidden economy estimate
for India, of 22.5 per cent for 1989–90, to scale up our ordinal hidden
economy series to arrive at the complete cardinal underground economy
sizes for India.
The data sources for these variables are listed in Appendix 3.1. Table
3.1 lists a detailed account of the different variables used for the MIMIC
specification and their full names since convenient abbreviations have
been used otherwise throughout this chapter. Before proceeding further,
we first test for the non-stationarity of the variables and the results are
discussed in the next section.
3.3.1 Unit Root Tests
The MIMIC approach might suffer from the standard problem of data
non-stationarity. In this regard, we have conducted a series of Augmented
Dickey Fuller (ADF) tests to check for the presence of unit roots in the


Table 3.2 Results from unit root tests (augmented Dickey-Fuller tests)
Variables Test-Statistics Selected Lag-length

Ln(Realgdp) −0.290 4
Grgdp −5.181 3
CashM3 −2.035 1
DcashM3 −4.298 0
LN(Rcurrr) −2.238 1
Drcurr −1681.543 0
Ln(CPI) −0.335 4
Inflation −6.194 3
Corgdp −2.856 0
Dirtgdp −4.199 3
Indgdp −2.228 0
Dindgdp −5.389 0
Npubemp 1.594 0
Dnpubemp −4.177 1
Note: Lag-lengths are selected using procedure suggested by Campbell and Perron (1991)
starting with a maximum lag length of 5. For the level series, we have included a trend term
along with the constant, however we do not include trend term for the differenced series.
Critical values for level with trend: –4.202 (1 per cent level), –3.525 (5 per cent level), –3.193
(10 per cent level), Critical values for the first-differenced without trend: –3.602 (1 per cent
level), –2.936 (5 per cent level), –2.606 (10 per cent level).
Source: MacKinnon (1991)
variables. The lag-length for the ADF tests is selected by the procedure
suggested by Campbell and Perron (1991). Among the casual variables
used in our analysis, indirect taxes over GDP and the number of people in
public employment were found to be I(1) variables at 10 per cent levels of
significance while among the indicator variables both the cash to M3 ratio
and the level of real currency seem to be I(1) variables. For all these four
variables, the first differences have been considered in our analysis. Table
3.2 gives the relevant unit root test results.
3.3.2 The MIMIC Model’s Estimates of the Hidden Economy of India
In this sub-section, we present the hidden economy estimates of four dif-

ferent models using varied combinations of causal and indicator variables.
The estimation results are shown in Table 3.3 along with the various
diagnostic statistics for our estimated models reported in Table 3.4. In
almost all the models, the diagnostic statistics give a satisfactory result
in terms of model fit. In these models, we have tried to study the hidden


Table 3.3 MIMIC model results and the parameter estimates
Variable Model 1 Model 2 Model 3 Model 4

Indicators
Growth rate of Real 0.41 0.39 0.42 0.63
GDP (Grgdp) (1.70)* (0.96) (1.69)* (1.09)
First difference of the 1.00 1.00
currency to M3 ratio
(DcashM3)
First difference of level 1.00 1.00
of real currency in the
economy (Drcurr)
Causes
Inflation −0.64 0.06 −0.63 0.04
(−5.06)*** (0.41) (−4.81)*** (0.28)
Ratio of total corporate 0.18 0.26
taxes to nominal GDP (1.43) (1.66)
(Corgdp)
Ratio of total direct −0.06 0.01
taxes to nominal GDP (−0.45) (0.06)
(Dirtgdp)
The first difference of −0.23 −0.26 −0.29 −0.32
the number of persons (−1.75)* (−1.56) (−2.22)** (−1.93)*
employed in the public
sector (Dnpubemp)
First-differenced ratio −0.01 0.02 0.00 0.03
of total indirect taxes (−0.09) (0.11) (0.00) (0.19)
to nominal GDP
(Dindgdp)
Note: t-statistics are reported in parentheses. * denotes significance at 10% level, ** at 5%

level and *** and 1% level using a two-tailed t-test.
economy determined by inflation, the first difference of indirect tax to

GDP ratio, the first difference of the number of public employees as either
the direct tax to GDP or the corporate tax to GDP ratios. The indicators
used include the growth of real GDP per capita and either the first differ-
ence of the currency to M3 ratio or the first difference of the real level of
currency in the economy. Models 1 and 3 focus on the use of the growth
rate of real GDP and the first difference of the real level of currency as the
two indicator variables whereas models 2 and 4 use the first difference of
the currency to M3 ratio instead of first difference of the real level of cur-
rency as the indicator variable. Given the normalizations in the estimation


Table 3.4 Diagnostic statistics of the estimated MIMIC model
Model 1 Model 2 Model 3 Model 4

Chi-Square 0.74 3.03 0.73 2.92
[p-value] 0.86 0.39 0.87 0.40
RMSEA 0.01 0.01 0.01 0.01
P-Value for Test of Close 0.88 0.45 0.88 0.46
Fit (RMSEA < 0.05)
RMR (Root mean square 0.03 0.06 0.03 0.06
residual)
Adjusted Goodness of 0.95 0.82 0.95 0.82
Fit (AGFI)
PGFI (Parsimony 0.14 0.14 0.14 0.14
Goodness of Fit)
of the MIMIC model, the estimated coefficients on the ‘growth in real

GDP’ indicator variable suggest not only a significant positive relationship
between the size of the hidden economy and growth in measured GDP, but
also suggest that the effect of the hidden economy on the rate of growth in
GDP is 0.4 to 0.6 times its effect on the first difference of the currency to
M3 ratio or the first difference of the real level of currency in the economy.
From Table 3.3 we can infer that in both models 1 and 3, inflation and
the number of public employees exhibit a negative and significant effect
on the size of the hidden economy in India. The contractionary effect of
the increase in number of public employees on the hidden economy might
be seen as a result of an increase in the number of secure and official full-
time jobs in society which might act as a kind of substitute for part-time
jobs or illegal occupations, all of which add to the size of unaccounted
incomes. With increase in inflation, the real incomes of families might
fall and this might prompt more people to join the official labour force.
However, inflation can enter either with positive or negative impact on the
hidden activities. In the face of inflationary pressures, whether the public
whose real incomes have declined will opt for the official labour force or
the underground economy, would also depend on other factors like tax
morality, culture and also future expectations among the populace. Both
the indirect tax to GDP ratio and the direct/corporate tax to GDP ratios
in India seem to exhibit an expected positive effect on the hidden economy
size. This seems to endorse the view in the literature that an increase in tax
burdens is one of the prime reasons for the ballooning of the underground
activities over time around the globe.
As we have already mentioned, the obtained ordinal series needs to be
converted to a cardinal series. This can be done by scaling the available


30
25
20
15
10
0
19 68
19 78
19 88
8
19 66
19 70
19 76
19 80
19 86
19 90
19 96
19 62
19 64
19 72
19 74
19 82
19 84
19 92
19 94
–9
–
–
–
–
–
–
–
–
–
–
–
–
67
77
87
97
65
69
75
79
85
89
95
61
63
71
73
81
83
91
93
19
Model 1 Model 2
Figure 3.1 Time path for the Indian hidden economy as a per cent of GDP
(1961/62–1997/98)
ordinal series using ‘benchmark estimates’ from other studies (Schneider

2000) or by estimating a cardinal series through some other approach,
say the physical indicator approach or the cash demand approach, and
using values from it to calibrate the ordinal series obtained by the MIMIC
approach. In the present chapter, we adopted the first option. We have
used Bhattacharyya’s (1999) hidden economy estimate for India, of 22.5
per cent for 1989–90, to scale up our ordinal hidden economy series to
arrive at the complete cardinal underground economy sizes for India. The
implied hidden economy series as a percentage of official GDP is plotted
in Figure 3.1 for both model 1 and model 2. Table 3.5 presents the size of
the hidden economy as a percentage of measured GDP. The size of the
hidden economy has increased from 8.99 per cent (1960/61) to 23.20 per
cent (1997/98) over the time period under study.
3.4 THE ROLE OF POLITICAL AND

INSTITUTIONAL FACTORS
In this section, we try to explain the growth in the size of the hidden
economy in India. Kaufmann (1999) considers knowledge and informa-
tion; leadership and collective action as prime weapons to tackle corrup-
tion. This translates to a need for better-defined civic institutions and
advocacy agents like the media. Stapenhurst (2000) provides a brief link


Table 3.5 Size of the hidden economy for India as a per cent of measured
GDP (1961/62–1997/98)
Year Size (estimated using Model 2)

1961/62 8.99
1962/63 9.59
1963/64 10.11
1964/65 10.65
1965/66 11.20
1966/67 11.52
1967/68 11.72
1968/69 12.09
1969/70 12.47
1970/71 12.83
1971/72 13.40
1972/73 14.32
1973/74 14.93
1974/75 15.39
1975/76 15.98
1976/77 16.59
1977/78 17.27
1978/79 17.65
1979/80 18.03
1980/81 18.52
1981/82 19.07
1982/83 19.68
1983/84 20.17
1984/85 20.65
1985/86 21.14
1986/87 21.55
1987/88 21.91
1988/89 22.20
1989/90 22.50
1990/91 22.79
1991/92 22.97
1992/93 23.86
1993/94 23.25
1994/95 23.28
1995/96 23.21
1996/97 23.01
1997/98 23.19


between an active media and the amelioration of both corruption and

also the shadow activities of the economy. Djankov et al. (2002) dem-
onstrate that the government ownership of the media is generally associ-
ated with less press freedom, fewer political and economic rights, and,
most importantly, inferior social outcomes in the areas of education and
health. Ahrend (2002) provides strong empirical evidence that suggests
that strengthening press freedom should be among the priorities fighting
against corruption.
India provides an interesting framework. India has been a democracy
since 1947. Periodic national elections have taken place since 1952. There
is a relatively free and independent press with significant time-series vari-
ation. Using these data, we are able to examine a connection between the
development of mass media, political and institutional factors and govern-
ment actions to cater for the needs of the citizens. In this connection, our
chapter can be viewed in line with the growing literature that uses data from
India to examine the role of institutional and political factors to explain
government responsiveness. Besley and Burgess (2000) demonstrate that
party ideology affects public policy. Besley and Burgess (2002) show that
state governments are more responsive to falls in food production and crop
damage in those states where newspaper circulation is higher and electoral
accountability is greater.7 To this extent, we ran the following regression:
Δht 5 a 1 bht-1 1 dnt-1 1 gElectiont 1m Coalitiont

1 x Majority Size Dummyt*Electiont 1 q Majority Sizet
1 ␾War Dummyt 1 lEmergency Dummyt 1 et (3.4)
where Dht represents the growth in the size of India’s hidden economy at
time period ‘t’, ht-1 is the lagged value of the size of the hidden economy,
nt-1 refers to the total per capita newspaper circulation in time period ‘t21’.
We would expect that an increase in the presence of civic institutions like
media have a contractionary effect on the growth of size of the hidden
economy for democratic countries. The variable Election consists of a
binary variable that equals one if a parliamentary election takes place in
financial year ‘t11’. One may think that during the time of the election the
government may try to curb the size of the hidden economy. So, we would
expect the coefficient on this variable to be negative. In Equation (3.4), we
have included two additional regressors: Majority Sizet and an interaction
term between Majority Sizet and Election variable. The variable Majority
Sizet denotes the proportion of Lok Sabha seats held by the government
at the centre. This may impact policies directly. A government that is
approaching an election may be more active if it has a slight majority


compared to a large majority. The interaction term (Election * Majority

Size) is introduced to measure this effect.
In recent years, India has witnessed government in the form of coali-
tion. Given this, we would also introduce the variable Coalition. This
variable is measured by the proportion of financial year ‘t’ where a coali-
tion government rules India. Our attempt aims to examine whether the
size of the hidden economy differs across the coalition governments and
the single-party governments. In Equation (3.4), we have also introduced
two additional regressors, namely War Dummy and Emergency Dummy.
On June 27, 1975, President Fakhruddin Ali Ahmed declared a state of
emergency due to ‘internal disturbances’ and this was lifted on 21 March
1977. Thus, Emergency Dummy equals 1 in financial years 1975–76 and
1976–77. Between financial years 1960–61 and 1997–98, three periods of
external conflict had happened. A conflict with China took place between
20 October 1962 and 9 November 1962; and conflicts with Pakistan took
place between 1 September 1965 and 23 September 1965, and between 3
December 1971 and 17 December 1971. Hence the variable War Dummy
takes the value of 1 in financial years 1962–63, 1965–66 and 1971–72. The
error term is denoted by et.
3.5 EMPIRICAL RESULTS
In this section, we present our empirical results from estimation of

Equation (3.4). The result is presented in Table 3.6.8 Our results show
that irrespective of the model lagged per capita newspaper circulation
exerts a negative significant impact on the growth in the size of the hidden
economy. Our results also demonstrate that the growth in the size of the
hidden economy is around 1 per cent significantly lower in the election
year compared to a no-election year. This indicates that during the time of
the election the government tries to curb the size of the hidden economy.
We provide the evidence that the growth in the size of the hidden economy
is positively associated with the majority size. The coefficient associated
with coalition governments is positive and significant. This may be due
to the fact that coalition governments are more corrupt since they are
short-lived anyway. Lastly, we also note that both the War Dummy and
Emergency Dummy enter with positive coefficients; however, the impact is
not significant. It is a common perception that during the emergency era,
a lot more restrictions were in place. This may explain the positive sign of
the coefficient.
We have used all elections. However, in India, elections can be dif-
ferentiated into scheduled elections and mid-term elections. Given the


Table 3.6 Regression results for the growth in the size of the hidden
economy (all elections)
Variable Coefficient t-statistics Coefficient t-statistics

Constant 0.091 3.456*** 0.087 7.267***
Lagged size of hidden economy −0.003 −0.196 – –
Election dummy −0.016 −1.718* −0.011 −2.397**
Election dummy*majority size 0.084 2.558** 0.088 3.653***
Majority size 0.075 2.794** 0.078 3.360***
Coalition government dummy 0.021 1.755* 0.022 2.481**
War dummy 0.001 0.118 0.001 0.126
Emergency dummy 0.003 0.692 0.003 0.770
Lagged per capita newspaper −1.014 −2.230* −1.093 −5.530***
circulation
No. of observations 36 36
R2 0.697 0.696
p-value p-value
F (8,27) [F (7,28)] 7.758 0.000 9.178 0.000
Ljung Box Q-Statistics (1) 0.013 0.909 0.011 0.916
Notes: t-statistics are calculated based on standard errors adjusted for heteroskedasticity
and autocorrelation.
* denotes significance at 10% level, ** at 5% level and *** at 1% level using a two-tailed
t-test. Ljung-Box Q Statistics for serial correlation are calculated based on estimated
residuals and reported for lag 1, and for lag 4. They are distributed as c2 distribution with 1
and 4 degrees of freedom, respectively.
time-span of our data, ten elections were held: six are scheduled while
the rest are mid-term in nature. Hence, we estimate Equation (3.4) using
scheduled elections only. We report our results in Table 3.7. Our reported
results in Table 3.6 remain the same. The election variable enters with a
negative coefficient. Second, the effects of coalition government fragmen-
tation do not depend on how the election year dummy is introduced. The
variable lagged per capita newspaper circulation continues to exert a nega-
tive significant impact on the growth in the size of the hidden economy.9
Given these results, we ran another regression where we included the
mid-term election as a separate variable along with the scheduled election
to differentiate the impact of these two different types of elections.10 Please
see Table 3.8 for this. Here we obtain that the coefficient of scheduled elec-
tion is negative and significant, however, mid-term election enters with a
positive insignificant coefficient. We also note that the interaction term
(Mid-Term Election * Majority Size) is negative and significant in the case


economy (scheduled elections only)

Constant 0.097 4.247*** 0.086 7.263***
Scheduled election dummy −0.011 −1.571 −0.012 −1.792**
Scheduled election 0.078 2.393** 0.085 3.013***
dummy*majority size
Majority size 0.076 2.893** 0.083 3.935***
Coalition government dummy 0.020 1.679* 0.024 2.371**
War dummy 0.002 0.474 0.002 0.521
Emergency dummy 0.004 0.977 0.003 0.859
Lagged per capita newspaper −0.914 −2.283** −1.091 −5.557***
circulation
R2 0.692 0.690
p-value p-value
F (8,27) [F (7,28)] 7.593 0.000 8.910 0.000
of mid-term elections and positive and significant with scheduled elections

(Scheduled Election * Majority Size). The variable Majority Size continues
to be positive and significant. The effects of coalition government remain
the same. Still, the variable lagged per capita newspaper circulation is
negative and significant.11
We also note that it would be interesting to analyze the stability of our
results regarding the political variables across the entire period, or whether
they occur only after 1977, the year when the Indian National Congress
Party (the largest party in India at that time) lost power for the first time
since Indian independence. However, given the sample size, we could
not experiment with this. We have tried to incorporate a dummy to see
whether this has any effect. In all cases, the coefficient associated with this
variable is negative but insignificant and our results in terms of other vari-
ables remain qualitatively unchanged.12 Furthermore, we do not assign


economy (differentiating scheduled and mid-term elections)

Constant 0.096 4.170*** 0.086 6.874***
Scheduled election dummy −0.011 −1.833* −0.012 −1.894*
Scheduled election 0.076 2.514** 0.083 3.210***
dummy*majority size
Mid-term election dummy 0.004 0.963 0.004 0.996
Mid-term election −0.075 −2.456** −0.076 −2.556**
dummy*majority size
Majority size 0.076 2.514** 0.083 3.940***
Coalition government dummy 0.017 1.276 0.020 1.900*
War dummy 0.002 0.440 0.002 0.476
Emergency dummy 0.004 0.876 0.003 0.765
Lagged per capita newspaper −0.913 −2.233** −1.084 −5.325***
circulation
R2 0.697 0.695
p-value p-value
F (10,25) [F (9,26)] 5.749 0.000 6.584 0.000
any role for government ideology as a determinant of growth in the size of

hidden economy. This is because most political parties in India are organ-
ized based on linguistic, regional, religious and caste affiliations.
3.6 SUMMARY AND POLICY CONCLUSIONS
The basic findings of the chapter can be summarized as follows. The size
of the hidden economy in India has shown an increasing trend over the
years, varying from 8.9 per cent (1960/61) to about 23 per cent (1997/98).
An increased growth of newspapers translate to cleaner governance, for


example, to a lower amount of shadow economy activities in the economy.

Our findings are robust irrespective of the empirical specifications. During
the time of the election, the government tries to curb the size of the hidden
economy. We also note that a government that is approaching election
can be more active if it has a slight majority compared to a large majority.
This is a first and tentative result and has to be confirmed with additional
empirical and theoretical research.
NOTES
1. See for example, the Economic Journal, 109, no. 456, June 1999, the feature ‘contro-
versy: on the hidden economy’.
2. The literature on the ‘shadow’, ‘underground’, ‘informal’, ‘second’, ‘cash’ or ‘parallel’,
economy is increasing. Various topics, on how to measure it, its causes and its effect on
the official economy are analyzed. See, for example, survey type publications by Frey
and Pommerehne (1984); Thomas (1992); Schneider (1994a, 1994b, 1997, 1998); Loayza
(1996); Pozo (1996); Lippert and Walker (1997); Johnson, Kaufmann and Shleifer
(1997), and Johnson, Kaufmann and Zoido-Lobatón (1998); and for an overall survey
of the global evidence of its size in terms of value added Schneider and Enste (2000).
3. This definition is used, for example, by Frey and Pommerehne (1984), Lubell (1991) and
Schneider (1994a).
4. See Zellner (1970), Goldberger (1972), Jöreskog and Goldberger (1975).
5. Kaldor (1956) tried to estimate the size of the hidden economy in India, by estimating
the income that avoided the income tax.
6. Giles (1999b) has also used the same indicator.
7. See also the work by Banerjee and Iyer (2004) in this connection.
8. We have used the size of hidden economy as calculated by model 1. However, if we use
the others, the reported results remain qualitatively unchanged.
9. We do not report these results, however, they are available on request from the authors.
10. We have used mid-term elections without the lag.
11. We do not report these results, however, they are available on request from the authors.
12. We do not report the results, however, they are available on request from the authors.
REFERENCES
Acharya, S.N. (1985), Aspects of the black economy in India. Report of a Study
by National Institute of Public Finance and Policy, Ministry of Finance,
Government of India.
Ahrend, R. (2002), Press freedom, human capital and corruption. Mimeo.
DELTA, Paris.
Aigner, D., F. Schneider and D. Ghosh (1988), Me and my shadow: estimating
the size of the US hidden economy from time series data. In W.A. Barnett; E.R.
Berndt and H. White (eds), Dynamic Econometric Modeling. Cambridge, MA:
Cambridge University Press, pp. 224–243.
Bagachwa, M.S.D. and A. Naho (1995), Estimating the second economy in
Tanzania. World Development, 23, 1387–1399.
Banerjee, A. and L. Iyer (2004), History, institutions, and economic performance:


the legacy of colonial land tenure systems in India. Mimeo. Harvard Business
School, USA.
Besley, T. and R. Burgess (2000), Land reform, poverty reduction, and growth:
evidence from India. Quarterly Journal of Economics, 115, 389–430.
Besley, T. and R. Burgess (2002), The political economy of government respon-
siveness: theory and evidence from India. Quarterly Journal of Economics, 117,
1415–1451.
economy. The Economic Journal, 109, 348–359.
Bhattacharyya, D.K. and S. Ghose (1998), Corruption in India and the hidden
economy. Economic and Political Weekly, October 31, 2795–2799.
Campbell, J.Y. and P. Perron (1991), Pitfalls and opportunities: what macroecon-
omists should know about unit roots. NBER Macroeconomics Annual, 141–201.
Chopra, O.P. (1982), Unaccounted income: some estimates. Economic and Political
Weekly, April 24, 739–744.
Djankov, S., C. McLeish, T. Nenova and A. Shleifer (2002), Who owns the media?
Mimeo, World Bank, Washington DC.
for measurement. Review of Income and Wealth, 30, 1–23.
Giles, D.E.A. (1999a), Modelling the hidden economy and the tax-gap in New
Zealand. Empirical Economics, 24, 621–640.
Giles, D.E.A. (1999b), The rise and fall of the New Zealand underground
economy: are the responses symmetric? Applied Economics Letters, 6, 185–189.
Giles, D.E.A. and P.J. Caragata (2001), The learning path of the hidden economy:
tax and growth effects in New Zealand. Applied Economics, 33, 1857–1867.
economy. Canadian Tax Foundation, Toronto.
Goldberger, A.S. (1972), Structural equation models in the social sciences.
Econometrica, 40, 979–1001.
Government of India, Indian Public Finance Statistics. Ministry of Finance,
Department of Economic Affairs.
Government of India, Statistical Abstract. Various issues.
Gupta, P. and R. Mehta (1981), An estimate of underreported national income.
Journal of Income and Wealth, 5, 109–113.
sition. Brookings Papers on Economic Activity, Fall, Washington DC.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998), Regulatory discretion
and the unofficial economy. American Economic Review, 88, 387–392.
Jöreskog, K.G. and A.S. Goldberger (1975), Estimation of a model with multiple
indicators and multiple causes of a single latent variable. Journal of the American
Statistical Association, 70, 631–639.
Jöreskog, K. and J. Sorbom (1993a), LISREL 8: Structural Equation Modeling
With the Simplis Command Language. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Jöreskog, K. and J. Sorbom (1993b), LISREL 8: User’s Reference Guide. Hillsdale,
NJ: Lawrence Erlbaum Associates.
Kaldor, N. (1956), Indian tax reform: report of a survey. Ministry of Finance,
Government of India.


Kaufman, D. (1999), Economic reforms: necessary but not sufficient to curb

corruption, in ‘curbing corruption’. World Bank, Washington DC.
Lippert, O. and M. Walker (eds) (1997), The underground economy: global
evidences of its size and impact. Vancouver, BC: The Frazer Institute.
Lubell, H. (1991), The informal sector in the 1980’s and 1990’s. Paris: OECD.
MacKinnon, J.G. (1991), Critical values for cointegration tests. In R.F. Engle and
C.W.J. Granger (eds), Long Run Economic Relationships. Oxford, UK: Oxford
University Press, pp. 267–276.
Mogensen, G.V., H.K. Kvist, E. Körmendi and S. Pedersen (1995), The shadow
economy in Denmark 1994: measurement and results. Study no. 3. The
Poso, S. (ed.) (1996), Exploring the Underground Economy: Study of Illegal
Research.
Reserve Bank of India, Handbook of Statistics on the Indian Economy, various
issues.
Economics, 88, 643–668.
Schneider, F. (1994b), Measuring the size and development of the shadow
economy. Can the causes be found and the obstacles be overcome? In H.
Brandstaetter and G. Werner (eds), Essays on Economic Psychology. Berlin,
Heidelberg: Springer Publishing Company, pp. 193–212.
Schneider, F. (1997), The shadow economies of Western Europe. Journal of the
Institute of Economic Affairs, 17, 42–48.
Schneider, F. (1998), Further empirical results of the size of the shadow economy of
17 OECD-countries over time. Paper presented at the 54th Congress of the IIPF
Cordoba, Argentina. Department of Economics, University of Linz, Linz, Austria.
Schneider, F. and D. Enste (2000), Shadow economies: size, causes, and conse-
quences, the Journal of Economic Literature, 38, 77–114.
Smith, J.D. (1985), Market motives in the informal economy. In W. Gaertner and
A. Wenig (eds), The Economics of the Shadow Economy. Heidelberg: Springer
Publishing Company, pp. 161–177.
Stapenhurst, R. (2000), The media’s role in curbing corruption. World Bank
Institute Working Paper, World Bank, Washington DC.
Tanzi V. and L. Schuknecht (1997), Reconsidering the fiscal role of government:
the international perspective. American Economic Review, 87, 164–168.
Thomas, J.J. (1992), Informal Economic Activity. LSE, Handbooks in Economics,
Zellner, A. (1970), Estimation of regression relationships containing unobservable
independent variables. International Economic Review, 11 (3), 441–454.


APPENDIX 3.1 DATA

Annual Data on India’s currency in circulation, money supply (M3),
real gross domestic product (GDP), population, and consumer price
index (CPI) have been compiled from various issues of the Handbook of
Statistics on the Indian Economy, published annually by the Reserve Bank
of India. The data of the tax variables used as causal factors in the empiri-
cal MIMIC analysis, have been obtained from several issues of the Indian
Public Finance Statistics, Ministry of Finance, Department of Economic
Affairs, Government of India. The data on the number of public employ-
ees have been collected from some volumes of the publication Statistical
Abstract of the Government of India. Data on India’s total newspaper cir-
culation were obtained from Press in India, published by the Registrar of
Newspapers on behalf of the Ministry of Information and Broadcasting,
Government of India. The political variables have been arranged and
compiled from a wide range of sources. India’s election and the data on
coalition have been constructed with the help of information on elections
from the book India Decides, Elections 1952–1995 and from the website of
Election Commission of India (www.eci.gov.in).


4 Size, development and perception of
the shadow economy in Switzerland
Christoph A. Schaltegger
4.1 INTRODUCTION
Size, development, causes and problems of the shadow economy have

been the cause of controversy for years. In an interview on 10 January
2000, Milton Friedman argued: ‘The black market was a way of getting
around government controls. It was a way of enabling the free market to
work. It was a way of opening up, enabling people’ (Public Broadcasting
Service, 2000). On the other hand, Doris Leuthard, minister of economic
affairs in Switzerland, intensified her fight against black market activities
with the words:
Moonlighters are cheating the government by evading taxes and social security
contributions. [. . .] I am aware of the fact that moonlighting can never be exter-
minated completely. It is in the nature of the shadow economy that authori-
ties can never eliminate all forms of illegal work. Nevertheless, a new law is
important for two reasons: First of all, moonlighting represents a threat for the
protection of the employees. Often, black workers face lower wages and have
to work under precarious and often uncertain conditions. He or she is exposed
to the arbitrariness of the employer. Second, the new law will secure our accom-
panying measures to the bilateral treaty with the EU in the case of ‘freedom of
movement and residence’ (Leuthard, 2007).
Independent of different points of view regarding the impact of the

shadow economy on the welfare of society, according to several surveys
the shadow economy grew strongly in almost all OECD countries during
the last 30 years (see for example Thomas, 1992; Johnson, Kaufmann
and Zoido-Lobatón, 1998a, b; or Schneider and Enste, 2000, 2002). Since
2004, the size of the shadow economy has been more or less stable; some
countries even showed a decrease in the size of the unofficial economy.
With the current global economic crisis, the size of the shadow economy
is expected to grow again, however (Schneider, 2009). Very much the
same applies for the German-speaking countries. In Switzerland, the size
150


The shadow economy in Switzerland 151
of the shadow economy is traditionally estimated well below 10 per cent

of the official gross domestic product. However, the extent of the shadow
economy has nearly tripled during the last 30 years. Each twelfth franc
earned in the Swiss economy passes off the official economy with negative
effects for the social security contributions raised (Schneider, Torgler and
Schaltegger, 2008).
This chapter focuses on the Swiss case. After a review of the literature
(Section 4.2), we report on recent developments in Switzerland to tackle
the black economy (Section 4.3). Section 4.4 summarizes the results of
empirical studies concerning the size and the development of the shadow
economy in Switzerland (Section 4.5). Thereafter, we discuss the results on
the public perception of moonlighting in Switzerland. Section 4.6 offers
some concluding remarks.
4.2 REVIEW OF THE LITERATURE
The problem of moonlighting is not new in either the literature or politics.

In Switzerland, moonlighting and tax evasion were already subject to a
report by the federal council to the parliament in 1962.1 Inspired by exten-
sive scientific work at the University of Zurich, Hans Schmid and Kaspar
Villiger, both members of Parliament, asked for a second, more compre-
hensive report in 1983 and in 1986. At the same time, the Swiss Science
Foundation supported studies by Weck-Hannemann, Pommerehne and
Frey (1984, 1986) to shed more light on the hidden economy. They con-
cluded that the size of the shadow economy in the early 1980s accounted
for between 3 and 6 per cent of GDP in Switzerland with an increasing
trend from a level of 1 per cent in the early 1960s (see Table 4.1).
This study formed the basis for a comprehensive report by the federal
council to the parliament in 1987.2 The latest report by the federal council
was released in 2002,3 this time with concrete proposals for a new law,
which came into force as of 2008.
Very much the same applies to the literature. Among the first, Tanzi
(1980, 1982, 1983) offered estimates of the underground economy in
the United States between the years 1930–80. One of the first survey
publications was conducted by Frey and Pommerehne (1984). Since
then, an ever increasing amount of empirical and theoretical studies
has emerged. Schneider and Enste (2002) provide a recent and compre-
hensive survey. Despite the enormous extent of the literature, there is
still much controversy4 about definitions, empirical methods and policy
recommendations.5


Table 4.1 Size of the shadow economy in percent of GDP in 1960 and
1978
1960 1978
Sweden 5.4 13.2
Belgium 4.7 12.1
Denmark 3.7 11.8
Italy 4.4 11.4
Netherlands 5.6 9.6
France 5.0 9.4
Norway 4.4 9.2
Austria 4.6 8.9
Canada 5.1 8.7
Germany 3.7 8.6
USA 6.4 8.3
UK 4.6 8.0
Finland 3.1 7.6
Ireland 1.7 7.2
Spain 2.6 6.5
Switzerland 1.1 4.3
Japan 2.0 4.1
Source: Weck-Hannemann, Pommerehne and Frey (1984).
4.2.1 Definition
One of the most widely used definitions of the shadow economy uses
‘all economic activities which contribute to the officially calculated (or
observed) Gross National Product, but are currently unregistered’ (Feige,
1989, 1994). Smith (1994) proposes another definition: ‘market-based
production of goods and services, whether legal or illegal that escapes
detection in the official estimates of GDP’. Schneider (2005) offers a more
systematic and broad view on legal and illegal underground activities (see
Table 4.2).
4.2.2 Measurement
There is also much controversy about the adequate measurement of the

shadow economy. There are basically two concepts: a direct and an indi-
rect measurement approach. Surveys are examples of direct methods.6
From surveys we know that moonlighting is deeply anchored in the
population. An inquiry by the German institute TNS EMNID from the


Table 4.2 A taxonomy of types of underground economic activitiesa
Type of Monetary Transactions Non Monetary Transactions

Activity
Illegal Trade with stolen goods; drug Barter of drugs, stolen goods,
Activities dealing and manufacturing; smuggling etc. Produce or
prostitution; gambling; smuggling; growing drugs for own use.
fraud; etc. Theft for own use.
Tax Evasion Tax Avoidance Tax Evasion Tax Avoidance
Legal Unreported Employee Barter of All do-it-
Activities income from discounts, legal services yourself work
self-employment; fringe benefits and goods and neighbor
Wages, salaries help
and assets from
unreported work
related to legal
services and
goods
a
year 2003 says that over half of all interviewed persons (55 per cent) are
not concerned when poor households earn some additional money in the
shadow economy, for example, by cleaning. A quarter of the Germans (26
per cent) also see no problem in moonlighting, if this helps to construct
a residential building for private purposes. The fact that, for example,
journeymen earn some additional black money is, for 23 per cent of the
persons asked in East Germany and for 17 per cent in West Germany,
absolutely unproblematic. Only 36 per cent of all Germans asked reject
moonlighting in principle (Schneider, Torgler and Schaltegger, 2008).
In a survey by the Rockwool Foundation, Copenhagen, even every tenth
person confessed to having worked in the black market during the past 12
months (Feld and Larsen, 2005). Mostly it does not stop with helping out
a friend spontaneously, a single time. Rather each moonlighter works on
average approximately 428 hours in the year in the black market – which
corresponds to a quarter of the official working time. Experts, who rely on
survey data, estimate that in Germany moonlighting corresponds to some
4 per cent of the official GDP. However, Schneider and Enste (2002) point
to various disadvantages of survey methods, namely the dependence on
the respondents’ willingness to cooperate and the problem of translating
the answers into precise monetary numbers of the shadow economy.


Indirect approaches measure the extent of the shadow economy by

indicators. One such method estimates the difference between national
expenditures and income statistics as a proxy for the shadow economy.
The argument is that shadow activities can circumvent their ‘traces’ in the
income statistics but cannot hide expenditures. Petersen (1982) and Smith
(1985) use the approach in their studies for Germany and Great Britain.
However, discrepancies in official statistics are not only due to hidden
activities but also reflect all errors in the official statistics that occur in all
countries. The same can be said about the approach that estimates discrep-
ancies between the official and actual labour force (Thomas, 1992).
The transaction approach developed by Feige (1994) uses the quantity
equation MV 5 PT: total volume of payments in a society must equal
the total volume of monetary transactions. The discrepancy between esti-
mated payments (MV) and estimated transactions (PT) provides an idea
of unrecorded transactions and thus the shadow economy (Feige, 1994).
The method is technically very demanding and is therefore not used by
many authors.
The currency demand approach particularly used by Tanzi (1980, 1983)
but also by Weck-Hannemann, Pommerehne and Frey (1986) makes the
assumption that transactions in the hidden sector take place in cash. An
increase in the size of the shadow economy will therefore increase the
demand for currency. When estimating the currency-demand function,
controlling for other factors such as income changes, interest rates and
habits, the ‘excess demand’ gives an indication of the size of the shadow
economy.
The dynamic multiple-indicators multiple-causes model (DYMIMIC)
is basically a structural equations model specifying causal relationships
to the unobserved variable, which is assumed to be influenced by a set of
indicators. In our case, the unobserved variable is the size of the shadow
economy. The shadow economy is assumed to be influenced by a set
of causes like the tax burden, the regulation density or the existing tax
morale, the effects of which are captured by the structural relationship of
the shadow economy on predicting variables like monetary, labour market
or economic growth indicators. There is a great amount of literature on
the DYMIMIC approach and the possible indicators and causes of the
shadow economy (OECD, 2002; Schneider, 2005).
4.2.3 Size and Development of the Shadow Economy
The calculations of the extent of the shadow economy presented here

are based on a combination of the currency demand approach with the
DYMIMIC procedure. Table 4.3 shows the results of this combined


Table 4.3 The size of the shadow economy in per cent of the official GDP for 21 OECD-countries from 1989/90 to 2009
OECD 89/90 94/95 97/98 99/00 01/02 2003 2004 2005 2006 2007 2008 2009
Australia 10.1 13.5 14.0 14.3 14.1 13.7 13.2 12.6 11.4 11.7 10.6 10.9

Belgium 19.3 21.5 22.5 22.2 22.0 21.4 20.7 20.1 19.2 18.3 17.5 17.8
Canada 12.8 14.8 16.2 16.0 15.8 15.3 15.1 14.3 13.2 12.6 12.0 12.6
Denmark 10.8 17.8 18.3 18.0 17.9 17.4 17.1 16.5 15.4 14.8 13.9 14.3
Germany 11.8 13.5 14.9 16.0 16.3 17.1 16.1 15.4 15.0 14.7 14.2 14.6
Finland 13.4 18.2 18.9 18.1 18.0 17.6 17.2 16.6 15.3 14.5 13.8 14.2
France 9.0 14.5 14.9 15.2 15.0 14.7 14.3 13.8 12.4 11.8 11.1 11.6
Greece 22.6 28.6 29.0 28.7 28.5 28.2 28.1 27.6 26.2 25.1 24.3 25.0
England 9.6 12.5 13.0 12.7 12.5 12.2 12.3 12.0 11.1 10.6 10.1 10.9
Ireland 11.0 15.4 16.2 15.9 15.7 15.4 15.2 14.8 13.4 12.7 12.2 13.1
Italy 22.8 26.0 27.3 27.1 27.0 26.1 25.2 24.4 23.2 22.3 21.4 20.0
155
Japan 8.8 10.6 11.1 11.2 11.1 11.0 10.7 10.3 9.4 9.0 8.8 9.5
Netherlands 11.9 13.7 13.5 13.1 13.0 12.7 12.5 12.0 10.9 10.1 9.6 9.9
New Zealand 9.2 11.3 11.9 12.8 12.6 12.3 12.2 11.7 10.4 9.8 9.4 10.2
Norway 14.8 18.2 19.6 19.1 19.0 18.6 18.2 17.6 16.1 15.4 14.7 15.3
Austria 6.9 8.6 9.0 9.8 10.6 10.8 11.0 10.3 9.7 9.4 8.1 8.5
Portugal 15.9 22.1 23.1 22.7 22.5 22.2 21.7 21.2 20.1 19.2 18.7 19.5
Sweden 15.8 19.5 19.9 19.2 19.1 18.6 18.1 17.5 16.2 15.6 14.9 15.4
Switzerland 6.7 7.8 8.1 8.6 9.4 9.5 9.4 9.0 8.5 8.2 7.9 8.3
Spain 16.1 22.4 23.1 22.7 22.5 22.2 21.9 21.3 20.2 19.3 18.7 19.5
USA 6.7 8.8 8.9 8.7 8.7 8.5 8.4 8.2 7.5 7.2 7.0 7.6
Unweighted average 12.7 16.2 16.8 16.8 16.7 16.5 16.1 15.6 14.5 13.9 13.3 13.8
Notes: Calculations are based on a combination of the currency demand approach with the DYMIMIC procedure.

Source: Schneider, Torgler and Schaltegger (2008); Schneider (2009)
27/07/2011 16:16
method. The span in the 21 OECD countries listed reaches from nearly
9 per cent to approximately 30 per cent. South-European countries with
approximately 25–28 per cent are at the top, followed by Scandinavian
countries with 17–19 per cent. In the midfield are countries such as
Germany and France with some 14–16 per cent, while Switzerland and
Austria exhibit a comparatively small shadow economy with approxi-
mately 10 per cent of the official GDP.
Table 4.3 additionally shows that the extent of the shadow economy has
clearly increased in most countries over the last 20 years. On the OECD
average, the shadow economy increased by more than 24 per cent until
2000. However, in the recent past the trend changed to some extent: Table
4.3 shows that the shadow economy has been stabilizing or even declin-
ing since the end of the 1990s in most OECD countries. The unweighted
average of the shadow economy in the 21 OECD countries in the year
1999/2000 amounted to 16.8 per cent, whereas this declined to 13.3 per
cent in the year 2008. With fiscal packages to stimulate the economy in
2009, Schneider (2009) expects the shadow economy to increase again up
to 13.8 per cent on the OECD average.
A broader overview can be found in Schneider (2005) with estimates for
the shadow economy all over the world for 145 countries. There are very
large worldwide differences. On the African continent, Schneider (2005)
estimates the size of the shadow economy as 43 per cent of the official
GDP in 2002/2003 (unweighted average). For Asian countries, the average
size of the shadow economy is estimated as 30 per cent of the official GDP.
For Central and South American countries, the size accounts for 43 per
cent and 40 per cent for transition countries in the former Soviet Union
and Central Europe.
4.2.4 Causes
The literature addresses the causes for the growth of the informal sector
through different channels: economic, sociological and psychological factors
(Schneider and Enste, 2000). According to Schneider (2001), the causes of
the shadow economy can be divided into the following three categories:
● Factors in connection with the general expansion of the state

activity,
● Factors in connection with changes in the job market,
● Factors in connection with general changing values in society.
The assumption is that the decision to engage in activity in the shade

of the official economy can be explained by individual cost–benefit


considerations. The higher the costs of working in the official economy

and the smaller the costs of working in the unofficial economy, the more
attractive the shadow economy will be. Many studies show that the fol-
lowing eight causes represent crucial factors in explaining the shadow
economy.
1. Tanzi (1983, 1999), Giles (1999) and Schneider and Enste (2002)
come to the conclusion that the tax burden and social security con-
tributions exert a significant influence on the extent of the shadow
economy. It is immediately obvious that taxes and social security
contributions distort labour–leisure choices and hence create a bias
towards the untaxed shadow economy. A rise in the tax burden
increases the incentives for performing economic activities in the
shade of the official economy. Though, the problem not only holds
true for high-income earners, it is also important for low-income
earners. Social security systems, which cause marginal tax rates of
100 per cent and more, are not only confronted with the problem
of the ‘poverty trap’, in order not to lose transfer eligibility, the
transfer recipients have strong incentives to search for work in the
informal economy (for an overview of the Swiss case, see Leu et al.,
2008).
2. The regulatory framework of an economy exerts a crucial influence
on the extent of the shadow economy. Regulations cause both costs to
those who have to respect them as well as to those who have to super-
vise these rules and legal provisions. In order to escape the costs of
regulations, there is an incentive to move into the unregulated shadow
economy. Interestingly enough, it is rather the increased costs of the
law enforcement that explain the rising size of the shadow economy
than the regulatory density, according to Johnson, Kaufman and
Zoido Lobatón (1998b).
3. The perceived burden with taxes and social security contributions
are further crucial factors of influence for the extent of the shadow
economy. Sociological experiments show that the perceived burden
can be as important as the true burden, as well as the relative changes.
Thus, frequent changes of reforms push many into the shadow
economy because the reliability and the stability of the political system
and the legal framework are questioned. Something similar is also
valid for the complexity of the tax system. Schneider and Neck (1993)
show that the complexity of the tax system can more than compensate
for the influence of the absolute tax burden and the regulatory density
on the shadow economy. Complex tax systems are a crucial explana-
tory factor for the extent of the shadow economy.


4. Changing values of a society are often mentioned as an important

reason for the increase of the shadow economy. In this respect ethical
and moral convictions towards unofficial activities together with
aspects such as tax morality or acceptance of national standards are
of importance (Torgler, 2004; Kirchgässner, 2006).
5. Labour market regulations such as work time reduction, early retire-
ment or minimum wages are recognized in various studies as crucial
aspects of the shadow economy. The more strict the labour market
regulations, the stronger the incentives for moving into the shadow
economy (Deutsches Institut für Wirtschaftsforschung DIW, 1998;
Hunt, 1999).
6. The provision and quality of public goods are important for the
explanation of the size of the shadow economy. If the public goods
including legal rules or the behaviour of the authorities represent
an adequate return of the government for the taxes, the loyalty
of the citizens is higher and consequently the size of the shadow
economy smaller (Schneider and Enste, 2000; Frey and Feld 2002,
within the context of tax evasion). As May, Pyle and Sommers
(2002) point out, it is important to distinguish between transition
and non-transition countries. For transition economies, political
stability is a crucial aspect for the size of the unofficial economy.
In contrast, regulatory aspects, greater voice and accountability,
heightened government effectiveness and a stronger rule of law
matter very little for countries in transition as opposed to non-
transition countries.
7. The level of personal income exerts a significant influence on the extent
of the shadow economy. On one hand, high-income earners look for
work in the shade of the official economy because their tax burden is
very high (this is valid also for the so-called marriage penalty). On the
other hand, the incentive to work in the shadow economy is strong for
the low-wage segment due to high marginal taxes in connection with
social transfers (Schneider and Enste, 2000).
8. In the long run, a snowball effect is observed with the shadow
economy. If neighbours and friends work in the shadow economy,
the willingness to do likewise is higher as many experimental studies
show: an imitation effect steps in (Bajada, 1999).
Some studies point out that the institutional conditions represent also an
important factor for this development. In this respect, the basic condi-
tions of the political decision-making procedure are crucial. These define
the rules for the politico-economic players and the relationship between
citizens and the government (Schneider and Enste, 2000, 2002; Tanzi,


2000; Torgler and Schneider, 2007). All in all, the extent and the causes
of the shadow economy have been quite well investigated during the last
few years. An important insight of this literature is: if we want to suc-
cessfully fight the shadow economy, we must concentrate on the causes.
Pursuit and higher penalty clauses fight the symptoms and are therefore
limited in scope. In addition, such measures are costly and can cause the
opposite of what is intended. Often, a more effective strategy is a general
simplification of the laws, namely of the tax and transfer system, the
delimitation of the state activities, particularly in the case of regulations
as well as an intensified participation of the citizens in the decision-
making procedures of the government. Such reforms form a promising
strategy for success to limit the shadow economy (Schneider, Torgler and
Schaltegger, 2008).
4.3 RECENT DEVELOPMENTS IN SWITZERLAND
Against the background of worldwide increasing sizes of the shadow

economy, Swiss politics has intensified its fight against moonlighting. At
the end of the 1990s the Federal Council assigned the Secretariat of State
for Economic Affairs (SECO), to prepare a package of adequate measures.
The result was the federal law to fight the black economy, which
came into force on 1 January 2008 (Bundesrat, 2002).7 According to the
Secretariat of State for Economic Affairs the law aims at three goals:
(1) new incentives through administrative relief during the registration
of work with the social security and tax authorities, (2) enforced sanc-
tions against moonlighting, in addition to new control authorities in the
cantons, and (3) the sensitization of the population to the problems of
moonlighting.8
More precisely, the new law simplifies the accounting procedure
for household and temporarily limited activities with the social secu-
rity administration and by withholding tax in order to decrease
bureaucratic complexity. It also forces the cantons to designate their
own control authority and these authorities are obliged to exchange
their results. Furthermore, moonlighting sanctions are being inten-
sified; companies involved in black activities are excluded from
public procurement or from financial subsidies and assistance of the
government. Finally, the general public was informed about the new
law by public campaign activities.9 Thus, the package of measures to
tackle the shadow economy covers a mix of sensitization, incentives and
repression.


4.4 SIZE AND DEVELOPMENT OF THE SHADOW

ECONOMY IN SWITZERLAND
What was and will be the effect of these new measures?10 To answer the
question, we have to take a broader look at the data. As we have seen
in Section 4.3, in the German-speaking countries, the shadow economy
strongly increased during the last 30 years. Table 4.4 and Figure 4.1 show
the development of the estimated size of the shadow economy for the
three German-speaking countries of Germany, Austria and Switzerland
during the period 1975–2009. For Germany, the calculations show nearly
a tripling, while the official GDP only grew approximately 62 per cent.
The shadow economy thus grew far faster than the official economy. In
Austria, the unofficial economy grew from 2 per cent of the official GDP
to some 8 per cent. Something similar also occurred in Switzerland: while
Table 4.4 The shadow economy in Germany, Austria and Switzerland

from 1975 to 2009
Year Germany Billion € Austria Billion € Switzerland Billion CHF.

1975 29.6 0.9 12
1980 80.2 2.0 14
1985 102.3 3.9 17
1990 147.9 7.2 22
1995 241.1a 12.4 25
1996 257.6a 14.6 27
1997 274.7a 16.0 29
1998 280.7a 16.9 30
1999 301.8a 18.2 32
2000 322.3a 19.8 35
2001 329.8a 21.1 37.5
2002 350.4a 21.8 38.7
2003 370.0a 22.5 39.4
2004 356.1a 23.0 39.5
2005 346.2a 22.0 38.7
2006 345.5a 21.2 37.0
2007 349.0a 20.80 36.8
2008 346.8a 19.92 35.4
2009 351.8a 20.5 36.4
Notes: a Since 1995 the numbers apply to a unified Germany.

Calculations are based on a combination of the currency demand approach with the
DYMIMIC procedure.
Source: Schneider, Torgler and Schaltegger (2008) and Schneider (2009)


18
16
Shadow economy in % of GDP
14
12
10
4 Germany
Austria
2
Switzerland
0
1975 1985 1995 1997 1999 2001 2003 2005 2007 2009
Figure 4.1 The development of the shadow economy in Germany, Austria

and Switzerland from 1975 to 2009 in per cent of GDP
the shadow economy increased considerably from some 3 per cent to 8 per
cent, the official economy grew only by approximately 38 per cent. This
development documents that both the willingness for moonlighting and its
influence on the economy have risen strongly.
The extent of the shadow economy grew strongly during the 1970s,
1980s and 1990s. From 2003/04 on, the size of the shadow economy in per
cent of the official GDP decreased for the first time in all three German-
speaking countries to some extent until the recent economic down-turn.
In Switzerland, the shadow economy recently increased slightly according
to the calculations by Schneider (2009) from 35.4 billion CHF in the year
2008 to 36.4 billion CHF in the year 2009. This corresponds to an increase
of one percentage point. Once more, all three German-speaking countries
face a similar development. The increase can be attributed to the global
financial and economic crises and the government actions to support
aggregate demand.
In the case of Switzerland, the date of the final approval of the new law
to fight the black economy is of interest; this was given by parliament on
17 June 2005. Due to the enlarged government competencies including the
effect of the fight of the shadow economy and relief for household services
in the official economy, the shadow economy decreased in the year 2004/05
and then onwards until the recent economic recession (see Table 4.3 and
Figure 4.1). Note that the effect is observable even before the enactment


Table 4.5 Effects of the federal law in Switzerland, effective since January
2008
Measures Decrease of the shadow

economy; basis 2005
Administrative easing of social security contributions – 900 million CHF
and with the withholding tax by introduction of a
simplified declaration and accounting procedure for
smaller activities (e.g. household, temporary limited
activities)
Obligation of the cantons to designate a cantonal – 300 million CHF
control authority with intensified control competence
Obligation to exchange control results between the – 200 million CHF
authorities and organs involved
Strengthened sanctions; e.g. exclusion of public – 300 million CHF
procurement and shortening of financial subsidies
and assistances
Source: Schneider, Torgler, Schaltegger (2008)
of the new law in 2008. The shadow economy decreased by some 3.0 per
cent of GDP and reached a value of 8 per cent of the official GDP. In
2009, moonlighting is expected to rise again to 8.3 per cent of GDP with
the impact of the three stabilization packages of the government to fight
the economic recession, which increased the potential for moonlighting
(Schneider, 2009).11
Table 4.5 shows the effects of the main measures of the new federal
law as published in Schneider, Torgler and Schaltegger (2008) based on
the DYMIMIC approach. The simulations for Switzerland use a similar
approach to that used by Schneider (2004) to calculate the differential
effects of different policy measures in Germany. The results show that the
administrative relief has by far the largest effect on the reduction of the
shadow economy.
What can be said concerning the size of the shadow economy in dif-
ferent economic sectors? It is immediately understandable that differ-
ent economic sectors are affected by the shadow economy differently.
Obviously, in some jobs the discretionary leeway of hiding income or
working without making out an invoice is larger than in other jobs.
Weck-Hannemann, Pommerehne and Frey (1984) tried to get a picture
of the structure of the shadow economy in Switzerland. They con-
ducted a survey among experts from public administration, unions, busi-
ness associations and other interest groups. The survey concluded that


Table 4.6 Size of the shadow economy, Swiss economic sectors, 2006
Economic sectors in % of GDP Billion CHF

Construction industry and crafts enterprise 36% 13.3
(inclusive repairs)
Other trade and industrial companies 17% 6.3
(automobiles, machinery, etc.)
Services industry (hotel, restaurants, etc.) 18% 6.7
Maintenance and entertainment industries 14% 5.2
Other industrial companies and services 15% 5.5
(assistance, hairdressing, baby-sitting)
Whole shadow economy 100% 37
Source: Schneider, Torgler and Schaltegger (2008)
moonlighting is most common in house-keeping and cleaning services,

in the hotel and restaurant industry, the building and construction and
the maintenance sectors. Less common is moonlighting in trade, educa-
tion, culture, recreation, health and beauty businesses as well as in textile,
clothing and furniture industries, according to the experts interviewed.
Little moonlighting is assumed by the experts in the metal and machin-
ery, food, chemical and electricity industries. All in all, according to
the answers, an overall size of 3 per cent of GDP is estimated by Weck-
Hannemann, Pommerehne and Frey (1984).
Table 4.6 shows estimates of the size of the shadow economy shown in
Schneider, Torgler and Schaltegger (2008). The results differ somewhat
to the survey results in the early 1980s by Frey, Weck-Hannemann and
Pommerehne (1984). Econometric estimates by Schneider (2009) indicate
that the construction industry and crafts enterprises are more exposed to
black activities than other branches of the economy in Switzerland. The
construction industry and handicrafts are the largest shadow economy
industries, followed by important ‘other trades and industrial companies’,
and services enterprises (hotel, restaurants, etc.). Thereafter, services
(for example, hairdressing, baby-sitting) as well as the maintenance and
entertainment industries are exposed to the shadow economy.
Are there any important differences within Switzerland? Figure 4.2
shows the development of the shadow economy for the 26 Swiss cantons
over the time from 1990 to 2000. Schneider (2009) has calculated the
cantonal shadow economies by three steps: first, a currency demand equa-
tion for Switzerland was estimated. Second, the overall size of the Swiss
shadow economy was derived from the currency-demand equation for the
years 1990, 1995 and 2000. The resulting values in per cent of GDP are:


Jura
Genf
Neuenburg
Wallis
Waadt
Tessin
Thurgau
Aargau
Graubünden
St. Gallen
Appenzell I.Rh.
Appenzell A.Rh.
Basel-Stadt 2000
Basel-Landschaft 1995
Basel-Stadt 1990
Solothrun
Freiburg
Zug
Glarus
Nidwalden
Obwalden
Schwyz
Uri
Luzern
Bern
Zürich
0.00% 2.00% 4.00% 6.00% 8.00% 10.00% 12.00%
Source: Schneider, Torgler and Schaltegger, 2008
Figure 4.2 Size of the shadow economy among Swiss cantons, 1990–2000
1990: 6.17 per cent; 1995: 6.41 per cent; 2000: 7.98 per cent. In a third step
the values have been disaggregated according to the distribution of can-
tonal economic sectors – that are assumed to be differently affected by the
shadow economy – with the following five sectors:
1. Construction, craftsmanship including repairing 36%

2. Other craftsmanship and industrial firms (cars, machinery, 17%
etc.)
3. The whole service sector in hotels, restaurants, also 18%
catering, etc.
4. Entertainment sector, prostitution, gambling, etc. 14%


5. Other craftsmanship and all household services, cleaning, 15%

gardening, ironing, babysitting, etc.
Sectors 1, 2, 3 and 4 especially vary from canton to canton. The results

show an increase in almost all cantons. At the same time, however, strong
differences between the cantons are to be observed. Up till now the causes
for the differences between the cantons and the different developments in
the cantons have not been investigated more thoroughly. This remains a
promising field for further research.
4.5 PUBLIC PERCEPTION OF MOONLIGHTING IN

SWITZERLAND
How about the perception of the Swiss population? Do the Swiss perceive
the growing shadow economy over the last 30 years as a problem? Initially,
we would assume that government’s efforts for a new and stricter law to
enforce correct declaration is based on a broad public support. In fact,
according to a survey conducted among the Swiss population, almost half
of the interviewed persons (49 per cent) and the employers (43 per cent)
take moonlighting actually as a problem for the Swiss economy (see Table
4.7; GFS Bern, 2007). However, one third of the respondents answer that
moonlighting represents rather no problem. If we analyze the structure
of answers, some interesting results show up: men, young individuals
under 40 years, persons with high education and/or a very high household
income, employed persons as well as Swiss-Germans are significantly more
strongly inclined to view the problem less seriously. Thus, the attitude of
the inhabitants towards moonlighting depends on gender, age, education
level, income and employment status, as well as the language region in
Switzerland. With the employers, the problem perception is connected
with the firm size, education, political ideology and the language region.
Employers from large enterprises, those with a high education level, with
political left-wing positions, as well as, again, the ones from German-
speaking Switzerland judge moonlighting in Switzerland compared with
the reference groups as less problematic (GFS Bern, 2007).
The results in Table 4.7 correspond with the answers to the specific
survey question by GFS Bern (2007). In the question about the expo-
sure of different economic branches, the construction industry appears
again above all other industries. In the second place in the sectors in
which moonlighting is particularly common, the catering and the hotel
industry shows up followed by the land and forestry industry and private
households (GFS Bern, 2007, p. 34). Moonlighting is clearly noticed as


Table 4.7 Public perception of the shadow economy in Switzerland

according to a survey conducted by GFS Bern
Problems of the Swiss economy: the shadow economy in comparison
Survey question: The Swiss economy has to tackle different challenges. How
important is the shadow economy on a scale between 0 (no problem at all) and 10
(very important) according to your opinion?
Scale Answers in % of total Answers in % of total
interviewed persons interviewed employers
(n5915) (n5111)
0 (no problem) 3% 6%
1–2 10% 9%
3–4 18% 19%
5 20% 23%
6–7 20% 16%
8–9 15% 15%
10 (very important) 14% 12%
Source: GFS Bern (2007): report to the secretariat of state for economic affairs for the
public campaign on the shadow economy in Switzerland
a branch-specific phenomenon, whereby work- and personnel-intensive

industries with small qualification requirements seem to be particularly
concerned.
With the total value of the shadow economy and with the help of addi-
tional data such as the usual hourly wages in different economic sectors
it is possible to calculate the amount of working hours that would have
been necessary to carry out the black economy work (see also Schneider,
2004). Taking it one step further, it is also possible to calculate how many
workers would be employed, if the entire shadow economy could be con-
verted into official full-time job places.
Table 4.8 shows, according to such estimations, how in Switzerland
the number of these fictitious ‘full time moonlighters’ and foreigners,
who are illegally active in the shadow economy, developed during the
period 1995–2008. However, it has to be mentioned that the average black
economy worker in Switzerland only gains approximately two thirds of
his or her income by activities in the shadow economy (supplementary
income moonlighters).
In Germany, the number of these calculated ‘full-time moonlight-
ers’ amounted to 7.3 million employees in 1995 and increased up to 8.2
million in 2008. In addition, the illegally employed foreign workers are


Table 4.8 Development of full-time moonlighters in Switzerland, 1995 to

2008
In 1000 employees
Full-time domestic Illegally employed foreigners
moonlighters
1995 391 55
1996 426 61
1997 456 67
1998 462 69
1999 484 74
2000 517 79
2001 543 84
2002 556 88
2003 565 90
2004 560 89
2005 520 82
2006 493 78
2007 490 77
2008 471 74
Source: Schneider, Torgler and Schaltegger (2008)
not to be neglected: their number rose from 878 000 persons in the year
1995 to 955 000 persons in the year 2008. Also in Austria the increase was
considerable: in 1995, 575 000 full-time moonlighters were calculated, in
the year 2008 the number had risen to 679 000. The number of illegally
employed foreign persons increased from 75 000 persons in 1995 up to
93 000 persons in the year 2008. The same simulations can be made for
Switzerland. In the year 1995, in Switzerland, 391 000 employees were
engaged in the shadow economy, a number that rose up to 471 000
persons in 2008. With illegally employed foreign persons, this fictitious
value amounts to 74 000 persons for 2006 compared with 55 000 persons
in 1995 (Schneider, 2004; Schneider, Torgler and Schaltegger, 2008).
Similar results for Germany have been obtained by a survey by Feld and
Larsen (2005). The average time spent on moonlighting accounted in 2001
for 8 hours and 14 minutes per person per year and decreased to 7 hours
and 30 minutes in 2004.
Despite these facts and developments, the Swiss population does
not perceive the shadow economy as a major problem for their economy
in general and the labour market in particular compared with other
problems (see Table 4.9). According to the survey by GFS Bern (2007),


Table 4.9 Ranking of the problems for the Swiss labour market according
to a survey conducted by GFS Bern
Problems of the Swiss economy/labour market: the shadow economy in

comparison
Survey question: The Swiss economy / labour market has to tackle different
challenges. How important is the shadow economy in comparison to other
problems on a scale between 0 (no problem at all) and 10 (very important)
according to your opinion?
Scale Share of answers between Share of answers between
scale 6 to 10 in % of total scale 6 to 10 in % of total
population asked in the employers asked in the
survey (n 5 915) survey (n 5 111)
Bullying and stress in 64 56
the office
Moonlighting 49 43
Wage dumping by 50 37
foreign employees
Unemployment 45 27
Job security 32 29
Source: GFS Bern (2007, p. 39).
moonlighting is regarded as clearly less problematic than bullying

and stress in the office, which is regarded as a very large problem for two
of every three Swiss inhabitants. About 90 per cent of the respondents
think that moonlighting is harmful for employees due to insufficient
insurance protection and 80 per cent of the respondents think that the
shadow economy is negatively affecting the social security system. Also,
80 per cent answer that the shadow economy induces distortions at the
expense of those who work in the official economy. However, 69 per cent
of the interviewed persons agree that the shadow economy helps those
who cannot find a job in the official economy, and 50 per cent think that
everybody has worked at least once in the unofficial economy.
Although only a minority thinks moonlighting is actually a serious
problem for the labour market if compared with other problems and
the understanding of moonlighting is very problematic, a majority of
the inhabitants (57 per cent) describe moonlighting nevertheless as a
serious offence, which has to be penalized (61 per cent). One third of the
individuals asked have, however, no problems with moonlighting (see
Table 4.10).


Table 4.10 Is working in the shadow economy a severe offence? Survey

results for Switzerland according to gfs
Problems of the Swiss economy: the shadow economy in comparison
Survey question: What is your opinion:

(a) is moonlighting a severe offence?
(b) is moonlighting a severe offence and should it be penalized severely?
Scale (a) (b)
Not at all 14 15
Rather no 22 16
Don’t know 7 8
Rather yes 27 33
Absolutely yes 30 28
Source: GFS Bern (2007), p. 44
4.6 CONCLUSIONS
The shadow economy grew strongly in the last 30 years in almost all
OECD countries. Since 2004, however, this seems to have stabilized, some
countries even show a decrease in the shadow economy. With the current
economic downturn and the fiscal stimulus packages to support aggregate
demand, a rise of the shadow economy is to be expected, however.
In Switzerland, the size of the shadow economy is estimated to be under
10 per cent of the official GDP and is therefore well under the OECD
average of about 14 per cent. However, the extent of the shadow economy
increased considerably during the last 30 years, too. It is not surprising that
Swiss politics also calls for an intensified fight against moonlighting. The
result is the federal law to fight moonlighting, which came into force on
1 January 2008. It aims to simplify the accounting procedure for smaller
activities (for example, household and temporary limited activities) with
social security administration and introducing a withholding tax in order to
decrease the bureaucratic burden. It also forces the cantons to designate and
equip control authorities and the authorities and organizations involved
must exchange their control results. Intensified sanctions were also imple-
mented: the exclusion of public procurement or the shortening of public
subsidies and financial assistance for companies involved in black economy
activities. Finally, the general public was sensitized in an information cam-
paign about the new law. The package of measures against moonlighting
thus encompasses a mix of sensitization, incentives and repression.


According to first empirical results, these measures have proven to

work. The shadow economy decreased in the past few years to some
extent even though the recent government actions to stabilize the sluggish
economy in 2008 and 2009 have a potential to re-increase the extent of the
shadow economy.
Despite these political actions and even though a majority of the popu-
lation thinks moonlighting is problematic, the interviewed persons do not
perceive the shadow economy as a major problem for the economy and the
labour market in comparison to other problems, even though a majority
of the population qualifies moonlighting as a severe offence.
In any case, the causes of the shadow economy are various. The govern-
ment continues to expand its influence on the economy. The job market
changes at the same time. In addition, a change of values is observable.
Thus, the shadow economy remains a challenge for the Swiss govern-
ment as well as for many other governments. Recent studies point out
that the institutional framework represents an important factor in this
development. Therefore, the basic conditions of the political decision-
making procedures are crucial. These define the rules for the economic
players and the relationship between citizens and government. If we want
to fight the black economy successfully, we have to tackle the causes.
Pursuit and higher penalty clauses fight the symptoms – successfully, if
designed properly – but are limited in scope. Such measures are moreover
costly and can sometimes cause the opposite of what is intended. Often
more effective is a general simplification of the regulatory framework as
well as an intensified participation of citizens on government decisions.
Such reforms form a promising strategy for success to tackle the shadow
economy.
NOTES
1. See report of the Federal Council in response to Motion Eggenberger; BBl 1962 I 1057.
2. See report of the Federal Council from 9 June 1987 in response to Postulat Schmid
83.395 from 16 March 1983, and Interpellation Villiger 86.409 from 20 March 1986;
BBl 1987 II 1217.
3. See Report of the Federal Council from 16 January 2002; BBl 2002 3605.
4. See, for example, Dixon (1999).
5. Compare the different opinions of Tanzi (1999), Thomas (1999) and Pedersen (2003).
6. Another approach uses data from tax audits by the tax administrations to calculate the
so-called ‘tax gap’ (for empirical estimates for Switzerland see Frey and Feld, 2002, or
Pommerehne and Weck-Hannemann, 1989). In fact, this approach measures the size
of tax evasion and is therefore not directly comparable with measures of the shadow
economy.
7. See the federal law: Bundesgesetz über Massnahmen zur Bekämpfung der
Schwarzarbeit (Bundesgesetz gegen die Schwarzarbeit, BGSA) SR 822.41.


8. See the official homepage: http://www.evd.admin.ch/themen/00129/00769/index.

html?lang5de
9. See homepage of the campaign activity: www.keine-schwarzarbeit.ch. Examples of the
public campaign advertisements are in Appendix 4.1.
10. A first review on the effectiveness of the new law was released on 18 May 2009 (Bericht:
Umsetzung des Bundesgesetzes über Massnahmen zur Bekämpfung der Schwarzarbeit;
1.1.2008-31.12.2008).
11. For an empirical assessment of the Swiss stabilization policy since 1950, see Schaltegger
and Weder (2010).
REFERENCES
Bajada, C. (1999), Estimates of the underground economy in Australia. The

Economic Record, 75, 369–384.
Bundesrat (2002), Botschaft zum Bundesgesetz über Massnahmen zur
Bekämpfung der Schwarzarbeit. Bern, 16 January 2002.
DIW (Deutsches Institut für Wirtschaftsforschung) (1998), Arbeitszeitpräferenzen
in West- und Ostdeutschland 1997, Potenzial für Verkürzung der Arbeitszeit
gesunken. DIW-Wochenbericht, 65, 667–677.
Dixon, H. (1999), Controversy: on the use of the ‘hidden economy’ estimates.
Feige, E.L. (ed.) (1989), The Underground Economies. Tax Evasion and Information
Distortion. Cambridge, UK, Cambridge University Press.
Supplement to Public Finance/ Finances Publiques, 49, 119–136.
Feld, L. and C. Larsen (2005), Black activities in Germany in 2001 and in 2004.
Rockwool Foundation, Copenhagen.
Frey, B.S. and W.W. Pommerehne (1984), The hidden economy: state and pros-
pect for measurement. Review of Income and Wealth, 30/1, 1–23.
Frey, B.S. and L. Feld (2002), Deterrence and morale in taxation: an empirical
analysis. CESifo Working Paper no. 760.
GFS Bern (2007), Schlussbericht zur Planungsstudie Kampagne ‘Schwarzarbeit
in der Schweiz’ Bern. Available at http://www.keine-schwarzarbeit.ch/thema/
studien/index.html?lang5de
Giles, D.E.A. (1999), Measuring the hidden economy: implications for economet-
ric modelling. Economic Journal, 109, 370–380.
Hunt, J. (1999), Has work-sharing worked in Germany? Quarterly Journal of
Economics, 89, 117–148.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998a); Corruption, public
finances and the unofficial economy. World Bank Discussion Paper.
Johnson, S., D. Kaufmann and P. Zoido-Lobatón (1998b); Regulatory discretion
and the unofficial economy. American Economic Review, 88, 387–392.
Kirchgässner, G. (2006), Steuermoral und Schwarzarbeit. In D. Enste and F.
Schneider (eds), Jahrbuch Schattenwirtschaft: Zum Spannungsfeld von Politik
und Ökonomie, pp. 109–134.
Leu, R.E., M. Gerfin, T. Müller, G. Kirchgässner, C. Knöpfel and A. Spermann
(2008), Erwerbsabhängige Steuergutschriften: Möglichkeiten und Auswirkungen
einer Einführung in der Schweiz. Zürich: Rüegger.
Leuthard, D. (2007), Speaking note 11 November 2007, for ‘Bundesgesetz gegen


Schwarzarbeit’. Available at http://www.evd.admin.ch/themen/00129/00769/

index.html?lang5de
May, J.W., W. Pyle and P.M. Sommers (2002), Does governance explain unofficial
activity? Applied Economics Letters, 9, 537–539.
OECD (2002), Measuring the Non-Observed Economy: A Handbook. Paris:
OECD.
Scandinavia: a measurement based on questionnaire service. Study No. 10, The
Petersen, H.-G. (1982), Size of the public sector, economic growth and the infor-
mal economy: development trends in the Federal Republic of Germany. Review
of Income and Wealth, 28/2, 191–215.
Pommerehne, W.W. and H. Weck-Hannemann (1989), Einkommensteuer-
hinterziehung in der Schweiz: Eine empirische Analyse. Schweizerische
Zeitschrift für Volkswirtschaft und Statistik, 125, 515–556.
Public Broadcasting Service (PBS) (2000), Interview with Milton Friedman,
10 January 2000. Available at http://www.pbs.org/wgbh/commandingheights/
shared/minitext/int_miltonfriedman.html#2
Schaltegger, C.A. and M. Weder (2010), Fiskalpolitik als antizyklisches
Instrument? Eine Betrachtung der Schweiz. Perspektiven der Wirtschaftspolitik,
11, 146–177.
Schneider, F. (2001), Arbeit im Schatten: Einige theoretische und empirische
Überlegungen über die Schattenwirtschaft. Perspektiven der Wirtschaftspolitik,
2, 425–439.
Schneider, F. (2004), Arbeit im Schatten: Eine Wachstumsmaschine für Deutschland?
Wiesbaden: Gabler Verlag.
know? European Journal of Political Economy, 21/4, 5, S. 598–642.
Schneider, F. (2009), Weltwirtschaftskrise führt wieder zum Anstieg der
Schattenwirtschaft in Deutschland. Johannes Kepler Universität Linz.
Schneider, F. and D.H. Enste (2000), Shadow economies: size, causes, and conse-
University Press.
Schneider, F. and R. Neck (1993), The development of the shadow economy under
changing tax systems and structures: some theoretical and empirical results for
Austria. Finanzarchiv, 50, 344–369.
Schneider, F., B. Torgler and C.A. Schaltegger (2008), Schattenwirtschaft,
Steuerhinterziehung und Steuermoral in der Schweiz. Zurich: Rüegger.
Smith, J.D. (1985), Market motives in the informal economy. In W. Gaertner and
A. Wenig (eds), The Economics of the Shadow Economy. Heidelberg: Springer
Publishing Company, pp. 161–177.
Smith, P. (1994), Assessing the size of the underground economy: the statistics
Canada perspectives. Canadian Economic Observer, Catalogue No.: 11-010,
3.16-33, at 3.18.
Spiro, P.S. (1993), Evidence of a post-GST increase in the underground economy.
Canadian Tax Journal/ Revue Fiscale Canadienne, 41(2), 247–258.
Tanzi, V. (1980), The underground economy in the United States: estimates and
implications. Banca Nazionale del Lavoro, 135(4), 427–453.


Tanzi, V. (1982) (ed.): The Underground Economy in the United States and Abroad,
Lexington, MA: Lexington Books.
Tanzi, V. (1983), The underground economy in the United States: annual esti-
mates, 1930–1980. IMF Staff Papers, 30, 283–305.
Tanzi, V. (2000), Policies, Institutions and the Dark Side of Economics.
Cheltenham, UK: Edward Elgar.
Thomas, J.J. (1992), Informal Economic Activity. LSE, Handbooks in Economics,
Thomas, J.J. (1999), Quantifying the black economy: measurement without theory
yet again? Economic Journal, 109, 381–389.
Torgler, B. (2004), Tax morale. Dissertation, University of Basel.
Torgler, B. and F. Schneider (2007), Shadow economy, tax morale and institu-
tional quality: a panel analysis. CREMA Working Paper Series 2007-02.
Weck, H., W.W. Pommerehene and B.S. Frey (1984), Schattenwirtschaft.
München: Vahlen.
Weck, H., W.W. Pommerehene and B.S. Frey (1986), Die heimliche Wirtschaft.
Struktur und Entwicklung der Schattenwirtschaft in der Schweiz. Bern and
Stuttgart: Haupt.


APPENDIX 4.1 EXAMPLES OF PUBLIC

CAMPAIGN ACTIVITY
Figure A4.1.1 Examples of public campaign activity


PART II
Regional variation in the size and

development of the shadow economy



5 Regional variations in the nature
of the shadow economy: evidence
from a survey of 27 European Union
member states
Colin C. Williams and Jan Windebank
5.1 INTRODUCTION
For much of the last century, a popular and recurrent belief was that the
shadow economy was disappearing and becoming a minor residue exist-
ing only in a few marginal enclaves of the modern economy (Geertz, 1963;
Lewis, 1959). This modernisation thesis, however, has been increasingly
refuted. It is now recognised that the shadow economy is relatively wide-
spread and growing relative to the legitimate declared economy in many
global regions (Schneider and Enste, 2000, 2002; ILO, 2002a, b; OECD,
2002; Feige and Urban, 2008; Schneider, 2008; Charmes, 2009; Jütting
and Laiglesia, 2009; Rodgers and Williams, 2009). Indeed, a recent
OECD report finds that out of a global working population of some 3
billion, around two-thirds (1.8 billion) work in the shadow economy
(Jütting and Laiglesia, 2009). Such work, therefore, is far from being a
small residual realm. It is a prominent feature of the contemporary global
economy. Given this, the aim in this chapter is to further contribute to
understanding this realm by moving beyond the dominant focus in much
of the current literature on the variable size of the shadow economy and
instead, unravelling the nature of the shadow economy and how this
varies spatially. To do this, the findings of a survey based on 26 659 face-
to-face interviews conducted in 27 EU member states during 2007 will be
here reported.
To commence, the first section will briefly review the extant literature
on the shadow economy and in doing so, display that most studies have
sought to measure the variations in its magnitude. Rather few studies have
unravelled the nature of work in the shadow economy and those that do
so are largely small-scale intensive studies of particular places, populations
177


and/or sectors. The result is that little is known about the spatially variable
nature of the shadow economy since most of the current studies are not
comparable. To bridge this gap, therefore, the second section introduces
an extensive cross-national survey designed by a team of researchers
(including one of the authors of this chapter) that used the same method,
questionnaire and definition to evaluate the nature of the shadow economy
in the 27 member states of the European Union. The third section will then
report the findings with regard to the character of the shadow economy in
different regions of the EU. The outcome will be a comprehensive and up-
to-date understanding of the spatial variations in the nature of the shadow
economy. The final concluding section then reviews the findings and their
implications for public policy.
Before commencing, however, the shadow economy needs to be defined.
With some 45 different nouns and 10 adjectives used to denote this realm,
including the ‘cash-in-hand’, ‘undeclared’, ‘shadow’, ‘informal’, ‘black’
and ‘underground’ economy/sector/work (Williams, 2004), it might be
thought that there would be considerable confusion over how to define
this phenomenon. This, however, is not the case. Across academic and
policy-making circles alike, a strong consensus has begun to emerge on
how to define the shadow economy (European Commission, 1998, 2007;
Schneider, 2008; Sepulveda and Syrett, 2007; Williams, 2004). Shadow
work is nearly always delineated in terms of what is absent from, or insuf-
ficient about, it compared with declared work, and the strong consensus
is that the only absence from, or insufficiency about, shadow work is
that this remunerated production and/or sale of licit goods and services
is not declared to the authorities for tax, social security and/or labour
law purposes when it should be declared (European Commission, 2007;
OECD, 2002; Renooy et al., 2004; Williams and Windebank, 1998). If
work possesses additional absences or insufficiencies, then it is not usually
defined as part of the shadow economy. If the goods and/or services are
illegal (for example, drug-trafficking), for example, then this is defined as
‘criminal’ activity rather than shadow work since the only illicit aspect of
shadow work is that the income received is not declared when it should be
declared. If the activity is not remunerated, similarly, it is not the shadow
economy but the unpaid sphere.
For analytical purposes, furthermore, this chapter divides the shadow
economy into two types: wholly ‘undeclared’ jobs where none of the
income is declared to the state, as is the case in waged employment or
own-account jobs conducted on a wholly off-the-books basis; and ‘under-
declared’ formal employment where an official registered employee is
paid by their formal employer two wages, an official declared wage and
a supplementary unofficial undeclared wage, sometimes known as an


Regional variations in the shadow economy 179
‘envelope wage’. As will become apparent, although the study of the

shadow economy has traditionally largely focused only on the former type
of shadow work, a full and comprehensive understanding of the nature of
the shadow economy cannot be achieved unless both types are analyzed.
5.2 PREVIOUS RESEARCH ON THE NATURE OF

THE SHADOW ECONOMY
Even a cursory glance at the extensive literature on the shadow economy
quickly reveals that the vast majority has adopted a ‘size matters’ per-
spective focused on measuring the magnitude and growth of this realm.
On the whole, those measuring its size and how this varies across dif-
ferent populations have relied on indirect measurement methods using
proxy or surrogate indicators, such as the amount of cash in circulation
(Feige, 1979), the number of very small enterprises in existence (ILO,
2002) or even the amount of electricity consumed (Friedman et al., 2000;
Kaufmann and Kaliberda, 1996; Lacko, 1999), or adopted more refined
methods using multiple indicators (Schneider and Enste, 2000, 2002).
The problem, however, is that such indirect methods have been widely
argued to provide little information of the nature of the shadow economy
(European Commission, 2007; OECD, 2002; Thomas, 1992; Williams and
Windebank, 1998).
In recent years, therefore, there has been a growing recognition that if
the nature of the shadow economy is to be better understood, data on its
character will need to be collected using direct surveys. This is the conclu-
sion of both OECD experts in their handbook on methods of measuring
shadow work (OECD, 2002), the most recent communication of the
European Commission on undeclared work (European Commission,
2007), evaluations by national governments (e.g., ONS, 2005) as well as
a host of academic evaluations of direct and indirect methods (Smith,
1986; Thomas, 1992; Williams, 2004, 2006a; Williams and Windebank,
1998).
Until now, nevertheless, most direct surveys of the shadow economy
have been relatively small-scale intensive studies of specific localities and/
or populations (Barthe, 1985; Fortin et al., 1996; Howe, 1988; Lemieux,
Fortin and Frechette, 1994; Leonard, 1994; McCrohan, Smith and Adams
1991; Neef, 2002; Pahl, 1984; Pavlovskaya, 2004; Round and Williams,
2008; Round, Williams and Rodgers 2008; Smith and Stenning, 2006;
Warde, 1990; Williams, 2004, 2006a; Williams and Round, 2007). Using
different definitions and methods, these have largely produced data that
are not comparable. The result is that it has been difficult to develop


an understanding of the spatial variations in the nature of the shadow

economy. Extensive cross-national surveys of the nature of shadow work
are largely absent. One exception is a European Barometer survey (Rose,
2005; Wallace and Haerpfer, 2002; Wallace, Harpfer and Latcheva 2004;
Williams, 2006b) of East–Central European nations. This, however,
included only a very limited number of questions on the shadow economy
as part of a much wider social and economic survey. As such, it was unable
to fully unravel how the nature of the shadow economy alters across dif-
ferent countries and populations. For example, it did not explicitly address
whether employees are paid additional envelope wages by their employ-
ers, or the proportion of shadow work that is waged employment and the
share that is own-account work. The outcome is that, until now, little evi-
dence has been available that allows analysts to evaluate how the nature of
the shadow economy varies spatially.
Nevertheless, the mostly small-scale intensive studies so far conducted
do provide some useful clues of the broad range of types of shadow
work that need to be differentiated when seeking to understand the
spatial variations in the nature of the shadow economy. The conven-
tional focus when studying shadow work was upon waged employment
that is wholly hidden from, and/or unregistered by, the state and often
conducted under degrading, low-paid and exploitative ‘sweatshop’-
like conditions by marginalised populations who participated in such
endeavour out of necessity due to no other options being open to them
(Castells and Portes, 1989; Davis, 2006; Sassen, 1997), as exemplified
by ‘sweatshop’-like work in the garment manufacturing sector (Bender,
2004; Espenshade, 2004; Ross, 2004). The outcome was that a narrow
representation of shadow work emerged that portrayed it as peripheral
employment existing at the bottom of a hierarchy of types of employ-
ment whose participants shared similar characteristics to ‘downgraded
labour’: they receive few benefits, low wages and have poor working
conditions (for example, Castells and Portes, 1989; Gallin, 2001; Portes,
1994; Sassen, 1997; Ybarra, 1989).
However, over the past few decades, numerous studies of particular
places, sectors and populations across the globe have led to a recogni-
tion, on the one hand, of a range of additional types of shadow waged
employment and, on the other hand, the existence of more autonomous
kinds of shadow work. Indeed, it has now become widely recognised
that the majority of shadow work in many places is conducted on an
own-account or self-employed basis (Chen, 2004; ILO, 2002b; Williams,
2006).
A small stream of literature, especially with regard to East–Central
Europe, has also increasingly recognised that besides ‘undeclared’ jobs


wholly hidden from the state for tax, social security and labour law
purposes, there are ‘under-declared’ formal jobs where formal employ-
ees receive from their formal employer two wages, one declared and
one undeclared ‘envelope wage’. This has been identified in Latvia
(Sedlenieks, 2003; Žabko and Rajevska, 2007), Lithuania (Karpuskiene,
2007; Woolfson, 2007), Romania (Neef, 2002), Russia (Williams and
Round, 2007) and Ukraine (Williams, 2007). All, however, are again
small-scale qualitative surveys and not strictly comparable. In Latvia,
for example, Sedlenieks (2003) reports 15 face-to-face interviews in
Riga whilst the study in Lithuania by Woolfson (2007) is an in-depth
case study of one person, albeit a cause célèbre. The Ukraine evidence,
meanwhile, is based on a survey of 600 households in just three localities
(Williams, 2007), whilst the evidence from Russia is based on interviews
with 313 households in three districts of Moscow (Williams and Round,
2007).
It is not only shadow waged employment that has been re-read by
recognising the existence of under-declared as well as undeclared employ-
ment. Shadow work conducted on a self-employed or own-account basis
has also started to be reconceptualised (Neef, 2002; Round, Williams and
Rodgers 2008; Smith and Stenning, 2006; Williams and Round, 2007,
Williams, Round and Rodgers, 2007). Until now, most of the literature
has simply assumed that shadow work (whether of the waged or self-
employed variety), is conducted under anonymous business-like relations
for profit-motivated purposes. Recently, however, a small tributary of
thought has started to question this by arguing that shadow work, espe-
cially on an own-account basis, is sometimes conducted for closer social
relations such as kin, neighbours, friends and acquaintances, as well as for
purposes other than purely financial gain. So far, however, this re-reading
of shadow work as embedded in networks of familial and community
support has been highlighted only in a limited range of West European
countries, such as France, Sweden and the UK (Cornuel and Duriez, 1984;
Persson and Malmer, 2006; Williams, 2004; Williams and Windebank,
1998). Whether it is more broadly applicable in other European contexts,
such as East–Central Europe and Southern Europe, has not been largely
investigated.
In sum, most studies of the nature of the shadow economy have been
small-scale intensive studies of particular places, populations and sectors
that are not strictly comparable. The result is a significant gap in our
knowledge base concerning the spatial variations in the nature of the
shadow economy. Here, in consequence, the intention is to fill this gap by
reporting a cross-national survey of the nature of the shadow economy in
the European Union.


5.3 EXAMINING THE SHADOW ECONOMY IN THE

EUROPEAN UNION
Recognising the lack of primary data on the nature of the shadow
economy, in late 2005, the European Commission’s Employment Analysis
Division of the DG Employment and Social Affairs commissioned a team
of researchers (which included one of this chapter’s authors) to evalu-
ate the feasibility of conducting a direct survey (European Commission,
2005). Following its design (TNS Infratest et al., 2006), this survey was
implemented as Special Eurobarometer No. 284 (‘Undeclared work in the
European Union’), as part of wave 67.3 of Eurobarometer. The fieldwork
was conducted during May and June 2007.
Using the same basic sampling method as Eurobarometer surveys
in general, 26 659 face-to-face interviews were conducted in the 27 EU
member states, ranging from 500 in smaller member states to 15001
in larger EU countries. In all countries, a multi-stage random (prob-
ability) sampling method was applied. Within each, that is, a number of
sampling points were drawn with probability proportional to popula-
tion size (for total coverage of the country) and to population density
according to the Eurostats NUTS II (or equivalent) and the distribution
of the resident population in terms of metropolitan, urban and rural
areas. In each of the selected sampling units, a starting address was then
drawn at random. Further addresses (every nth address) were subse-
quently selected by standard ‘random route’ procedures from the initial
address. In each household, meanwhile, the respondent was drawn at
random (following the ‘closest birthday rule’). All interviews were con-
ducted face-to-face in people’s homes and in the appropriate national
language with adults aged 15 years and over. For data collation, CAPI
(computer assisted personal interview) was used in countries where it
was available.
In all countries, furthermore, a national weighting procedure was
employed for data analysis purposes that used marginal and intercellular
weighting by comparing the sample with the universe description taken
from Eurostat population data and national statistical offices. All results
in this chapter are based on this weighting procedure. In each country, this
weighting procedure ensures that the gender, age, region and size of local-
ity of the sample were proportionate to the universe. As a result, all data
reported below are nationally representative. The data were collated and
analysed using the Statistical Package for Social Scientists (SPSS) while
the statistical significance of findings was measured at the 0.05(*), 0.01 (**)
and 0.001 (***) levels.
To gather data, the face-to-face interview schedule adopted a gradual


approach so far as approaching the more sensitive questions were con-

cerned. It firstly asked respondents for their opinions and attitudes
regarding the shadow economy and having established some rapport,
then moved in the second section onto questions regarding their personal
purchase of goods and services on a wholly shadow basis in the last 12
months along with their reasons for doing so. The third section then
asked questions on whether they had received ‘envelope wages’ from
their regular formal employer in the past 12 months. Respondents were
asked, ‘Sometimes employers prefer to pay all or part of the regular
salary or the remuneration for extra work or overtime hours cash-in-
hand and without declaring it to tax or social security authorities. Did
your employer pay you all or part of your income in the last 12 months
in this way?’. To decipher whether this was for regular work, overtime
payments or both, interviewees were then asked ‘Was this income part of
the remuneration for your regular work, was it payments for overtime, or
both?’ and they were then asked for the percentage of their gross yearly
income from their main formal job received on an undeclared basis.
Fourth and finally, questions were asked regarding their supply of wholly
shadow work including the type of work they conducted, for whom
and why they had undertaken this shadow work. Below, the results are
reported.
5.4 GEOGRAPHICAL VARIATIONS IN THE

NATURE OF THE SHADOW ECONOMY IN THE
EU-27
What is the nature of the shadow economy in the EU-27? Who conducts
such work? And in what sectors and businesses does such work prevail?
Examining the results of the 26 659 face-to-face interviews, Table 5.1
examines the prevalence of both wholly shadow jobs (here referred to as
‘undeclared’ work) where all the income is concealed from the state for tax,
social security or labour law purposes, as well as ‘under-declared’ formal
jobs where an official employee receives from their regular employer two
wages, a declared wage and an undeclared (‘envelope’) wage. Its headline
finding is that just under one in ten (9 per cent) of the surveyed popula-
tion participated in either undeclared or under-declared work in the 12
months prior to interview, of which 4 per cent engaged solely in unde-
clared work, a further 4 per cent solely in under-declared work and 1 per
cent in both undeclared and under-declared work. The shadow economy,
in consequence, is not confined to the margins of the European economic
landscape.


Table 5.1 Participation in undeclared and under-declared work in the

European Union: by sector, business and population group
Country % participating during past 12 months in:

Any Solely Solely Both No
undeclared undeclared under- undeclared undeclared
or under- work declared and under- or under-
declared work declared declared
work work work
All 9 4 4 1 91
Sector:
Construction 17*** 6 7 4 83
Industry 8 3 4 1 92
Household 6 3 2 1 94
services
Transport 9 3 5 1 91
Personal 7 3 3 1 93
services
Retail 8 2 4 2 92
Repair services 16 9 3 4 84
Hotel, 9 3 3 3 91
restaurant &
café
Agriculture 12 4 7 1 88
No. of employees:
1–20 13*** 4 6 3 87
21–50 9 4 4 1 91
51–100 11 7 3 1 89
101–500 8 4 2 1 92
5011 5 3 1 1 95
Gender:
Men 12*** 5 5 2 88
Women 6 2 3 1 94
Occupation:
Managers 7** 3 3 1 93
White collar 8 3 4 1 92
workers
Manual 10 4 4 2 90
workers
Age:
15–24 16*** 7 5 4 84
25–39 10 4 5 1 90
40–54 7 3 3 1 93
551 5 2 2 1 95



work work work
Gross income/
month:
<€500 15*** 4 8 3 85
500 – 1000.99 11 4 5 2 89
1001 – 2000.99 9 4 3 2 91
2001 – 3000.99 8 6 1 1 92
30011 7 3 3 1 93
Notes: Statistical significance: *** 5 0.001 (0.1%)
Source: Eurobarometer no. 284 survey, 2007
Participation rates in undeclared and under-declared work, however, vary

according to both the type of business in which people work and by popu-
lation group. Starting with the economic sectors, and as previously identi-
fied in smaller-scale studies (for example, Pedersen, 2003; Small Business
Council, 2004; Williams, 2006), the shadow economy is particularly rife in
the construction sector, as well as repair services which includes a range of
home improvement and maintenance trades, and agriculture. It is also clus-
tered amongst employees working in smaller businesses. Some population
groups, moreover, engage in the shadow economy more than others. Men,
manual workers, younger aged people and lower-income groups are signifi-
cantly more likely to participate in the shadow economy than other social
groups. Importantly, however, even if participation is higher in some busi-
nesses and population groups than others, the shadow economy appears
to be a ubiquitous phenomenon. Businesses of all sizes and sectors, and all
population groups, participate to some extent in shadow work.
Is it the case, however, that the nature of undeclared work is the same
everywhere? Or can spatial variations in the nature of undeclared work be
identified? If so, how does its character vary? To answer these questions,
and given the small numbers of respondents involved, the results from the
27 EU member states are below grouped into four broad geographical
regions:


● Continental Europe, UK and Ireland (Belgium, Germany, France,

Ireland, Luxembourg, Netherlands, Austria and the UK);
● Eastern and Central Europe (Bulgaria, Czech Republic, Estonia,
Latvia, Lithuania, Hungary, Poland, Romania, Slovenia and
Slovakia);
● Southern Europe (Cyprus, Greece, Spain, Italy, Malta and
Portugal), and
● Nordic countries (Denmark, Finland, Sweden).
When this is done, Table 5.2 reveals some clear variations in the nature
of the shadow economy. Starting with the commonality of engagement
in shadow work, the finding is that in East–Central Europe nearly one in
five (18 per cent) of the surveyed population had participated in shadow
work in the previous 12 months compared with 12 per cent in Nordic
nations, 8 per cent in Southern Europe and 5 per cent in Continental
Europe.
Turning to the nature of shadow work undertaken in these EU regions,
however, some marked differences exist. In East–Central Europe and
Southern Europe, the vast majority of the shadow work conducted tends
to be under-declared work (61 per cent and 70 per cent, respectively, of all
shadow work) whilst in Nordic nations and Continental Europe, shadow
work is largely wholly undeclared work (60 per cent and 83 per cent of
all shadow work) and under-declared waged employment is much less
prevalent.
Besides such clear regional variations, there are also variations within
regions. For example, in East–Central Europe, and as Table 5.2 displays,
there is a group of East–Central European countries where participation
is more widespread, including Romania (where 35 per cent had engaged
in shadow work in the last 12 months), Latvia (25 per cent) and Bulgaria
(19 per cent), and in which the vast majority of shadow work is in the
form of under-declared waged work rather than wholly undeclared work.
In other East–Central European nations such as Slovenia, the Czech
Republic and Estonia, however, where overall participation rates in the
shadow economy are slightly lower, a smaller proportion of the shadow
economy is composed of under-declared work and the majority is wholly
undeclared work. In consequence, intra- as well as inter-regional vari-
ations exist in the nature of the shadow economy within the European
Union.
In consequence, and to further unpack these geographical variations
in the nature of the shadow economy, firstly, how wholly undeclared
work varies spatially will be evaluated and secondly, the spatialities of
‘under-declared’ waged employment.


Table 5.2 Participation rates in undeclared and under-declared work in

the European Union: by EU region and country

work work work
All 9 4 4 1 91
East–Central Europe 18 4 11 3 82
Romania 35 2 30 2 65
Latvia 25 7 11 7 75
Bulgaria 20 4 10 6 80
Estonia 18 7 4 7 84
Hungary 15 7 4 4 85
Poland 15 4 9 3 85
Lithuania 15 3 6 6 86
Slovakia 12 5 5 2 88
Czech Rep 9 6 1 2 91
Slovenia 8 4 2 2 92
Nordic nations 12 10 1 1 88
Denmark 20 17 1 2 80
Sweden 12 10 1 1 88
Finland 7 4 2 1 93
Continental Europe 5 3 1 1 95
Netherlands 11 8 1 2 89
Belgium 9 3 3 3 90
Austria 9 5 2 2 91
France 7 6 0 1 93
Ireland 6 3 1 1 94
West Germany 4 3 1 0 96
East Germany 3 1 1 1 97
Great Britain 3 1 1 1 97
Luxembourg 0 0 0 0 100
Southern Europe 8 2 5 1 82
Italy 10 2 7 1 90
Spain 7 1 4 2 93
Portugal 7 2 3 2 93
Cyprus 6 0 6 0 94
Greece 5 2 1 2 95
Malta 0 0 0 0 100


5.4.1 Geographical Variations in the Nature of Undeclared Work

(Wholly Shadow Jobs)
In recent decades, it has been widely recognised that undeclared work

can be conducted either on a waged employment or own-account
basis, and studies have begun to analyse the ratio of undeclared waged
employment to undeclared self-employment in different places. The
ILO (2002b), for example, identify that some 70 per cent of undeclared
work in sub-Saharan Africa is conducted by the self-employed, 62 per
cent in North Africa, 60 per cent in Latin America and 59 per cent in
Asia. Until now, no comparable figures have been available for the
European Union. This survey, however, reveals that some 78 per cent
of undeclared work is conducted on a self-employed basis and just 22
per cent as waged employment. Compared with other global regions, in
consequence, shadow work appears to be more commonly conducted
on an own-account basis in the EU than these other global regions.
Whether this is due to the more effective and efficient regulation of the
employment-place in Europe than elsewhere, or whether other factors
explain this trend, is an open question that will need to be answered in
future research.
Following earlier studies in west European nations which identi-
fied that undeclared own-account work is often conducted for closer
social relations (Cornuel and Duriez, 1984; Persson and Malmer, 2006;
Williams, 2004), the finding of this survey is that across the EU-27 as a
whole, some 55 per cent of all undeclared work is conducted for closer
social relations, namely for kin, friends, neighbours and acquaintances,
indicating that work in the shadow economy is more embedded in social
networks of familial and community solidarity in the EU than previously
considered.
Nevertheless, the nature of undeclared work is not everywhere the
same. As Figure 5.1 displays, there are variations between EU regions
in not only the ratio of undeclared waged employment to undeclared
own-account work, but also in terms of the degree to which undeclared
work is embedded in closer social networks of familial and commu-
nity solidarity rather than more anonymous market-like relations. In
both Southern Europe and East–Central Europe, that is, a greater
proportion of undeclared work is conducted as waged employment (24
and 23 per cent, respectively) than in Nordic nations and Continental
Europe (15 and 18 per cent). In Nordic nations and Continental
Europe, meanwhile, a higher proportion is conducted on an own-
account basis and a much greater share of this is conducted for closer
social relations such as kin, friends, acquaintances and neighbours.


All EU-27
Southern
East-Central
Continental
Nordic
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Waged employment Self-employment for closer social relations

Self-employment for other private persons or households Don’t know
Figure 5.1 Variations in the nature of undeclared work in the European

Union: regional variations
Indeed, in Nordic and Continental European nations, some 70 and 63

per cent of all shadow work is embedded in closer social networks of
familial and community solidarity whilst in East–Central Europe and
Southern Europe, only 43 and 40 per cent is of this type. It appears, in
consequence, that the meaning of undeclared work varies across these
different EU regions.
It is not only at the level of EU regions that the nature of undeclared
work varies according to the social relations involved. As Figure 5.2
reveals, there are also urban–rural variations. In urban areas a greater
proportion of undeclared work takes the form of waged employment
whilst in more rural areas, not only is a greater share undertaken on
an own-account basis but also for kin, neighbours and acquaintances.
Indeed, nearly one-third (31 per cent) of all undeclared work in large
towns takes the form of waged employment compared with just 20 per
cent in small or middle-sized towns and merely 13 per cent in rural areas.
Meanwhile, two-thirds (66 per cent) of undeclared jobs in rural areas are
conducted for closer social relations but only half (51 per cent) in small
and medium- sized towns and 45 per cent in large towns. The outcome
is that undeclared work in rural areas is more embedded in networks of
familial and community support, whilst in urban areas it is more con-
ducted as waged employment probably under social relations more akin
to anonymous market-like work.


Large town
Small or middle-sized town
Rural area or village
All EU-27
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Waged employment
Self-employment for closer social relations
Self-employment for other private persons or households
Not known
Figure 5.2 Variations in the nature of undeclared work in the European

Union: an urban–rural comparison
It is not just the nature of wholly undeclared work which displays

marked spatial variations. The nature of under-declared work also varies
geographically.
5.4.2 Geographical Variations in the Nature of Under-declared

(‘Envelope’) Work
Of the 26 659 face-to-face interviews conducted in the 27 EU member

states, 11 885 were with formal employees, of which some 1 in 20 (5
percent, or 616 employees) had received envelope wages from their formal
employer in the past year amounting on average to over two-fifths (43 per
cent) of their gross total wage. Of these, 29 per cent received such payments
for their regular work, 27 per cent for extra work or overtime and 36 per
cent for both their regular and overtime work. Extrapolating to the EU as
a whole, this intimates that some 11 million of the 210 million employees
in the EU might be in receipt of envelope wages: some 3 million for their
regular work, 3 million for overtime or extra work and 4 million for both
their regular and overtime work.
Is this practice evenly spread across the EU or is it concentrated in par-
ticular countries and regions? And does its character differ across the EU
in terms of whether it is paid for regular or extra work? Table 5.3 reveals
some clear differences across EU regions in the prevalence and character
of under-declared work. This practice of formal employers paying their
formal employees an additional ‘off-the-books’ wage is more common in


Table 5.3 Prevalence and types of envelope wages: by EU geographical

region
Country No. waged % who Envelope wage paid as remuneration for:

employees receive
Regular Over- Both Refusal
surveyed envelope
work time/ regular and 1 don’t
wage
extra overtime know
work work
Continental 3,907 2*** 18*** 65 14 3
Europe
East-Central 4,511 11 39 21 37 3
Europe
Southern Europe 1,933 4 18 43 26 13
Nordic countries 1,534 3 7 70 9 14
EU-27, weighted 11,885 5 29 27 36 8
average
Source: Eurobarometer survey no. 278, 2007
East–Central Europe where 11 per cent of the formal employees surveyed

report that they receive such an envelope wage compared with just 2 per
cent in Continental Europe, 4 per cent in Southern Europe and 3 per cent
in Nordic countries. There are also differences in the nature of under-
declared employment across these four regions. Some two-thirds of enve-
lope wage payments in both Continental Europe and Nordic countries (65
per cent and 70 per cent respectively) are for overtime or extra work. In
East–Central Europe and southern Europe, however, envelope wages are
paid more usually for regular work or for both regular work and overtime.
The outcome, as Table 5.4 reveals, is that even if only just over a third
(36 per cent) of the employees surveyed are in East–Central Europe, over
two-thirds (68 per cent) of those reporting that they receive envelope
wages are in this EU region and 84 per cent of employees are receiving
envelope wages for their regular employment. As such, the practice of
under-declared employment, especially for regular employment, is heavily
concentrated in East–Central Europe.
Nevertheless, and as displayed in Table 5.5, there are marked cross-
national variations in both the extent to which employers use the practice
of envelope wage payments and the nature of this practice within each EU
region. In Continental Europe, for example, the prevalence of envelope


Table 5.4 Distribution of envelope waged work across EU regions: by

type of envelope waged work
Country % of % of all % of all receiving an envelope

waged receiving wage for:
employees an
Regular Overtime/ Both regular
surveyed envelope
work extra work and overtime
wage
work
Continental 36 14*** 8*** 27 6
Europe
East-Central 36 68 84 43 81
Europe
Southern Europe 16 12 7 16 11
Nordic countries 12 6 1 14 2
EU-27 100 100 100 100 100
Source: Eurobarometer survey, 2007
wages ranges from 6 per cent of employees in Belgium to 1 per cent in

the United Kingdom, Luxembourg, France and Germany. Similarly, in
Southern Europe, its prevalence ranges from 7 per cent of employees
in Italy to 1 per cent in Malta. It is within East–Central Europe, however,
that the clearest segmentation of nations can be identified with regard to
under-declared employment. On the one hand are a group of countries in
which under-declared employment is common, and envelope wages are
used more to reimburse regular hours with such wages on average amount-
ing to around half of formal employees’ wages (Romania, Latvia, Bulgaria,
Lithuania and Poland). On the other hand are a group of countries in which
under-declared employment is less common, paid more for overtime or extra
work and such wages amount on average to around a quarter of employees’
wage packets (Czech Republic, Estonia, Hungary, Slovenia and Slovakia).
The result is that over half (54 per cent) of all under-declared employment
is concentrated in just five East–Central European countries – Romania,
Latvia, Bulgaria, Lithuania, Poland – that contain less than 16 per cent of
the EU adult population. Indeed, three-quarters of employees receiving
envelope wages for regular work are in these five countries. Despite this
concentration of under-declared employment in a narrow range of coun-
tries, nevertheless, such a practice remains a ubiquitous phenomenon. It is
prevalent in all member states of the EU.


Table 5.5 Percentage of employees in employment receiving envelope

wages in the past 12 months, by nation
Country No. of % all % of gross Envelope wages paid for:

waged employees income
Regular Overtime/ Both Refusal
employees receiving received
work extra work regular & 1 don’t
surveyed envelope as
overtime know
wages envelope
work
wages
EU27 11,885 5 43 29 27 36 8
East-Central 4,511 12 50 40 16 41 3
Europe
Romania 453 23*** 70*** 48*** 9 41 2
Latvia 511 17 46 47 18 34 1
Bulgaria 446 14 44 46 15 37 2
Lithuania 446 11 48 44 10 46 0
Poland 337 11 53 35 15 50 0
Estonia 457 8 31 37 20 32 11
Hungary 392 8 24 19 50 27 4
Slovakia 537 7 25 39 43 18 0
Slovenia 431 5 23 13 40 28 19
Czech 501 3 14 13 46 41 0
Republic
Southern 1,933 6 37 15 25 38 22
Europe
Italy 401 7 63 12 17 44 27
Spain 423 5 19 18 36 32 14
Portugal 421 4 42 19 15 33 33
Cyprus 210 4 11 9 54 37 0
Greece 292 3 31 29 54 17 0
Malta 186 1 20 100 0 0 0
Continental 3,907 2 24 20 61 16 3
Europe
Belgium 428 6 14 18 66 10 6
Austria 519 4 19 14 48 30 8
Ireland 458 2 26 11 66 0 23
Netherlands 515 2 18 0 75 15 10
Germany 657 1 23 35 65 0 0
France 507 1 54 11 61 28 0
Luxembourg 223 1 11 0 100 0 0
UK 600 1 1 27 53 20 0
Nordic nations 1,534 3 7 7 69 13 11
Finland 491 3 8 6 80 9 5
Sweden 546 3 7 6 80 9 5
Denmark 497 2 6 8 44 16 32
Source: Eurobarometer survey no.284, 2007


5.5 CONCLUSIONS
This chapter has sought to understand the regional variations in the
nature of the shadow economy in the European Union. Until now,
although a great deal of attention has been paid to measuring cross-
national variations in the magnitude of the shadow economy, much less
attention has been paid to how the nature of the shadow economy varies
cross-nationally and between different EU regions. The aim of this chapter
has been to begin to fill this gap by reporting the findings of a 27-nation
survey based on 26 659 face-to-face interviews.
Before summarising the findings, a cautionary note is required. Given
the potential sensitivity of the topic and the pilot nature of the survey
methodology, the above findings need to be treated with some caution.
Indeed, in the past, some have raised concerns about the honesty of partic-
ipants on such a sensitive topic. The feedback from interviewers, however,
is that a lack of honesty was not an issue in this survey. Interviewers
reported fair or very good cooperation on the part of the respondent in
88 per cent of the interviews. In only 2 per cent was cooperation deemed
to be bad. It seems, therefore, that although this work might be hidden
from the authorities for tax, social security or labour law purposes, few
respondents hide it from interviewers. This reinforces previous surveys
that reveal how respondents openly talk about their undeclared work
(Leonard, 1994; MacDonald, 1994; Pahl, 1984; Williams, 2006). It is none-
theless the case that some participants might not know whether they are
engaged in shadow work since they may be unaware of their employer’s
reporting practices. The implication, therefore, is that although conduct-
ing direct surveys of shadow work are not perhaps as problematic as many
previously assumed, some caution is nevertheless required.
Nonetheless, this direct survey provides some important and fresh
insights into the spatial variations in the nature of the shadow economy
in the European Union. First, it has revealed that throughout the EU-27,
nearly one in ten people have engaged in the shadow economy over
the past 12 months. In the past, and perhaps due to the way in which
the declared and undeclared realms were deemed separate and discrete
spheres, the emphasis was on shadow work wholly hidden from the state
for tax, social security and labour law purposes. In this chapter, however,
it has been revealed that just as many people engage in under-declared as
undeclared work. Despite this, few have so far sought to incorporate both
under-declared and undeclared work in order to provide a fuller under-
standing of the spatial variations in the nature of the shadow economy.
This chapter has begun to bridge that gap.
In doing so, it has revealed that unless such under-declared employment


is incorporated, then only a partial understanding of the shadow economy

will be achieved. Of the one in ten (9 per cent) of the surveyed population
engaged in either undeclared or under-declared work in the 12 months
prior to interview, 4 per cent engaged solely in undeclared work, 4 per
cent solely in under-declared work and 1 per cent in both undeclared
and under-declared work. There are, nevertheless, some marked spatial
variations in the nature of the shadow economy. In East–Central Europe,
nearly one in five (18 per cent) of the surveyed population had partici-
pated in shadow work in the previous 12 months compared with 12 per
cent in Nordic nations, 8 per cent in Southern Europe and 5 per cent in
Continental Europe. The nature of the shadow economy, however, sig-
nificantly differs between these EU regions. In East–Central Europe and
Southern Europe, that is, the vast majority of shadow work (61 per cent
and 70 per cent respectively) is in the form of under-declared employment,
whilst in Nordic nations and Continental Europe the majority of shadow
work (60 per cent and 83 per cent) is wholly undeclared work, mostly of
the own-account variety.
Examining wholly undeclared jobs, therefore, this chapter reveals
marked spatial variations in not only the ratio of undeclared waged
employment to undeclared own-account work, but also the degree to
which undeclared work is embedded in closer social networks of familial
and community solidarity. In Southern and East–Central Europe, as well
as urban areas, a greater proportion of undeclared work is conducted
as waged employment. Meanwhile, in Nordic nations and Continental
Europe, as well as rural areas, not only is a greater proportion conducted
on an own-account basis but a much higher share is for closer social rela-
tions such as kin, friends, acquaintances and neighbours. In consequence,
the meaning of undeclared work varies across different EU regions and
between urban and rural areas.
Turning to under-declared work, meanwhile, this chapter has simi-
larly revealed that although some 1 in 20 employees had received addi-
tional undeclared ‘envelope wages’ in the previous 12 months from their
employer and that these undeclared payments on average amounted to
over two-fifths (43 per cent) of their gross wage, this wage practice again
displays marked spatial variations. Even though it is relatively more
common in East–Central Europe, it is a ubiquitous practice, although
such envelope wages are paid more for overtime or extra work in Nordic
nations and Continental Europe, but more for regular employment in
East–Central and Southern Europe. In consequence, by showing how
the shadow economy means different things in different regions and
places, this chapter reveals the need for a more sensitive and nuanced
appreciation of the spatial variations in its nature.


This finding has significant implications. Until now, commentators

from Southern Europe and Nordic nations for example, have often talked
past each other when discussing the shadow economy and what needs to
be done about it. This chapter reveals how this might be largely because
the lived practices of shadow work are very different in these two regions.
Hopefully, therefore, this chapter will help overcome these sometimes
heated debates and discussions between scholars from different EU
regions. If greater appreciation can occur that there are different lived
practices in different EU regions, then perhaps many commentators might
start appreciating why misunderstandings arise.
Hopefully, furthermore, the finding that there are marked spatial
variations in the nature of the shadow economy will also help advance
policy-making in this realm. It clearly displays that a ‘one size fits all’
policy approach is unlikely to be appropriate since the shadow economy
differs markedly across the European Union. Instead, different policies
will be required in different regions and places due to the varying types of
shadow work in different regions. If this paper therefore simulates further
investigation of its spatial variations and how it can be variously tackled
in different EU regions, then it will have achieved its objectives. What is
certain, however, is that the shadow economy can no longer be treated as
some minor residue of little importance in Europe and beyond. This large
ubiquitous phenomenon, with its spatially variable nature, now needs to
be brought more centre-stage in economic analyses, its complex and vari-
able nature better understood and the resultant policy implications more
fully explored.
ACKNOWLEDGEMENTS
The authors would like to thank the Employment Analysis Division in

the Employment and Social Affairs DG of the European Commission
for providing access to the Special Eurobarometer survey no. 278 data-
base so that the analysis in this chapter could be undertaken. The normal
disclaimers of course apply.
REFERENCES
Barthe, M.A. (1985), Chômage, travail au noir et entraide familial. Consommation,

3(1), 23–42.
Bender, D.E. (2004), Sweated Work, Weak Bodies: Anti-Sweatshop Campaigns and
Languages of Labor. New Brunswick: Rutgers University Press.


Castells, M. and A. Portes (1989), World underneath: the origins, dynam-

ics and effects of the informal economy. In A. Portes, M. Castells and L.A.
Benton (eds), The Informal Economy: Studies in Advanced and Less Developing
Countries. Baltimore, MD: Johns Hopkins University Press, pp. 11–40.
Charmes, J. (2009), Concepts, measurement and trends. In J.P. Jütting and J.R.
Laiglesia (eds), Is Informal Normal? Towards More and Better Jobs in Developing
Countries. Paris: OECD, pp. 27–62.
economy and the formal regulatory environment. Paper presented at EGDI and
UNU-WIDER Conference ‘Unlocking human potential: linking the informal
and formal sectors’, 17–18 September, Helsinki.
Cornuel, D. and B. Duriez (1985), Local exchange and state intervention. In N.
Redclift and E. Mingione (eds) Beyond Employment: Household, Gender and
Subsistence. Oxford, UK: Basil Blackwell, pp. 56–83.
Davis, M. (2006), Planet of Slums. London: Verso.
Espenshade, J. (2004), Monitoring Sweatshops: Workers, Consumers and the Global
Apparel Industry. Philadelphia: Temple University Press.
European Commission (1998), Communication of the Commission on Undeclared
Work. Available at: http://europa.eu.int/comm/employment_social/empl_esf/
docs/com98-219_en.pdf.
European Commission (2005), Feasibility of a Direct Survey and Pilot Study of
Undeclared Work in the European Union. Brussels: European Commission.
European Commission (2007), Stepping up the Fight Against Undeclared Work.
Brussels: European Commission.
Feige, E.L. (1979), How big is the irregular economy? Challenge, November/
December, 5–13.
Feige, E.L. (1999), Underground economies in transition: non-compliance and
institutional change. In E.L. Feige and K. Ott (eds), Underground Economies in
Transition: Unrecorded Activity, Tax Evasion, Corruption and Organized Crime.
Aldershot: Ashgate, pp. 11–28.
Feige, E.L. and I. Urban (2008), Measuring underground (unobserved, non-
observed, unrecorded) economies in transition countries: can we trust GDP?
Journal of Comparative Economics, 36, 287–306.
Fortin, B., G. Garneau, G. Lacroix, T. Lemieux and C. Montmarquette (1996),
L’Economie Souterraine au Quebec: Mythes et Realités. Laval : Presses de
l’Université Laval.
Gallin, D. (2001), Propositions on trade unions and informal employment in time
of globalization. Antipode, 19(4), 531–549.
Geertz, C. (1963), Old Societies and New States: The Quest for Modernity in Asia
and Africa. Glencoe, IL: Free Press.
Howe, L. (1988), Unemployment, doing the double and local labour markets in
Belfast. In C. Cartin and T. Wilson (eds), Ireland from Below: Social Change and
Local Communities in Modern Ireland. Dublin: Gill and Macmillan, pp. 140–163.
ILO (2002a), Decent Work and the Informal Economy. Geneva: International
Labour Organization.
ILO (2002b), Women and Men in the Informal Economy: A Statistical Picture.
Geneva: International Labour Organization.


Jütting, J.P. and J.R. Laiglesia (2009), Employment, poverty reduction and devel-
opment: what’s new? In J.P. Jütting and J.R. Laiglesia (eds), Is Informal Normal?
Towards More and Better Jobs in Developing Countries. Paris: OECD, pp. 17–26.
Karpuskiene, V. (2007), Undeclared work, tax evasion and avoidance in Lithuania.
Paper presented at colloquium of the Belgian Federal Service for Social Security
on Undeclared Work, Tax Evasion and Avoidance, Brussels, June.
Kaufmann, D. and A. Kaliberda (1996). Integrating the unofficial economy into
the dynamics of post-socialist economies: a framework for analyses and evi-
dence. In B. Kaminski (ed.) Economic Transition in Russia and the New States of
Eurasia. London: M.E. Sharpe, pp. 139–164.
Lacko, M. (1999), Electricity intensity and the unrecorded economy in post-
socialist countries. In E.L. Feige and K. Ott (eds), Underground Economies in
Transition: Unrecorded Activity, Tax Evasion, Corruption and Organized Crime.
Aldershot: Ashgate, pp. 141–166.
Lemieux, T., B. Fortin and P. Frechette (1994), The effect of taxes on labor supply
in the underground economy. American Economic Review, 84(1), 231–254.
Leonard, M. (1994), Informal Economic Activity in Belfast. Aldershot: Avebury.
Lewis, A. (1959), The Theory of Economic Growth. London: Allen and Unwin.
MacDonald, R. (2004), Fiddly jobs, undeclared working and the something for
nothing society. Work, Employment and Society, 8(4), 507–530.
McCrohan, K., J.D. Smith and T.K. Adams (1991), Consumer purchases in
informal markets: estimates for the 1980s, prospects for the 1990s. Journal of
Retailing, 67(1), 22–50.
Neef, R. (2002), Aspects of the informal economy in a transforming country: the
case of Romania. International Journal of Urban and Regional Research, 26(2),
299–322.
OECD (2002), Measuring the Non-observed Economy. Paris: OECD.
ONS (2005), Identifying Sources on Entrepreneurship and the Informal Economy.
London: Office of National Statistics.
Pahl, R.E. (1984), Divisions of Labour. Oxford: Blackwell.
Pavlovskaya, M. (2004), Other transitions: multiple economies of Moscow house-
holds in the 1990s. Annals of the Association of American Geographers, 94:
329–351.
Pedersen, S. (2003) The Shadow Economy in Germany, Great Britain and
Scandinavia: A Measurement Based on Questionnaire Surveys. Copenhagen: The
Rockwool Foundation Research Unit.
Persson, A. and H. Malmer (2006), Purchasing and Performing Undeclared Work
in Sweden. Part 1: Results from Various Studies. Malmo: Skatteverket.
Portes, A. (1994), The informal economy and its paradoxes. In N.J. Smelser and R.
Swedberg (eds), The Handbook of Economic Sociology. Princeton, NJ: Princeton
University Press, pp. 121–143.
Renooy, P., S. Ivarsson, O. van der Wusten-Gritsai and R. Meijer (2004),
Undeclared Work in an Enlarged Union: An Analysis of Shadow Work: An In-
depth Study of Specific Items. Brussels: European Commission.
Rodgers, P. and C.C. Williams (2009), The informal economy in the former Soviet
Union and in Central and Eastern Europe. International Journal of Sociology,
39(2), 3–11.
Rose, R. (2005), Insiders and Outsiders: New Europe Barometer 2004. Studies
in Public Policy 404, Centre for the Study of Public Policy, University of
Strathclyde, Glasgow.


Ross, R.J.S. (2004), Slaves to Fashion: Poverty and Abuse in the New Sweatshops.
Ann Arbor, MI: University of Michigan Press.
Round, J. and C.C. Williams (2008), Everyday tactics and spaces of power:
the role of informal economies in post-Soviet Ukraine. Social and Cultural
Geography, 9(2), 171–185.
Round, J., C.C. Williams and P. Rodgers (2008), Corruption in the post-Soviet
workplace: the experiences of recent graduates in contemporary Ukraine. Work,
Employment and Society, 22(1), 149–166.
Sassen, S. (1997), Informalization in advanced market economies. Discussion paper
20. Issues in Development, International Labour Office, Geneva.
Schneider, F. (ed.) (2008), The Hidden Economy. Cheltenham, UK: Edward Elgar.
Schneider, F. and D.H. Enste (2000), Shadow economies: size, causes and conse-
Schneider, F. and D.H. Enste (2002), The Shadow Economy: An International
Survey. Cambridge, UK: Cambridge University Press.
Sedlenieks, K. (2003), Cash in an envelope: corruption and tax avoidance as an
economic strategy in contemporary Riga. In K.-O. Arnstberg and T. Boren
(eds), Everyday Economy in Russia, Poland and Latvia. Stockholm: Almqvist &
Wiksell, pp. 89–104.
Sepulveda, L. and S. Syrett (2007), Out of the shadows? Formalisation approaches
to informal economic activity. Policy and Politics, 35(1), 87–104.
Small Business Council (2004), Small Business in the Informal Economy: Making
the Transition to the Formal Economy. London: Small Business Council.
Smith, A. and A. Stenning (2006), Beyond household economies: articulations and
spaces of economic practice in postsocialism. Progress in Human Geography,
30(2), 1–14.
Smith, S. (1986), Britain’s Shadow Economy. Oxford, UK: Clarendon.
Thomas, J.J. (1992), Informal Economic Activity. London: Harvester Wheatsheaf.
TNS Infratest, Rockwool Foundation and Regioplan (2006), Feasibility study on
a direct survey about undeclared work VC/2005/0276. Directorate General of
Employment and Social Affairs, European Commission, Brussels.
Wallace, C. and C. Haerpfer (2002), Patterns of participation in the informal
economy in East-Central Europe. In R. Neef and M. Stanculescu (eds), The
Social Impact of Informal Economies in Eastern Europe. Aldershot, UK:
Ashgate.
Wallace, C., C. Harpfer and R. Latcheva (2004), The Informal Economy in East-
Central Europe 1991–1998. Vienna: Institute for Advanced Studies.
Warde, A. (1990), Household work strategies and forms of labour: conceptual and
empirical issues. Work, Employment and Society, 4(4), 495–515.
Williams, C.C. (2004), Cash-in-hand Work: The Underground Sector and Hidden
Economy of Favours. Basingstoke, UK: Palgrave Macmillan.
Williams, C.C. (2006), The Hidden Enterprise Culture: Entrepreneurship in the
Underground Economy. Cheltenham, UK: Edward Elgar.
Williams, C.C. (2007), Tackling undeclared work in Europe: lessons from a study
of Ukraine. European Journal of Industrial Relations, 13(2), 219–237.
Williams, C.C. and J. Round (2007), Beyond negative depictions of informal
employment: some lessons from Moscow. Urban Studies, 44(12), 2321–2338.
Williams, C.C., J. Round and P. Rodgers (2007), Beyond the formal/informal
economy binary hierarchy. International Journal of Social Economics, 34(6),
402–414.



Economies: Implications for Work and Welfare. London: Routledge.
Woolfson, C. (2007), Pushing the envelope: the ‘informalization’ of labour in
post-communist new EU member states. Work, Employment and Society, 21(3),
551–564.
Ybarra, J.-A. (1989), Informalisation in the Valencian economy: a model for under-
development. In A. Portes, M. Castells and L.A. Benton (eds), The Informal
Economy: Studies in Advanced and Less Developing Countries. Baltimore, MD:
Johns Hopkins University Press, 216–227.
Žabko, M.A. and F. Rajevska (2007), Undeclared work and tax evasion: case of
Latvia. Paper presented at colloquium of the Belgian Federal Service for Social
Security on Undeclared Work, Tax Evasion and Avoidance. Brussels, June.


6 Regional patterns of the shadow
economy: modelling issues and
evidence from the European Union
Helmut Herwartz, Friedrich Schneider and
Egle Tafenau
6.1 INTRODUCTION
Shadow economy has so far drawn attention mainly on the level of coun-
tries. Its extent has been estimated relying on country-specific time series
and surveys or cross-sectional and panel data for several economies. The
regional aspects of the shadow economy have received less attention,
though this could deliver additional important information for policy
decisions and has also advantages for purposes of estimation and model
diagnosis.
In considering the merits of the regional estimates of the shadow
economy for policy decisions, two main benefits arise. First, information
about the differences in the extent of the shadow economy helps govern-
ments to set accurate targets in combating shadow economy or detecting
tax fraud. This is especially the case if it is possible to gather information
on region-specific sources of hidden economic activity. Second, supporting
the design of regional policies, the estimates of the shadow economy can
be employed to derive estimates of total economic activity in each region.
Concerning the estimation itself, gains arise from the employment of
regional data if the extent of the shadow economy is measured based on
aggregated data, for example with the help of the Multiple Indicators
Multiple Causes (MIMIC, see also Chapters 1 and 2) approach. The extent
of the shadow economy is modelled in this framework as an outcome of
causal variables like taxes, size of the governmental sector and labour
market conditions. At the same time, shadow economy clearly affects
these phenomena. For instance, Herwartz & Theilen (2010) provide a
theoretical model for the joint determination of hidden economic activity
and tax rates and institutions under cross-national financing. Thus, esti-
mating the extent of the shadow economy based on the MIMIC approach
201


is likely subject to an endogeneity problem. As argued below in Section

6.3.1 for specific variables, endogeneity moderates if regional data is used.
The actual level of such variables in the regions is not only affected by
local decisions but also by those of the central government that rely less on
the specific developments in individual regions. Moreover, if the regional
units of an empirical analysis are not administrative units, the endogeneity
problem is relieved further.
This chapter has three aims. First, we discuss the peculiarities in the
modelling of the extent of the shadow economy at a regional scale and
adjust the MIMIC approach to analyse this phenomenon across regions.
Second, the extent of the shadow economy is derived for the NUTS 2
regions1 of the European Union (EU), including spatial adjustments to the
model. The EU regional policies are directed to the NUTS 2 level regions
and, therefore, knowledge about the different economic phenomena at this
regional level is of special interest. Third, the gap between potential and
actual tax revenues is quantified based on the shadow economy estimates
that are derived from the MIMIC model implementation. In this chapter,
shadow economy is understood as all legal and illegal economic activi-
ties that should be included in the gross domestic product (GDP) but are
hidden from authorities for avoiding taxes or regulations.
The few currently available estimates of the shadow economy at
regional levels are derived throughout for individual countries and,
thus, lack comparability within integrated economic entities like the EU.
Chaudhuri, Schneider and Chattopadhyay (2006) derive the shadow
economy quotas for Indian regions by applying the MIMIC approach
separately for each region. Based on surveys, Mummert and Schneider
(2001) analyse the extent of the unofficial economy in German regions and
Mróz (2005) documents regional estimates for Poland. To our knowledge,
no regional shadow economy estimates are available that are comparable
across a set of countries.
In this respect, the EU constitutes an interesting case. First, the correct
measurement of the total economic activity as measured by the GDP is
of particular interest at the regional level. A large share of the EU budget
is spent on regional development and convergence targets through the
Structural Funds and the Cohesion Fund. In determining the eligibility
of a region for funding from these funds, the GDP per capita is the most
important criterion. Though the national statistical agencies are required
to include in the GDP both official and unofficial economic activities
(ESA, 1995), the countries conform with this obligation to varying
extents (United Nations, 2008). By underreporting economic activity
in certain regions it is possible to gain eligibility for extensive financial
resources.


Modelling regional patterns of the shadow economy 203
Second, as the financial supports from the EU budget require co-

financing, the ability to use these resources for development projects of
a region depends on the own financial capabilities of the regions. This,
in turn, depends on the one hand on the extent and quality of economic
activities undertaken in the respective regions and, on the other hand, on
the tax compliance behaviour of the local residents. Thus, the analysis of
the tax gaps – the difference between the potential and actual tax revenues
– helps to evaluate the ability to allure sufficient resources for stimulating
regional development.
Our empirical analysis reveals that in some countries the variation of
the extent of shadow activities is considerable, such that regionally dif-
ferentiated policies might be required to combat shadow economy. The
calculated national average shadow economy quotas are generally con-
sistent with previous estimates from country level analyses. The tax gaps,
implied by estimated shadow economy quotas, vary considerably across
the regions even if corrected for the size of population. Moreover, the tax
gaps tend to be larger in wealthier regions. Therefore, if a government
aims to increase tax revenues by detecting shadow activities, it might be
reasonable to concentrate on relatively rich areas.
The chapter is organised as follows. The measurement issues to deter-
mine the extent of the shadow economy at a regional scale are discussed
in Section 6.2. It is first argued that to derive internationally comparable
regional estimates of shadow economic activity the MIMIC approach is
the only applicable method as a consequence of data limitations. Second,
the MIMIC method is sketched with a particular focus on spatial interac-
tions and their effects in regression models. Third, explicit approaches to
implement the spatial interactions in the context of the MIMIC model are
given.
In Section 6.3 the causes and indicators of the shadow economy that are
considered in the MIMIC model are introduced along with their empiri-
cal counterparts. Further, the parameter estimation results for different
specifications of the MIMIC model are discussed.
Section 6.4 sketches the scaling of the shadow economy quotas from
the preliminary index obtained by implementing the MIMIC model.
Further, the estimates for the extent of the shadow economy are given for
both regional and aggregated country levels. In addition, the variation of
shadow economic activity within countries is discussed.
Model implied tax gaps are discussed in Section 6.5. The estimation of
the potential tax revenues is one of the main uses of the shadow economy
estimates. It is argued that the tax gaps are not necessarily largest in the
regions with most extensive shadow economy, but rather in rich regions.
The final section offers a brief summary of the chapter.


6.2 MEASURING THE SHADOW ECONOMY

6.2.1 Considerations for Regional Estimation
Measuring the shadow economy is difficult on three main grounds. First,

the variety and lack of rigour of the definitions of the shadow economy
leave unclear what is trying to be measured. Second, all methods that have
been used for measuring the shadow economy can be criticised. Third,
in choosing the methods, econometricians are often constrained by data
availability. A comprehensive review of the methods that are used for
estimating the extent of the shadow economy is presented in Schneider
and Enste (2000).
Differences in the shadow economy estimates can be ascribed to all
three issues. For example, based on direct household surveys only the
aspects of hidden labour market activities can be analysed. Though some
authors consider the results of such surveys to accurately reveal the situ-
ation in this part of the shadow economy (Williams, 2004), others argue
that the extent of hidden labour activities is underestimated as agents are
usually reluctant to reveal their unofficial activities (Schneider and Enste,
2000; Giles, 1999).
The subject of measurement is wider in the case of indirect methods.
These methods, based on discrepancies between national incomes and
expenditures, electricity consumption and cash in circulation are also
supposed to include shadow transactions between firms or illegal activi-
ties. However, data reliability and the stability of the relationships across
time and countries limit the scope of indirect measurements. Moreover,
since methods based on electricity consumption or currency demand rely
implicitly on the invariance of economic relationships influencing electric-
ity or money demand, several authors argue that the size of the shadow
economy is overestimated when implementing these methods (Schneider
and Enste, 2000; Giles, 1999).
Also the model-based MIMIC approach applies to a rather wide
definition of the shadow economy, possibly including illegal activities. In
this framework, the size of the shadow economy is modelled as a latent
variable. The strength of the MIMIC method is its reliance on both the
causes and indicators of the shadow economy. Of course, this method also
has weaknesses. There are several variables that could be interpreted as
causes of the shadow economy but also as indicating its size. Moreover,
in order to derive easily interpretable results, it is necessary to obtain at
least one or two shadow economy estimates based on other methods or
previous research. Thus, the reliability of the final results also depends
on these estimates. According to Breusch (2005) the scaling procedures


themselves are unreliable. In addition, the method has been criticised for
delivering estimates of the potential rather than the actual extent of the
shadow economy (Eilat and Zinnes, 2000). A dispute on the suitability of
the MIMIC method for measuring the size of the shadow economy can be
found in Breusch (2005) and Dell’Anno and Schneider (2006).
The direct methods are, in general, not applicable for international com-
parisons as such surveys are not conducted in all countries and the survey
questions are not standardised. In addition, the questions about unofficial
work are included in the surveys only occasionally, making the derivation
of time series even for just one country difficult. However, currently, the few
available country-specific shadow economy estimations at a regional scale
are based mainly on surveys (Mummert and Schneider, 2001; Mróz, 2005).
Due to the incomparability of the survey data, cross-country analyses
of the extent of shadow economic activity rely on indirect methods or the
latent variable approach. Obviously, the methods to obtain internation-
ally comparable shadow economy estimates at a regional scale are also
constrained to those. However, compared to the country-level analyses,
additional limitations arise.
First, compared to the country level, data availability is more severely
constrained at regional scales. On the one hand, there are variables like
electricity consumption that are collected only on the national scale,
though also regional figures would be possible. On the other hand, for
indicators like money demand it is extremely difficult to derive reliable
estimates at a regional level. Thus, both the currency demand and electric-
ity consumption method cannot be implemented to measure the extent
of the shadow economy at a regional scale. In choosing an estimation
method for the size of the shadow economy, availability of the necessary
indicators plays a crucial role.
Second, regional data is often characterised by spatial autocorrelation.
Omission of spatial effects can bias the estimation results or inference as
it has been discussed for example in the convergence literature (Ertur,
Le Gallo and Baumont, 2006). This also applies to the shadow economy
measurement. In case of the MIMIC approach, both the cause and the
indicator variables may be related to hidden economic activity in the
neighbouring regions and, thus, respective adjustments of the estimation
procedure are necessary.
In the analysis of regional issues there arises the question of an appro-
priate level of regional disaggregation. The choice depends on the question
at hand, but also on data availability. In case of the EU, the NUTS 2 level
regions are mostly used in regional analyses. This is motivated by the EU
regional policy that concentrates on this level and, as a consequence, the
data collection is also most comprehensive for the NUTS 2 regions.


6.2.2 The MIMIC Approach
According to Giles (1999) and Schneider and Enste (2000), the MIMIC
approach is currently considered to be most reliable for estimating the
extent of the shadow economy as it simultaneously exploits the infor-
mational content of both its causes and indicators. The first application
of MIMIC modelling to measure shadow economy was Frey and Weck-
Hanneman (1984), but some aspects of the MIMIC model had been used
earlier to determine the relative size of the shadow economy in different
countries (Frey and Weck, 1983).
MIMIC modelling is based on Zellner’s (1970) approach of estimating
regressions with unobserved independent variables, later generalised by
Jöreskog and Goldberger (1975). It belongs to a group of models that
consist of linear structural relationships allowing unobservable variables.
In the present context, the unobservable (latent) variable is the extent of
the shadow economy.
The core of a MIMIC model consists of a structural equation and meas-
urement equations. The structural equation relates the cause variables to
the unobserved latent variable
hr 5 grxr 1 zr (6.1)
where hr stands for the latent variable (shadow economy) in region r and
the q 3 1 vector xr 5 (x1r, x2r, . . ., xqr) r collects the causes of shadow activi-
ties. The corresponding parameters are denoted by g 5 (g1, g2, . . ., gq) r
and zr is an error term. All variables are measured as deviations from their
means, thus, E [ xr ] 5 0 and E [ hr ] 5 E [ zr ] 5 0. Moreover, the error term
zr is assumed to be uncorrelated with the causes, E [ xrzr ] 5 0, its vari-
ance is Var [ zr ] 5 c, and the covariance matrix of the cause variables is
E [ xr xrr ] 5 F.
The measurement equations relate indicators to the latent variable,
yr 5 lhr 1 er (6.2)
where yr 5 (y1r, y2r, . . ., ypr) r is a p 3 1 vector of indicators of the latent

variable and l 5 (l1, l2, . . ., lp) r are the model parameters quantifying the
impact of the shadow economy onto its indicators. The white noise error
terms in the measurement equations are denoted as er 5 (e1r, e2r, . . ., epr) r.
Again, E [ yr ] 5 E [ er ] 5 0 and E [ xrerr ] 5 0, E [ erhr ] 5 0, E [ erzr ] 5 0 by
assumption. Finally, the covariance structure of the disturbances in the
measurement Equation (6.2) is given by E [ ererr ] 5 Qe.
In order to estimate the model, the theoretical or implied covariance


matrix of the observed variables is derived based on the reduced form of

the model by substituting the latent variable from Equation (6.1) into the
measurement Equation (6.2).
yr 5 lgrxr 1 nr, nr 5 lzr 1 er. (6.3)
The estimation of the model relies on fitting the implied covariance

matrix of the observed variables to the sample covariance, such that
the difference between the two matrices is minimised. The implied cov-
ariance matrix of the observed variables is S (q) 5 E [ zr zrr ] . The vectors
q 5 (lr, gr, c, vech (F) r, vech (Qe) r) r and zr 5 (yrr xrr) r collect all model
parameters2 and the observable variables (causes and indicators) of the
model, respectively. By nature of Equation (6.3), the implied covariance
matrix is
l (grFg 1 c) lr 1 Qe
S (q) 5 c d.
lgrF
(6.4)
Fglr F
The sample covariance matrix S is
S5 c d.
E [ yryrr ] E [ yr xrr ]
(6.5)
E [ xryrr ] E [ xr xrr ]
The covariance matrix of the cause variables is fixed to be equal to the

sample counterpart, F 5 E [ xr xrr ] . Still, the rest of the parameters is not
identified. To identify the parameters, one of the elements in l has to be
fixed. Then, there remain p 1 q 1 1/2 p(p 1 1) free parameters that have to
be estimated based on 1/2(p 1 q)(p 1 q 1 1) equations.
The model parameter q estimated by minimising a distance measure
with respect to q. Common distance criteria, Unweighted Least Squares
(ULS), Generalized Least Squares (GLS) and Maximum Likelihood
(ML), are formalized as follows:
FULS 5 tr e [ S 2 S (q) ] 2 f ,
1
(6.6)
2
FGLS 5 tr e [ I 2 S (q) S21 ] f ,

1
(6.7)
2
FML 5 ln 0 S (q) 0 1 tr { SS21 (q) } 2 log 0 S 0 2 (p 1 q) . (6.8)
The ML objective function is derived under the presumption of a

multivariate normal distribution. In the case of the GLS criterion, the


underlying assumptions are less restrictive. The observations have to be

independent and identically distributed with any distribution character-
ised by moderate kurtosis (Bollen, 1989). Both methods are suitable for
inferential purposes if the distributional assumptions are satisfied. The
ULS objective function is not derived from a distributional assumption.
Thus, inferential statistics cannot be obtained directly (Backhaus et al.,
2006) but can be derived by means of resampling methods.
6.2.3 Spatial Effects
As discussed in Section 6.2.1, regional economic phenomena are often

spatially correlated or influenced by economic processes in neighbouring
regions. In response to this, there has arisen an extensive literature on the
impact of spatial interactions in econometric estimation and the ways to
improve the estimation procedures by considering the spatial effects.
Anselin (1988) motivates several potential triggers of empirically
observed spatial economic dependencies. First, the regional units are
artificial and might not correspond to areas that form a comprehensive
economic unit. Moreover, such economic areas might overlap. These
issues can invoke measurement errors. Second, there are spatial exter-
nalities and spillovers, arising for example from the mobility of agents
and trade linkages between regions. Third, relationships linking economic
phenomena might not be uniform across space. LeSage and Pace (2009)
mention, in addition, omitted or unobserved variables as the reasons of
spatial dependencies.
Regarding quantitative modelling, the spatial econometrics literature
differentiates between spatial heterogeneity and spatial autocorrelation.
Spatial heterogeneity refers to the instability of the functional form of a
model or its parameters across spatial units. It can usually be corrected
for by assuming that the model parameters depend on some characteris-
tics of the regions like location or magnitude. Spatial autocorrelation has
similarities to the serial correlation often found in time series: a variable
is correlated with its lagged value. However, this phenomenon is multi-
dimensional in the spatial context: a spatially lagged variable has to be
constructed taking into account the magnitude of the variable in several
regions as the spatial units are economically connected to several other
regions. Ignoring the interregional linkages may result in biased inference
or estimation.
According to Anselin (1988) special estimation techniques are necessary
in the case of two spatial specifications of the standard regression models.
First, the dependent variable might be determined also by its own value
in other regions, the spatially lagged dependent variable. Such a model is


called spatial autoregressive or, shortly, spatial lag model. Second, eco-
nomic phenomena might be characterised by spatially autoregressive dis-
turbances – the spatial error model. While ignoring the spatial structure of
the errors influences only the efficiency of ordinary least squares estimates
and, thus, results in biased inference, neglecting the effect of the spatially
lagged dependent variable leads to biased estimation.
In addition, spatial structures may arise through the explanatory
variables of an empirical model: an economic phenomenon in a region
is influenced not only by variables characteristic to that region but also
by corresponding characteristics of other regions. This problem can be
solved by augmenting the model with spatially lagged explanatory vari-
ables. Such a model can be estimated by means of standard econometric
methods. However, the problem of multidimensionality of the spatially
lagged variables also occurs here.
To overcome the multidimensionality problem, it is common to intro-
duce a spatial weight matrix that describes to what extent the effects from
each region are taken into account. In this matrix the regional weights are
collected for each geographic entity. Thereby a spatially lagged variable is
calculated as a weighted average of the respective variable in the relevant
regions.
Several approaches are applicable to determine the weights. A simple
and widely followed approach relies on the contiguity of regions or
the distance between regions (choosing the weight matrix is discussed,
for instance, in Getis and Aldstadt, 2004; Ord, 2010). In the case of
the former, all regions that have a border with the region of interest
can be considered as relevant and are assigned a positive weight. The
weights can be equal or unequal, for example, proportional to the
common border line. The non-contiguous regions receive zero weights.
Moreover, it is also feasible to consider, in addition, the neighbours of
neighbouring regions as relevant. In that case the regions that locate
further away may get smaller weights in the derivation of the spatially
lagged variable.
In the case of a weight matrix based on distances, several techniques
can be applied. A simple approach is to introduce a cut-off distance
D: a region is considered in the calculation of the spatial lag only if the
distance between the centres of the two regions is less than D km. The
regions at a distance of more than D km receive zero weight. The positive
weights can be assigned to the relevant regions such that nearby regions
have a higher weight than the regions locating further away. A version of
a distance-based weight matrix is also used in Section 6.3.3 below. The
respective derivation of the weights is described in detail in Appendix
A6.1.1.


6.2.4 Spatial Effects in the MIMIC Model
Interactions between regions can also be expected to exist in the case of the
shadow economy. Similar to basic regression models, here the interactions
can show up in several ways. With regard to the effects in the structural
equation, the following scenarios can be distinguished.
1. The latent variable is influenced by its values in the related regions:
hr 5 rh [ W ] rh 1 grxr 1 zr,¬r 5 1,. . ., R, (6.9)
where R is the number of regions in the sample, rh and W are the

spatial autoregressive coefficient and weight matrix, respectively, and
[W]r denotes the rth row of W. The vector h collects the latent variable
for all R regions.
2. The error term in the structural equation is spatially autoregressive:
hr 5 grxr 1 zr,¬zr 5 rz [ W ] rz 1 yr, (6.10)
where rz is the spatial autocorrelation coefficient, yr a white noise

error term and the vector z collects the spatially autoregressive model
disturbances for all R regions.
3. Apart from region specific cause variables corresponding cross
regional measures also impact on the latent variable:
hr 5 grxr 1 rx ([ W ] r X) r 1 zr. (6.11)
Here rx is a q 3 1 spatial parameter vector and X the R 3 q matrix

collecting the sample information on cause variables.
A similar classification applies for the measurement equations.
1. Indicator variables are spatially autoregressive:
yr 5 ry ([ W ] rY) r 1 lhr 1 er, (6.12)
where ry is a p 3 p diagonal matrix of spatial autoregressive coeffi-

cients. The R observations of the p indicator variables are collected in
the R 3 p matrix Y.
2. Spatially autoregressive disturbances:
yr 5 lhr 1 er,¬er 5 re ([ W ] re) r 1 xr, (6.13)


where re is a p 3 p diagonal matrix of spatial autoregressive coef-

ficients, e the R 3 p matrix of the original error terms er and xr the
model disturbances that conform with the assumptions of the basic
MIMIC model.
3. The indicator variables reflect not only a local latent variable, but also
that in related regions:
yr 5 lhr 1 rh ([ W ] rh) r 1 er, (6.14)
where rh is a p 3 1 vector of the spatial effects coefficients (for each

measurement equation the spatial effects coefficient might differ) and
h the vector of the latent variable for all R regions.
The specifications including the spatially autoregressive latent variable

or spatially autoregressive error terms are rather demanding to imple-
ment as the model cannot be adjusted in such a way that the covariance
matrix of the observed variables implied by the model is constant across
the observations. The specifications (6.11) and (6.12) with spatial lags of
the observed variables can be expressed in a form that does not conflict
with this assumption. Therefore, these two specifications are considered
for the empirical assessment of shadow economic activity in Section 6.3.3.
For purposes of the model implementation, the specifications (6.11) and
(6.12) are adjusted. Notice that the Equation (6.11) is equivalent to
hr 5 gr (xr 2 |
rx ([ W ] rX) r) 1 zr, (6.15)
where |rx is a q 3 q diagonal matrix of spatial adjustment coeficients with

| ) 5 rr . Equation (6.12) can be converted to
gr (2r x x
yr 2 ry ([ W ] rY) r 5 lhr 1 er, (6.16)
In Equations (6.15) and (6.16) it is implicitly assumed that the latent

variable depends on spatially adjusted cause measures or is reflected in the
spatially adjusted indicator variables. In both cases the spatially adjusted
variables are used to determine the sample covariance matrix S. This
requires in a first step the estimation of the spatial adjustment coefficients
in ry or |
r x.
For the empirical applications below further restrictions are imposed
since the spatial adjustment coefficients cannot be estimated directly.
First, an identical spatial adjustment coefficient is set for all causal or
indicator variables. For the indicator variables individual spatial adjust-
ment coefficients are allowed in one specification of the model. Second,


the spatial adjustment coefficients are derived by varying them in the

range from –1 to 1. In full detail the estimation procedure is described in
Appendix 6.1.
6.3 ESTIMATION OF THE MIMIC MODEL
6.3.1 The Model
Generally, for the adaptation of the MIMIC model to quantify the hidden
sector the division of variables into indicators and causes is ambiguous.
The classification of a variable as an indicator or a cause depends on the
econometrician’s opinion on the direction of the linkages. Unfortunately,
the theoretical underpinnings are not very helpful in this aspect. The
choice of causes and indicators has been extensively discussed in Chapter
1 of this handbook. Our model set-up relies on the approach followed in
Dell’Anno, Gómez-Antonio and Alañon-Pardo (2007). Their analysis is
based on time series data for France, Spain and Greece. A similar model
has been used in Dell’Anno (2007) for measuring the extent of shadow
economic activity in Portugal.
On the side of potential causes of the shadow economy, three groups of
variables are distinguished. The first group consists of direct and indirect
taxes. In general, higher tax rates are expected to encourage hidden eco-
nomic activity. However, if instead of the imposed tax rates effective tax
rates are used – measured as the share of collected tax revenues in the tax
base – the data also reflect tax compliance. Therefore, depending on the
employed measure, the relation of the tax variable to the shadow economy
might be ambiguous.
Providing a second group of cause variables, the extent and quality of
public services or the scope of public control (on tax evasion, adherence
to regulations) is employed. Such variables should be negatively related
to the extent of shadow activities. Agents are more willing to pay taxes if
their contributions are used for good public services or the risk of being
detected is high if they hide economic activity.
Related to the tax rates and the extent of public services is the tax
morale – reflecting the agents’ attitude towards paying taxes. In countries
and regions where the agents perceive tax avoidance and tax evasion as an
improper behaviour the extent of the shadow economy is expected to be
smaller. If the tax morale is high in a country, relatively high tax rates can
be set without the threat that a considerable part of the economic activ-
ity will be hidden as a consequence. Good tax compliance can be moti-
vated by providing high-quality public services. Though the justifiability


of cheating on taxes has been studied in the World Value Surveys, for
many countries the most recent survey dates back to 1999. As especially
in Central and Eastern European countries the economic situation has
changed considerably since then, we consider the data to be unreliable for
measuring tax morale in later years and omit the tax morale variable from
the model.
Consisting of unemployment and self-employment rates, the third
group of causes indicates labour market conditions. Both variables are
expected to have a positive effect on the extent of the shadow economy.
Unemployed persons have more time to engage in shadow activities in
comparison with full-time employees. For the self-employed, it is easier to
hide their true incomes than is the case for employed persons.
Considering the indicators of the shadow economy, it is difficult to
find variables that directly correspond with the extent of the shadow
economy. In the literature measures related to cash demand, GDP and
labour force participation are often considered. For the EU model in
Section 6.3.3, the reported GDP per capita and the rate of labour force
participation are employed since data on cash demand are hardly avail-
able at a regional level of disaggregation. Notably, it is generally unclear
whether the extent of the shadow economy should have a positive or
a negative effect on these variables (Dell’Anno, Gómez-Antonio and
Alañon-Pardo, 2007). However, as the shadow activities are more likely
to be omitted from the GDP statistics, we expect that the GDP per capita
and shadow economy quota are negatively related. GDP per capita is
also used for the purpose of normalisation having a fixed coefficient of
–1.
Similarly, we expect the shadow economy to impact negatively on the
labour force participation rate, which is calculated with consideration of
labour force surveys and, thus, covers both shadow and official labour
force. The agents participating in the black labour market might be less
willing to report their activities. Moreover, since criminal activities are
also included in the shadow economy, the recipients of criminal income
might not work or not reveal their activities.
We acknowledge that the considered indicator and cause variables might
involve an endogeneity problem. For example, in the case of taxes, finan-
cial authorities in countries with a larger share of the shadow economy
need to impose higher tax rates to finance public services. However, as
the nominal tax rates are set at the country level and estimation is done
at the regional level, the endogeneity problem diminishes. Likewise, the
size of the public sector can be influenced by the extent of the shadow
economy: due to low tax revenues the public sector has to decrease. Again,
this problem is likely to be smaller at a regional than at the national level.


Though the budgets of the sub-national administrations are influenced

by tax revenues collected within such administrative units, there are in
addition central government expenditures flowing into regions. Such flows
smoothen the disparities in budgetary possibilities, which might have
arisen through shadow economy or low economic activity in a region.
Analogous arguments apply if the size of the public sector is measured in
terms of public sector employment. Local administrations have to fulfil
not only the duties taken by themselves but also those that are laid on
them by the central government. Moreover, the central government has
employees in different regions and such employment is less influenced by
the local tax returns than the public sector employment at country level by
nationally collected taxes. Generally, the endogeneity problem diminishes
if the analysed regions do not correspond to administrative units or have
only a minor administrative importance. This is the case for the NUTS 2
regions in some countries (Sweden, Hungary, Finland, Germany).
6.3.2 Data
A list of cause and indicator variables is provided in Table 6.1 along

with the regional level at which these measures are available. The data
are retrieved from the Eurostat regional database, except for the tax
wedges (drawn from Eurostat, 2006) and the value added tax (VAT) rates
(European Commission, 2009). The year of the analysis is 2004, chosen as
the year with overall least missing data.
Considering the correspondence of the indicators to their theoretical
counterparts, some comments are in order. Notably, the imposed tax rates
reflect neither the progressiveness of the tax system nor the differences in
the allowed deductions. The effective tax rate (paid taxes) captures these
issues to some extent. However, the paid taxes variable is affected by peo-
ple’s tax paying behaviour, and also by the effectiveness of tax collection
activities of the financial institutions. An improvement in these factors
would rather reduce than increase the extent of the shadow economy, as
opposed to the effect of the taxes themselves.
Another variable whose interpretation is not straightforward is the
share of public employment. It has been thought to indicate the goodness
of public services or the scope of public control. In fact, this measure can
also reflect the ineffectiveness of the public sector. If this were the case its
expected relationship with the shadow economy is positive since people
are not motivated to pay taxes for financing an overly large public sector.
Possibly, the relationship is also dependent on the country under observa-
tion. For example Dell’Anno, Gómez-Antonio and Alañon-Pardo (2007)
find contradictory evidence depending on the country.


Table 6.1 Causes and indicators of shadow economy
Variables Measures Regional level

Causes
Direct tax rates – Paid taxesa NUTS 2
– Tax wedgeb Country
Indirect tax rates Value added tax (VAT) rate Country
Size of public sector Employment by economic activity: NUTS 2
NACEc sectors L to Q / Total
Unemployment Unemployment rate (15 years and over) NUTS 2
Self-employment Number of self-employed / Labour force NUTS 2
Indicators
GDP per capita GDP in purchasing power standards NUTS 2
(PPS) per inhabitant
Labour force Economic activity rate (15 to 64 years) NUTS 2
participation
Notes:
a
Calculated using the data from households accounts: Paid taxes 5 (current taxes on income,
wealth, etc. 1 social contributions) / balance of primary income. The balance of primary
income is the income earned by the households (profits from self-employment, wages,
property income) minus property costs (rents, interests, etc.), without subtracting taxes and
adding transfers. The tax measures come from the secondary distribution of income account
of households. It relies on the balance of primary income, subtracts the taxes paid and adds
transfers received by the households, in order to derive the disposable income. Thus, the
paid taxes measure shows the share of paid direct taxes in households’ gross income.
b
Difference between the labour cost for an employer and the net wage his or her employee
takes home. Calculated for a single worker without children at two-thirds of average
earnings.
c
Classification of Economic Activities in the European Community.
The initial sample includes all NUTS 2 regions of the 27 EU member

states. Mainly due to missing data, some of these have been removed.
First, the overseas territories are omitted. Second, Bulgaria, Cyprus,
Luxembourg, Malta and Romania have been dropped due to missing
data for some of the variables. For the same reason, two regions of the
United Kingdom, North Eastern Scotland and Highlands and Islands,
are excluded. Finally, Denmark and Slovenia are not represented with
their NUTS 2 regions, but with the country as a whole since only country
level data are available. After these adjustments, 238 regions remain in the
sample. In the case of the remaining missing observations the data for cor-
responding NUTS 1 regions are used, if available, or national data if the
data for NUTS 1 regions are also missing.
The descriptive statistics documented in Table 6.2 show that there is
a lot of variation in the observed variables across the EU regions. This


Table 6.2 Descriptive statistics
Variable Min Max Mean Standard

deviation
Causes
Paid taxes, % 23 70 37.8 7.6
Tax wedge, % 20.0 48.9 40.4 6.3
VAT, % 0 25 18.8 2.8
Share of public employment, % 17.8 66.1 30.3 6.2
Unemployment rate, % 2.4 24.9 9.0 5.1
Self-employment rate, % 5.8 42.9 15.8 7.6
Indicators
GDP (PPS) per capita, € 7600 72500 21698 7602
Labour force participation, % 52.6 81.0 70.1 5.6
supports the presumption of considerable variation in the latent variable

shadow economy. The smallest variation is detected in the labour force
participation rates and in the nominal tax rates data. For the VAT rate,
it has to be noted that the lowest tax rates are to be found in some island
regions where the respective countries allow reduced rates. The extreme
example is Åland in Finland, where the VAT rate is set to zero.
In the analysis, all variables are used in the form of their relative differ-
ence from sample averages. This obtains a comparable scale for all vari-
ables and facilitates the rescaling and interpretation of the latent shadow
economic activity.
Considering the set up of the model and the chosen variables, it is
straightforward to expect that spatial interactions play a role in deter-
mining the extent of the shadow economy in a region. In the case of the
causes, the direction of the relationship between a variable and the shadow
economy is presumably the same, independent of whether the variable is
measured for the region under observation or in its neighbouring regions.
For example, when agents take the decision to start a job in the shadow
economy, they take into account the official labour market situation not
only in the local region, but also in these regions that are in a commut-
ing range. It can be expected that high unemployment rates in the nearby
regions have a positive effect on the extent of the shadow economy in a
region of interest as it reduces the probability of finding an official job,
even if commuting.
In the case of the indicators, spatial interactions might be somewhat
more complicated. For the GDP per capita, it is expected that shadow
economy is smaller in a region if the neighbouring regions have a high


GDP per capita. Notably, relatively high GDP per capita might signal
good labour market conditions and, thus, reduces the agents’ willingness
to take up a shadow activity. This consideration would mean that the
‘effective’ GDP per capita indicating the extent of the shadow economy
should be adjusted upwards, implying a negative spatial effects coefficient
r; see Equation (6.16).
With respect to the labour force participation, the effect can be twofold.
On the one hand, it is probably more difficult to get an official job in a
neighbouring region with a high labour force participation rate due to suf-
ficient domestic labour supply. On the other hand, a high rate of labour
market participation can indicate good labour market conditions. Thus,
the labour force participation rate should be adjusted downwards or
upwards.
6.3.3 The Estimated Model
To estimate the model, we have tried to use alternatively ML, GLS and
ULS algorithms outlined in Subsection 6.2.2. The results from the GLS
estimation are documented in Table 6.3.3
Including all cause and indicator variables, the full model refers to
model 1. In models 2–7, there is always one of the cause variables omitted,
to assess the robustness of the results. Due to the possible effects from the
neighbouring regions, in models 8, 9 and 10 spatial effects are taken into
account. The procedure is described in detail in Appendix 6.1. In models
8 and 9 the indicator variables are spatially adjusted: in model 8 with
an identical spatial effects coefficient and in model 9 with two distinct
parameters for each indicator variable. In model 10, it is assumed that the
cause variables have to be adjusted for spatial effects, using a homogenous
spatial effects parameter.
All of the estimated models have a very good fit, as indicated by the
high values of the adjusted goodness-of-fit index. As an overall diagnostic
of model fit and validity of specification a x2 test statistic conforms with
the hypothesis that the model implied and sample covariance matrix are
identical.
In the following discussion of the parameters’ significance we mostly
refer to a 5 per cent significance level. We acknowledge that the infer-
ence based on the GLS estimation can be biased if the spatial effects are
omitted from the model. However, as the ULS estimation combined with
resampling delivers qualitatively similar results, the bias is obviously not
too large. Moreover, the inferential conclusions based on the spatially
adjusted models are similar.
In light of the former discussion, the parameter estimates have their


Table 6.3 MIMIC parameter estimates
218
Variables M1 M2 M3 M4 M5 M6 M7 M8 M9 M10
Causes
Paid taxes −0.598* −0.501* −0.536* −0.518* −0.722* −0.700* −0.593* −0.522* −0.597*
(−5.89) (−5.01) (−5.06) (−5.39) (−6.51) (−6.56) (−5.59) (−5.01) (−5.92)
Tax wedge 0.363* 0.089 0.463* 0.304* 0.548* 0.233* 0.351* 0.247* 0.321*
(4.05) (0.92) (4.26) (3.46) (5.79) (2.37) (4.22) (2.81) (3.52)
VAT 0.548* 0.526* 0.640* 0.577* 0.492* 0.643* 0.610* 0.615* 0.548*
(5.64) (5.23) (6.33) (6.05) (5.25) (6.44) (6.22) (6.29) (5.71)
Share of public employment 0.193* −0.072 0.113 0.205* 0.267* 0.045 0.163* 0.078 0.171*
(2.73) (−0.94) (1.45) (2.50) (4.00) (0.58) (2.28) (1.02) (2.27)
Unemployment rate 0.128* 0.227* 0.190* 0.124* 0.151* 0.164* 0.138* 0.148* 0.133*
(4.13) (6.03) (6.05) (3.37) (4.41) (4.45) (4.74) (4.39) (4.26)

Self-employment rate 0.154* 0.188* 0.127* 0.176* 0.129* 0.165* 0.171* 0.150* 0.144*
(4.53) (4.91) (3.81) (4.57) (3.92) (4.84) (4.70) (4.89) (4.76)
27/07/2011 16:16
Indicators
GDP (PPS) per capita −1.0 −1.0 −1.0 −1.0 −1.0 −1.0 −1.0 −1.0 −1.0 −1.0
Labour force participation −0.284* −0.263* −0.255* −0.294* −0.267* −0.325* −0.249* −0.330* −0.459* −0.271*
(−6.77) (−7.10) (−7.37) (−6.33) (−6.56) (−6.69) (−6.69) (−6.96) (−6.80) (−7.41)

Statistics
Degrees of freedom 25 18 18 18 18 18 18 25 25 25
x2 33.07 30.88 14.05 32.69 18.91 19.48 18.25 32.07 22.73 23.82
RMSR 0.004 0.004 0.003 0.005 0.003 0.003 0.002 0.004 0.004 0.003
AGFI 0.950 0.942 0.974 0.939 0.965 0.963 0.966 0.951 0.965 0.964
R2h 0.539 0.693 0.362 0.688 1.031 –0.036 0.548 0.864 0.983 0.973
Spearman’s rank correlation 1.000 0.907 0.956 0.906 0.986 0.929 0.946 0.998 0.991 0.992
of h
Notes: Own calculations. GLS estimation results. Models 8–10 adjust for spatial effects. t-statistics in parentheses. * denotes 0 t-statistic 0 . 1.96.
219
GDP per capita is the normalising variable with its coefficient fixed to –1. RMSR – root mean squared residual, AGFI – adjusted goodness-of-fit
index, R2h 5 1 2 c/Var [ h ] – coefficient of determination for the latent variable.

27/07/2011 16:16
expected signs and are significant. However, in discussing the model set up
and data the effect of the measures of paid taxes and public employment
were highlighted to be potentially ambiguous. The share of paid direct
taxes in the income of the households can indicate not only tax burden
and tax compliance, but also the tax collecting efficiency of the financial
authorities. If the tax compliance or tax collecting efficiency argument
dominates, paid taxes should have a negative impact on the extent of the
shadow economy, as indicated by our estimates. This negative effect is
somewhat magnified by including the imposed tax rate and the size of the
public sector in the model (compare the results from model 1 with those
from models 3 and 5). If in addition unambiguous measures of tax paying
morale and governmental efficiency would be included, the coefficient
would probably be smaller. Such indicators are, however, not available
for the whole sample.
The interaction of the paid taxes measure and the size of the public
sector is further revealed in model 2, where the effect of the public sector
employment turns negative (though insignificant) after excluding paid
taxes from the model. In this case the share of the public sector in employ-
ment reflects both government inefficiency (positive effect to shadow
economy) and the extent of its services, including the capacity for assuring
tax compliance (negative effect to shadow economy).
The tax wedge and the VAT rate have a positive effect on the shadow
economy quotas as expected. Especially in the case of the VAT the effect is
rather stable across alternative model specifications. The same applies to the
public employment variable, despite the ambiguity of its interpretation. The
only exception is model 2 as discussed above. Thus, the share of the public
sector in employment can be argued to measure public sector inefficiency.
Both labour market variables, the unemployment and the self-
employment rate, have a significantly positive relation with the extent of
the shadow economy. However, their effect is smaller than the effect of
the taxes. Notably, such a direct comparison is possible as the size of the
coefficients is relative to the normalising variable GDP per capita.
From the indicator variables the labour force participation rate is nega-
tively related with the index of hidden economic activity. As mentioned,
GDP per capita is chosen as the normalising variable. The correspond-
ence of the signs of other parameter estimates to their expected direction
justifies the normalisation to a negative parameter value.
Omitting one of the cause variables from the model specification has
only minor effects on the parameter estimates as can be seen from models
2–7. In most cases there are no changes in either the signs or in the sig-
nificance of the estimates. The exceptional cases are the specifications 2,
3 and 7. In models 3 and 7, the single difference is the non-significance


of the share of public employment. Omitting the regionally available tax

measure, the share of direct taxes in households incomes, in addition to the
effect of the tax wedge becomes non-significant (model 2).
In models 8–10 cause or indicator variables have been adjusted for
spatial effects. For the specifications 8 and 10 an identical spatial effects
coefficient is employed for all indicator and cause variables, respectively.
The estimation procedure delivered as the most appropriate spatial effects
coefficients –0.06 and 0.10 for indicator and causes variables, respectively.
When the spatial effects coefficient is allowed to differ across the two indi-
cators, GDP per capita of the neighbouring regions turns out to be irrel-
evant (r 5 0). Moreover, the argument of good labour market conditions
dominates in case of labour force participation as shown by the negative
spatial effects coefficient r 5 –0.18.
The inclusion of spatial effects reduces the estimation uncertainty of the
latent variable as indicated by the high values of the coefficient of determi-
nation for the latent variable, R2h, in the spatial models. Thus, combining
the MIMIC approach with spatial effects enables us to obtain more precise
estimates for the extent of the shadow economy. However, the structural
parameter estimates linking causes and indicators with the latent variables
are not strongly influenced by the inclusion of spatial effects. As the only
exception, in model 9 public employment loses its significance. Also the
resulting ordering of the regions according to the extent of the shadow
economy (discussed in the next section) is highly correlated in the spatially
adjusted and spatially unadjusted models, as indicated by the high values
of the Spearman’s rank correlation in Table 6.3. In the following discus-
sion we consider model 9 to be most reliable due to its lowest estimation
uncertainty as indicated by the coefficient of determination for the latent
variable, R2h.
6.4 SHADOW ECONOMY ESTIMATES
6.4.1 Deriving Shadow Economy Estimates
Approaches used in the literature

Due to the normalisation, the MIMIC method obtains only a pre-
liminary index for the extent of the shadow economy. To derive explicit
shadow economy quotas several approaches have been employed in
the literature. Whatever the specific method, one or two base values are
needed for anchoring.4 These can be taken from previous research (Frey
and Weck-Hanneman, 1984; Chaudhuri, Schneider and Chattopadhyay,
2006; Dell’Anno, 2007; Dell’Anno, Gómez-Antonio and Alañon-Pardo,


2007; Buehn and Schneider, 2008) or from an alternative set of estimates

obtained for example by means of the currency demand approach (Giles,
1997; Schneider, 2007).
So far the MIMIC modelling of the shadow economy has been imple-
mented in time series (Giles, 1997; Dell’Anno, 2007; Dell’Anno, Gómez-
Antonio and Alañon-Pardo, 2007; Buehn and Schneider, 2008) or in a
pooled cross section time series context (Frey and Weck-Hanneman,
1984; Schneider and Bajada, 2005; Schneider, 2007). If the latent variable
is assumed to be directly related to the percentage share of the shadow
economy in the GDP, one needs two base values: one to fix the overall
extent of the shadow economy and one for the step size (Frey and Weck-
Hanneman, 1984). Only one base value is required if the latent variable can
be interpreted as the growth rate of the shadow economy (this is the case
if it is directly related to the GDP growth). Then, the remaining shadow
economy quotas are determined by means of integration (Schneider and
Enste, 2000; Dell’Anno, Gómez-Antonio and Alañon-Pardo, 2007). These
two approaches are probably the simplest, but not the only ones. The
derivation of the shadow economy estimates from the MIMIC index has
caused controversies and there is still no uniformly accepted approach
(Breusch, 2005; Dell’Anno and Schneider, 2006).
The reliability of the derived shadow economy estimates depends, thus,
on the choice of the anchoring procedure and the external or self-derived
anchoring values. To reduce the bias arising from measurement errors in
the base value, Dell’Anno and Schneider (2003) have used the average of
all available estimates for a year and a country. In addition, the credibility
of the shadow economy estimates derived from a MIMIC model is affected
by the estimation errors inherent in the preliminary index obtained from
Equation (6.1). However, this issue has rarely earned a remark, with the
exception of Giles and Tedds (2002).
The anchoring techniques that process base values for one or two arbi-
trary individual observations are sensitive not only to measurement errors
in the base values, but also to the chosen observations. To address the latter
dimension of sensitivity, an alternative choice of the base values is pro-
posed and adopted in this work. It takes the average extent of the shadow
economy and its variation in the sample as the points of reference. Even
though the sample mean and the variation are subjected to (smaller) meas-
urement errors, the results are likely less sensitive to a modification of initial
parameters as in the case of anchoring with item specific a priori quotas.
An anchoring technique based on sample moments

If all variables in the model are measured in the form of relative differences
from the sample average, it is reasonable to assume that this also holds


for the latent variable. The preliminary latent variable index h^ r calculated
from Equation (6.1) can be either assumed to be proportional to it or to
show the relative difference itself. In the first case one obtains
SEr 2 SE
h^ r 5 a , (6.17)
SE
where SEr is the region’s true percentage share of the shadow economy in
the officially measured GDP, SE is the corresponding sample mean and a
denotes a proportionality factor. Solving (6.17) for SEr yields
h^ r
SEr 5 SEa1 1 b. (6.18)
a
Thus, in order to transform the preliminary index to the shares of the

shadow economy in GDP, the average extent of the shadow economy and
the proportionality factor a need to be known. We assume that the sample
average and standard deviation of the shadow economy quota are given.
Then, based on Equation (6.18), the variance of the extent of the shadow
economy is
SE 2Var [ h^ r ]
Var [ SEr ] 5 . (6.19)
a2
Solving (6.19) for the proportionality factor a gives
Var [ h^ r ]
Å Var [ SEr ]
a 5 SE . (6.20)
The mean and the variance of the shadow economy estimates are derived
based on country level data as described in Section 6.4.2. The variance
of the latent variable Var [ h^ r ] is calculated as the sample variance of the
preliminary index.
The second possibility is to assume that the preliminary index reflects
the relative difference from the mean of the latent variable, that is, the
proportionality factor is unity, a 5 1. In this case, the shadow economy
quotas read as
SEr 5 SE (1 1 h^ r) . (6.21)
It is noteworthy that in Equation (6.21) only the sample average is

needed for calculating the shadow economy quota. This approach is


comparable to the adjustment method in Schneider and Enste (2000) and

Dell’Anno, Gómez-Antonio and Alañon-Pardo (2007).
6.4.2 Shadow Economy Estimates
The quantification of the MIMIC model can be used to extract a pre-

liminary index for the shadow economy by means of Equation (6.1). The
estimate for the variance of the error term in this equation is around
ĉ ≈ 0.03 in models without adjusting for spatial effects and between
0.001 and 0.006 in models with spatially adjusted variables. The pre-
liminary index of the latent variable is approximately between 20.65
and 0.65, with the range depending on the model specification. Thus, as
evidenced by the ratio of the standard error of the error term in the struc-
tural equation "c^ to one of the extreme values of the preliminary index,
the potential error featuring shadow economy quotas is in fact large
but it is smaller in the case of the spatial models. However, as discussed
below, the ordering of regions according to shadow economy quotas
appears plausible (see Appendix 6.2) and is overall in line with related
country level studies. For instance, the Western European regions are
generally characterised by smaller shadow economy quotas than the
Eastern European regions.
In deriving the shadow economy quota from the preliminary index,
SE 5 17.2% and Var [ SE ] 5 5.4% are used as the average and the stand-
ard deviation of the shadow economy quotas, respectively. These quanti-
ties originate as weighted averages from the national estimates for the EU
countries from Schneider (2007) (also documented here in Table 6.4). To
derive the weighted average and standard deviation, it is assumed that
each region within a country features the same shadow economy quota as
the country itself. Then, regional GDPs are used as weights to calculate
the moments.
In Table 6.4 the country level shares of the shadow economy in GDP
are documented for the spatially uncorrected full model (model 1) and
the spatial models (8–10). The national shadow economy quotas are
calculated as weighted averages of the regional shadow economy quotas,
using GDP as weights. All estimates are determined by means of the
anchoring procedure that relies on two reference parameters. In the case
of the spatially uncorrected model, besides the GLS results also the ULS
estimates are provided. For model 9 with individual spatial adjustment of
indicators complementary estimates are provided for the anchoring proce-
dure that relies only on the sample mean. For the purpose of comparison
results of Schneider (2007) (average quotas 2003/04 and 2004/05) are also
documented.


Table 6.4 The national averages of the shadow economy in 2004, in % of

GDPa
Country M1 M1b M8c M9d M9d,e M10f Schneider

(2007)g
Netherlands 9.2 9.7 9.5 9.7 12.2 9.3 11.6
Denmark 10.8 12.5 11.4 12.0 13.8 11.1 16.4
United 11.8 13.4 11.9 12.1 13.8 11.7 11.0
Kingdom
Austria 13.9 14.3 14.0 14.2 15.2 14.1 9.7
Germany 15.1 14.5 14.9 14.6 15.5 15.2 15.7
France 16.2 15.9 16.0 15.8 16.3 16.3 13.5
Sweden 15.8 14.2 16.0 15.9 16.3 15.7 16.8
Ireland 14.3 19.0 14.8 15.9 16.4 15.1 14.5
Finland 15.9 16.9 16.1 16.4 16.7 16.1 16.1
Slovenia 16.2 17.7 16.3 16.6 16.8 16.5 27.8
Estonia 16.3 18.7 16.2 16.6 16.8 16.7 38.7
Spain 16.5 18.7 16.5 16.9 17.0 16.9 20.9
Belgium 17.4 15.1 17.3 16.9 17.0 15.9 20.0
Czech 16.7 17.3 16.9 17.1 17.1 16.2 18.8
Republic
Portugal 18.8 19.4 19.0 19.3 18.6 19.0 20.8
Latvia 21.0 21.6 20.6 20.4 19.3 21.2 39.9
Italy 20.4 18.7 20.6 20.5 19.4 20.4 24.0
Greece 20.6 20.3 20.8 20.9 19.7 20.7 26.9
Slovak 22.0 24.9 21.9 22.3 20.6 22.3 18.7
Republic
Hungary 22.3 21.2 22.4 22.4 20.6 22.3 24.8
Lithuania 22.9 22.8 22.6 22.4 20.6 23.1 30.8
Poland 28.8 30.3 28.9 29.2 25.2 28.9 27.8
Notes: Own calculations.

a
Based on GLS estimates, if not indicated otherwise. Weighted average of regional
shadow economy estimates, with the regional GDP shares as the weights. The anchoring
procedure using average 5 17.2 and standard deviation 5 5.4 has been applied if not
indicated otherwise. The countries are ordered according to the shadow economy quota
obtained from model 1 (GLS, two parameter anchoring procedure).
b
Based on ULS estimates.
c
Indicators spatially adjusted, with an identical spatial effects coefficient.
d
Indicators spatially adjusted, each with its own spatial effects coefficient.
e
The anchoring procedure using only average 5 17.2.
f
Cause variables spatially adjusted.
g
Average over Schneider’s (2007) estimates for 2003/04 and 2004/05.


Table 6.5 Some former estimates of the shadow economy quota, % of GDP
Country Estimate Year Method Source

Greece 26–27 2002 MIMIC Dell’Anno et al.
(2007)
Italy 16.1–18.1 2004 ISTAT (2008)
Latvia 20.8 2003 Expenditures Bernotaitė and
Piskunova (2005)
Latvia 24–25 2003 Money demand Brēķis (2007)
Lithuania 20.8 2004 Survey of market LFMI (2005)
participants
Slovenia 18.8–23.0 2004 Employment Nastav and Bojnec
discrepancy (2007)
Comparing the ULS with GLS results reveals that the obtained shadow
economy quotas are not very sensitive with respect to the choice from
these estimation methods. The only country with a remarkable difference
is Ireland. The estimates from the spatial models are similar to the results
of the model without any spatial adjustment. Also the two anchoring
procedures deliver fairly similar results, with the two-parameter method
obtaining a slightly higher variance of the shadow economy quotas across
countries (and regions) in comparison with the procedure that processes
only the mean. As a consequence of the extremely high shadow economy
shares documented in Schneider (2007) for the Baltic countries the esti-
mated standard deviation is likely to exceed the true moment.
Juxtaposing the shadow economy estimates from the different models as
presented in Table 6.4 to those of Schneider (2007) reveals a high similarity
of the results. Conditional on model 9, the estimates for Finland, Ireland,
Germany, Poland, Sweden and the United Kingdom are especially salient
in this respect. The largest differences are obtained for Estonia, Latvia and
Slovenia (more than 10 percentage points), but also for Denmark, Greece
and Lithuania (more than 4 percentage points). Comparing the results
with some alternative estimates in the literature (see Table 6.5), it can be
concluded that in some cases our estimates probably underestimate the
extent of hidden economic activity (Slovenia, Greece), while in some other
cases the results in Schneider (2007) are likely to overestimate their true
counterparts (Latvia, Lithuania, Slovenia).
The regional estimates of the shadow economy quotas are graphically
displayed in Figures 6.1–6.3 and listed in Appendix 6.2.5 Figure 6.1 corre-
sponds to the results of model 9 with individual spatial adjustments of the
indicators and the two-parameter anchoring procedure. In Figure 6.2 the


>30
25–30
20–25
15–20
10–15
<10
Figure 6.1a Shadow economy (% of GDP) in the EU NUTS 2 regions,

based on model 9. The derivation of the shadow economy
quotas based on the sample mean and variance
results from the same model are shown, but the derivation of the shadow
economy quotas relies only on the sample average. The shadow economy
quotas following from the GLS estimation without any spatial effects
(model 1) and the two-parameter anchoring are displayed in Figure 6.3.
Panel (a) of each figure reveals that the share of the shadow economy is


>20
10–20
0–10
–10–0
<–10
Figure 6.1b Shadow economy’s deviation from the country average (%)
in the EU NUTS 2 regions, based on model 9. The derivation
of the shadow economy quotas based on the sample mean and
variance
highest in Eastern and Southern Europe. Panel (b) of each figure illustrates
the variation in the extent of the shadow economy in each country and
reveals that there is also a marked regional variation in several Western
European countries.


>30
25–30
20–25
15–20
10–15
<10

quotas based on the sample mean
According to the estimates based on two-parameter anchoring, the

regions with the highest shadow economy shares, in excess of 30 per cent
of the GDP, are in Poland (see Figure 6.1a). The region with the highest
shadow economy quota is Swietokrzyskie in the south-east of the country.
Such high shadow economy quotas are not found for any other country.


>20
10–20
0–10
–10–0
<–10
of the shadow economy quotas based on the sample mean
However, among the regions with shadow economy quotas exceeding 25

per cent of GDP are five Greek, three Italian and one Slovak regions. Thus,
in some Southern European regions hidden economic activity is likely
more widespread than in many Eastern European regions. Compared to
shadow economy quotas implied by the non-spatial model (Figure 6.3a),


>30
25–30
20–25
15–20
10–15
<10

quotas based on the sample mean and variance
the differences are small. However, if the anchoring procedure relies only
on the sample mean and it is, thus, assumed that the latent variable corre-
sponds to the true relative deviations of the shadow economy quotas from
the average, in no region the shadow economy is estimated to reach 30 per


>20
10–20
0–10
–10–0
<–10
of the shadow economy quotas based on the sample mean and
variance
cent of the GDP and only some Polish regions exceed the 25 per cent mark
as evidenced in Figure 6.2(a).
The Netherlands’ regions are mainly found at the other end of the cross
regional distribution of shadow economy quotas. In the Netherlands, the


model implied shadow economy quotas are mostly less than 10 per cent
(anchoring with mean and variance) or 12.5 per cent (anchoring with
mean) of GDP, followed by regions in the United Kingdom. The smallest
model-implied extent of hidden activity is found for the Finnish island
region Åland in which the VAT rate is set to zero. As the VAT rate has a
large effect in our estimation results the implied shadow economy quota
is close to zero for this region. In one of the models, the estimate is even
negative and, therefore, we consider the shadow economy estimate for
Åland to suffer from an estimation bias. The second exceptional region
among those with minor importance of unofficial activity is the Greek
region Ionia Nisia. Behind this result is the highest realization of the paid
taxes variable in the whole sample, being more than twice as high as for the
remaining regions of Greece. Therefore, this region can also be considered
to be an outlier and the corresponding shadow economy estimate should
be treated with care.
Looking at the variation of the extent of the shadow economy within
countries, panel (b) of Figures 6.1–6.3 again reveals similarities across
alternative approaches to model estimation. The main differences arise
from the anchoring procedure. Measured as percentage deviations from
the country mean the within-country variation is smaller in the case of the
one parameter procedure. It turns out that the poorer regions of a country
tend to exhibit a larger extent of the shadow economy than the richer
regions. For example, the regions in East Germany have higher shadow
economy quotas than those in West Germany.6 In Finland the peripheral
and poorer northern part of the country has higher shadow economy
quotas than the relatively rich southern areas of the country. In the case of
Italy the pattern from panel (a) of Figures 6.1–6.3 is strengthened: the rich
north-east regions show relatively small shares of the shadow economy in
the GDP, while in the southern regions the extent of the shadow economy
is more than 20 per cent above the country average.
There are also exceptions to the overall tendency that rich regions have
a smaller extent of the shadow economy than the poor ones. The two most
notable examples are Brussels in Belgium and Inner London in the United
Kingdom. Both of them are among the richest regions not only in the
respective country, but in the whole of Europe. With a shadow economy
quota around 20 per cent Brussels features the highest extent of model-
implied unofficial activity in Belgium. The estimated 15 per cent quota
for Inner London is the second largest share of the shadow economy in
the United Kingdom. Behind these results are the relatively high local
unemployment rates. In fact, Brussels and Inner London have the highest
unemployment rate in Belgium and the UK, respectively.
In Table 6.6, the coefficients of variation of the shadow economy quota


Table 6.6 The coefficients of variation of the shadow economy in 2004a
Country M1 M1b M9c M9c,d M10e Number

of regions
Austria 0.04 0.04 0.03 0.02 0.03 9
Belgium 0.15 0.18 0.14 0.09 0.18 11
Czech Republic 0.05 0.12 0.04 0.03 0.06 8
Germany 0.12 0.16 0.12 0.07 0.11 39
Denmark 0 0 0 0 0 1
Estonia 0 0 0 0 0 1
Spain 0.15 0.14 0.15 0.10 0.14 16
Finland 0.15 0.12 0.14 0.09 0.14 5
France 0.09 0.05 0.07 0.05 0.08 22
Greece 0.18 0.15 0.18 0.12 0.18 13
Hungary 0.05 0.04 0.04 0.03 0.04 7
Ireland 0.08 0.03 0.07 0.04 0.07 2
Italy 0.15 0.15 0.13 0.09 0.15 21
Lithuania 0 0 0 0 0 1
Latvia 0 0 0 0 0 1
Netherlands 0.06 0.05 0.05 0.03 0.08 12
Poland 0.07 0.08 0.07 0.05 0.07 16
Portugal 0.11 0.08 0.11 0.08 0.11 5
Sweden 0.03 0.04 0.03 0.02 0.03 8
Slovenia 0 0 0 0 0 1
Slovak Republic 0.13 0.19 0.14 0.10 0.13 4
United Kingdom 0.10 0.07 0.09 0.05 0.10 35

a
Based on GLS estimates, if not indicated otherwise. Based on the weighted average
and variance, the regional GDP shares have been used as the weights. The anchoring
procedure using average 5 17.2 and standard deviation 5 5.4 has been applied if not
indicated otherwise.
b
c
d
e
are documented. The coefficient of variation shows how much the shadow
economy quotas vary in a country across its regions. It is calculated as
the ratio of the weighted standard deviation to the weighted average of
the shadow economy quotas where regional GDP shares are used as the
weights. The table confirms the results from panel (b) of Figures 6.1–6.3:
there is a considerable variation in the shadow economy quotas within
some countries. Most outstanding in this respect are Belgium, Germany,
Spain, Finland, Greece, Italy and the Slovak Republic. The governments of


these countries should consider seriously whether ‘one-size-fits-all’ policies

are reasonable when taking measures against shadow economy or if they
should be regionally diversified. Relatively low variation in the shadow
economy quotas are observed in Austria, the Czech Republic, Hungary
and Sweden. Once again, these results are stable across different models.
6.5 TAX GAPS

Shadow economy estimates have been used for numerous purposes. First,
they are employed in understanding the interplay of unofficial economic
activity with other phenomena like corruption, institutional quality,
bureaucracy and taxes (Dreher, Kotsogiannis and McCorriston, 2009;
Friedman et al., 2000; Chaudhuri, Schneider and Chattopadhyay, 2006;
Johnson, Kaufmann and Zoido-Lobatón, 1998). This allows policy recom-
mendations for combating the shadow economy. Also, its effect on eco-
nomic growth (Eilat and Zinnes, 2002) or, more generally, its interactions
with the official economy (Schneider, 2008) have been analysed. Second,
they can be used to quantify the actual GDP of a country or actual per
capita incomes. For example, Feige and Urban (2008) rely on adjusted
GDP estimates to evaluate the success of transition for the post-communist
countries during the 1990s. In Herwartz, Schneider and Tafenau (2010)
corrected regional GDP estimates are employed to examine the accuracy of
the current pattern of eligibility for regional support from the EU. Third,
shadow economy estimates have been employed as an input to determine
the magnitude of uncollected tax revenues (Giles, 1997). This helps to decide
on the extent of measures to motivate tax compliance and detect tax fraud.
In this section, the regional estimates of the gap between potential and
actual tax revenues are derived for the NUTS 2 disaggregation of the EU.
In order to estimate the tax gap, we rely on the simplifying assumption
that the share of lost tax revenues in collected tax revenues is equal to
the share of the shadow economy in the GDP. The collected tax revenues
are obtained from the secondary distribution of the income account of
households, as the sum of current taxes on income, wealth and so forth,
and social contributions (from Eurostat, 2008). Thus, it is closely related
to the paid taxes measure employed for model estimation. Using this vari-
able builds upon the implicit suggestion that the share of participants in
the shadow activities is equal across different income and wealth groups.
On the one hand, high-income earners face higher average and marginal
tax rates and contribute more than proportionately to the tax revenues.
Moreover, they are likely less engaged in the shadow activities. As a
consequence the calculated tax gap might be overestimated. On the other


Table 6.7 Tax gaps of the EU countries in 2004
Country Aggregated tax gap, Tax gap per capita, Tax gap,
Mio. Euro Euro % in GDP
Austria 9536 1167 4.1
Belgium 15759 1513 5.4
Czech Republic 3280 321 3.7
Germany 94079 1140 4.3
Denmark 8380 1550 4.3
Estonia 284 210 2.9
Spain 28245 695 3.4
Finland 6646 1272 4.4
France 75525 1246 4.5
Greece 6975 631 3.8
Hungary 3905 387 4.7
Ireland 4050 999 2.7
Italy 72783 1251 5.2
Lithuania 643 187 3.5
Latvia 347 150 3.1
Netherlands 15655 961 3.2
Poland 10411 273 5.1
Portugal 5331 532 3.7
Sweden 15695 1746 5.5
Slovenia 985 493 3.6
Slovak Republic 1346 250 4.0
United Kingdom 55574 942 3.1
Notes: Own calculations. Based on the shadow economy estimates from the GLS
estimation with individual spatial effects coefficients of the indicator variables (model 9).
hand, however, the underlying collected taxes measure does not include
corporate and indirect taxes, which might result in underestimation of the
tax gap.
The tax gaps are derived from the shadow economy quotas implied by
model 9. The tax gaps at the country level are documented in Table 6.7.
The aggregated tax gaps are calculated as the country sums of regional
total tax gaps, given in Appendix 6.2. As a result, if the shadow activi-
ties could be transferred into the official economy, considerable amounts
of additional tax revenues could be collected or the tax rates could be
reduced due to a lower share of the shadow economy as discussed in
Section 6.3.1.
Relating the tax gaps to the official GDP account shows that hiding
economic activity has remarkable country specific implications. The tax


2000
1500
Tax gap per capita, Euros
EU
DE
IT
1000
ES
500
0
0 1 2 3 4 5 6 7 8
GDP per capita, 10,000 Euros
Figure 6.4 The GDP (PPP) per capita and the tax gap per capita in the
EU NUTS 2 regions
gaps relative to the official GDP vary between 2.7 and 5.5 per cent. By
construction, the share of the tax gap in the GDP depends on the one hand
on the level of taxes in the country and on the other hand on the extent of
the shadow economy. Therefore, the smallest relative tax gaps are calcu-
lated for countries like Ireland, Estonia, Latvia and the United Kingdom
with low direct tax rates. At the other end are Poland, Italy, Belgium and
Sweden, characterised either by very high shadow economy quotas or high
tax rates.
It is intuitive that in large and wealthy regions absolute tax gaps
might be larger than in small and poor regions, even if the former are
characterised by a lower extent of the shadow economy. Also, the tax
gap per capita tends to be larger in richer regions, as shown in Figure
6.4. This holds for both the whole sample and several countries, as, for
instance, Germany, Italy and Spain displayed separately in the figure.
Accordingly, the highest absolute tax gaps per capita are not necessarily
found in the regions with the largest extents of hidden economic activity.
Based on Figure 6.5 the relationship is rather negative, though for indi-
vidual countries the evidence is mixed. For Italy and Spain the negative
relationship seems to hold, but for Germany the relationship is rather
unclear.


2500
EU
DE
IT
2000 ES
Tax gap per capita, Euros
1500
1000
500
0
5 10 15 20 25 30 35
Share of shadow economy, % in GDP
Figure 6.5 The share of shadow economy in the GDP and tax gap per
capita in the EU NUTS 2 regions
Considering country specific correlation coefficients, it is confirmed that

generally the tax gap per capita is larger in wealthy than in poor regions,
as evidenced in Table 6.8. With regard to the correlations between tax gap
per capita and the extent of the shadow economy, such a clear statement
cannot be made. For most of the countries no statistically significant cor-
relation can be detected and for the few remaining economies the results
are contradictory: both negative and positive correlations can be found.
Therefore, the extent of the shadow economy should not be the only crite-
rion to decide from which regions the largest additional tax revenues could
be potentially attained. If the governments aim to reduce the tax gaps, it
might be reasonable to concentrate on the richer regions even if in those
regions the extent of hidden economic activity is typically modest.
Juxtaposing the tax gaps relative to the GDP with the GDP per capita
and the extent of the shadow economy, the emerging patterns correspond
to the expectation that the relative tax gaps are larger in regions with lower
GDP per capita and extensive hidden economic activity. This is evidenced
in Figures 6.6 and 6.7 for the cross-country sample and confirmed in Table
6.8 by the correlation coefficients for individual countries. However, in
several economies the relationship between the shadow economy and rela-
tive tax gap is insignificant.


Table 6.8 Correlation of the tax gap with the extent of the shadow
economy and GDP per capita
Country Tax gap per capita Tax gap relative to Number of

GDP regions
Shadow GDP per Shadow GDP per
economy capita economy capita
Austria 0.132 0.755** 0.409 −0.850*** 9
Belgium 0.111 0.553* −0.085 −0.875*** 11
Czech Republic 0.098 0.964*** 0.481 −0.918*** 8
Germany 0.094 0.744*** 0.688*** −0.808*** 39
Denmark 1
Estonia 1
Spain −0.533** 0.916*** 0.916*** −0.820*** 16
Finland 0.950** −0.656 0.998*** −0.875* 5
France 0.074 0.846*** 0.891*** −0.631*** 22
Greece 0.113 0.403 0.389 −0.595** 13
Hungary −0.290 0.988*** 0.779** −0.612 7
Ireland 2
Italy −0.874*** 0.966*** 0.950*** −0.923*** 21
Lithuania 1
Latvia 1
Netherlands −0.186 0.520* 0.346 −0.863*** 12
Poland −0.467* 0.958*** 0.732*** −0.583** 16
Portugal −0.788 0.985*** 0.232 −0.586 5
Sweden 0.510 0.953*** 0.353 −0.632* 8
Slovenia 1
Slovak Republic −0.828 0.988** 0.998*** −0.922* 4
United Kingdom 0.318* 0.852*** 0.337** −0.514*** 35
Notes: Own calculations. Based on the shadow economy estimates from the GLS
estimation with individual spatial effects coefficients of the indicator variables (model 9). *,
** and *** denote statistical significance at 0.1, 0.05 and 0.01 level, respectively.
The tax gaps per capita are shown in Figure 6.8. Comparing this map
with the maps of the extent of the shadow economy, differences are imme-
diately obvious and confirm the discussion above: tax gaps per capita tend
to be larger in richer countries. The largest tax gaps per capita are found
in Belgium and Sweden, but also in the richest regions of Europe (Inner
London, ^Ile de France, Hamburg). In Italy potentially more additional
tax revenues can be obtained in the northern than in the southern regions
of the country.


8
EU
DE
7 IT
ES
Share of tax gap, % in GDP
2
0 1 2 3 4 5 6 7 8
GDP per capita, 10,000 Euros
Figure 6.6 The GDP (PPP) per capita and the tax gap relative to the
GDP in the EU NUTS 2 regions
8
EU
DE
7 IT
ES
Share of tax gap, % in GDP
2
5 10 15 20 25 30 35
Share of shadow economy, % in GDP
Figure 6.7 The share of the shadow economy and tax gap in the GDP in
the EU NUTS 2 regions


>1250
1000–1250
750–1000
500–750
<500
Figure 6.8 The tax gaps per capita (€) in the EU NUTS 2 regions, based
on model 9. The derivation of the shadow economy quotas
based on the sample mean and variance
6.6 SUMMARY
Estimating the extent of the shadow economy at a regional level poses

additional challenges as compared with such an exercise across coun-
tries. First, as less data are collected at regional than at country level, the


empirical analyses are constrained with data availability – especially if

international comparisons are of interest. Second, data limitations con-
strain the set of methods which are applicable to analyse shadow economic
activity. Third, quantitative models should account for spatial linkages
as economic processes taking place in a region are strongly influenced by
processes in spatially close regions.
In this chapter the estimates of shadow economy quotas in the regions
of the EU at the NUTS 2 classification level are provided for 2004. These
estimates are used to analyse tax gaps – differences between actual and
potential tax revenues. It is the first contribution where MIMIC modelling
is combined with some aspects of spatial econometrics. In addition, a new
anchoring procedure is proposed which is supposedly less sensitive with
respect to the chosen base values than existent approaches.
Confirming previous research, the estimation results reveal that the
extent of the shadow economy exhibits considerable variation across the
EU economies. Providing a new dimension of heterogeneity, the extent
of hidden economic activity is diagnosed to vary markedly across regions
within countries. With a few exceptions, relatively poor regions tend to
have a higher share of the shadow economy than rich ones.
In interpreting the results, it has to be kept in mind that there might be
substantial measurement errors, though in the spatial models the estimated
error of the latent variable is considerably smaller than in the non-spatial
counterparts. Also, due to the unavailability of data, some variables
that potentially affect the size of the shadow economy are omitted from
the model. Above all, unambiguous measures of tax paying morale or
government efficiency could improve the analysis. However, the country-
level estimates for the extent of the shadow economy are rather similar
to corresponding results from earlier studies. Moreover, the ordering of
the regions according to the extent of the shadow economy and most of
the shadow economy estimates are reasonable. The results are remark-
ably robust with respect to distinct model specifications and estimation
techniques, including the consideration of spatial effects. Even though the
estimates might show the potential of shadow activities rather than their
true extent, policymakers should pay attention to their regional variation.
One of the main uses of the shadow economy estimates is the assessment
of potentially lost tax revenues. Such an exercise is undertaken also in this
chapter. An additional advantage of the regional estimates of the shadow
economy is the possibility of capturing this information in designing poli-
cies to combat shadow economy and stimulate development. The finding
that the extent of the shadow economy tends to be larger in poor regions
poses a twofold problem for regional policy issues if local co-financing of
regional development projects is required as for example in the EU. First,


such regions have limited resources just because they are poor. Second, the
high share of shadow activities in the local GDP curtails the tax revenues.
As it is difficult to achieve sufficient quality of public services in such a
situation, the motivation to reveal the economic activities to governmental
institutions remains low.
In designing regional policies in support of such regions it is important
to consider their effect on both the official and the shadow economy.
Though it might be reasonable to accept shadow activities to some extent
in order to keep people in the labour market – even if unofficial – it is nec-
essary to avoid participation in the unofficial economy becoming a widely
accepted phenomenon. In order to prevent such a development, it is nec-
essary to improve the possibilities of finding a job in the official labour
market, but also to improve public services. In particular, it is important
that tax payers approve the way their tax contributions are used. That
would improve the attractiveness of the region both as a place to live or
a location for business activities. This, in turn, can enhance the region’s
economic development.
Our analysis of tax gaps shows that not only the poor but also the rich
regions deserve attention. Within countries, there is no stable relationship
between the extent of the shadow economy and the size of tax gaps per
capita. Thus, governments should not rely solely on the shadow economy
quotas if they aim to detect substantial tax fraud and to collect additional
tax revenues. However, in analysing the relation between the tax gap and
the wealth of the regions it appears that larger additional tax revenues
could be gained from rich regions. This is consistent with the argument
that relatively wealthy people might have more possibilities to avoid taxes
than the poor. For example, Renooy (1990) has found for the Netherlands
that the middle-income groups are the most active shadow economy
participants.
Therefore, in deciding the extent to which measures for reducing
shadow economy and detecting such activities should be implemented,
differentiated approaches might be required across regions that are char-
acterised by distinct levels of development. Considering that a whole
complex of policies can influence the motivation to participate in the offi-
cial or shadow economy and that their effects might be counterproductive,
it appears necessary to develop region specific policy designs. In Chapter
5 of this handbook Williams and Windebank conclude similarly that
regional characteristics should be considered in designing the policies to
tackle shadow work.
Depending on the region, different transactions can dominate in the
shadow activities. Examples are labour market activities, transactions
between the firms or criminality. Knowing the main sources of the shadow


economy and tax fraud for each region would help to improve policy
design in combating the shadow economy. Such valuable complemen-
tary information could be derived from surveys conducted in individual
countries.
There are also possibilities to improve the shadow economy estima-
tion based on the MIMIC approach. First, including a variable like tax
morale in the model would increase the reliability of the results. Second,
the indicators for the extent of the shadow economy could be reconsid-
ered as the currently used GDP per capita and labour market participa-
tion rates might be only weakly related to shadow economy. Acceptable
indicators could be related to criminality rates, to detected tax fraud or to
the value of durables like cars compared to official incomes. Both of these
suggestions call for additional data collection at the regional level.
NOTES
1. NUTS – The Nomenclature of Territorial Units for Statistics. The NUTS system is
developed for statistical purposes and the statistical units do not correspond necessarily
to administrative units. The size of the NUTS 2 regions is required to be 0.8–3 million
inhabitants.
2. The operator vech denotes half-vectorisation, that is, it transforms a symmetric matrix
into a vector that collects the unique elements of the matrix.
3. ML estimation failed due to singularity problems. ULS coupled with bootstrap inference
delivers results similar to those of GLS. The only qualitative difference is the insignifi-
cance of tax wedge, share of public employment and self-employment rate in the ULS
estimation. ULS estimation results are available from the authors on request.
4. Such procedures are usually called calibration or benchmarking in the literature. Breusch
(2005) strongly criticises both of these terms, suggesting ‘anchoring’ as a better term.
5. We are grateful to Lorena Gola and Cornelius Peters for preparing the maps.
6. This result contradicts the results of Mummert and Schneider (2001) for 1995 and 1999:
using survey data, they concluded that the shadow economy quota was higher in the
western part of Germany than in the eastern part.
REFERENCES
Anselin, L. (1988), Spatial Econometrics: Methods and Models: Studies in

Operational Regional Science. Dordrecht: Kluwer Academic Publishers.
Backhaus, K., B. Erichson, W. Plinke and R. Weiber (2006), Multivariate
Analysemethoden. Eine anwendungsorientierte Einführung, 11th edn. Berlin:
Springer.
Bernotaitė, R. and A. Piskunova (2005), An expenditure-based estimate of
Latvia’s shadow economy. SSE Riga Working Papers 2005:5 (73). Available at
http://www.sseriga.edu.lv/library/working_papers/FT_2005_5.pdf.
Bollen, K.A. (1989), Structural Equations with Latent Variables. New York: John
Wiley & Sons.


Brēķis, E. (2007), Latvia’s underground economy modelling with an emphasis

on tax policy. Summary of the promotion paper, University of Latvia, Riga.
Available at http://www3.acadlib.lv/grey/grey1.idq?CiRestriction5economy.
Breusch, T. (2005), Estimating the underground economy using MIMIC models.
Available at http://129.3.20.41/eps/em/papers/0507/0507003.pdf.
Buehn, A. and F. Schneider (2008), MIMIC models, cointegration and error cor-
rection: an application to the French shadow economy. IZA Discussion Paper
No. 3306. Available at http://ftp.iza.org/dp3306.pdf.
Chaudhuri, K., F. Schneider and S. Chattopadhyay (2006), The size and develop-
ment of the shadow economy: an empirical investigation from states of India.
Journal of Development Economics, 80, 428–443.
Dell’Anno, R. (2007), The shadow economy in Portugal: an analysis with the
MIMIC approach. Journal of Applied Economics, X(2), 253–277.
Dell’Anno, R., M. Gómez-Antonio and A. Alañon-Pardo (2007), The shadow
economy in three Mediterranean countries: France, Spain and Greece. A
MIMIC approach. Empirical Economics, 33(1), 51–84.
Dell’Anno, R. and F. Schneider (2003), The shadow economy of Italy and other
OECD countries: what do we know? Journal of Public Finance and Public
Choice, XXI(2–3), 97–120.
Dell’Anno, R. and F. Schneider (2006), Estimating the underground economy by
using MIMIC models: a response to T. Breusch’s critique. Working Paper No.
0607, Department of Economics, Johannes Kepler University of Linz. Available
at http://www.econ.jku.at/papers/2006/wp0607.pdf.
Finance, 16(6), 773–796.
Eilat, Y. and C. Zinnes (2000), The evolution of the shadow economy in transi-
tion countries: Consequences for economic growth and donor assistance.
CAER Discussion paper No. 83. Available at http://pdf.usaid.gov/pdf_docs/
PNACK691.pdf.
Eilat, Y. and C. Zinnes (2002), The shadow economy in transition countries: friend
or foe? A policy perspective. World Development, 30(7), 1233–1254.
Ertur, C., J. Le Gallo and C. Baumont (2006), The European regional conver-
gence process, 1980–1995: do spatial regimes and spatial dependence matter?
International Regional Science Review, 29(1), 3–34.
ESA (1995), European System of Accounts ESA 1995. Available at http://circa.
europa.eu/irc/dsis/nfaccount/info/data/esa95/en/esa95en.htm.
European Commission (2009), VAT Rates Applied in the Member States of
the European Community. DOC/2401/2009 – EN, European Commission,
Taxation and Customs Union, Brussels. Available at http://ec.europa.eu/
taxation_customs/resources/documents/taxation/vat/how_vat_works/rates/vat_
rates_en.pdf.
Eurostat (2006), Taxation trends in the European Union: Data for the EU Member
States and Norway. Research in Official Statistics, Office for Official Publications
of the European Communities, Luxembourg. Available at http://ec.europa.eu/
taxation_customs/resources/documents/taxation/gen_info/economic_analysis/
tax_structures/2009/2009_full_text_en.pdf.
Eurostat (2008), General and regional statistics. Available at http://epp.eurostat.
ec.europa.eu/portal/page?_pageid50,1136162,0_45572076&_dad5portal&_sch
ema5PORTAL.


Feige, E.L. and I. Urban (2008), Measuring underground (unobserved, non-

observed, unrecorded) economies in transition countries: can we trust GDP?
Journal of Comparative Economics, 36(2), 287–306.
Frey, B.S. and H. Weck (1983), Estimating the shadow economy: a ‘naive’
approach. Oxford Economic Papers, 35(1), 23–44.
Frey, B.S. and H. Weck-Hanneman (1984), The hidden economy as an ‘unob-
Getis, A. and J. Aldstadt (2004), Constructing the spatial weights matrix using a
local statistic. Geographical Analysis, 34(2), 90–104.
Giles, D. (1997), The hidden economy and the tax-gap in New Zealand: a latent
variable analysis. Available at http://web.uvic.ca/econ/wp9807.pdf.
Giles, D. (1999), Measuring the hidden economy: implications for econometric
modeling. The Economic Journal, 109(456), F370–F380.
Giles, D.E. and L.M. Tedds (2002), Taxes and the Canadian underground
economy. No. 106, Canadian Tax Paper, Canadian Tax Foundation, Toronto.
Herwartz, H., F. Schneider and E. Tafenau (2010), One share fits all? Regional var-
iations in the extent of shadow economy in Europe. Mimeo, University of Kiel.
Herwartz, H. and B. Theilen (2010), Does the EU financing system contribute to
shadow economic activity? Mimeo, University of Kiel.
ISTAT (2008), La misura dell’economia sommersa secondo le statistiche ufficiali.
Instituto nazionale di statistica. Available at http://www.istat.it/salastampa/
comunicati/non_calendario/20080618_00/testointegrale20080618.pdf.
and the unofficial economy. The American Economic Review, 88(2), 387–392.
Statistical Association, 70(351), 631–639.
LeSage, J. and R.K. Pace (2009), Introduction to Spatial Econometrics, Statistics:
Textbooks and Monographs. Boca Raton, FL: CRC Press.
LFMI (2005), The 15th survey of the Lithuanian economy 2004/2005 (2).
Lithuanian Free Market Institute. Available at http://www.freema.org/index.
php/research/the_15th_survey_of_the_lithuanian_economy_20042005_2/2768.
McMillen, D.P. (1996), One hundred fifty years of land values in Chicago: a non-
parametric approach. Journal of Urban Economics, 40(1), 100–124.
Mróz, B. (2005), The shadow economy in Poland and its socio-economic implica-
tions. In S. Bukowski (ed.), The Dilemmas of Economic Policy, Development and
Regional Economic Integration. Radom, Poland: WSFiB, pp. 167–176. Available
at http://www.wydawnictwa.wsfib.edu.pl/Polska_w_UE/mr%F3z.pdf.
united Germany? FinanzArchiv, 58(3), 286–316.
Nastav, B. and Š. Bojnec (2007), Shadow economy in Slovenia: the labour
approach. Managing Global Transitions, 5(2), 193–208.
Ord, J.K. (2010), Spatial autocorrelation: a statistician’s reflections. In L. Anselin
and S. Rey (eds), Perspectives on Spatial Data Analysis. Advances in Spatial
Science. Heidelberg: Springer, chapter 12, pp. 165–180.
Renooy, P.H. (1990), The informal economy: meaning, measurement and social
significance. PhD thesis, Universiteit van Amsterdam, Amsterdam.


estimates for 145 countries. Economics: The Open-Access, Open-Assessment
E-Journal, 1(2007–9). Available at http://www.economics-ejournal.org/
economics/journalarticles/2007-9.
Schneider, F. (2008), The shadow economy in Germany: a blessing or a curse for
the official economy. Economic Analysis and Policy, 38(1), 89–111.
Schneider, F. and C. Bajada (2005), An international comparison of under-
ground economic activity. In C. Bajada and F. Schneider (eds), Size, Causes
and Consequences of the Underground Economy: An International Perspective.
Aldershot, UK: Ashgate, chapter 5, pp. 73–106.
quences. Journal of Economic Literature, 38(1), 77–114.
United Nations (2008), Non-observed economy in national accounts: survey of
country practices. United Nations Economic Commission for Europe, New
York and Geneva.
Williams, C.C. (2004), Cash-in-Hand Work. The Underground Sector and the
Hidden Economy of Favours. Basingstoke, UK: Palgrave Macmillan.
Zellner, A. (1970), Estimation of regression relationships containing unobservable
independent variables. International Economic Review. 11(3), 441–454.


APPENDIX 6.1 ADJUSTING FOR SPATIAL EFFECTS

In models 8, 9 and 10 it is assumed that the indicators or causes are meas-
ured with an error due to spatial interdependencies. Thus, instead of the
original variables zi (a R 3 1 vector of the R observations of the ith observed
variable, i [ { 1, 2, . . ., p 1 q } ), the variables |
z i 5 (I 2 riW) zi are used in
the models, with | z i denoting the vector of spatially adjusted observations
for the ith observed variable, ri the spatial effects coefficient corresponding
to the ith variable and W the spatial weight matrix. In the following sec-
tions the derivation of the weight matrix and the procedure of choosing an
appropriate parametrization of the spatial effects are described.
A6.1.1 The Weight Matrix
The weight matrix is based on a distance matrix, derived from distances

among NUTS 3 regions.1 In order to obtain the distances between the
NUTS 2 regions, the land area shares of NUTS 3 regions within each
NUTS 2 region are calculated. In the case of missing data, those are
replaced with Audr /Nr, where Nr is the number of NUTS 3 regions in a
NUTS 2 region r and Aud r denotes the share of the NUTS 2 area for which
the area data at the NUTS 3 level are missing (‘undefined area’). Thus,
r 5 1 if area data are unavailable for all NUTS 3 regions within a
Aud
NUTS 2 region r. If for some NUTS 3 regions data are available but for
some others within the same NUTS 2 region not, Aud r 5 1 2 Ar , with Ar
d d
denoting the share of the NUTS 2 region r’s area for which area data at
NUTS 3 level are available.
The distances drs between the NUTS 2 regions r and s are calculated as
a a ark asl dkl if¬r 2 s

drs 5 • k [r l[s (6.22)
0 if¬r 5 s,
where ark and asl denote the land area share of NUTS 3 regions k and l in
NUTS 2 regions r and s, respectively. The original distance between the
NUTS 3 regions k and l is denoted by dkl.
In order to convert the distance matrix into a weight matrix, the so-
called tri-cube function is used. This function has also been used by
McMillen (1996) and reads as
drs 3 3
a1 2 a b b
wrs 5 •
if¬drs , D and r 2 s
D (6.23)
0 if¬drs $ D or r 5 s.


Notably the tri-cube approach implies that the region s is considered to

have a spatial effect on region r (wrs . 0) only if its distance from region r
is at most D kilometres. Therefore, the number of regions exerting a spatial
effect varies across the regions.2 For the implementation of MIMIC models
the weight matrix W has been row-standardised, such that g swrs 5 1.
A6.1.2 Spatial Effects in the SEM
Three sets of models have been estimated.
1. Assuming that all variables (both indicators and causes) should be

adjusted for spatial effects.
2. Assuming that only the indicator variables (yr) should be adjusted for
spatial effects.
3. Assuming that only the cause variables (xr) should be adjusted for
spatial effects.
In all scenarios, it is initially assumed that for all of the spatially

affected variables the same spatial effects coefficient can be used for the
adjustment, that is an identical r is used for all transformed variables:
|
z i 5 (I 2 rW) zi. After spatial adjustment the data are processed as the
non-weighted data in the unweighted model: all variables are centred and
divided by the mean. Then the MIMIC estimation is applied.
In choosing the most appropriate model, the coefficient of determina-
tion for the latent variable (calculated as R2h 5 1 2 c^ /Var [ h^ ] ) is used as
the selection criterion.3 The cut-off distance D and the spatial parameter r
are found in the following steps.
1. R2h is maximized both with respect to r and D.4 The cut-off distance
D and the spatial effects coefficient are assumed to be in the ranges
30 km # D # 150 km (step size 10 km) and 21 < r < 1, with step size
0.1, respectively.
2. The grid for the cut-off distance is reduced to 5 km in the neighbour-
hood of the maximum obtained from step 1. Again, the model with
the maximum R2h is chosen. The corresponding D is taken as the
optimal cut-off distance (D*).
3. Step size 0.02 is chosen for r in the neighbourhood of the r obtained in
step 2 and R2h is maximized over this grid to find the most appropriate r*.
The most appropriate model is considered to be the one that uses the
weight matrix with cut-off distance obtained in step 2 and the spatial
effects coefficient from step 3.


In the case of model 9 in Table 6.3, the two indicators are assumed to
have distinct spatial effects coefficients. It has been assumed that the cut-
off distance is in both cases as in the model with identical spatial effects
coefficient (model 8). Then, the model has been estimated for all pairwise
combinations of r from 20.5 < r < 0.5, with step size 0.05. Again, the deci-
sion criterion is R2h. After that we refine the grid to 0.02 in the surrounding
of the initial values of the r parameter. Throughout, GLS model estima-
tion is initiated with ULS estimates as the starting values. For spatial ULS
models the starting values are taken from the unweighted model.
In Table 6.3 the model with all variables being weighted according to
the same weighting scheme is not presented due to unreasonable param-
eter estimates. In the case of models 8 and 9 (indicators spatially adjusted,
model 8 using the same r and model 9 distinct values of r for each indica-
tor), the cut-off distance D* is 90 km. This appears reasonable, considering
that it could be interpreted as a commuting distance. In model 8 the spatial
effects coefficient r* 5 20.06. In model 9, r* 5 20.00 for the GDP per
capita indicator and r* 5 20.18 for the labour force participation rate.
In model 10 where only the cause variables are spatially adjusted, the
cut-off distance D* is 50 km and the selected spatial effects coefficient is
r* 5 20.10.


APPENDIX 6.2 SHADOW ECONOMY IN THE EU

NUTS 2 REGIONS
Table A6.2.1 Shadow economy quotas (% in GDP) in the EU NUTS 2

regions in 2004a
Regionb Region’s name M1 M1c M9d M9d,e M10f TGg TGpch

AT31 Oberösterreich 13.0 13.9 13.6 14.8 13.3 1511 1086
AT12 Niederösterreich 13.7 13.7 14.0 15.1 13.9 1860 1189
AT22 Steiermark 13.9 13.7 14.1 15.2 14.1 1250 1047
AT33 Tirol 13.9 14.1 14.2 15.3 14.2 761 1102
AT34 Vorarlberg 13.6 14.8 14.2 15.2 13.9 425 1180
AT13 Wien 14.2 15.1 14.5 15.4 14.4 2215 1375
AT32 Salzburg 14.4 13.9 14.5 15.4 14.5 620 1181
AT21 Kärnten 14.3 14.4 14.6 15.5 14.5 584 1045
AT11 Burgenland (A) 15.1 15.1 15.2 15.9 15.3 310 1116
BE24 Prov. Vlaams Brabant 15.0 12.3 14.5 15.5 12.7 1846 1785
BE21 Prov. Antwerpen 14.8 13.6 14.8 15.6 13.2 2442 1461
BE23 Prov. Oost-Vlaanderen 15.2 13.0 14.9 15.7 12.9 2066 1500
BE22 Prov. Limburg (B) 16.4 13.9 15.8 16.3 15.2 1096 1356
BE25 Prov. West-Vlaanderen 16.3 13.4 15.8 16.3 16.1 1581 1392
BE31 Prov. Brabant Wallon 17.9 14.1 17.2 17.2 15.6 708 1951
BE35 Prov. Namur 18.5 14.4 17.2 17.2 18.2 660 1456
BE34 Prov. Luxembourg (B) 18.3 15.5 17.6 17.5 18.1 327 1286
BE33 Prov. Liège 19.1 16.7 18.3 18.0 18.9 1525 1478
BE32 Prov. Hainaut 19.3 17.0 18.4 18.0 19.0 1728 1344
BE10 Région de Bruxelles- 21.5 19.5 20.8 19.6 19.5 1779 1775
Capitale/Brussels
Hoofdstedelijk
Gewest
CZ03 Jihozápad 15.4 16.1 15.9 16.4 15.7 348 297
CZ05 Severovýchod 15.7 16.9 16.3 16.6 15.9 418 282
CZ06 Jihovýchod 16.2 17.5 16.8 17.0 16.5 487 298
CZ01 Praha 17.5 15.0 17.1 17.1 15.6 538 460
CZ02 Střední Cechy 16.7 16.7 17.1 17.2 15.0 383 335
CZ07 Střední Morava 16.7 18.7 17.5 17.4 17.0 363 296
CZ04 Severozápad 17.2 20.0 18.1 17.8 17.6 352 313
CZ08 Moravskoslezsko 17.5 20.8 18.4 18.0 17.8 391 311
DE11 Stuttgart 12.8 13.3 12.8 14.3 13.0 4724 1181
DE71 Darmstadt 13.0 12.6 12.8 14.3 13.1 4849 1288
DEB3 Rheinhessen-Pfalz 13.6 12.0 12.8 14.3 13.6 2170 1076
DE21 Oberbayern 13.7 11.8 13.1 14.5 13.7 5705 1358
DE23 Oberpfalz 13.8 12.8 13.1 14.5 13.9 1074 985
DE12 Karlsruhe 13.6 13.1 13.2 14.6 13.8 3111 1141


Table A6.2.1 (continued)

DE91 Braunschweig 13.9 13.7 13.3 14.7 14.0 1739 1049
DEB2 Trier 14.5 12.3 13.4 14.7 14.5 492 958
DE14 Tübingen 14.1 13.2 13.5 14.8 14.2 1994 1109
DE22 Niederbayern 14.1 13.3 13.5 14.8 14.2 1173 983
DEB1 Koblenz 14.4 12.8 13.5 14.8 14.5 1548 1015
DEC0 Saarland 14.3 13.5 13.6 14.8 14.3 1064 1007
DE25 Mittelfranken 14.0 13.9 13.7 14.9 14.2 1985 1162
DE26 Unterfranken 14.3 13.6 13.7 14.9 14.4 1405 1046
DE13 Freiburg 14.6 13.5 14.0 15.1 14.7 2379 1089
DEF0 Schleswig-Holstein 14.8 13.2 14.0 15.1 14.8 3140 1112
DE27 Schwaben 14.6 13.8 14.1 15.2 14.7 1958 1100
DE73 Kassel 15.2 13.2 14.1 15.2 15.2 1235 979
DEA2 Köln 15.2 13.0 14.1 15.2 15.2 5386 1237
DEA1 Düsseldorf 14.8 14.4 14.2 15.2 14.9 6342 1211
DE72 Gießen 15.3 13.6 14.3 15.3 15.3 1123 1056
DE92 Hannover 15.0 14.0 14.3 15.3 15.1 2464 1137
DEA3 Münster 15.4 13.8 14.4 15.3 15.4 2811 1071
DE24 Oberfranken 14.8 15.2 14.6 15.5 15.0 1199 1082
DE93 Lüneburg 15.3 14.0 14.6 15.5 15.3 1937 1140
DEA5 Arnsberg 15.6 15.5 15.0 15.8 15.8 4305 1140
DE94 Weser-Ems 16.1 14.4 15.1 15.9 16.1 2516 1019
DEA4 Detmold 16.1 15.4 15.4 16.0 16.2 2400 1157
DEG0 Thüringen 16.4 17.3 16.1 16.5 16.5 2226 939
DE42 Brandenburg-Südwest 17.5 17.9 17.1 17.1 17.6 1540 1095
DE60 Hamburg 17.9 16.6 17.1 17.1 17.9 2645 1526
DED2 Dresden 18.0 19.2 17.7 17.5 18.1 1721 1030
DE50 Bremen 18.8 18.0 17.8 17.6 18.8 889 1342
DE30 Berlin 18.7 17.7 17.9 17.7 18.6 4338 1279
DE41 Brandenburg-Nordost 18.7 19.9 18.3 17.9 18.8 1342 1155
DEE0 Sachsen-Anhalt 18.5 20.5 18.3 17.9 18.7 2605 1037
DED1 Chemnitz 18.4 20.4 18.4 18.0 18.6 1536 982
DED3 Leipzig 18.6 20.0 18.4 18.0 18.7 1156 1076
DE80 Mecklenburg- 18.8 20.4 18.5 18.1 18.9 1851 1070
Vorpommern
DK00 Denmark 10.8 12.5 12.0 13.8 11.1 8380 1550
EE00 Estonia 16.3 18.7 16.6 16.8 16.7 284 210
ES30 Comunidad de Madrid 13.1 15.0 13.2 14.6 13.6 4446 772
ES24 Aragón 15.1 16.0 15.3 16.0 15.5 841 679
ES51 Cataluña 15.2 18.0 15.8 16.3 15.7 5421 808
ES22 Comunidad Foral de 16.1 16.7 16.2 16.5 16.4 460 798
Navarra
ES12 Principado de Asturias 16.0 17.9 16.4 16.7 16.4 674 637



ES23 La Rioja 16.1 17.3 16.6 16.8 16.4 201 690
ES62 Región de Murcia 15.9 19.5 16.7 16.9 16.5 702 549
ES21 Pais Vasco 16.5 18.5 16.8 16.9 16.9 1759 838
ES52 Comunidad Valenciana 16.1 19.2 16.8 17.0 16.6 2835 636
ES53 Illes Balears 17.6 19.1 17.9 17.6 18.0 714 756
ES13 Cantabria 17.6 19.5 18.1 17.8 18.0 384 702
ES42 Castilla-la Mancha 18.7 19.5 18.8 18.3 18.9 1048 570
ES41 Castilla y León 19.2 19.9 19.3 18.6 19.4 1701 692
ES11 Galicia 19.3 21.4 19.9 19.0 19.6 1768 653
ES61 Andalucia 19.6 22.9 20.1 19.1 20.0 4686 617
ES43 Extremadura 22.1 23.8 22.2 20.5 22.4 604 565
FI20 Åland 1.8 0.6 –0.1 5.8 2.1 0 –5
FI18 Etelä-Suomi 14.5 15.9 15.2 15.9 14.8 3478 1352
FI19 Länsi-Suomi 17.2 17.9 17.6 17.5 17.3 1620 1222
FI1A Pohjois-Suomi 17.9 18.4 18.2 17.8 18.0 758 1202
FI13 Itä-Suomi 19.6 19.7 19.6 18.8 19.6 789 1181
FR42 Alsace 13.8 14.9 13.9 15.0 14.1 2094 1165
FR10 Île de France 15.1 15.3 14.8 15.6 15.2 19026 1673
FR23 Haute-Normandie 15.2 15.6 14.9 15.7 15.4 1977 1095
FR24 Centre 15.8 15.0 15.3 16.0 15.9 2828 1136
FR43 Franche-Comté 15.6 15.8 15.4 16.0 15.8 1244 1091
FR71 Rhône-Alpes 15.6 15.8 15.4 16.0 15.8 7173 1211
FR51 Pays de la Loire 15.9 15.5 15.6 16.2 16.0 3707 1096
FR22 Picardie 15.9 16.8 15.8 16.3 16.1 2100 1116
FR26 Bourgogne 16.4 15.9 16.0 16.4 16.5 1837 1133
FR41 Lorraine 16.4 17.2 16.1 16.5 16.6 2579 1104
FR52 Bretagne 16.5 15.3 16.1 16.5 16.6 3258 1069
FR25 Basse-Normandie 16.7 15.5 16.2 16.5 16.8 1552 1075
FR62 Midi-Pyrénées 16.8 15.3 16.3 16.6 16.9 3051 1123
FR53 Poitou-Charentes 16.8 16.2 16.4 16.7 16.9 1855 1088
FR72 Auvergne 16.7 16.0 16.5 16.7 16.8 1483 1113
FR21 Champagne-Ardenne 17.2 16.6 16.7 16.9 17.3 1582 1180
FR30 Nord – Pas-de-Calais 16.9 17.8 16.7 16.9 17.1 4308 1070
FR82 Provence-Alpes-Côte 17.8 16.4 17.1 17.1 17.8 5963 1262
d’Azur
FR63 Limousin 18.2 16.1 17.5 17.4 18.1 855 1181
FR61 Aquitaine 18.3 16.7 17.6 17.4 18.3 3812 1245
FR81 Languedoc-Roussillon 20.0 17.2 18.8 18.3 19.9 2883 1164
FR83 Corse 24.9 17.0 21.3 19.9 24.3 357 1295
GR22 Ionia Nisia 5.9 10.2 9.5 12.1 6.3 117 535
GR30 Attiki 18.1 18.4 18.2 17.9 18.3 2914 737
GR42 Notio Aigaio 18.0 16.6 18.3 17.9 18.0 167 552



GR41 Voreio Aigaio 20.1 16.0 19.5 18.8 19.9 110 542
GR43 Kriti 21.5 19.5 21.9 20.3 21.5 348 581
GR12 Kentriki Makedonia 22.8 22.4 23.2 21.1 22.8 1150 602
GR11 Anatoliki Makedonia, 23.9 23.2 24.5 22.0 23.9 353 579
Thraki
GR14 Thessalia 24.4 22.5 24.5 22.0 24.4 386 525
GR21 Ipeiros 25.4 22.9 25.3 22.5 25.3 190 558
GR23 Dytiki Ellada 25.1 24.0 25.4 22.6 25.1 372 507
GR24 Sterea Ellada 25.1 25.0 25.7 22.8 25.1 335 600
GR25 Peloponnisos 26.7 23.7 26.8 23.5 26.6 282 472
GR13 Dytiki Makedonia 26.6 26.5 27.2 23.8 26.6 251 852
HU21 Közép-Dunántúl 20.1 20.8 20.8 19.6 20.3 405 364
HU22 Nyugat-Dunántúl 20.9 20.9 21.3 19.9 21.0 365 365
HU10 Közép-Magyarország 22.7 20.5 22.4 20.7 22.6 1689 597
HU32 Észak-Alföld 22.7 21.8 22.7 20.9 22.7 399 258
HU31 Észak-Magyarország 22.7 23.1 23.0 21.0 22.7 361 283
HU23 Dél-Dunántúl 23.4 22.2 23.3 21.2 23.3 295 301
HU33 Dél-Alföld 23.4 21.9 23.4 21.3 23.3 391 289
IE02 Southern and Eastern 13.9 18.8 15.6 16.1 14.7 3088 1039
IE01 Border, Midlands and 16.0 19.7 17.5 17.4 16.7 963 889
Western
ITD2 Provincia Autonoma 18.1 15.1 17.7 17.5 18.0 631 1276
Trento
ITC4 Lombardia 17.1 16.6 17.8 17.6 17.2 13684 1466
ITD1 Provincia Autonoma 18.6 16.0 18.3 18.0 18.5 675 1423
Bolzano-Bozen
ITD4 Friuli-Venezia Giulia 18.1 16.3 18.3 18.0 18.1 1580 1313
ITD3 Veneto 18.3 17.4 18.9 18.3 18.4 6179 1324
ITD5 Emilia-Romagna 18.8 17.0 19.2 18.5 18.8 6063 1474
ITC1 Piemonte 18.9 17.7 19.4 18.6 19.0 5958 1383
ITC2 Valle d’Aosta/Vallée 20.5 15.6 19.6 18.8 20.3 173 1413
d’Aoste
ITE3 Marche 20.0 18.2 20.2 19.2 20.0 1863 1230
ITE4 Lazio 21.2 17.8 20.6 19.4 21.0 7944 1516
ITC3 Liguria 21.2 17.6 20.7 19.5 21.0 2109 1329
ITE1 Toscana 20.6 18.1 20.7 19.5 20.5 4897 1365
ITF1 Abruzzo 21.5 18.8 21.3 19.9 21.3 1344 1042
ITE2 Umbria 21.9 18.2 21.4 20.0 21.7 1045 1226
ITF2 Molise 24.5 21.6 24.0 21.7 24.3 330 1027
ITG2 Sardegna 24.2 22.9 24.1 21.8 24.1 1762 1072
ITF5 Basilicata 24.3 22.7 24.2 21.8 24.2 578 966
ITF4 Puglia 24.2 23.8 24.4 21.9 24.1 3934 970



ITF6 Calabria 25.8 23.4 25.1 22.4 25.6 1755 873
ITF3 Campania 26.0 24.3 25.6 22.8 25.8 5528 960
ITG1 Sicilia 26.7 24.6 26.0 23.0 26.5 4750 947
LT00 Lithuania 22.9 22.8 22.4 20.6 23.1 643 187
LV00 Latvia 21.0 21.6 20.4 19.3 21.2 347 150
NL31 Utrecht 9.0 8.7 9.2 11.9 9.2 1241 1064
NL33 Zuid-Holland 8.8 9.4 9.2 11.9 9.0 3354 970
NL23 Flevoland 8.6 10.6 9.5 12.1 9.0 342 942
NL42 Limburg (NL) 8.8 10.0 9.5 12.1 7.3 935 822
NL41 Noord-Brabant 8.9 10.1 9.6 12.2 9.2 2251 935
NL32 Noord-Holland 9.5 9.4 9.9 12.4 9.7 2822 1089
NL21 Overijssel 9.4 10.2 10.0 12.4 9.7 949 856
NL22 Gelderland 9.7 9.9 10.0 12.4 9.9 1867 948
NL11 Groningen 10.0 9.8 10.2 12.6 9.0 504 877
NL13 Drenthe 9.8 10.4 10.3 12.6 8.9 428 887
NL34 Zeeland 9.9 10.1 10.3 12.7 8.5 353 929
NL12 Friesland (NL) 11.1 10.8 11.3 13.3 11.3 609 947
PL12 Mazowieckie 26.7 26.7 26.9 23.6 26.7 2060 401
PL22 Śląskie 26.2 29.7 26.9 23.6 26.4 1349 286
PL52 Opolskie 27.7 29.7 28.2 24.4 27.8 220 208
PL63 Pomorskie 28.8 30.8 29.1 25.0 28.8 571 261
PL32 Podkarpackie 28.8 29.1 29.2 25.1 28.7 432 205
PL21 Małopolskie 29.0 29.8 29.5 25.3 29.0 787 243
PL41 Wielkopolskie 29.2 31.0 29.7 25.5 29.2 905 270
PL11 Łódzkie 29.7 31.2 30.3 25.8 29.8 664 257
PL43 Lubuskie 30.9 33.1 31.1 26.4 30.9 245 243
PL62 Warmińsko-Mazurskie 30.8 32.8 31.1 26.4 30.9 317 221
PL51 Dolnośląskie 30.7 33.8 31.3 26.5 30.8 859 298
PL61 Kujawsko-Pomorskie 30.9 33.0 31.3 26.5 30.9 514 249
PL34 Podlaskie 31.2 30.1 31.5 26.6 31.0 265 220
PL42 Zachodniopomorskie 31.1 34.1 31.5 26.7 31.2 441 261
PL31 Lubelskie 31.7 30.2 31.9 26.9 31.4 479 219
PL33 Świętokrzyskie 32.9 32.8 33.3 27.8 32.8 304 236
PT17 Lisboa 16.7 17.8 16.9 17.0 17.0 1977 719
PT11 Norte 19.0 20.9 20.1 19.1 19.3 1645 442
PT16 Centro (PT) 20.8 19.5 21.3 19.9 20.9 1131 478
PT18 Alentejo 21.5 21.4 21.3 19.9 21.7 353 458
PT15 Algarve 22.0 20.5 21.8 20.3 22.1 224 549
SE21 Småland med Öarna 15.3 13.5 15.4 16.0 15.2 1242 1553
SE23 Västsverige 15.2 14.2 15.4 16.0 15.1 3007 1671
SE31 Norra Mellansverige 15.3 14.0 15.4 16.0 15.2 1252 1516
SE33 Övre Norrland 15.8 13.3 15.4 16.0 15.6 771 1516



SE12 Östra Mellansverige 15.6 14.4 15.6 16.2 15.5 2465 1629
SE32 Mellersta Norrland 16.1 13.6 15.8 16.3 15.9 594 1600
SE11 Stockholm 16.1 14.2 16.2 16.5 16.0 4137 2218
SE22 Sydsverige 16.7 15.3 16.8 16.9 16.6 2228 1707
SI00 Slovenia 16.2 17.7 16.6 16.8 16.5 985 493
SK01 Bratislavský kraj 18.7 18.7 18.4 18.0 18.9 241 400
SK02 Západné Slovensko 20.9 24.2 21.5 20.1 21.3 424 227
SK03 Stredné Slovensko 24.3 28.3 24.8 22.2 24.6 328 243
SK04 Východné Slovensko 25.1 29.8 25.7 22.8 25.4 354 226
UKD2 Cheshire 10.1 12.1 10.6 12.8 9.3 903 911
UKC1 Tees Valley and 10.9 13.3 11.1 13.2 11.4 793 689
Durham
UKD3 Greater Manchester 10.9 12.8 11.1 13.2 10.1 1945 768
UKE4 West Yorkshire 10.6 13.2 11.1 13.2 9.9 1616 765
UKF2 Leicestershire, Rutland 10.6 12.9 11.1 13.2 11.1 1375 861
and Northants
UKJ1 Berkshire, Bucks and 10.5 12.7 11.1 13.2 11.1 2468 1162
Oxfordshire
UKE3 South Yorkshire 11.0 12.8 11.2 13.2 10.3 939 734
UKG2 Shropshire and 10.7 12.9 11.2 13.2 11.3 1163 774
Staffordshire
UKF1 Derbyshire and 11.0 12.8 11.3 13.3 10.2 1594 789
Nottinghamshire
UKH2 Bedfordshire, 10.9 12.5 11.3 13.3 11.4 1870 1151
Hertfordshire
UKM2 Eastern Scotland 11.2 12.7 11.3 13.3 11.7 1693 883
UKJ3 Hampshire and Isle of 11.3 12.5 11.4 13.4 11.7 1701 944
Wight
UKK1 Gloucestershire, 11.2 12.6 11.4 13.4 11.6 2060 932
Wiltshire and
Bristol/Bath area
UKG1 Herefordshire, 11.2 12.6 11.5 13.4 10.3 1168 938
Worcestershire and
Warks
UKL2 East Wales 11.5 12.5 11.6 13.5 11.9 836 779
UKM3 South Western 11.4 13.6 11.6 13.5 11.9 1824 799
Scotland
UKH3 Essex 10.9 13.3 11.7 13.6 11.5 1712 1042
UKD5 Merseyside 12.0 13.1 11.8 13.7 11.2 989 728
UKG3 West Midlands 11.2 14.4 11.8 13.7 10.5 1960 758
UKH1 East Anglia 11.7 12.7 11.8 13.6 12.2 1944 867



UKC2 Northumberland, Tyne 12.1 13.3 11.9 13.7 12.5 1037 746
and Wear
UKD4 Lancashire 12.0 12.8 11.9 13.7 11.1 1078 749
UKE1 East Yorkshire 11.8 13.8 12.0 13.8 12.3 704 791
and Northern
Lincolnshire
UKD1 Cumbria 11.5 13.6 12.1 13.9 12.0 403 817
UKF3 Lincolnshire 11.8 13.8 12.2 13.9 12.3 546 811
UKE2 North Yorkshire 12.7 12.1 12.4 14.0 13.0 730 952
UKJ2 Surrey, East and West 12.3 12.7 12.4 14.0 12.6 3056 1185
Sussex
UKJ4 Kent 12.1 13.3 12.4 14.0 12.5 1612 999
UKL1 West Wales and The 12.5 13.5 12.4 14.1 12.9 1307 698
Valleys
UKI2 Outer London 12.1 13.7 12.5 14.1 11.0 5564 1238
UKK2 Dorset and Somerset 12.9 12.6 12.7 14.2 13.2 1073 886
UKK4 Devon 13.5 13.3 13.3 14.6 13.8 864 787
UKN0 Northern Ireland 13.8 13.6 13.5 14.8 14.1 1364 797
UKI1 Inner London 14.0 15.4 14.2 15.2 12.9 5277 1806
UKK3 Cornwall and Isles of 14.8 14.4 14.7 15.5 15.1 406 785
Scilly

a
Based on GLS estimates, if not indicated otherwise. The anchoring procedure using
average 5 17.2 and standard deviation 5 5.4 has been applied if not indicated otherwise.
b
The two first letters of the code refer to the country.
c
d
e
f
g
Tax gap (Million €), based on estimates from model 9.
h
Tax gap per capita (€), based on estimates from model 9.
NOTES
1. We are grateful to Artem Korzhenevych for providing this matrix.
2. An alternative approach would be to fix the number of closest regions that are considered
to have a spatial effect to each region. As the geographical size of the NUTS 2 regions
varies a lot, we believe that the distance approach is more valid for characterising the
spatial extent of the spatial effects.
3. Using the general goodness of fit index or the value of the objective function is not appro-
priate as the sample covariance matrix depends on the data transformation and, thus, these
measures are not comparable across models estimated with differently transformed data.
4. In order to avoid random peaks, the moving average of R2h across alternative values of r
with step size 3 is maximised. The same procedure is applied in step 2.



PART III
Shadow economy, illicit work and related

activities



7 The shadow economy and do-it-
yourself activities: what do we
know?
Andreas Buehn and Alexander Karmann
7.1 INTRODUCTION
Why study the shadow economy and do-it-yourself (DIY) activities?

Shadow economic activities increasingly capture the interest of scholars,
policymakers, journalists and the public alike. In broad terms, the shadow
economy covers a wide range of economic activities that are not taxed,
regulated or reported to authorities. These activities take place outside
a society’s legal system and are thus not recorded in national (income)
accounts. Two main aspects make the shadow economy an interesting
and relevant research object for economists. First, according to the over-
whelming majority of the available empirical evidence, the size of the
shadow economy has been growing in most countries over the last decades.
For example, Schneider, Buehn and Montenegro (2010) estimate that its
average size amounted to approximately 35 per cent and 34 per cent of the
gross domestic product (GDP) in developing and transition countries of
the former Soviet Union and Eastern Europe in 2005, respectively. While
the shadow economies in developing and transition countries are relatively
large, the shadow economy is much smaller in developed countries: on
average, it amounted to ‘only’ 17 per cent of GDP by 2005 (Schneider,
Buehn and Montenegro, 2010). These figures nevertheless suggest that the
shadow economy has reached a remarkable size in almost all countries
around the world. Consequently, it has attracted much interest in both,
the (public) media and the academic literature alike. A second aspect is
that effective economic policymaking requires accurate information about
shadow economic activities. Detailed information enables governments
to first effectively measure their extent and to study the determinants and
then to determine how resources may be allocated to deter these activities.
Moreover, tracking the development of shadow economic activities over
261


time gives some advice on how successful these efforts have been and helps
governments to improve or adjust their policies.
While much is know about the size of the shadow economies in dif-
ferent parts of the world, their determinants, and impacts, the literature
has paid less attention to DIY activities. One reason is that in developed
countries, DIY activities are less sizable and dynamic compared to the
shadow economy. Often, they are seen as positive and creative spare time
activities such as fancywork and redecoration and, consequently, have
not attracted much attention in the academic literature. By contrast, the
shadow economy is treated in a negative way and most societies attempt
to reduce or – at least – control it through punishment or by creating
economic conditions supporting growth of the official economy.
In developing countries, however, DIY activities are an important part
of life. As a way of making a living outside the formal economy, either as
an alternative to it, or as a means of supplementing the formally earned
income, they often provide subsistence to families. For this reason, early
studies on the informal economy in developing countries focused on all
parts of the labour force outside the formal labour market, that is (small)
self-employed individuals (for example, Hart, 1970), employment in
unregistered enterprises (for example, International Labour Office (ILO),
1972), and the shadow economy. These studies conclude that the main
determinant for the existence of informal economic activities in developing
countries is the provision of subsistence to families.
DIY activities obviously matter to different degrees in developing and
developed countries. Nevertheless, they are by definition part of the infor-
mal economy that in broad terms covers a wide range of untaxed, unregu-
lated and unreported economic activities. For this reason, DIY activities
should be taken into account when the shadow economy is studied. Recent
descriptive evidence suggests that DIY activities have reached an impor-
tant size not only in developing but also in developed countries and are
thus a noteworthy research object. We analyze shadow economic and DIY
activities simultaneously in a structural equation model and present first
empirical evidence about the relationship between both informal economic
activities, a research object that has not yet been addressed in the literature.
While Section 7.2 discusses and defines the shadow economy and
DIY activities, Section 7.3 provides descriptive evidence about the
shadow economy and DIY activities. Theoretical considerations as to
why individuals turn to shadow economic and DIY activities are given
in Section 7.4. Section 7.5 briefly discusses how both activities may be
related to each other and presents the empirical application testing this
relationship using Germany as a case study. Section 7.6 summarizes and
concludes.


The shadow economy and do-it-yourself activities 263
7.2 NOTIONS AND CLASSIFICATIONS

The character of informal economic activities such as shadow economic
and DIY activities differs depending on the level of a country’s develop-
ment. In developed countries, informal economic activities often refer
to tax evasion, the employment of undeclared labour, such as illegal or
undocumented immigrants, and the smuggling of illegal goods such as
drugs. In developing countries informal economic activities are, however,
often the source of employment for a significant share of the labour force,
due to the weakness of the formal sector to create a sufficient number of
legal jobs. Consequently, the discussion of informal economic activities
in developed countries often focuses on the shadow economy and centres
on rising unemployment, problems of financing public expenditures, tax
evasion and anti-social behaviour, while informal economic activities in
developing countries are considered a central aspect of economic and
social dynamics essential to promote economic development and growth.
The literature has established many different criteria to characterize
activities taking place in the informal economy. A closer inspection reveals
that these criteria are rather heterogeneous: determinants, consequences
and sometimes even the character of the informal activity studied are used
to distinguish informal economic activities from formal ones. For example,
Feige (1981, 1990) and Tanzi (1982, 1986) use the incentive to evade taxes
to classify informal economic activities while Harding and Jenkins (1989)
use consequences for employees such as lack of social benefits and sub-
minimum wages. Other authors (for example, Hart, 1970; De Soto, 1989)
use the legal status or the character of the activity – unregistered or self-
employed – as criterion. The approach of using a selection of criteria to
characterize informal economic activities seems – at a first glance – to be
not very exact and systematic. However, some basic keywords occur again
and again: undeclared labour, tax evasion, unregulated or unlicensed
enterprises, illegality or criminality are the criteria typically considered to
distinguish informal economic activities from formal ones. Table 7.1 lists
these criteria in alphabetical order together with a brief characterization
and influential representatives who used the respective criterion when
defining informal economic activities.1
Using a list of some basic criteria to characterize informal economic
activities has pros and cons. The advantage of this approach is that it
enables researchers to distinguish between very different activities that
all comprise the informal economy. For example, it allows distinguish-
ing between goods and services produced and consumed within the
household, forms of illegal employment, tax evasion and social secu-
rity fraud, and even criminal economic activities like smuggling drugs.


Table 7.1 Criteria of informal economic activities
Criterion Example/explanation Author(s)

Government Lack of government ILO (1972), Feige (1981, 1989),
regulation regulation Harding and Jenkins (1989)
Illegality General illegal activities Feige (1981, 1989), Harding and
Jenkins (1989), Renooy (1990)
Labour Undeclared labour, lack ILO (1972), Harding and Jenkins
market of social benefits, sub- (1989), Renooy (1990)
minimum wages
National Escaping measurement in Feige (1981), Tanzi (1982, 1986),
statistics national statistics because Renooy (1990)
of accounting conventions,
non- or under-reporting
Professional Self-employment, family Hart (1970, 1973), ILO (1972),
status workers, domestic servants Swaminathan (1991)
Registration Unregistered or unlicensed De Soto (1989), Swaminathan
enterprises (1991)
Survival Widespread in developing Banerjee (1982), Swaminathan
countries, less important (1991)
for developed countries
Unreported Unreported with the Allingham and Sandmo (1972),
income/tax intension to evade taxes Feige (1981, 1990), Tanzi (1982,
evasion 1986), Frey (1989), Alm (1991)
The disadvantage is that it makes providing an explicit, comprehensive

definition of informal economic activities a challenging if not impossible
task. For this reason, researchers gave up trying to formulate a unique
definition of informal economic activities but rather present a definition in
accordance with the problem at hand. The literature has nevertheless come
to an agreement or compromise that the informal economy comprises all
goods and services which normally should be added to the calculation of
the national product but are not part of it because firms are (1) not legally
registered as businesses or goods and services are produced in home
production and distributed within the household sector (DIY economy);
(2) establish informal working contracts with their employees or do not
comply with tax laws and regulations (shadow economy); or (3) transac-
tions are not revealed to authorities because goods and services are illegal
(Thomas, 1992). Examples of firms operating in the informal economy are
thus small-scale production units such as family-based businesses, micro-
enterprises with at most five employees, and large formally registered
firms that establish informal working contracts to avoid payment of social


security contributions, evade taxes or get around certain labour market

regulations such as minimum wages and safety regulations.
Shadow economic and DIY activities can be assigned to the informal
economy using one or the other criterion presented in Table 7.1. Shadow
economic activities are informal because using undeclared labour in
order to save on labour costs and social security contributions, evading
taxes and the non- or under-reporting of income are unlawful and thus
informal. The shadow economy thus includes all market-based, lawful
produced goods and services that are deliberately concealed from public
authorities to avoid payment of income, value added or other taxes and
social security contributions; to get around certain legal labour market
standards, such as minimum wages, maximum working hours and safety
standards; or to avoid administrative procedures, such as filling in forms
and statistical questionnaires. Another important characteristic of shadow
economic activities is that they are not recorded in the official national
statistics. This characteristic of informal economic activities in general
and shadow economic activities in particular, is also the reason why DIY
activities – although being neither unlawful nor illegal – are by definition
part of the informal economy.
DIY activities include all market-based goods and services which are
produced in home production in order to avoid gross wage payments,
including taxes and social security contributions, in the official economy
or to avoid any net wage payments in the shadow economy. That is, DIY
activities are primarily undertaken to avoid labour costs either in the offi-
cial or in the informal economy. It is also important to note that the main
difference between DIY and shadow economic activities is that the former
are entirely legal. Although DIY activities are entirely legal or lawful
they are not accounted in the official statistics and are thus – like shadow
economic activities – informal.
Scholars analyzing the shadow economy often focus on the shadow
economy in the narrow sense, that is, on the hidden or clandestine pro-
duction/consumption of goods and services that are generally lawful.2
These activities are typically referred to as the legal part of the shadow
economy (see, for example, Schneider and Enste, 2000). DIY activities
are, however, typically neglected in the analysis. For this reason, not
much is known about the relationship between shadow economic and
DIY activities. On the one hand, shadow economic and DIY activities
can be seen as substitutes. If it is too risky to demand shadow economic
services – for fear of being caught and incurring fines and/or punishment
– for example, individuals may undertake DIY activities. On the other
hand, shadow economic and DIY activities may complement each other,
that is, individuals demand shadow economic services in addition to the


production of goods and services by themselves assuring a certain quality

standard and efficient execution. For example, an individual may choose
to renovate her home herself but may hire a handyman informally for
tasks she does not know how to do or cannot do well. In this way, she is
supplementing (or complementing) her own DIY activities with shadow
economic ones. In Section 7.5, we address this question and present a
structural equation model to analyze the relationship between these two
types of informal economic activities. Before we turn to the empirical
analysis of the question of whether shadow economic and DIY activi-
ties are substitutes or complements, the next section first presents some
descriptive statistics about the size and development of shadow economic
and DIY activities.
7.3 THE SHADOW ECONOMY AND DIY

ACTIVITIES: SOME DESCRIPTIVE EVIDENCE
7.3.1 Size and Trends of the Shadow Economy
Most societies attempt to control the shadow economy through various

measures such as punishment, prosecution and enlightenment in order
to spur growth in the official economy. Gathering information about
the extent of the shadow economy, who is engaged in shadow eco-
nomic activities, the frequency of these activities and the magnitude of
them, is crucial for making effective and efficient decisions regarding
the allocations of a country’s resources. Unfortunately, it is very dif-
ficult to get information about shadow economy activities on the goods
and labour market, because participants do not wish to be identified.
Although substantial literature3 exists on single aspects of the shadow
economy and comprehensive surveys have been written by Schneider
and Enste (2000), and Feld and Schneider (2010), the subject is still quite
controversial.
Nevertheless, the literature has gained some insight about the devel-
opment and the size of the shadow economies in developing, Eastern
European and Central Asian (mostly the former transition countries) and
high-income OECD countries in the recent past. According to Schneider,
Buehn and Montenegro (2010) who estimated the size of the shadow
economies in 162 countries around the world over the period 1999 to
2007, the shadow economy has reached a remarkably large size of an
average value of 32.9 per cent of official GDP in that period. The average
size of the shadow economies in all of these 162 countries (developing,
Eastern European and Central Asian and high-income OECD countries)


decreased modestly from 34.1 per cent of official GDP in 1999 to 31.0 per
cent of official GDP in 2007.
The size of the shadow economies in 98 developing countries decreased
from 36.6 per cent of official GDP in the year 1999 to 35.1 per cent in the
year 2007. The three countries with the smallest shadow economies are
China, Singapore and Vietnam with an average country size of 12.8, 13.0
and 15.2 per cent.4 The highest shadow economies includes Peru, Panama
and Bolivia; with a size of 58.7, 63.5 and 66.6 per cent of GDP. The large
shadow economies in developing countries are often only to some extent
an issue of tax burden and regulation, given the simple fact that the limited
local economy means that citizens are often unable to earn a living wage
in a legitimate manner. Working in the shadow economy is often the only
way of achieving a minimal standard of living. It should also be noted
that the average size of the Asian shadow economies are smaller than the
shadow economies of African and Latin American countries.
Estimations of the size and trend of the shadow economies in the tran-
sition countries has been undertaken since the late 1980s starting with
the works of Kaufmann and Kaliberda (1996), Johnson, Kaufmann and
Shleifer (1997) and Lackó (2000). They all use the physical input (electric-
ity) method and come up with quite large estimates. Alexeev and Pyle
(2003) as well as Belev (2003) critically evaluated the aforementioned
studies arguing that the estimated sizes of the shadow economies are to
a large extent a historical phenomenon and partly determined by institu-
tional factors. Schneider, Buehn and Montenegro (2010) estimated that
the size of the shadow economies in the Eastern European and Central
Asian countries decreased from 36.9 per cent in 1999 to 33.7 per cent
in 2006. The three countries with the smallest shadow economy in this
geographical region are the Slovak and Czech Republics and Hungary
with an average size over the period 1999 to 2006 of 18.1 per cent, 18.4
per cent, and 24.4 per cent. The highest shadow economies are found in
Moldova, Ukraine and Georgia; with 44.5 per cent, 49.7 per cent, and
65.9 per cent.
Compared to the developing as well as Eastern European and Central
Asian countries, the size of the shadow economies in the high income
OECD countries is much smaller. It was 17.7 per cent of official GDP
in 1999, and decreased to 16.6 per cent in 2007 (Schneider, Buehn and
Montenegro, 2010). Some high income OECD countries (like Italy) show
ups and downs, others (like Belgium and Australia) show a steady decrease
in their shadow economies. The countries with the lowest shadow econo-
mies among the high-income OECD countries are Switzerland, the United
States and Luxembourg; with an average size of the shadow economy over
the period 1996 to 2007 of 8.5 per cent, 8.6 per cent, and 9.7 per cent. Italy,


Table 7.2 Average size and trend of the shadow economies in different
parts of the world
Region of the world Years

1999 2000 2001 2002 2003 2004 2005 2006
Developing countries 37.0 36.7 36.8 36.8 36.5 36.1 35.5 35.1
Transition countries 36.9 36.3 36.1 35.8 35.3 39.8 34.3 33.7
OECD countries 17.7 17.4 17.3 17.3 17.3 17.1 17.0 16.8
Source: Schneider, Buehn and Montenegro (2010)
Greece and Mexico experience the highest shadow economies among the
high-income OECD countries with an average size over the period 1996
to 2007 of 27.0 per cent, 27.5 per cent, and 30.0 per cent. Table 7.2 sum-
marizes the descriptive evidence about the size of the shadow economies
around the world according to the most recent and comprehensive study
of Schneider, Buehn and Montenegro (2010).
7.3.2 Size and Distribution of DIY Activities
While much is known about the size of the shadow economies around the
world and the determinants driving shadow economic activities, econo-
mists did not pay much attention to the economics of the household until
the publication of Garry Becker’s book The Treatise of the Family (Becker,
1981). In his book, Becker explains the behaviour of family members with
models of constrained optimisation. Introducing two kinds of human
capital – household human capital and market human capital – he showed
that specialization, not only in the division of labour but also in human
capital investment between household and market activities, is optimal
for the family.5 While Becker (1981) laid the foundations of the economics
of the household, research in the last two decades has mostly focused on
intra-household bargaining and the effects of the tax system and the supply
of public childcare on the child-related labour decision of women. Less
attention had been paid to the economics of DIY activities or home pro-
duction in general, that is, to questions such as what are important deter-
minants influencing DIY activities, who is particularly likely to engage in
these activities and how sizable are they.6 Since labour supply is not merely
a choice between leisure and paid (taxed) work but also influenced by
the consumers’ productivity with respect to non-market household DIY
production (Becker, 1965), DIY and other non-market household produc-
tion can make the consumer better off – even at a usually lower level of


productivity – than when working in the paid, but taxed, labour market
and purchasing goods and services that are subject to taxation. Hence, it
can be rational for households to engage in DIY activities.
A recent comprehensive questionnaire-based survey on DIY activities
in Germany, Great Britain, Denmark, Norway and Sweden has been con-
ducted by the Rockwool Foundation Research Unit (Brodersen, 2003).
To assure comparability, all country-specific surveys used a set of nearly
identical questions on DIY activities and were surveyed between 1993 and
2001 in Denmark and between 1998 and 2001 in the other four countries.
Evaluation of the questionnaires showed that age is a very important
factor affecting the households’ likelihood of carrying out DIY activities
in the home. In all countries surveyed, the likelihood of engaging into
DIY activities within the last 12 month falls with age. Another important
determinant is whether the respondent was an owner-occupier or not.
Those who owned their own homes are the most likely to carry out repairs
and improvements themselves. The report further shows that the volume
and value of DIY activities in relation to home repairs and maintenance
is much higher for owner-occupiers than for tenants. The third important
variable is marital status which is significant in all countries in that married
or cohabitating respondents were more likely to carry out DIY activities.
In addition to these variables, which show significant influence in all
countries, a number of variables such as income and occupation are impor-
tant determinants in several but not all countries. Income is found to be
significant in all countries except Great Britain. In the Scandinavian coun-
tries of Norway and Sweden, a positive correlation between income and the
likelihood of DIY activities is observed, while the correlation is negative in
Germany.7 The respondents’ occupation plays a role in Denmark, Sweden
and Germany, where, in particular, skilled workers and the self-employed
carry out home repairs and maintenance. Skilled workers are significantly
more likely to carry out this type of work in all three countries, while self-
employed – compared to other occupational groups – are significantly less
likely to carry out DIY activities in Denmark and Sweden. In Germany,
the educational level also plays a role in that students were more likely to
carry out DIY activities than other occupational groups. Regional effects
are found in Great Britain, Sweden and Germany. As one would expect
– and what is also confirmed by the macroeconomic analysis presented in
our chapter (see Section 7.5) – the likelihood of engaging in DIY activities
is higher in the Eastern part of Germany, that is, in the former GDR.
The Danish survey of 2001 on DIY activities also included questions
on the type of work carried out within the last 12 months. With respect
to minor repairs and maintenance, painting and wallpapering are by far
the most common types of DIY activities: 63 per cent of the households


asked said they did this type of work themselves. In those cases where
wallpapering had been done by a firm, 21 per cent of households did the
painting themselves. Brodersen (2003) argues that these high figures for
DIY activities are probably due to the fact that, in particular, painting
neither requires special knowledge nor expensive materials. Whether the
quality of the DIY activity is satisfactory is, of course, another question.
However, if one’s own skills are not enough to assure a certain quality,
individuals may demand ‘professional’ shadow economic services in addi-
tion to their own DIY activities. In this case shadow economic and DIY
activities are complementary to each other.
Other activities which do not require much special knowledge and are
thus typically carried out by households themselves are activities such
as fitting a new kitchen, new ceiling, cupboards and other carpentry.
More advanced activities such as plumbing, heating or doing work in the
bathroom are – at least in the Danish case – typically left to specialists.
These activities, which often require expensive special tools or special
knowledge, deter households from doing them themselves and are thus
substitutive to DIY activities.
More insight into the time households allocate to DIY activities in 15
European countries is provided in the Harmonized European Time Use
Survey (HETUS) (HETUS, 2007) developed by Statistics Sweden. Using
data from Eurostat, Statistics Sweden estimated the mean time a household
spends on DIY activities during an average day. Figure 7.1 shows the distri-
bution of four types of DIY activities, handicraft, gardening, construction
and repairs, and informal help to other households, in these 15 countries.
It can be seen that households spent relatively little time doing handicraft.
Significantly, more time is spent doing gardening, construction and repairs,
and even informal help to other households, though households spent most
of their time doing gardening. Among countries, the distribution of handi-
craft, construction and repairs, and informal help to other households is
relatively homogeneous, although Italians and Spanish devoted on average
much less time to construction and repairs than their European counter-
parts. Important differences can be observed with respect to gardening
and informal help to other households. For these activities the distribution
is much more heterogeneous, Bulgarians and Slovenians spent twice (in
some cases, for example, Finland, Spain and the United Kingdom, even
three or four times) as much time gardening than individuals in the other
13 European countries. The reason for this may be seen in the relative
importance of agriculture in the Bulgarian and Slovenian economies.
Assuming that individuals split an average day into 16 hours of work
and 8 hours of recovery, Figure 7.2 shows how much of their 16 hours
of work individuals in the different European countries devoted to the


Belgium 0:03 0:14 0:11

Bulgaria 0:09 0:30 0:09 0:08
Estonia 0:08 0:18 0:15 0:13
Finland 0:06 0:07 0:12 0:11
France 0:04 0:13 0:18 0:09
Germany 0:04 0:15 0:11 0:09
Italy 0:04 0:11 0:03 0:10
Latvia 0:05 0:19 0:09 0:11
Lithuania 0:06 0:13 0:09 0:13
Norway 0:06 0:10 0:14 0:08
Poland 0:03 0:11 0:10 0:15
Slovenia 0:04 0:28 0:13 0:08
Spain 0:05 0:06 0:04 0:09
Sweden 0:02 0:11 0:12 0:08
United Kingdom 0:03 0:10 0:11 0:09
Handicraft Gardening Construction and repairs Informal help to other households
Source: HETUS (2007), own calculations
Figure 7.1 Distribution of DIY activities
7.0%
6.0%
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
Belgium
Bulgaria
Estonia
Finland
France
Germany
Italy
Latvia
Lithuania
Norway
Poland
Slovenia
Spain
Sweden
United Kingdom
Source: own calculations
Figure 7.2 Time allocated to DIY activities within 16 hours of work


four types of DIY activities. It is obvious that Bulgarians, Estonians and

Slovenians devote on average the most time of their 16 hours day to DIY
activities, they spent around 6 per cent or almost one hour on DIY activi-
ties. This is around twice as much as individuals allocate to DIY activities
in Belgium, Italy and Spain. In the latter countries, individuals devote only
less than 3 per cent or 30 minutes to DIY activities. This is the lowest level
among the 15 European countries included in the Harmonized European
Time Use Survey. All other countries such as Germany, France or the
United Kingdom show levels of DIY activities that fall between the lower
and upper bound of 3 per cent and 6 per cent.
Some of the surveys surveyed by the Rockwool Foundation Research
Unit, that is, the Danish surveys of 1993/94, 2000 and 2001 as well as the
German survey of 2001 also asked respondents about their motives for
doing DIY activities. It turned out that the main motives for DIY activi-
ties are economic ones. In both countries, around 80 per cent of those
asked said that it was mainly or partly to save money, while less than 25
per cent said it was mainly because they enjoyed doing this type of work.
Danes, who enjoyed DIY activities, carried out significantly more work
themselves than those who were motivated by saving money. To conclude
on the determinants of DIY activities, the surveys have shown that first
economic motives play an important role in the context of DIY activities
and second that individuals significantly spent more time on DIY activities
if they enjoyed it.
Analyzing household expenditure surveys on materials for DIY repairs
and maintenance as well as turnover in DIY stores and paint and wall-
paper shops, the estimated total value of DIY activities in the form of home
repairs, maintenance and improvements corresponds to approximately
2.5–3 per cent of GDP in Denmark, Norway and Sweden and 1–1.5 per
cent in Germany and Great Britain. These figures correspond to around 10
full time jobs a year per 1000 inhabitants in the Scandinavian countries and
to 3–4 full time jobs per 1000 inhabitants in Germany and Great Britain
if all the DIY activities were carried out as invoiced work by a firm. These
differences may be explained by the relatively high tax burden and mar-
ginal tax rates in the Scandinavian countries compared to Germany and
Great Britain. Other, national factors may also play a noteworthy role.
For example, low growth rates in Germany during the survey period (for
example 0.2 per cent in 2001), the need for renovations and modernization
of large parts of the Danish housing stock, temporary subsidies for main-
tenance and improvements of homes, or differences in building traditions
were also likely to contribute to the observed differences. In Section 7.5 we
present macroeconomic estimates for the size and trends of DIY activities
in Germany contrasting the microeconomic evidence provided by HETUS.


Moreover, we explore the relationship of DIY activities to the shadow

economy using a structural equation model.
7.4 THEORETICAL CONSIDERATIONS ABOUT THE

SHADOW ECONOMY AND DIY ACTIVITIES
While the previous section has presented some descriptive evidence about
the size and trends of the shadow economy and DIY activities, this section
discusses their major determinants. We start with the shadow economy
and then turn to DIY activities.
7.4.1 Determinants of Shadow Economic and DIY Activities
Tax burden
Almost all studies show that the overall tax burden is among the main
causes for the existence of the shadow economy. Since taxes distort rela-
tive prices they affect individuals’ behaviour, in particular their labour–
leisure choices, and thus stimulate labour supply in the shadow economy
where taxation can be avoided. The bigger the difference between the total
cost of labour in the official economy and the after-tax earnings (from
work), the greater is the incentive to avoid this difference and to work
in the shadow economy. Since this difference depends broadly on social
security contributions and the overall tax burden, they are key features
of the existence and the increase of the shadow economy. Analogously to
the shadow economy, higher taxes may produce an incentive to produce
goods and services by oneself relatively cheaply rather than to buy goods
and services in the official economy which is relatively costly.
Empirical results on the influence of the tax burden on the shadow
economy is provided – to name some references – in the studies of Schneider
(1994, 2005, 2007) and Johnson, Kaufmann and Zoido-Lobatón (1998);
they all found statistically significant evidence for the influence of taxation
on the shadow economy. This strong influence of indirect and direct taxa-
tion on the shadow economy is further demonstrated by discussing empiri-
cal results in the cases of Austria and the Scandinavian countries. For
Austria, the driving force for shadow economic activities is the direct tax
burden (including social security payments); it has the largest influence,
followed by the intensity of regulation and complexity of the tax system.
A similar result was found by Schneider (1986) for the Scandinavian coun-
tries (Denmark, Norway and Sweden). In all three countries various tax
variables – the average direct tax rate, the average total tax rate (indirect
and direct tax rate) and marginal tax rates – show the expected positive


correlation and are highly statistically significant. These findings are sup-
ported by studies of Kirchgässner (1983, 1984) for Germany, by Klovland
(1984) for Norway and Sweden, and Karmann (1990) for Germany too.
Measuring the tax (and social security contribution) burden is not an
easy task, because the tax and social security systems are vastly differ-
ent among countries. In order to have some general comparable proxies,
the literature applies various measures for the tax (and social security
contributions) burden: (1) direct/indirect taxes as a proportion of total
overall taxation, (2) share of direct/indirect taxes: direct/indirect taxes in
percentage of the GDP, and (3) size of government: general government
final consumption expenditures in per cent of GDP, which includes all
government current expenditures for purchases of goods and services.
Sometimes, authors use the fiscal freedom index, which is a subcomponent
of the Heritage Foundation’s economic freedom index, as a proxy for the
tax burden. This index ranges from 0 to 100, where 0 is least fiscal freedom
and 100 maximum degree of fiscal freedom, and measures the fiscal burden
in an economy by including the top tax rates on individual and corporate
income.
Intensified regulations are another important factor which reduces the
freedom (of choice) for individuals engaged in the official economy. In
particular, labour market regulations such as minimum wages or dis-
missal protection, trade barriers, such as import quotas, and labour
market restrictions for foreigners, such as restrictions regarding the free
movement of foreign workers, increase the costs of labour and, thus, the
incentives to reduce costs by escaping to the shadow economy. Johnson,
Kaufmann and Zoido-Lobatón (1998) find significant empirical evidence
of the influence of (labour) regulations on the shadow economy; and the
impact is clearly described and theoretically derived in other studies, for
example, for Germany (Deregulation Commission, 1990/91). Johnson,
Kaufmann and Shleifer (1997) present a model that predicts that countries
with more general regulation of their economies tend to have a higher
share of the unofficial economy in total GDP. Johnson, Kaufmann and
Zoido-Lobatón (1998) conclude that it is the enforcement of regulation
which is the key factor for the burden levied on firms and individuals, and
not the overall extent of regulation – mostly not enforced – which drives
firms into the shadow economy. Friedman et al. (2000) reach a similar con-
clusion. In their study every available measure of regulation is significantly
correlated with the share of the unofficial economy and the estimated sign
of the relationship between their measures of regulation and the shadow
economy is unambiguously positive: more regulation is associated with


a larger shadow economy. These findings show that governments should

put more emphasis on improving enforcement of laws and regulations,
rather than increasing their number. Some governments, however, prefer
this policy option (more regulations and laws), when trying to reduce the
shadow economy, mostly because it leads to an increase in power for the
bureaucrats and to a higher rate of employment in the public sector. With
respect to DIY activities, however, we later argue that these activities
are not primarily driven by business and labour market regulations but
individual constraints such as unemployment and wages.
To measure the intensity of regulation or the impact of regulation on
the decision of whether to work in the official or unofficial economy is a
difficult task. The literature thus uses different measure of regulation: (1)
public sector employees and (2) indices measuring the intensity of regula-
tion in an economy such as the business freedom index of the Heritage
Foundation or the regulatory quality index of the World Bank. These
indices include measures of the incidents of market-unfriendly policies,
such as price controls or inadequate bank supervision, as well as percep-
tions of the burdens imposed by excessive regulation in areas such as
foreign trade and business development.
Labour market
It has been shown in a number of studies (Bajada and Schneider, 2005;
Enste and Schneider, 2006; Feld and Schneider, 2010) that the situa-
tion of the official economy also plays a crucial role for labour supply
in the informal economy. A booming official economy offers numer-
ous employment opportunities which are usually not available if the
economy faces a recession. In the latter case people will try to com-
pensate their losses of income through additional economic activities.
Numerous studies have identified labour market determinants as an
important driving force for informal economic activities.8 The reason
for the consideration of labour market variables in the analysis of infor-
mal economic activities is, that unemployed refers to the share of the
labour force that is without work but available for and seeking employ-
ment and thus are likely to supply labour in the shadow economy or
engage in DIY activities.
Consensus exists that unemployment in OECD countries is caused to a
large extent by high labour costs. Obviously, the higher the unemployment,
the less employment opportunities exist in the official economy and, thus,
the higher are incentives to engage in informal economic activities. If this
relationship holds true, one would expect a positive correlation between
unemployment and informal economic activities, that is, the higher the
unemployment, the greater the shadow economy and/or DIY activities.


Additionally, DIY activities may enhance the self-esteem of the unemployed,

thereby further stimulating DIY activities. On the other hand, a negative
relationship may hold because the unemployed have less money for purchas-
ing goods and services, in the official and also in the unofficial economy.
It is also apparent that the higher the average gross hourly earnings
in the official small trade sector, the higher the costs for individuals who
demand such services. Given that individuals are able to do these activities
themselves, they may substitute demand both in the official small trade
sector and in the unofficial sector – which run the risk of punishment
and fines – for DIY activities.9 In order to account for the labour market
incentives, the literature uses the unemployment rate and hourly wages in
empirical analyses.
To summarize, we argue that the shadow economy is primarily driven
by institutional factors like taxation and regulation, while DIY activities
respond to supply-side variables like unemployment and wages. Besides
these determinants, we include per capita real disposable income as a
control variable. Here, a positive relationship is assumed. Since real dis-
posable income is positively correlated with the demand for goods and
services in general, we hypothesize that the higher the real disposable
income, the greater the demand not only in the official but also in the
unofficial economy and, hence, the larger the shadow and DIY economies.
7.4.2 Indicators of Shadow Economic and DIY Activities
Shadow economic and DIY activities cannot be directly measured. For

this reason, the literature uses proxies or indicators which reflect these
activities to a certain degree and represent transmission channels of
informal economic activities to the official economy. An obvious trans-
mission channel is the effects of shadow economic activities on economic
growth.
An increase of the shadow economy can lead to reduced state revenues
which in turn reduce the quality and quantity of publicly provided goods
and services. Ultimately, this can lead to an increase in the tax rates for
firms and individuals in the official sector, quite often combined with a
deterioration in the quality of the public goods (such as the public infra-
structure) and of the administration, with the consequence of even stronger
incentives to participate in the shadow economy. Johnson, Kaufmann and
Zoido-Lobatón (1998) present a simple model of this relationship. Because
the quantity and quality of the public infrastructure are key elements for
economic growth, an increasing shadow economy – ceteris paribus – results
in lower growth rates of the official economy. This negative view of the
shadow economy is also held by, for example, Loayza (1996).


An alternative view – held by some authors (for example, Asea 1996,

Tanzi 1999) – is that shadow economic activities are something positive
and creative, responding to the demand for services and small-scale manu-
facturing. Thus, the shadow economy adds a dynamic component to the
economy promoting the creation of new markets and enhancing entrepre-
neurship. This in turn can lead to more competition and higher efficiency
both stimulating economic growth.
All in all, the effects of the shadow economy on official growth figures
are ambiguous. In addition to official economic growth figures, the lit-
erature uses different monetary indicators, for example, M0 over M1, M0
over M2, or M0 over GDP, as a proxy for shadow economic activities.
With respect to these indicators, the idea is that people who engage in
shadow economy transactions do not want to leave traces. Using cash
for payments in the unofficial economy eliminates the ‘paper trail’ and
protects principal and agent. Hence, most shadow economy activities
are reflected in an additional use of cash. In other words, excessive cash
holdings, that is, cash holdings which exceed equilibrium money demand
determined by the transaction motive, mirror shadow economic activities.
While both indicators of shadow economic activities are well-established
in the literature, this is not the case for unobservable informal economic
activities of the household (DIY activities). A suitable indicator for such
activities might be turnover in DIY stores because inputs for DIY activi-
ties are typically – at least in developed and emerging economies –bought
in DIY stores. Hence, inventory turnover at DIY stores may reveal levels
of DIY activity (Karmann, 1990; Brodersen, 2003).
Alternative indicators for DIY activities might be labour market
indicators such as the labour force participation rate or the number of
hours worked in the official economy. Because these are measures of the
labour supply for the production of goods and services during a speci-
fied period in the official economy one would expect a negative relation-
ship to informal economic activities. This means that – ceteris paribus
– the more common the DIY activities, the lower the labour supply in
the official economy. The average hours worked per week in the official
economy can thus be used as an indicator for DIY activities.10 If these
activities increase ceteris paribus, the number of hours worked in the
official economy will reduce. Therefore, one would expect a negative
relationship between DIY activities and working hours in the official
economy, often measured by the average number of hours worked per
week. We must however admit that no consensus exists in the literature
about indicators/proxies for DIY activities, which is obviously due to
the fact that, until now, the literature has not paid much attention to
these activities.


7.5 EMPIRICAL ANALYSIS

While the previous two sections have provided an overview about the size
of shadow economic and DIY activities as well as some theoretical con-
siderations regarding the most important determinants of both activities,
this section analyzes the relationship between the shadow economy and
DIY activities. The first pioneering measurements of DIY activities for
Germany were undertaken by Karmann (1990). This analysis was extended
in Buehn, Karmann and Schneider (2009) where the authors use two types
of structural equation models (SEM) to estimate the size and trends of
the shadow economy and DIY activities in Germany. In order to analyze
empirically whether the shadow economy and DIY activities are substi-
tutes or complements, we modify the SEM presented in Buehn, Karmann
and Schneider (2009) by explicitly focusing on the relationship between the
two latent variables of shadow economy and DIY activities. Hence, in the
first step we modify our previous empirical analysis for shadow economic
and DIY activities in Germany by explicitly estimating the link between
both and finally attempt to draw some general conclusions.
7.5.1 Data
The data for the empirical analysis cover the period 1970 to 2005 on an
annual basis. Following Buehn, Karmann and Schneider (2009) and our
theoretical considerations in Section 7.4, we use public revenue data (per-
centage of GDP) provided by the OECD (Organization for Economic
Co-operation and Development) for the approximation of tax and social
security contribution burdens.11 For the regulatory burden, we use the
number of individuals employed in public service (percentage of total
population) excluding employees by the railways and postal service.12
For unemployment and wages we use the number of unemployed in the
German economy and the average gross hourly earnings of male workers
in the small trade sector. All data are taken from the Federal Statistical
Office of Germany. In addition to these determinants, we include per
capita real disposable income – taken from the Deutsche Bundesbank – as
a control assuming a positive relationship. Since real disposable income
is positively correlated to demand for goods and services in general, a
higher per capita real disposable income stimulates demand not only
in the official but also in the unofficial economy which would increase
shadow economic and DIY activities. Because German reunification
offered remarkable opportunities not only in the formal but also in the
informal economy, we include a zero-one dummy taking one in the years
1991 and 1992 to control for different behavioural patterns in Eastern


and Western Germany and structural changes to the German economy

as a result of German reunification in 1990. We expect a positive correla-
tion between the dummy variable and shadow economic as well as DIY
activities.
Buehn, Karmann and Schneider (2009) applied a vector error correc-
tion model to adjust currency in circulation for the introduction of the
euro in 2002 and foreign demand for the deutsche mark stemming from
Eastern and southeast Europe because of unstable political circumstances
in the early 1990s, the war in Kosovo and the Bulgarian financial crisis of
1996/97. We use their predicted times series for domestic currency in cir-
culation in Germany as an indicator for the shadow economy. The second
indicator variable is the growth rate of real GDP taken from the Federal
Statistical Office of Germany.
For DIY activities, we use the average hours worked per week and
turnover in DIY stores as indicators. While the former are taken from
the Federal Statistical Office of Germany, the latter were provided by
A.C. Nielsen Company GmbH starting in 1978 when the A.C. Nielsen
Company GmbH conducted the first annual survey on turnover in DIY
stores in Germany. To complete the time series for the entire period 1970–
2005, Buehn, Karmann and Schneider (2009) regressed its annual growth
rates on a constant term and on a linear time component and calculated
estimates from 1970 to 1977. They were then used to predict the level of
turnover in DIY stores for the years 1971 to 1978. For a detailed descrip-
tion of all data transformations and a complete list of sources see Buehn,
Karmann and Schneider (2009).
7.5.2 Shadow Economic and DIY Activities: Substitutes or Complements?
Descriptive evidence for Germany

To begin with, Figure 7.3 plots the size and development of the shadow
economy and DIY activities in Germany as estimated by Buehn, Karmann
and Schneider (2009). While the size of the shadow economy is displayed
on the left hand vertical axis, the size of DIY activities is displayed on the
right hand vertical axis. Figure 7.3 shows a remarkable increase in the
shadow economy over the past 25 years, reaching 17.40 per cent of official
GDP in 2005. The German reunification in 1990 triggered a steep rise
in the shadow economy during the reconstruction period that followed.
After East Germany caught up to West Germany, growth in the shadow
economy slowed down considerably.
DIY activities increased from 4.05 per cent of official GDP in 1970 to
4.94 per cent in 1995 and remained more or less stable until 2005. Like
shadow economic activities, DIY activities also experienced a big push


20% 6%
18%
5%
16%
14%
4%
12%
10% 3%
8%
2%
6%
4%
1%
2%
0% 0%
1971 1975 1979 1983 1987 1991 1995 1999 2003
Shadow economy DIY economy
Note: The times series estimates for the sizes and trends of the shadow and DIY economy
are taken from Buehn, Karmann and Schneider (2009).
Figure 7.3 Shadow and DIY economy in Germany in % of GDP

(1970:2005)
following the German reunification – even though the dynamics were not
as pronounced. On the whole, between 1970 and 2005, DIY activities grew
more slowly than did the shadow economy. However, the catch-up process
in Eastern Germany after reunification offered remarkable opportunities in
both sectors of the informal economy. However, shadow economic activi-
ties appear to develop more dynamically than household DIY activities.
Although Figure 7.3 shows an upward trend – more pronounced for the
shadow economy – of shadow economic and DIY activities, their exact
relationship is per se not clear. They may function as complements or sub-
stitutes under different circumstances. If it is too risky to demand shadow
economic services – for fear of being caught and incurring fines and/or
punishment – individuals may be more willing to undertake DIY activities
instead. In this case, both would be substitutes. However, shadow eco-
nomic and DIY activities may complement each other, that is, individuals
demand for ‘professional’ shadow economic services in addition to their
own production of goods and services in the household, may be to assure
a certain quality standard as well as an efficient and faster execution. To
analyze this theoretical reasoning empirically, we examine this question in
the next section by using an SEM.


The empirical model and hypotheses

An SEM can provide useful information about the relationship between
the shadow economy and DIY activities. Such an SEM models the shadow
economy and DIY activities as two distinct latent variables and explores
their relationship using the covariance structures between these latent
variables’ observable causes and indicators.
Formally, the SEM consists of two parts: the structural equation
model and the measurement model. The structural equation model can be
represented by:
h 5 Bh 1 Gx 1 ß, (7.1)
where each xi, i 5 1, . . ., q in vector xr 5 (x1, x2, . . ., xq) is a potential

cause of one of the two latent variables h1, h2 constituting vector h.
The individual coefficients gr 5 (g1, g2, . . ., gq) in the (2 3 q) matrix
G describe the relationships between the two latent variables and their
causes. Each latent variable is determined by a set of exogenous causes.
The error terms in vector ß represents the two unexplained components of
the latent variables, the (2 3 2) covariance matrix for which is abbrevi-
ated by Y. F is the (q 3 q) covariance matrix of the causes. The (2 3 2)
coefficient matrix B provides information about the interaction between
DIY activities and the shadow economy.
The measurement model links the latent variable to its multiple observ-
able indicators, that is, it is assumed that the latent variable deter-
mines its indicators. The measurement model provides information that
single-indicator models do not. It is specified by:
y 5 Lh 1 e, (7.2)
where yr 5 (y1, y2, . . ., yp) is the vector of indicators for corruption and the
shadow economy, L is a (p 3 2) matrix of regression coefficients, and e is
a p vector of white noise disturbances, the (p 3 p) covariance matrix for
which is given by Qe.
Inserting the reduced form of Equation (7.1), h 5 (I 2 B) 21Gx 1
(I 2 B) 21ß, into Equation (7.2) yields a reduced form regression model
where the endogenous variables yj, j 5 1, . . ., p are the indicators of the
latent variable h1 and h2 and the exogenous variables xi, i 5 1, . . ., q are
their causes. This model can be written as:
y 5 Px 1 z, (7.3)


where P 5 L (I 2 B) 21G is a (p 3 q) matrix and z 5 L (I 2 B) 21z 1 e.

The error term z in Equation (7.3) is a p vector of a linear transformation
of the white noise error terms ß and of e, resulting from the structural
equation and the measurement model, that is, z (0, W) . The covariance
matrix W is given by:
Cov (z) 2 W 5 E 5 L [ (L (I 2 B) 21ß 1 e) (L (I 2 B) 21ß 1 e) ]
5 L (I 2 B) 21Y (I 2 B) 21Q 1 e.
The parameters of the model are estimated using the observed variables’
variances and covariances by estimating an SEM’s covariance matrix
S (q) , S^ 5 S (q^ ) , that is as close as possible to the sample covariance
matrix of the observed causes and indicators.13 dentification and estima-
tion of the model is, however, not possible without placing restrictions on
certain model parameters. Among others, a restriction often imposed on
the model is that one indicator of each latent variable, that is, one element
of each column vector lp in L, set to an a priori value. In this way the
researcher also establishes an interpretable scale for the latent variable
(Bollen, 1989, pp. 91, 183).14
The SEM used here to analyze the relationship between the shadow
economy and DIY activities has the following specific characteristics:
x1
x2
c 0 x3
c d 5 c d c d 1 c 1 d c 1d ,
h1 0 b12 # h1 g g2 g3 0 0 # Gx4W 1 ß
h2 0 0 h2 0 0 0 g4 g5 c g ß2
q
x5
( (7.4)
xq
y1 1 0 e1
y2 l2 0 e2
y3 l3 0 e3
2 W c d 1 Ge4W,
h1
Gy4W 5 G 0 # (7.5)
h2
y5 0 l5 e5
( ( ( (
yp 0 lp ep
where h1 and h2 are the latent variables of the shadow economy and
DIY activities. Equations (7.4) and (7.5) represent the structural and


measurement models. The estimation of the parameter b12 describes the

effect of h2 (DIY activities) on h1 (the shadow economy) and, thus, reveals
the exact – substitutive or complementary – relationship between DIY
activities and the shadow economy.15
As explained above, it is not clear whether shadow economic and DIY
activities are complements or substitutes. Hence, we do not formulate
a concrete hypothesis about the interaction between these activities but
want to answer this question in our empirical analysis. With respect to
the causes and indicators of both activities we follow the literature and
our own theoretical considerations of Section 7.4 and formulate testable
hypotheses. These are:
(H1) An increase in tax and social security burdens increases shadow

economic and DIY activities, ceteris paribus.
(H2) The more the German economy is regulated, the greater the incen-
tive to work in the shadow economy, ceteris paribus.
(H3) The higher the per capita real disposable income, the higher
demand for goods and services in general and, thus, the larger the
shadow and DIY economies, ceteris paribus.
(H4) The higher unemployment in the official economy, the more
individuals engage in DIY activities, ceteris paribus.
(H5) Higher average gross hourly earnings for craftsmen lead to an
increase in the volume of DIY activities, ceteris paribus.
(H6) German reunification in 1990 offered remarkable opportunities in
the shadow and DIY economies. Hence, both economies grew as
a consequence of different behavioural patterns in East and West
Germany and structural changes due to reunification.
(H7) The more common DIY activities, the lower average hours worked
per week, ceteris paribus.
We do not formulate a hypothesis regarding the shadow economy’s effects

on the official economy because of the ambiguous guidance offered in the
previous theoretical and empirical literature.
As explained above, identification and estimation of an SEM model
requires imposing constraints on the model. A well-established way to
ensure identification of the model is to set the coefficient of one indicator
for each latent variable to a non-zero value.16 For the shadow economy,
we choose currency in circulation and set it to 1. This is common practice
in the literature because cash protects principal and agent in their shadow
economic activities by eliminating the ‘paper trail’. Thus, the higher the
shadow economy, the more currency circulates, ceteris paribus. Because
we are dealing with two latent variables simultaneously, it is also necessary


to fix the scale for the other unobserved variable, DIY activities. This was
done as follows.
According to the Federal Statistical Office of Germany capital produc-
tivity in construction business was 1.89 in 1991 (the approximate mid-
point of our observation period).17 As capital productivity measures the
ratio of output to capital input and we employ the capital input of DIY
activities, that is, turnover in DIY stores, as an indicator for the unob-
served variable DIY, that is, the output, the use of capital productivity as a
fixed (scaling) parameter might be appropriate. Assuming that the capital
productivity in construction business is nearly equal to that of DIY activi-
ties, we set the coefficient of the indicator variable turnover in DIY stores
to this level with a numerical value of 2.
Shadow economic and DIY activities: substitutes or complements?

According to our theoretical reasoning presented in Section 7.4, tax and
social security contribution burdens as well as the intensity of regula-
tion are the main elements to determine shadow economic activities.
DIY activities, however, are best explained through tax and social secu-
rity contributions, unemployment and average gross hourly earnings.
Furthermore, we use a zero-one dummy variable to control for structural
changes of the German economy as a result of German reunification in
1990 and per capita real disposable income as further control.
The result of the SEM estimations for the shadow economy and DIY
activities in Germany, applying the maximum likelihood estimator, is
shown in Figure 7.4. The goodness of fit statistics show satisfactory statis-
tical properties.
With regard to the shadow economy, all variables are statistically sig-
nificant at conventional significance levels. With regard to DIY activities,
we cannot confirm statistically significant influence of the tax and social
security contributions burden and the real disposable income on DIY
activities. In the following we summarize our findings from the estimation
of the presented models and address our proposed hypotheses:
(H11H2) The intensity of regulation and tax and social security con-
tribution burdens are statistically significant and positively
related to the shadow economy, having the expected positive
sign. We cannot confirm that the tax burden is a driving factor
for individuals to engage in DIY activities.
(H3) Per capita real disposable income is highly statistically sig-
nificant and positively related to the shadow economy. One
possible explanation is that the higher the disposable income
of households, the higher the demand for goods and services.


Regulation
14.86** Domestic currency

Real disposable in circulation
income
1.13*** 1.00 (fixed)
0.22 Shadow
economy 0.36***
Reunification 0.15** Growth rate of
dummy real GDP
0.03*** –2.97*
Average hours
0.14** worked per
Tax and social Do-it-yourself –0.12*
0.02 week
security burden economy
2.00 (fixed)
0.04**
Real turnover in
Unemployment –0.29* do-it-yourself
stores
Average gross hourly

earnings
Notes: *, **, and *** denote significance at the 10 **per cent, 5 **per cent, and 1 **per cent
significance level. The number of observations is 36. The estimated model has 37 degrees of
freedom which are determined by 0.5 (p 1 q) (p 1 q 1 1) – t; with p 5 number of indicators;
q 5 number of causes; t 5 the number for free parameters. The chi-square is 18.47 and the
corresponding p-value is 0.99. If the model fits the data perfectly and the parameter values
are known, the sample covariance matrix equals the covariance matrix implied by the model.
The null hypothesis of perfect fit corresponds to a p-value of 1. The root mean squared error
of approximation (RMSEA) measures the model’s fit based on the difference between the
estimated and the actual covariance matrix. RMSEA values smaller than 0.05 indicate a
good fit (Browne and Cudeck 1993). The estimated model’s RMSE is 0.00.
Figure 7.4 SEM for the shadow economy and DIY activities
Demand rises not only in the official but also, in part, in

the shadow economy, leading to a higher observed level of
shadow economic activity. This implies that both sectors are
complementary to each other. DIY activities, however, do not
respond to changes of the per capita disposable income.
(H4) With respect to unemployment we observe a statistically
significant positive coefficient to DIY activities confirming
our earlier hypothesis that a higher level of unemployment
increases these activities, ceteris paribus. It seems that unem-
ployed look for a meaningful activity by engaging into DIY
activities which may additionally enhance their self-esteem.18


(H5) Hourly earnings in the small trade sector do influence DIY

activities. The parameter estimate shows that higher hourly
earnings reduce DIY activities, ceteris paribus. A possible
explanation is the following: if craftsmen earn higher wages in
the official economy, they substitute spare time DIY activities
for the supply of services in the official economy. Moreover,
higher wages lead to higher income, all other things being
equal. This may increase the demand for craftsmen’s services
in the official economy and thereby decrease their DIY activi-
ties. Consequently, this reduces the size of the DIY economy.
(H6) The dummy variable is, as expected, significantly positively
related to both latent variables. This result reflects dif-
ferent behavioural patterns in East and West Germany
confirming Brodersen’s results that the likelihood of carry-
ing out DIY activities is higher in the new federal states
(Neue Bundesländer) than in the old federal states (Alte
Bundesländer). The significant coefficients of the dummy vari-
able also indicate the catching-up process of the Eastern to
the Western part of Germany after reunification, even in the
informal part of the German economy.
(H7) We can confirm that DIY activities affect the indicator vari-
able average hours worked per week, that is, all other things
being equal, increasing DIY activities reduce the number of
average hours worked per week. The more individuals engage
into DIY activities the less labour they supply in the official
economy.
The estimated coefficient on the growth rate of real GDP indicator vari-
able is statistically significant suggesting a positive relationship between
the shadow economy and growth of the official economy supporting the
views held, for example, by Asea (1996) and Tanzi (1999).
Our findings regarding both latent variables, the shadow economy and
DIY activities, confirm most of the previous findings of earlier theoreti-
cal and empirical research. For this reason, we consider interpreting the
estimated coefficient of the relationship between DIY activities and the
shadow economy. The SEM presented in Figure 7.4 shows a statistically
significant positive effect of DIY activities on the shadow economy. Thus,
for the case of Germany, we find empirical evidence for the hypothesis
that both informal economic activities share a substitutive relationship.
It seems that individuals substitute shadow economic activities with DIY
activities possibly because they fear the risk of detection and possible
punishment. They do not show a demand for shadow labour in addition


to their own DIY activities aimed to complement their own household

activities, thereby guaranteeing an efficient execution and assuring quality
standards. Such complementary behaviour would lead to a positive coef-
ficient in the SEM testing empirically the relationship between DIY activi-
ties and the shadow economy. On the contrary, our results suggest that
the opposite is true. Individuals who produce goods and services in home
production in the DIY economy do not demand complementary services
of the shadow economy.
7.6 SUMMARY AND CONCLUSIONS
In this chapter, we presented an SEM to analyze the relationship between

the shadow economy and DIY activities. Using the appropriate determi-
nants for both informal economic activities, we found a highly statistically
significant influence of regulation as well as tax and social security burdens
on the shadow economy. For DIY activities, we observed a highly statisti-
cally significant positive influence of unemployment and a negative one for
wages. Moreover, we found a regional catching-up effect for the shadow
economy and DIY activities in the sense that the German reunification
contributed significantly to both activities. While the shadow economy is
primarily driven by institutional factors like taxation and regulation, DIY
activities respond to supply-side variables like unemployment and wages.
Our results suggest that shadow economic activities are contingent on
governmental behaviour while DIY activities are driven by individual con-
straints. According to our calculations shadow economic and DIY activi-
ties have a negative relationship to each other, that is, they are substitutes,
at least in Germany.
What can we learn from this chapter and what type of policy conclu-
sions can be drawn from the results? First of all, the descriptive statistics
presented in Section 7.3 reveal that the shadow economies in developed,
transition and developing countries are quite different in size. In develop-
ing and transition countries, the shadow economy is, on average, almost
twice as large as in developed countries. While its enormous size in transi-
tion countries is a historical phenomenon, by and large, and often deter-
mined by institutional factors, it has a different reasoning in developing
countries: very limited possibilities in the local economy make it almost
impossible for citizens to earn their living in a legitimate manner. Working
in the shadow economy is often the only way of achieving a minimal
standard of living or subsistence. In developed countries, however, the
shadow economy is often an issue of the tax burden and regulation. A
similar pattern seems reasonable for DIY activities. Second, the SEM


presented in Section 7.5 shows that DIY activities in a developed country,

in particular, in Germany, are supply-side driven. They increase if unem-
ployment rises and if wages reduce. The choice between paid work, DIY,
and spare time suggests that DIY activities are associated with enjoyment
and self-satisfaction (see also Gronau, 1977). This is clearly demonstrated
in the questionnaire based surveys on DIY activities and home production
of Brodersen (2003). In developing countries however, DIY activities are a
central aspect of an economy’s economic and social dynamics.
If one wants to reduce the shadow economy and/or DIY activities fewer
regulations and lower tax and social security contribution burdens seem
to be efficient means of shifting more activity into the official economy.
While for shadow economic activities the result follows directly from
our findings, the argument for DIY activities runs as follows: Reducing
both the intensity of regulation and the amount of contributions to the
social security system typically results in lower unemployment and higher
net wages. Moreover, these measures reduce production costs for goods
and services in the official economy which have increased over the last
decade. Development of the economy towards higher service orientation
has increased prices and individuals could not afford as may goods and
services as they did. Consequently, DIY activities increased. Reducing the
intensity of regulation and tax and social security contribution burdens
would then reduce the costs of goods and services and, thus, individual’s
incentives to engage in DIY activities. This might be a desirable goal of
economic policy for the following reason: professional shadow workers
and/or workers employed in the official economy may produce goods and
services in a more efficient way than DIY workers usually do. Reducing
the DIY activities and stimulating the official economy would then not
only promote economic growth but also increase efficiency.
Though our results should be regarded as first steps, we have demon-
strated that both shadow and DIY activities serve as substitutes in a devel-
oped country. Whether this relationship holds true for a broader set of
(developed) countries and for the developing countries at all is, however, a
still unsolved question left for future research.
NOTES
1. Table 7.1 makes no claim to be a complete literature review on the issue of defining
informal economic activities. Excellent, comprehensive reviews on this matter are pre-
sented in Thomas (1992), Schneider and Enste (2000) and Gërxhani (2004).
2. The following terms are used as mere synonyms in the literature: black, concealed,
informal, non-observed, parallel, shadow, subterranean, underground or unrecorded
economy.


3. The literature about the ‘shadow’, ‘underground’, ‘informal’, ‘second’, ‘cash-’ or ‘paral-
lel’ economy is increasing. Various topics, on how to measure it, its causes and effects
on the official economy are analyzed. See for example, survey type publications by Frey
and Pommerehne (1984), Thomas (1992), Schneider (1994, 1997, 2005, 2007), Loayza
(1996), Pozo (1996), Lippert and Walker (1997), Johnson, Kaufmann, and Shleifer
(1997), Johnson, Kaufmann and Zoido-Lobatón (1998), Belev (2003), Pedersen (2003)
and Gërxhani (2004). For an overall survey of the global evidence of the size of the
shadow economy see Schneider and Enste (2000) and Alm, Martinez-Vazquez and
Schneider (2004), Kazemier (2005), Bajada and Schneider (2005), Enste and Schneider
(2006) and Feld and Schneider (2010).
4. It should be mentioned that Mainland China and Vietnam are still communist countries
with partly market economies, so that the figures of these two countries may be biased.
5. From the formal model one can not draw the conclusion which household member
should specialize in which activity. However, Becker (1981) argues that women have
biological commitment to the production and feeding of children and should thus spe-
cialize in household activities and invest in household human capital.
6. However, labour market models that study household labour supply and welfare taking
into account marketable domestic production and intra-household income sharing are
presented, for example, in Chiappori (1997) and Donni (2008).
7. The positive correlation between income and the likelihood of carrying out DIY activi-
ties in Norway and Sweden is surprising as one would expect that higher income means
a higher hourly wage, and thus a greater loss of hourly earnings for each hour spent
on DIY activities. That is, the higher the income the greater are the opportunity costs
of DIY activities, ceteris paribus and, thus, the greater incentives to substitute DIY
activities with the purchase of goods and services supplied in the official economy. Our
empirical results for Germany, presented in Section 7.5, support this view.
8. Schneider and Enste (2000) provide a comprehensive overview.
9. The results of Brodersen (2003) presented in Section 7.3.2 support this view.
10. Most recent empirical studies show that employees do not voluntarily accept a reduc-
tion in working hours. If governments and/or labour unions reduce working hours,
workers will use the unexpected extra leisure time for productive activities, for example,
to work in the shadow economy. See Schneider and Enste (2000) for a detailed
discussion.
11. The main components of public revenues are the income, value added and sales taxes,
social security contributions and payroll taxes.
12. The reason for the exclusion of these employees is that the postal service (Deutsche Post
AG) was state-run until 1995 and the railway service’s major share holder is still the
German government. Hence, at least until 1995, both companies did not operate under
market conditions.
13. q is a vector that contains the parameters of the model and S (q) is the covariance
matrix as a function of q implying that each element of the covariance matrix is a func-
tion of one or more model parameters.
14. An alternative is to set the variance of each unobservable variables to one. However,
setting one element of to an a priori value is often more convenient for economic inter-
pretation and thus typically done (Dell’Anno and Schneider, 2009).
15. We have also estimated the effect of h1 (the shadow economy) on h2 (DIY activities),
i.e., parameter b21. This link however turned out to be insignificant.
16. The choice of the indicator fixing the scale of the latent variable does not affect the
estimated coefficients (Bollen 1989).
17. For similar arguments, see also Karmann (1990).
18. We cannot confirm any significant correlation between unemployment and activities in
the shadow economy.


REFERENCES
Alexeev, M. and W. Pyle (2003), A note on measuring the unofficial economy in
the Former Soviet Republics. Economics of Transition, 11(1), 1–23.
Alm, J. (1991), A perspective on the experimental analysis of taxpayer reporting.
The Accounting Review, 66, 577–593.
Alm, J., J. Martinez-Vazquez and F. Schneider (2004), Sizing the problem of the
hard-to-tax. Working Paper. Georgia State University, USA.
Asea, P.K. (1996), The informal sector: baby or bath water? Carnegie-Rochester
Conference Series Public Policy, 45, 163–171.
Underground Economy: An International Perspective. Aldershot, UK: Ashgate
Publishing Company.
Banerjee, N. (1982), Survival of the poor. In H.I. Safa (ed.). Towards a Political
Economy of Urbanization in Third World Countries, Delhi: Oxford University
Press, pp. 175–183.
Becker, G.S. (1965), A theory of the allocation of time. Economic Journal, 75(299),
493–517.
Becker, G.S. (1981), The Treatise on the Family. Cambridge, MA: Harvard
University Press.
Belev, B. (2003), The Informal Economy in the EU Accession Countries: Size,
Scope, Trends and Challenges to the Process of EU Enlargement. Sofia: Center
for Study of Democracy.
Bollen, K.A. (1989), Structural Equations with Latent Variables. New York: Wiley.
Brodersen, S. (2003), Do-it-yourself work in north-western Europe. Maintenance
and improvement of homes. Study No. 11. The Rockwool Foundation Research
Unit, Copenhagen.
Browne, M.W. and R. Cudeck (1993), Alternative ways of assessing model fit. In
K.A. Bollen and J.S. Long (eds), Testing Structural Equation Models, Newbury
Park, CA: Sage.
Buehn, A., A. Karmann and F. Schneider (2009), Shadow economy and do-it-
yourself activities: the german case. Journal of Institutional and Theoretical
Economics, 165(4), 701–722.
Chiappori, P.-A. (1997), Introducing household production in collective models of
labor supply. Journal of Political Economy, 113, 1277–1306.
De Soto, H. (1989), The Other Path: The Invisible Revolution in the Third World.
New York: Harper and Row.
Dell’Anno, R. and F. Schneider (2009), A complex approach to estimate the
shadow economy: the structural equation modelling. In M. Faggini and T.
Lux (eds), Coping with the Complexity of Economics. Heidelberg: Springer, pp.
110–130.
Deregulation Commission (1991), Opening of markets and competition. Report
presented to the German Federal Government, Bonn.
Donni, O. (2008), Labor supply, home production, and welfare comparison.
Enste, D. and F. Schneider (2006), Wie groß ist die Schattenwirtschaft? Des Rätsels
Lösung. Wirtschaftsdienst–Zeitschrift für Wirtschaftspolitik, 86(2), 185–191.


Feige, E.L. (1981), The UK’s unobserved economy: a preliminary assessment,

Journal of Economic Affairs, 1, 205–212.
Distortion. Cambridge, UK: Cambridge University Press.
Feige, E.L. (1990), Defining and estimating underground and informal economies:
the new institutional economics approach. World Development, 18, 989–1002.
Feld, L. and F. Schneider (2010), Survey on the shadow economy and unde-
clared earnings in OECD countries. Invited Paper written for publication in
the German Economic Review, Department of Economics, University of Linz,
Linz, Austria.
Frey, B.S. (1989), How large (or small) should the underground economy be?
In E.L. Feige (ed.). The Underground Economy: Tax Evasion and Information
Distortion, Cambridge, UK: Cambridge University Press, pp. 111–129.
for measurement. Review of Income and Wealth, 30(1), 1–23.
Gërxhani, K. (2004), The informal sector in developed and less-developed
countries: a literature survey. Public Choice, 120(3–4), 267–300.
Gronau, R. (1977), Leisure, home production, and work: the theory of the
allocation of time revisited. Journal of Political Economy, 85(6), 1099–1123.
Harding, P. and R. Jenkins (1989), The Myth of the Hidden Economy: Towards
a New Understanding of Informal Economic Activity. Philadelphia: Open
University Press.
Hart, K. (1970), Small scale entrepreneurs in Ghana and development planning.
Journal of Development Studies, 6, 104–120.
HETUS (2007), Harmonized European time use survey, population and welfare
statistics. Statistics Sweden, Stockholm, Sweden.
International Labor Office (1972), Employment, Income and Equality: A Strategy
for Increasing Productivity in Kenya. Geneva: ILO.
sition. Brookings Papers on Economic Activity, 2, 159–221.
and the unofficial economy. The American Economic Review, 88(2), 387–392.
Karmann, A. (1990), Schattenwirtschaft und ihre Ursachen: Eine empirische
Analyse zur Schwarzwirtschaft und Selbstversorgung in der Bundesrepublik
Deutschland. Zeitschrift für Wirtschafts- und Sozialwissenschaften (ZWS),
110(3), 185–206.
Kaufmann, D. and A. Kaliberda (1996), Integrating the unofficial economy into
the dynamics of post socialist economies: a framework of analyses and evidence.
In B. Kaminski (ed.), Economic Transition in Russia and the New States of
Eurasia, London: M.E. Sharpe, pp. 81–120.
Kazemier, B. (2005), The underground economy: a survey of methods and esti-
mates. Discussion paper, Statistics Netherlands, Voorburg, Netherlands.
Kirchgässner, G. (1983), Size and development of West German shadow economy
1955–1980. Zeitschrift für die gesamte Staatswissenschaft, 139(2), 197–214.
erarbeiteten Sozialprodukts. Allgemeines Statistisches Archiv, 68(4), 378–405.
Klovland, J. (1984), Tax evasion and the demand for currency in Norway and


Sweden: is there a hidden relationship? Scandinavian Journal of Economics,

86(4), 423–39.
Lackó, M. (2000), Hidden economy—an unknown quantity: comparative analysis
of hidden economics in transition countries 1989–95. Economics of Transition,
8(1), 117–149.
Lippert, O. and M. Walker (1997), The Underground Economy: Global Evidences of
its Size and Impact. Vancouver: The Fraser Institute.
Scandinavia: a measurement based on questionnaire service. The Rockwool
Foundation Research Unit Study, No. 10.
Pozo, S. (ed.) (1996), Exploring the Underground Economy: Studies of Illegal
Research.
Renooy, P.H. (1990), The Informal Economy: Meaning, Measurement and Social
Significance. Amsterdam: Netherlands Geographical Studies.
Economics, 88(4), 643–668.
Schneider, F. (1994), Can the shadow economy be reduced through major tax
Schneider, F. (1997), The shadow economies of Western Europe. Economic
Affairs, 17(3), 42–48.
estimates for 145 countries. Economics: The Open-Access, Open-Assessment
E-Journal, 1, 2007–9.
quences. The Journal of Economic Literature, 38(1), 77–114.
Schneider, F., A. Buehn and C. Montenegro (2010), Shadow economies all over
the world: new estimates for 162 countries from 1999 to 2007. World Bank
Discussion Paper.
Swaminathan, M. (1991), Understanding the informal sector: a survey. WIDER
WP 95, Finland.
Tanzi, V. (ed.) (1982), The Underground Economy in the United States and Abroad.
Lexington DC: Heath.
Tanzi, V. (1986), The underground economy in the United States: reply to com-
ments by Feige, Thomas, and Zilberfarb. IMF-Staff Papers, 33, 799–811.
Thomas, J.J. (1992), Informal Economic Activity, LSE, Handbooks in Economics.
Harvester Wheatsheaf: London.


8 The shadow economy in the
residential construction sector
Christopher Bajada*
8.1 INTRODUCTION
Considerable international literature has been devoted to estimating the

size and investigating the consequences of the shadow economy world-
wide. Much of this work has primarily focused on developing method-
ologies to produce aggregate estimates of the shadow economy so as
to gauge its size and to study its behaviour and implications for public
policy. The literature provides little information on the shadow economy
at the broad industry level, for example, estimates of the shadow
economy in the services, accommodation, construction, retail trade and
primary production sectors of the economy. One reason for this is that
the methodologies used to produce the aggregate estimates of the shadow
economy cannot be easily applied at the sector level without having the
appropriate sector level data (for example, the proportion of cash used
in the building construction sector per unit time relative to the total
cash in circulation is not available for use with the ‘monetary method’).
Although these aggregate studies provide important information on the
trends in shadow economy activity, we are unable to determine from
them whether one or more of these broad industry sectors dominate or
drive the overall level of shadow economy activity. Such information is
very important for developing industry specific strategies which could
prove more effective in reducing the overall size of the shadow economy
than would a broad policy measure that applies across all industries.
Despite this, a micro analysis of each sector is undertaken at a country
level by each country’s own tax department, as they typically assess on an
annual basis the risks of non-compliance from the various sectors of the
economy. On the basis of these ‘internal’ estimates, each tax department
would develop its annual compliance strategy to tackle various aspects
of shadow economy activity or to devote more resources in auditing
risky sectors of the economy. Unfortunately, such information is never
published and so we are unable to gauge from any of this information the
293


relative size and risks associated with the shadow economy in these broad
industry sectors.
There are a few studies, however, that have produced estimates (or
analysis) of the size (or characteristics) of the shadow economy in the
building construction industry in which we are most interested (for
example: for Australia: CETF, 1998, 2003; ABS, 2004; for Canada:
KPMG et al., 1997; O’Grady and Lampert, 1998; O’Grady, 1998, 2001;
Statistics Canada, 1994; AHBR, 2004; NSDF, 1997; Zanasi, 1996; for the
UK: DTI, 2002, 2003). The Ontario Constriction Secretariat estimates the
shadow economy in the building construction industry in Nova Scotia,
Canada, to be 26 per cent of the sector’s overall GDP (see O’Grady, 1998,
2001; AHBR, 2004). Statistics Canada, on the other hand, estimate the
size of the shadow economy in the building construction industry to be
10 per cent of the sector’s GDP, while the Atlantic Home Building and
Renovation Sector Council (AHBR, 2004) estimates the shadow economy
in Nova Scotia to be 25 per cent of the sector’s GDP, similar in size to the
estimates by the Ontario Constriction Secretariat.
For Australia, the Australian Taxation Office’s (ATO) Property,
Building and Construction Project provides information on the charac-
teristics of the shadow economy in the building construction industry but
not specifically any estimates of its size. The project involves a risk scoping
exercise based on high-level risks associated with unreported income,
incorrect claims of input tax credits and the inclusion of personal expense
claims along with legitimate business expenses. Whenever a business
reports income and expenses outside the (benchmark) norms, the ATO
sends out field officers to investigate: (1) why the operations of that busi-
ness are outside the norm; (2) the nature of the compliance concerns; and
(3) to capture intelligence that would be useful for investigating practices
on other construction sites (see CETF, 2003). These investigations do
not produce any publically published estimates of the shadow economy
in the building construction industry, but have led to a number of rec-
ommendations to help improve compliance in the sector more broadly.
The Australian Bureau of Statistics (2004) embarked on an exercise to
evaluate the accuracy and reliability of the national accounts, in which
they evaluated the maximum likelihood of under-reporting in each of the
broad industry sectors (including construction). This study however did
not directly produce an estimate of the shadow economy in the building
construction industry in Australia.
The objective of this chapter is to produce the first known estimates of
the shadow economy in the residential construction sector. The estimates
are derived from a survey of builders who primarily work in Sydney and
in the surrounding metropolitan area conducted in December 2007 and


The shadow economy in the residential construction sector 295
February 2008. The study of the shadow economy in the building indus-
try is important for a number of reasons. First, the construction sector
employs a considerably large number of people (approximately 990 000
individuals or 9 per cent of all individuals employed nationally) and is
a major driver of economic activity in Australia. The Sydney residential
construction sector for new dwellings represents approximately 50 per cent
of the total value of new houses constructed across the state of NSW and
approximately 8 per cent across Australia and is therefore a substantial
component of the overall building industry in Australia.
Second, the shadow economy creates unfair price competition, whether
it is in construction or elsewhere. The effects on competition may be sig-
nificant, particularly in an environment where it may be ‘culturally accept-
able’ to obtain goods and services from the shadow economy. Take, for
example, a business that sub-contracts labour from the shadow economy.
In doing so this business is able to avoid many of the regulatory costs
involved in the employment of these individuals although it may bear a
risk should anything happen to these people at the workplace. The firm is
then able to substantially reduce its costs and therefore its prices in order to
effectively (and successfully) compete with legitimate business operators.
Third, the shadow economy compromises consumer protection rights.
The consumers that purchase goods and services from the shadow
economy must accept minimal guarantees on the quality of these products.
These consumers may regard the risk worthwhile if they believe that they
have saved a substantial amount of money from purchasing these goods in
the shadow economy. Unfortunately, there are many goods and services
that are provided under strict health and safety guidelines set by the law,
that are designed to ensure the consumer receives adequate protection.
Most certainly, many of the goods and services available in the shadow
economy would not conform in the same way to these guidelines. The con-
sequences in the construction industry may be longer term whereby, say,
any structural deficiency in a building or renovation may not be identified
until many years later, possibly after the builder has long left the scene
leaving the consumer substantially out-of-pocket.
The remainder of this chapter is organised as follows. In Section 8.2, we
provide an overview of the construction sector in Australia but with an
emphasis placed on the Sydney component of the residential construction
sector. In Section 8.3, we provide an outline of the survey, the methodol-
ogy used and the profile of the builders who participated in the survey. In
Section 8.4, we provide the first known estimates of the shadow economy
in the Sydney residential building construction sector and its characteris-
tics. In Section 8.5, we compare the results for Australia and Canada and,
in Section 8.6, we conclude.


8.2 AN OVERVIEW OF THE CONSTRUCTION

SECTOR
So as to set the scene for estimating the size and evaluating the characteris-
tics of the shadow economy in the residential construction industry (RCI),
we begin with a brief overview of the construction activities and employ-
ment profile in the building industry. This information will complement
the data from the builders’ survey (see Section 8.5) which will be used to
estimate the size of the shadow economy in the Sydney RCI.
The data presented in this section provide some key indicators of the
likelihood that there exists a shadow economy in the RCI. The key indi-
cators include: (1) the size and mix of residential building construction
activity; and (2) the employment profile of people working in this sector.
The emphasis on point (1) will be on the mix between renovations and
repairs versus new dwelling construction, as the former is more likely to
involve cash and contribute to the overall size of the shadow economy.
The emphasis on point (2) will be on the nature and profile of employment
(for example, employees versus self-employed) and earning differentials
between those working in the RCI compared to the earnings of those
working in the other sectors of the economy.
8.2.1 The Size and Mix of Building Construction Activity
In Figure 8.1 we plot the total value of new house construction and
renovations and repairs as a percentage of the total value of residential
buildings under construction in NSW. Figure 8.1 identifies an inter-
esting trend in the mix of construction activity since the mid-1980s.
The value of new house construction has declined substantially, from
approximately 55 per cent in 1985 to approximately 35 per cent by
2008, while at the same time the value of renovations and repairs has
increased, but only marginally, from approximately 12 per cent in 1985
to approximately 16 per cent by 2008. The combination of these trends
has resulted in a significant narrowing of the gap between the value of
new dwelling construction and renovations and repairs. As activities in
renovations and repairs are more susceptible to influences from shadow
economy operators, this would suggest that the risks associated with an
increase in the size of the shadow economy has become much greater
over time and so, when estimating the size of the shadow economy, it
is important to take into account the changing mix that is observed in
Figure 8.1.
In Table 8.1 we provide a decomposition of the RCI into construc-
tion of new dwellings and the market for renovations and repairs for


70
60
50
40
%
30
20
10
0
Sep-84
Aug-85
Jul-86
Jun-87
May-88
Apr-89
Mar-90
Feb-91
Jan-92
Dec-92
Nov-93
Oct-94
Sep-95
Aug-96
Jul-97
Jun-98
May-99
Apr-00
Mar-01
Feb-02
Jan-03
Dec-03
Nov-04
Oct-05
Sep-06
Aug-07
Value: residential building: new houses
Value: residential building: alts and additions
Figure 8.1 New houses and alterations/additions – NSW (as a percentage

of the total value of residential buildings under construction)
Table 8.1 Sydney and NSW residential construction market
New New Total Value Value of Value of Value of Value

houses other dwel- of new new altera- total of
residen- lings houses other tions and residen- total
tial residen- additions tial building
building tial to building
building residential
building
NSW (% 14.4 29.1 19.2 15.7 26.5 29.2 20.5 22.9
of
Aust)
Sydney 42.3 75.4 58.7 49.9 80.0 73.6 65.3 69.9
(% of
NSW)
Sydney 6.1 22.0 11.3 7.8 21.2 21.5 13.4 16.0
(% of
Aust)


the period 2007–08. The RCI in Sydney makes up approximately 59

per cent of the total number of new dwellings in the state of NSW, of
which NSW makes up approximately 19 per cent of the total number
of new dwellings in Australia. As a proportion of all new housing con-
struction across Australia, the Sydney RCI comprises 11.3 per cent of
all new house constructions. In terms of the dollar value of new housing
constructions, the Sydney RCI represents approximately 50 per cent
of the total value of new houses constructed across NSW and approxi-
mately 8 per cent across Australia. For renovations and additions,
approximately 74 per cent of all such activities in NSW occur within the
Sydney RCI. This is equivalent to 21 per cent of all such construction
activities in Australia. These results imply that any significant findings
of illicit economy activity in the Sydney RCI are themselves a significant
part of overall illicit activities in the sector across Australia. This would
be particularly true if the incidence of shadow economy activity in the
RCI in other states is at least equivalent to the activities in the Sydney
market.
The results in Figure 8.1 and Table 8.1 are extremely important in
the analysis of the shadow economy as it is quite likely that the shadow
economy is more prevalent in renovations and repairs than it is in new
dwelling construction. Primarily this is because the cost associated with
the construction of a new house is substantially larger and in most cases
will involve a loan from which progress payments on the construction of
the house are made directly by the banks to the builder. The fact that in
most cases the construction of these houses is undertaken by large com-
panies where cash negotiations are likely to be limited is another reason
to expect a relatively smaller shadow economy from this component of
the RCI. Nevertheless, it is also quite possible that a new house may be
completed without much of the interior finishes (carpets, tiling, painting
and so forth). The homeowner may engage directly with the appropriate
tradesperson to save themselves the builder’s commission and benefit from
a cash discount that is offered by the tradesperson. Under such circum-
stances there are significant opportunities for shadow economy activity in
the area of new constructions. Despite these opportunities, however, such
transactions might only constitute a relatively small part of the overall
construction costs of a new dwelling for reasons we have previously indi-
cated. This contrasts with shadow economy opportunities that would be
available across the range of smaller jobs in renovations and repairs which
typically involve small to medium size builders who may have a greater
capacity and willingness to deal in cash.
Table 8.2 provides disaggregated information on the value of new dwell-
ing construction and renovations and repairs within the Sydney region.


Table 8.2 Value of approvals of new houses, other residential buildings

and alterations and additions – 2007/08
Section A percentage of each Section B percentage of each

region across all categories category across all regions
Value Value of Value of Total Value Value Value of Value of
of new altera- of of new altera- total resi-
new other tions and new other tions and dential
houses resi- additions houses resi- additions build-
dential to dential to ing
build- residential build- residential
ing building ing building
Inner 5.2 76.1 18.8 100 1.7 18.7 9.6 11.0
Sydney
East 29.8 27.1 43.0 100 9.1 6.1 20.2 10.1
South 54.2 31.5 14.3 100 30.2 13.0 12.3 18.6
West 34.1 54.7 11.2 100 25.1 29.8 12.7 24.5
North 31.8 40.8 27.5 100 34.0 32.4 45.3 35.7
Total 33.4 45.0 21.6 100 100.0 100.0 100.0 100.0
Source: 87310DO003_200911 Building Approvals, Australia, Nov 2009
In Section A of Table 8.2, we consider each of the values as a percentage

of all residential buildings across each region and in Section B of Table
8.2, the percentage of each category’s value across all regions. These data
provide important disaggregated regional information on the total value
of residential construction which, when combined with similar data on the
incidence of shadow economy activity (obtained from the survey data),
provides a more reliable measure of the size of the shadow economy in the
RCI.
In Section A of Table 8.2, the value of new residential construc-
tion across Sydney is approximately 78 per cent of all residential con-
struction while renovations and repairs is approximately 22 per cent.
Approximately 43 per cent of residential construction across eastern
Sydney is in renovations and repairs, while in the other areas it aver-
ages 18 per cent. In Section B of Table 8.2, the value of each category
of construction is compared across regions. For example, the value of
renovations and repairs in eastern Sydney represents 20 per cent of all
renovations and repairs while in the northern sector of Sydney the per-
centage is much larger (45.3 per cent). Inner Sydney, and the south and
west of Sydney represent 9.6 per cent, 12.3 per cent and 12.7 per cent of
all renovations and repairs, respectively.


9.5
8.5
8
%
7.5
6.5
6
Dec-84
Dec-85
Dec-86
Dec-87
Dec-88
Dec-89
Dec-90
Dec-91
Dec-92
Dec-93
Dec-94
Dec-95
Dec-96
Dec-97
Dec-98
Dec-99
Dec-00
Dec-01
Dec-02
Dec-03
Dec-04
Dec-05
Dec-06
Dec-07
Figure 8.2 Percentage of persons employed in the construction sector
– Australia
8.2.2 Employment in the Residential Construction Sector
In Figure 8.2, we plot the number of individuals employed in the construc-

tion sector as a percentage of individuals employed across all sectors of the
economy. From 1985 until 1997, the percentage of individuals employed
in the construction sector rose from approximately 7 per cent in 1985 to
over 9 per cent by 2007.
There has been over this time an increase of approximately 60 per cent
in the number of people employed across all industries, with the biggest
increases in accommodation, café and restaurants; property and business
services; cultural and recreational services and construction. Interestingly,
a number of these industries have been targeted by the Australian
Taxation Office (ATO) because they are predominantly labour-orientated
and susceptible to significant shadow economy activity. The upward trend
which is evident in Figure 8.2 also coincides with a marginal increase in
the total value of renovations and repairs in the RCI (see Figure 8.1 for
the NSW example). Taking into account non-residential construction as
potentially explaining the increases in employment, we find (not shown
here) that total non-residential construction as a percentage of the total
value of building activity has since 1985 declined from just over 60 per cent
to less than 50 per cent in 2007. On the other hand, residential construction


Table 8.3 Persons employed in the construction industry
2004 2005 2006 2007

’000 ’000 ’000 ’000
General Construction
Building construction 207.1 206.5 209 213.4
Non-building construction 43.4 57.1 48.1 73.6
General construction nfd 0.9 1 0.7 0.5
Total 251.4 264.6 257.8 287.5
Construction Trade Services
Site preparation services 40.3 47.8 57 54.7
Building structure services 93.2 91.9 86.8 82.2
Installation trade services 162.9 171.3 180.1 212.9
Building completion services 158.1 182.4 191.5 181.3
Other construction services 78.5 74.4 94.3 87.3
Construction trade services nfd 6 4.2 3.5 6.5
Total 539 572 613.2 624.9
Construction - nfd 11.3 18.2 18.1 24.9
Total Construction 801.7 854.7 889.1 937.3
Total All Industries 9 639.4 9 947.9 10 143.7 10 451.2
Notes: All data presented are for May Quarter; nfd – not further defined.
Source: Labour Force, Australia, Detailed, Quarterly, May 2007, cat. no. 6291.0.55.003
increased from 40 per cent to just over 50 per cent by 2007, contributing
somewhat to the increasing trend observed in Figure 8.2.
To shed light on this we consider the data in Table 8.3, in which we find
that the 16.9 per cent growth in employment over the period 2004 to 2007
has been contributed predominantly by employment which is directly
related to construction. Only 3 per cent of the growth in employment in the
construction sector was attributed to non-building construction employ-
ment. Interestingly, the growth in the total number of hours worked over
the period 2004 to 2007 was only 1.75 per cent (constructed from: Labour
Force, Australia, Detailed, Quarterly, May 2007, cat. no. 6291.0.55.003).
When we compare wage growth over this time period, earnings in the con-
struction sector grew faster compared with average earnings from all other
sectors. In 2004, earning in the construction sector represented 1.13 times
the average earnings across all other industries, while in 2007, it repre-
sented 1.23 times (constructed from: Average Weekly Earnings, Australia,
May 2007, cat. no. 6302.0). The combination of this data may indicate
that the shadow economy in the construction sector is providing some
cushion against the increase in competition for jobs across this sector.


Table 8.4 Status of employment in the building construction industry
Type of Employee Employer Own Contribu- All

Employment account ting family employees
worker worker
Construction (’000) 676 46.6 210.6 4.1 937.3
% of total 72.19 5.0 22.5 0.4 100
employment
All industries (’000) 9 230.9 273.2 917.9 29.1 10 451.2
% of total 88.3 2.6 8.8 0.3 100.0
employment
Source: Labour Force, Australia, Detailed, Quarterly, May 2007, cat. no. 6291.0.55.003
A key indicator of a sector’s risk that it may support a sizable shadow

economy is in the proportion of its workforce that is self-employed (for
example see Pissarides and Weber, 1989; and Mirus and Smith, 1997). In
Table 8.4, we report the various categories of employment in the building
construction industry. These include, employee, employer, own account
worker and contributing family worker categories.
There is a greater reliance on the self-employed to accurately record
and report their earnings. Employers have an obligation to report earn-
ings and withhold income tax on behalf of their employees, which they
remit on their employees’ behalf directly to the tax authorities. An
individual who is self-employed records their own transactions (whether
in one or more books) and remits to the government their tax obliga-
tions bounded by the level of income the person chooses to report. It is
important to realise, however, that it is also equally possible to be an
employee and to be part-paid in a collusive agreement with the employer
where this additional income is not reported to the tax authorities by
either party.
The percentage of individuals that are employees in the construction
sector is 72.2 per cent while in the other sectors of the economy the average
is 88.3 per cent. The proportion of individuals working on their ‘own
account’ represent 22.5 per cent in the construction sector compared with
only 8.8 per cent across all other industries. The significant difference in
the proportion of self-employed individuals is likely to increase the chance
of underreporting of income and thus the likelihood of a significantly
larger shadow economy. Although we have suggested that the proportion
of self-employed in any given sector is a likely indicator of the risk associ-
ated with the existence of a significant shadow economy, it may not always
be the case that those sectors with the highest proportion of self-employed


individuals have the largest shadow economy. This is a matter resolvable

only by empirical investigation.
8.3 DESCRIPTION OF THE SURVEY
Survey instruments for estimating the shadow economy have often been
employed as an alternative to the indirect methods of estimation that
include, for example, the monetary method, the MIMIC approach, physi-
cal input method, the RESET approach and so forth. The principle idea
underlying the majority of these surveys is that such interviews provide
a great deal of information that either would not be available from pub-
lished data or not sufficiently disaggregated to assist with estimating or
understanding a specific component of the shadow economy. In almost
all cases the interviewer assures the prospective respondent that the infor-
mation they provide will not be disclosed and that their identity will be
removed from the survey record to guarantee their anonymity. As one
might expect, because of the nature of the topic, most respondents are
likely to deny being involved in the shadow economy because of fear of
being detected and punished, despite the promises that might be made by
the interviewer. It might also be the case that some individuals claim to be
participating in the shadow economy when, in fact, they may not be. They
may choose to do so simply to sensationalise the discussion with the inter-
viewer. Just for these reasons alone, the results from such surveys are likely
to downward bias the true extent of participation in the shadow economy
and so it is also possible that the results published in this chapter are also
susceptible to the same biases. Nevertheless every effort has been made to
ensure that the extent of the potential bias is as small as possible.
The survey questions were framed in a way that the respondent was not
asked to provide their own personal experiences but rather their percep-
tions of what they observed in the course of their activities in the building
industry. By framing the questions in this particular way, we limit the
likelihood that an individual will provide incorrect information if they
themselves are participating in the shadow economy. By potentially con-
veying their own experiences in the third person, the extent to which the
respondents may underestimate the extent of the shadow economy may
be significantly reduced. By taking this approach we did observe that the
majority of the builders were quite willing to engage in detailed discussion
on the nature and methods by which individual builders operate in the
shadow economy. In fact, for a survey of this kind, the response rate was
quite high (see below).
The survey instrument used in this chapter combined the author’s own


questions and those used by the Atlantic Home Building and Renovation
Sector Council.1 The use of common questions allows a direct comparison
on how the builders in the two countries respond to various issues and
allows for direct comparisons of the size of the shadow economy. A com-
parison of the survey responses and the size of the shadow economy in the
residential construction sector in each of the two countries is presented in
Section 8.5 of this chapter.
8.3.1 Target Population and Methodology
The target population for this survey was professional builders involved
in the construction of new dwellings and those involved in renovations
and repairs in the residential construction sector. The sample of builders
for this survey was drawn from the list of names and contact telephone
numbers publicly available on the Master Builders Association website
and from advertisements of builders in the Sydney local area newspapers
(also published online). The survey was conducted by telephone and each
builder was invited to participate in the survey. If any builder was unable
to answer the questions at the time when they were initially contacted, they
were asked if there was a suitable time for them to be contacted. In the few
instances when this occurred, the builder requested that they be contacted
in the evening and with the exception of one builder, they all eventually
participated in the survey. Each telephone interview lasted between 25 and
30 minutes. Given the nature of the survey questions and the duration
time of the survey (of which each respondent was advised in advance), the
response rate (see below) was higher than initially expected and above the
usual response rates observed for similar surveys conducted elsewhere.
The survey was conducted over a two month period: the first during the
early part of December 2007 and again during February 2008, allowing
time for the builders to return from their holidays over the January period.
8.3.2 Characteristics of Respondents
A total of 112 builders participated in the survey which represents an

overall response rate of 58.9 per cent. Of those builders that responded,
92.8 per cent belonged to a professional association. Of all those build-
ers belonging to a professional association in the target population, 74.3
per cent agreed to participate in the survey. Of all the builders who did
not belong to a professional association, only 16 per cent agreed to par-
ticipate in the survey. The non-members, all of whom were sourced from
online advertisements in the Sydney local area newspapers, were gener-
ally more reluctant to participate in the survey. On the other hand, those


Table 8.5 Membership of professional association
Members Non-members Total

Sample size 140 50 190
Respondents 104 8 112
Response rate 74.3% 16% 58.9%
Table 8.6 Survey results: type of residential construction and building

membership
Type of Residential Survey Number Percentage Membership

Construction number and (%)
New construction (Majority) 19 17% 16, 84.2%
Alternations and additions 93 83% 88, 94.6%
(Majority)
New constructions and 48 42.9% 44, 91.7%
alterations and additions
combined
respondents belonging to a professional association were more willing to

engage in such discussions. In Table 8.5, we report the survey response
rates for members and non-members alike.
In Table 8.6, we disaggregate the results in Table 8.5 according to
the mix of activities in the RCI. Of all the respondents, 17 per cent were
engaged, for the majority of the time, in the construction of new dwellings
while the remaining 83 per cent were engaged in renovations and repairs.
A total of 48 (or 42.9 per cent) of respondents reported to be involved in
both the construction of new dwellings and the renovations and repairs
sector, of which 44 (or 91.7 per cent) of these respondents were members
of at least one professional building association.
The regional distribution of residential construction activity from the
builders’ survey is given in Table 8.7. According to the survey results,
much of the activities in new dwelling construction took place in the north
and west of Sydney. On the other hand, much of the activities in renova-
tions and repairs took place in the north and east of Sydney. The survey
results provide a good representation of the actual regional distribution
of building activity across the Sydney RCI (see the actual regional distri-
bution of housing approvals given in Section B of Table 8.2). Therefore,
inferences drawn from this survey should provide a reasonable indicator of
the (shadow economy) activities across the residential construction sector.


Table 8.7 Builders’ survey: residential construction by region
New Construction
North East South West Inner west Total
Total 17 11 11 15 5 59
% of total 28.8 18.6 18.6 25.4 8.5 100
Alterations and Renovations
Total 83 49 35 19 42 228
% of total 36.4 21.5 15.4 8.3 18.4 100
Residential Construction
Overall Total 100 60 46 34 47 287
% of total 34.8 20.9 16.0 11.8 16.4 100
In the following section, we estimate the size of the shadow economy in

the RCI, the perceptions of the existing strategies to address the shadow
economy and suggestions to reduce the extent of illicit activities in the
sector as proposed by a number of respondents.
8.4 ESTIMATES OF THE SHADOW ECONOMY IN

THE RESIDENTIAL CONSTRUCTION SECTOR
Each of the respondents was asked a series of questions that were grouped
into various categories to help understand a range of issues in relation to
the shadow economy. The various questions in the survey were grouped
according to the following broad classifications: (1) perceptions of the
characteristics and consequences of the shadow economy; (2) perceptions
of the behaviour of those working in the building industry; and (3) per-
ceptions of the size of the shadow economy. In the following sections, we
present each of these results.
8.4.1 Perceptions of the Characteristics and Consequences of the Shadow

Economy
In Figure 8.3, we provide a summary on how each of the respond-

ents reacted to a series of statements on the shadow economy and its


90.3
100.0
89.3
84.2
90.0
73.7
72.3
72.0
80.0
70.0
57.9
53.6
52.7
60.0
47.3
47.4
45.2
45.5
43.8
%
50.0
39.8
39.3
36.6
36.8
36.8
36.8
34.4
40.0
30.0
20.0
10.0
0.0
Q1 Q2 Q3 Q4 Q5 Q6 Q7
New construction sector Renovations sector Total
Notes:
Relevant questions:
Do you ‘agree’ or ‘disagree’ with the following statements:
Q1. Cash operators in the building industry create unfair price competition.
Q2. Cash operators force legitimate businesses to lower their prices
Q3. The shadow economy helps to recruit new workers in the building industry
Q4. Competition from cash operators lowers wages in the building industry
Q5. The shadow economy gives the whole industry a bad image for people making career
choices
Q6. Competition from the shadow economy makes it less attractive to recruit and train
new workers like apprentices
Q7. The shadow economy is more evident in smaller value renovation or building jobs.
Figure 8.3 Respondent’s views on the impact of the shadow economy in

the residential construction sector (% agreeing to each of the 7
statements)
implications for the building industry. This figure presents some interest-
ing results. When asked to respond whether ‘cash operators in the build-
ing industry create unfair price competition’, 72.3 per cent of builders
agreed with the statement. This is reflected in the response rate (56.3 per
cent) to the statement that ‘cash operators force legitimate businesses to
lower their prices’. Such pressures can have a significant implication on
the behaviour of legitimate operators. Given that the shadow economy
operators avoid having to meet the costs of licensing and other regulatory
obligations imposed on legitimate operators, the extent to which shadow
economy operators can lower their prices comes at a significant detriment
to the legitimate operator, who is either faced with the prospect of being
forced out of business or unwillingly made a participant in the shadow
economy in order to survive.


There was general consensus that ‘the shadow economy is more evident
in smaller value renovations or building jobs’ (approximately 90 per cent
of respondents agreed with this statement) supporting our earlier state-
ment that renovations and repairs are likely to attract a greater level of
shadow economy activity. Interestingly, however, it would appear that
the shadow economy is not seen as tarnishing the image of the sector. In
fact, when asked whether ‘the shadow economy gives the whole industry
a bad image for people making career choices’, only 45 per cent of build-
ers agreed. The perceptions that the shadow economy is a lure for those
considering working in the industry was relatively low (approximately 65
per cent did not believe it was an incentive).
8.4.2 Perceptions on the Behaviour of those Working in the Building

Industry
Each builder was asked a series of questions on the profile of those working
in the building industry, in particular on their use of cash, their qualification
and their honesty. In Figure 8.4 we provide a summary on how each of the
builders responded to various statements characterising those working in the
building industry. When asked whether ‘contractors and sub-contractors in
general supply services for cash’, 17.2 per cent of respondents agreed, while
approximately 21 per cent of respondents agreed that ‘contractors and
sub-contractors in general under invoice work and agree to accept part
payments in cash’. What is of most concern is that approximately one-third
of respondents reported that businesses in general do not pay the correct
amount of workers’ compensation to cover their ‘casual’ employees for
injuries that might be sustained while working on a construction site.
When asked whether ‘building material suppliers supply materials for
cash’, only 5.7 per cent of respondents agreed. This result is not surprising
given that the purchase of material is much more difficult to conceal since
the introduction of the Australian Business Number (ABN), which must
be quoted on purchase orders. In addition, without reporting the sale of
these materials the seller forfeits their input tax credit entitlement which is
intended to offset the GST tax paid on input supplies.
Builders were also asked whether they believed there are individuals
working in the building industry that are in receipt of unemployment
benefits, are unqualified but work in the industry as a second job or are
unqualified but pass themselves off as being qualified. In Table 8.8, we
present these results, which suggest that there are (although the extent
of which is not known) a number of individuals working in the build-
ing industry that are not qualified, and are therefore uninsured, and
defrauding the social security system.


32.3
35.0
31.6
28.5
30.0
26.8
25.0
20.7
19.3
18.4
20.0 17.2
17.0
%
15.0
10.4
10.0
5.7
4.7
5.0
0.0
Q1 Q2 Q3 Q4
Notes:
Relevant questions:
What percentage of the time do you think:
Q1. Contractors and sub-contractors in general in the building industry supply services
for cash in the building industry?
Q2. Contractors and sub-contractors in general in the building industry under invoice
work and agree to accept part payments in cash?
Q3. Building material suppliers supply material for cash?
Q4. Businesses in general in the building industry do not pay the correct amount of
workers’ compensation?
Figure 8.4 Respondent’s views on illicit activities in the residential

construction sector (% agreeing to each of the 4 statements)
Table 8.8 Perceptions of fraudulent behaviour in the construction industry
Yes No
Do you think there are people who work in the building 73.2% 26.8%
industry while also receiving unemployment benefits?
Do you think there are people working in the building 83.9% 16.1%
industry who are unqualified but work in the industry
as a second job?
Do you think there are people working in the building 94.6% 5.4%
industry who are unqualified but attempt to pass
themselves off as qualified?


Table 8.9 Percentage of total business activity in the shadow economy by

region
Respondent’s Majority Respondent’s Majority Activity:

Activity: New Homes Renovations & Repairs
Inner Sydney 25.0% 12.9%
East 11.1% 13.3%
South 18.8% 13.2%
West 10.9% 13.2%
North 14.1% 14.0%
8.4.3 Perceptions on the Size of the Shadow Economy
Each respondent was asked to provide their own estimate of the percent-
age of total business activity in the building industry that might be char-
acterised as in the shadow economy. The results, sorted by region and by
those working predominantly in the construction of new dwellings and
renovations and repairs, are given in Table 8.9. The results in Table 8.9
take into account each respondent’s perception of the size of the shadow
economy in each of the regions they reported to work in. If, for example,
a respondent stated they worked in all regions, their estimate was included
as a data point in each of those regions. In column 2 of Table 8.9, we
report the percentage of building activity believed to be in the shadow
economy by those builders predominantly engaged in the construction
of new dwellings. With the exception of Inner Sydney, the percentage
of building activity in the shadow economy varied between 11 per cent
and 19 per cent, with an overall average of 16.6 per cent, when taking the
aggregate estimates provided by the respondents across all regions. Of
those builders who indicated that 90 per cent or more of their activities
was in the construction of new dwellings, they reported the percentage
of the total business activity that might be characterised as in the shadow
economy at 13.8 per cent. The difference may be explained by the fact that
the majority of those builders primarily working in the construction of
new dwellings were also working in renovations and additions and so their
view may reflect a combination of what they observe in both areas.
In column 3 of Table 8.9 we report the percentage of building activity
believed to be in the shadow economy by those builders predominantly
engaged in renovations and repairs. The figures in column 3 of Table 8.9
are comparatively similar in size across the five regions with an average of
14 per cent when taking the aggregate estimates provided by the respond-
ents across all regions.


Table 8.10 Estimates of the shadow economy in the residential

construction industry
Shadow Economy 2006–07 2007–08 2008–09

(% of Sydney RCI output) 13.8%–16.0% 13.8 %–16.0% 13.9%–15.9%
(% of Sydney Building Industry 7.5%–8.7% 6.4%–7.4% 6.8%–7.9%
output)*
Note: (*) calculated based on the assumption that the extent of tax evasion outside of
Sydney is similar to that of Sydney residential construction market.
In Table 8.10, we estimate the size of the shadow economy for the period
2006–07 to 2008–09. The estimates are derived based on the percentage
of total business activity reported in the survey for the period 2007–08
using: (1) the lower bound estimate provided by those respondents heavily
engaged in the construction of new dwellings (more than 90 per cent of
the time); (2) the corresponding upper bound estimates for the respond-
ents working predominantly in the construction of new dwellings (more
than 50 per cent of the time); and (3) the average estimate for renovations
and additions. The estimates for the periods (2006–07 and 2008–09), are
based on the assumption that the results of the survey are relatively stable
and that the results would likely be similar in the year prior to and after
the survey period. Estimates of the shadow economy worldwide typically
show little variation in the size of the shadow economy over a very short
time interval, thus the assumption which we make here is also likely to
hold true. The results in Table 8.10 suggest that the shadow economy in
the residential construction sector of Sydney is quite significant, averaging
approximately 14 per cent of Sydney RCI output or approximately 7 per
cent of the entire value of the building industry in Sydney.
In Table 8.11, we report how the respondents evaluated the impact
of the shadow economy on their level of income as a result of competi-
tion from the shadow economy. In column 2 of Table 8.11 we report the
proportion of builders who indicated that their income had declined as
a result of the shadow economy. Over 52 per cent of builders working
predominantly in the construction of new dwellings (column 2) reported
an average fall in their income by 11.2 per cent (column 4). This compares
with just over 46 per cent (column 2) of builders predominantly working
in renovations and repairs who reported an average fall in their income by
14.9 per cent (column 4).
Each respondent was asked to give their view on whether the shadow
economy in the building construction industry had increased, decreased


Table 8.11 Impact on income from shadow economy operators in the

building industry
Did your income fall as a result Reduction in total

of competition from cash income as a
operators in the building result of competi-
industry tion from cash
operators
Yes No
Respondent’s majority 52.6% 47.4% 11.2%
activity: new homes
Respondent’s majority 46.2% 53.8% 14.9%
activity: renovations
and additions
Table 8.12 Perceptions of how the shadow economy has changed over the
previous two years
Over the past two years have transactions in the

shadow economy:
Increased? Decreased? Unchanged? Don’t know
(% agree) (% agree) (% agree) (%)
Respondent’s majority 31.6 5.3 42.1 21.0
activity: new homes
Respondent’s majority 23.7 12.9 36.6 26.8
activity: renovations &
additions
All activities (new homes, 25.0 11.6 37.5 25.9
renovations & repairs)
or stayed the same in the previous two years. The results are presented in
Table 8.12. On average, 25 per cent of respondents believed that the shadow
economy in the building industry had increased, while approximately 38
per cent believed that it had remained unchanged. Only a small proportion
(approximately 12 per cent) of those surveyed believed that the shadow
economy had declined. Another 26 per cent were not sure whether it had
increased or decreased. The results in Table 8.10 suggest the possibility of a
slight decrease in the size of the shadow economy from the previous year. In
Table A8.1.1 of Appendix 8.1 to this chapter, we report the various reasons
given by each of the respondents for why they believed the shadow economy
either increased or decreased over the two years prior to the survey.


Table 8.13 The GST and its impact on the shadow economy
Has the GT reduced the shadow economy in

the residential construction sector?
Yes No
Respondent’s majority activity: 52.6% 47.4%
new homes
renovations and additions
All activities (new homes, 27.9% 72.1%
renovations and repairs)
Table 8.14 Offers and requests for cash (instances of consumers and
builders requesting cash in return for lower prices – %)
Builder Consumer
new homes
All activities (new homes, renovations 31.5% 68.2%
and repairs)
When asked specifically whether the introduction of the GST has

reduced the size of the shadow economy in the building industry, the
majority of respondents believed it had not. These results are reported in
Table 8.13. When asked the question on whether the GST has reduced
the size of the shadow economy, 47 per cent of builders working predomi-
nantly in the construction of new dwellings believe it had not, while on
the other hand, 76 per cent of those working in renovations and additions
believe it has not. In part, this may simply be due to the fact that the pro-
portion of the shadow economy in the area of new dwelling construction
is relatively small compared with the shadow economy in renovations
and repairs. This, however, requires further analysis, which is beyond the
scope of this chapter.
Each respondent was asked to indicate whether it is the builder or
the consumer that typically requests cash payments in return for lower
prices. The results are presented in Table 8.14. Whether the respondent
worked primarily in the construction of new dwellings or in renovations
and repairs, the results are relatively consistent across the two groups of


builders. The results suggest that, in all cash transactions occurring in the
building industry, the consumer on average would initiate 68 per cent of
those transactions with the intention of receiving a discount. On the other
hand, approximately 32 per cent of all cash transactions in the building
industry are initiated by the builders offering consumers discounts if they
settle in cash. The implication of such a result would suggest that there is a
significant likelihood that an education campaign that targets consumers
about the risk of the shadow economy could have a significant impact on
reducing such activities in the building industry. This is consistent with the
perceptions of builders that education campaigns can have a significant
impact on reducing the size of the shadow economy (see below).
8.4.4 Perceptions on Various Strategies to Reduce the Shadow Economy
Each respondent was asked a series of questions on the effectiveness of

various strategies for reducing the size of the shadow economy. Figure
8.5 provides a summary of these results. The strategy most commonly
regarded as effective in reducing the size of the shadow economy is
education, particularly about the risks of having work completed by
shadow economy operators. This was closely followed by expanded
surveillance and enforcement efforts by the ATO and encouraging insur-
ance companies to refuse insurance for building works for which there
are no receipts from builders. The strategies of engaging the WorkCover
Authority of NSW and the Office of Fair Trading were less popular
as effective strategies. Having the surveillance and enforcement levels
of Centrelink expanded was regarded effective by only 43 per cent of
respondents.
Each of the builders surveyed were offered an opportunity to provide
their own suggestions on what they regarded as effective strategies for
reducing the size of the shadow economy. The feedback produced a
myriad of suggestions including a very common one of tightening regula-
tion in the owner builders’ market. A selected list of these suggestions is
given in Table A8.1.2 in the appendix to this chapter.
8.5 COMPARING CANADA’S SHADOW ECONOMY

IN THE RESIDENTIAL CONSTRUCTION
SECTOR
The Atlantic Home Building and Renovation Sector Council (2004) esti-
mated the size of the shadow economy in Nova Scotia to be 25 per cent
of the sector’s overall GDP. The estimates of the shadow economy for


74.2
73.2
80.0
68.4
63.2
63.2
70.0
55.4
55.4
54.8
53.8
53.8
53.6
60.0
47.4
45.2
42.9
50.0
32.3
31.6
31.6
%
40.0
30.4
21.1
30.0
20.5
18.3
20.0
10.0
0.0
Q1 Q2 Q3 Q4 Q5 Q6 Q7
Notes:
Relevant questions:
How effective do you think each of the following changes might be in reducing the shadow
economy in the building industry? (Response: ‘effective’ or ‘not effective’)
Q1. Expand surveillance and enforcement efforts by the Office of Fair Trading.
Q2. Expand surveillance and enforcement efforts by the ATO.
Q3. Expand surveillance and enforcement efforts by the WorkCover Authority of NSW.
Q4. Expand surveillance and enforcement efforts by Centrelink.
Q5. Educate people about the legal risks of working in the shadow economy.
Q6. Educate consumers about the risks of having work completed by shadow economy
operators.
Q7. Encourage insurance companies to refuse insurance for building works for which
there are no receipts from builders.
Figure 8.5 Respondent’s view on strategies to reduce the shadow economy

in the residential construction sector (% agreeing to each of
the seven statements that the strategy is effective)
Sydney residential construction sector (approximately 14 per cent of the

sector’s GDP) is substantially lower than that estimated for Canada. To
get a sense why the results for Nova Scotia are so different from those of
Sydney, we turn to a number of survey questions to analyse the way in
which the builder’s in each of the two countries responded.
In Table 8.15, we report how the respondents in Canada evaluated
the impact of the shadow economy on their level of income as a result of
competition from the shadow economy. Column 2 of Table 8.15 provides
the proportion of builders who indicated that their income had declined
as a result of the shadow economy. Over 58 per cent of builders working
predominantly in the construction of new homes (column 2) reported an
average fall in their income by 22.8 per cent (column 4), which is double
that reported by builders in Sydney (see Table 8.11). This compares (for


Table 8.15 Impact on income from shadow economy operators in the

building industry – Canada
Did your income fall as a Reduction in total

result of competition from income as a result
cash operators in the of competition
building industry from cash
operators
Yes No
New homes 58.2% 41.8% 22.8%
Renovations and repairs 46.5% 53.5% 25.0%
All activities (new homes, 52.5% 47.5% 23.8%
Source: Author’s calculations based on data from AHBR (2004)
Table 8.16 Perceptions of how the shadow economy has changed over the
previous two years – Canada
Over the past two years have transactions in

the shadow economy:
Increased? Decreased? Unchanged?
(% agree) (% agree) (% agree)
Respondent’s majority activity: 28% 20% 52%
new homes
Respondent’s majority activity: 44% 10% 46%
All activities (new homes, 35% 16% 49%
Source: Author’s calculations based on data from AHBR (2004)
Canada) with just over 46 per cent (column 2) of builders predominantly

working in renovations and repairs who reported an average fall in their
income by 25 per cent (column 4), which is 10 per cent more than that
reported by builders in Sydney (see Table 8.11).
In the Canadian survey each respondent was asked to give their
view on whether the shadow economy in the building construction
industry had increased, decreased or stayed the same in the previous
two years. The results are presented in Table 8.16. On average, 35 per
cent of respondents believed that the shadow economy in the building


industry had increased, while approximately 49 per cent believed that it

had remained unchanged. Only a small proportion (approximately 16 per
cent) of those surveyed believed that the shadow economy had declined.
ABHR (2004) provides a list of reasons given by the builder for why they
believe the shadow economy either increased or decreased over the two
years prior to the survey. The results mirror those expressed by the build-
ers in Sydney.
In addition, each Canadian respondent was asked a series of ques-
tions on the effectiveness of various strategies for reducing the size
of the shadow economy. Figure 8.6 presents a summary of these results.
The strategy most commonly regarded as effective in reducing the size of
the shadow economy is education, particularly about the risks of having
work completed by shadow economy operators. This was closely followed
by expanded surveillance and encouraging insurance companies to refuse
insurance for building works for which there are no receipts from builders.
The results in Figure 8.6 bear a striking resemblance to the way in which
the builders in Sydney responded to the same set of questions, notably
that education is by far the most effective way of reducing the size of the
shadow economy.
The strategies of expanding the surveillance and enforcement by gov-
ernment tax auditors; expanding the surveillance and enforcement of
Office of NS Worker’s Compensation Board and Building Inspectors; and
expanding the surveillance and enforcement by NS Department of Labour
produced very similar reactions from the Canadian builders, with a typical
response rate (percentage agreeing) of close to 50 per cent. The combination
of these results, and comparison to those for Australia, appears to confirm
the result that the shadow economy in the construction sector in Canada is
substantially higher than it is in the construction sector in Australia.
8.6 CONCLUSION
In this chapter, we have provided the first known estimate of the shadow
economy in the residential building construction sector in Australia. The
results were obtained using a survey of builders working in the Sydney
and the surrounding metropolitan area. The survey was conducted by a
telephone interview during 2007–08, lasting between 25 and 30 minutes.
The shadow economy in the Sydney residential construction sector was
estimated at between 14 per cent and 16 per cent of the Sydney RCI (or 6.4
per cent and 7.4 per cent of the sector’s total value of construction), which
represents 60 per cent of the estimated shadow economy in residential
construction in Nova Scotia, Canada.


90
82
80
76
73.2
74
80
70 55.4
55.4
53.6
60
52
51
50
48
46
46
42.9
50
40
%
39
40
30
20.5
20
10
0
Q1 Q2 Q3 Q4 Q5 Q6
Australia Canada (NHC) Canada (HRR)
Notes:
Relevant questions:
Methods to limit shadow economy activity
Q1 Australia Expand surveillance and enforcement by the ATO
Canada Expand surveillance and enforcement by government tax auditors
Q2 Australia Expand surveillance and enforcement of Workcover Authority
Canada Expand surveillance and enforcement of Office of NS Worker’s
Compensation Board and Building Inspectors
Q3 Australia Expand surveillance and enforcement by Centrelink
Canada Expand surveillance and enforcement by NS Department of Labour
Q4 Australia Educate people about the legal risks of working in the shadow economy
Canada educate people who work in the underground sector about the legal risks
they are taking
Q5 Australia Educate consumers about the risks of having work completed by shadow
economy operators
Canada Educate consumers about the risks of having work done through
underground activities
Q6 Australia Encourage insurance companies to refuse insurance for building works for
which there are no receipts from builders
Canada Encourage insurance providers to refuse to insure underground building
or renovation projects
Figure 8.6 Comparing the respondent’s view on strategies to reduce the shadow
economy in the residential construction sector in Australia and
Canada (% agreeing to each question that the strategy is effective)
Estimates of the size of the shadow economy across the various broad
sectors of the economy are important to know, but such information is
rarely available. Although we focus in this chapter specifically on the
shadow economy in the residential construction sector, the methodology


adopted here could equally be used to calculate the shadow economy in

the other broad sectors of the economy. With such information it would
be feasible to study how the changes in the aggregate estimates of the
shadow economy, which are available for more than 140 countries around
the world (see Bajada and Schneider, 2005), are influenced by the individ-
ual changes in the shadow economy in the various sectors of the economy.
Developing policy instruments that can deal with the industry specific
issues driving participation in the shadow economy could prove to be
more effective than a single blunt instrument, such as the GST or an busi-
ness number (e.g. ABN), in curtailing the size of the shadow economy. The
estimates provided in this chapter are a start for evaluating the shadow
economy in the various sectors of the Australian economy.
NOTES
* I would like to thank Alanna Hardman for exceptional research assistance.

1. Permission was sought and granted for the use of these questions.
REFERENCES
Atlantic Home Building and Renovations Sector Council (ABHR) and Praxis
Research Consulting (2004), Current Impact of the Underground Economy:
Residential Construction Sector in Nova Scotia. Nova Scotia, Canada: ABHR
and Praxis Research Consulting.
Australian Bureau of Statistics (ABS) (2004), The Underground Economy and
Australia’s GDP, National Accounts Feature Article, October.
Underground Economy, Aldershot, UK: Ashgate Publishing.
Cash Economy Task Force (CETF) (1998), Improving Tax Compliance in the Cash
Economy. Australia: Australian Taxation Office.
Cash Economy Task Force (CETF) (2003), The Cash Economy under the New Tax
System. Australia: Australian Taxation Office.
Department of Trade and Industry (DTI) (2002), Combating cowboy builders.
Consultation Paper.
Department of Trade and Industry (DTI) (2003), Constriction Statistics Annual
2003 Edition.
KPMG, Revay and Associates, Marc Denhez and Bridges/A.G.T. Consulting
(1997), Strategic analysis of underground employment in the construction
industry, December.
Mirus, R. and R.W. Smith (1997), Self employment, tax evasion and the under-
ground economy: micro based estimates for Canada. Working Paper No.1002,
Harvard Law School, International Tax Program, October, Cambridge, MA.
Nova Scotia Department of Finance (NSDF) (1997), The Underground Economy
in Residential Constriction. Canada: NSDF.


O’Grady, J. (1998), Estimates of Revenue Losses to Government as a Result of

Underground Practices in the Ontario Construction Industry: 1995-1997. Ontario,
Canada: Prism Economics and Analysis, Ontario Construction Secretariat.
O’Grady, J. (2001), Estimates of Revenue Losses to Government as a Result of
Underground Practices in the Ontario Construction Industry: 1998-2000 Updated
Estimates. Ontario, Canada: Prism Economics and Analysis, August, Ontario
Construction Secretariat.
O’Grady, J. and G. Lampert (1998), The Underground Economy in Ontario’s
Construction Industry: Estimates of its Size and the Revenue Losses to Government
and the WISB. Ontario, Canada: The Ontario Construction Secretariat.
Pissarides, C. and G. Weber (1989), An expenditure based estimate of Britain’s
black economy, Journal of Public Economics, 39(17), pp.17–32.
Schneider, F. and C. Bajada (2005). An international comparison of underground
economic activity. In C. Bajada and F. Schneider (eds), Size, Causes and
Consequences of the Underground Economy. Aldershot, UK: Ashgate Publishing.
Statistics Canada (1994), The size of the underground economy in Canada.
Catalogue 13-603E, No. 2.
Zanasi, L. (1996), The Underground Economy in Construction in Saskatchewan.
Canada: United Brotherhood of Carpenters and Joiners.


APPENDIX 8.1
Table A8.1.1 Factors affecting the size of the shadow economy
Over the past 2 years transactions in the shadow economy have INCREASED.
Why?
● Increases in the cost of living;
● Government charges and developer contributions constantly increasing;
● Client expects contractor to take off the 10 per cent GST;
● A growth in unlicensed operators;
● The cost of maintaining a builder’s license. This includes the cost of courses
and associated fees;
● A growing number of quotes that are well below cost and could not be from
legitimate operators;
● ‘GST – three letters. I’ve never been offered so much cash as I have since the
GST came in.’
● ‘If you work only on extensions or renovations and repairs, i.e. jobs under
$12 000, cash work is 70 per cent of all work. Eighty per cent of bricklayers are
unqualified and its 30 per cent of other contractors that are working illegally.’
● Immigration. ‘There has been an influx of immigrants. They have different
cultural views to transacting in cash.’
● ‘I get quotes all the time from people who want cash. I lose a job a week to
cash in my area. *What brought cash in was GST – because the customer
doesn’t want to pay it. The question is always “how much for cash?” I can get
most work, for example, 80 per cent of my work for cash if I want it. I just
can’t have it. I need jobs on my books.’
Over the past 2 years transactions in the shadow economy have DECREASED.
Why?
● ‘Not many people have cash. A lot of people are borrowing.’
● ‘It’s not worth it – you would lose your input credits.’
● The people that are getting cash are still paying GST on material cost. But
some probably pay cash for materials and receive cash for the job. It’s less in a
lot of areas in the building sectors. I think the GST has changed it. Everyone is
more accountable than they used to be.
● ‘I get worried about not paying the right amount of insurance for workers in
case they have an accident and sue. These are real concerns.’ A lot of other
builders think the same way so they declare the amount they receive.
● ‘I have been doing larger jobs. Larger jobs do not attract the cash element.
There is a contract in place and the bank finances the job. As a bigger operator
you can’t pay employees or mortgages in cash. Smaller jobs attract cash.’
● ‘It’s probably because it’s getting more difficult to do it. There are more and
more tax audits occurring and the ATO tends to target construction generally.’


Table A8.1.2 Suggestions for reducing the shadow economy
Respondent Suggestions
● ‘Increased awareness through industry associations. Regulation is often carried
out without regard to effectiveness. There are very few mechanisms designed to
measure the outcomes of regulation.’
● ‘Owner builders are an unregulated sector and this sector attracts cash activity.
It needs to be closed. We lose trades and workers to the attraction of cash in
this sector whilst decent businesses pay the costs of regulation.’
● ‘Increase random inspections for licenses and wages earned. If every job had a
contract and a warranty then no one would do cash. Owner builders do more
cash work.’
● ‘Owner builders should be educated and there should be fines levied on owner
builders who engage in tax evasion or cash prices because it pressures us.’
● ‘The ATO should monitor owner builders more because 90 per cent of their
work is done on cash. They should be eliminated. They are a major driver of
cash.’
● ‘In Residential – reduce owner builders doing their own work. I know a very
good qualified carpenter who is not qualified as a builder yet he gets owners to
go ‘owner builder’ for work they want done. As a carpenter he can’t get HW,
so he carves the job up. It is carved in components less than 12 000. Split jobs
– part cash. That’s how they do it. Reducing “owner builders” would reduce
cash definitely.’
● ‘Workcover is concerned with on site issues but what would make Workcover
more effective is more men on the ground. We are all scared of Workcover –
but they do not visit sites nearly enough. Again Workcover mainly deals with
larger sites so they won’t visit the cash job sites. The cash site is not seen by
many government agencies – it’s invisible.’
● ‘If OFT [Office of Fair Trading] got involved and found that builders were
doing jobs for cash and not paying tax – then the OFT can deregister people
on ethical grounds – not construction grounds. So if someone is doing jobs for
cash and evading tax then technically – they can be deregistered.’
● ‘Banks report cash cheques and who was cashing them.’
● ‘Education within the industry is important as is educating the customer. It’s
too easy for owner builders to get a license. Inspection (of owner builders
works) at certain intervals by accredited supervisors is a very good idea
(practicing certifying authority – undertaken by a Council or a private
authority). That would get rid of a lot of rubbish and cash in the industry.’
● ‘It’s the ATO’s job (to reduce the shadow economy). They should increase
audits at the lower end of the industry. Workcover needs more enforcement in
terms of workers’ compensation. Educate consumers about the potential for
injuries on site. If there is no workers’ compensation, the injured person could
potentially sue (the customer). *The ATO should send a letter to new owner
builders explaining all this.’


Respondent Suggestions
● ‘Educate the customer to look for a photo on a tradesman’s license (require
photo ID). People go around quoting license numbers that belong to the
cousin of a friend. Workcover have to get out there and see what’s going on.
They need to do more spot checks on for example, immigrant labour etc.’
● ‘The ATO should send a letter to owner builders who submit DAs. The title
of the property should have on it that the work was done by an owner builder.
Progress payments should be shown to insurers.’
● ‘You could educate them via the DA process in Councils – Councils release
owner builders’ names to ATO who later can follow up with owner builders. It
would be better to educate people before they started the work.’
● ‘Education to the public is the best bet. The consumer at the end of the day
is paying and they should be made aware of what’s going on in the industry.
Training the consumer – making them aware of the risks involved.’
● ‘Audit subcontractors/undertake labour audits. The cash is mostly in labour.’


9 Who is working illicitly and why?
Insights from representative survey
data in Germany
Dominik H. Enste
9.1 INTRODUCTION
There has been controversy over the last 25 years about the exact defini-
tion as well as appropriate estimation methods to measure the size of
the shadow economy (see for example Schäfer, 1984). The causes, con-
sequences and instructions for economic policy have also been studied
extensively (see for example Schneider and Enste 2000a, b, 2002; Enste
and Schneider, 2006a, b, 2007; Enste and Hardege, 2007). Contrary to the
former research topics, this analysis focuses on the following question:
what kind of people do work illicitly – and why?
In order to tackle the topic, a broad overview of basic knowledge is
provided in the next section (9.2): first, the terms ‘shadow economy’ and
‘illicit work’ will be defined followed by some information about their
size and structure. Section 9.3 presents an evaluation of socio-economic
data that have been derived from different surveys. Using the example of
Germany, whether the ‘usual suspects’ are most likely to act as supplier
or demander on the shadow market is considered. Some explanations
for these developments as well as some estimates on the expected value
loss are presented. Section 9.4 discusses strategies to reduce shadow
economic activities. Measures to fight illicit work, by means of more
frequent controls and rigorous punishment, have proven to be hardly suc-
cessful. Furthermore, the efficiency of arrangements for tax deductibility
of handicraft and home-related services is in doubt, too. In the current
public discussion, minimum wages are considered as a possible solution
to the problem, though this would lead to an increase of shadow economic
activities. This chapter concludes by outlining the (key) findings of this
analysis.
324


Who is working illicitly and why? 325
9.2 THEORETICAL CONSIDERATIONS ON ILLICIT

WORK AND SHADOW ECONOMY
9.2.1 Definitions
The starting point for scientific research is the fact that almost every state
has a dual economy. Besides economic activities that are registered offi-
cially in national accounts, some people prefer to offer or demand goods
and services in the nonofficial sector instead. Depending on what kind of
definition is chosen to describe this deviant behaviour, size and evaluation
of the shadow economy differ drastically. The boundaries are not clear-
cut, but the terms ‘illicit work’ and ‘shadow economy’ are quite frequently
used to describe the same context (see for example Graf, 2007, 2008; Koch,
2008).
According to international standards and because of difficulties in
determining its dimension, value creation in self-sufficient economies
is not covered in national accounts. Apart from this, the shadow and
underground economy are counted among the nonofficial economy.
Consequently, they are not ranked in regular statistics. The classification
of these three sectors is drawn upon characteristics of the legality of pro-
duced goods and services on the one hand and the fulfilment on the other
hand (See Table 9.1).
Table 9.1 Self-sufficient, shadow and underground economy
Self-sufficient Shadow Underground

economy economy economy
Good/service legal legal illegal
Fulfilment legal illegal illegal
Examples ● do-it-yourself ● illicit work ● concealment of
● barter (approx. €150 stolen goods
● voluntary billion) ● drug trafficking
service ● material ● prohibited
● babysitting procurement gambling
● neighbourly without check ● fraud
help (approx. €80 ● smuggling
billion) ● human trafficking
Size of sector approx. €1,000 approx. €230 billion approx. €100 billion
billion
Source: Enste and Schneider, 2007; own calculations based on certain assumptions. The
figures should be interpreted with care: the size the methods used are controversial (see
Kazemier, 2006).


The underground economy includes criminal activities: the traded goods

and services are prohibited by law and this also applies to their production
and their execution. Examples of such economic activities include drug
trafficking, illegal prostitution and smuggling.
Elsewise, in a self-sufficient economy neither the goods and services nor
the fulfilment can be characterised as illegal, as the following examples
demonstrate: do-it-yourself (DIY), barter, voluntary service, babysitting
and neighbourly help.
The shadow economy can be distinguished from both in terms of legal-
ity, too. Goods and services, which are offered and demanded via black
markets, are not illegal. But it is neither allowed to produce these goods
nor is it permitted to carry out those kinds of services. Examples include
moonlighting, as well as material procurement and served man hours
without check. Undeclared work accounts for the major part of value cre-
ation in this sector, including working time equating to approximately 150
billion euros and material procurement without check, which accounts for
about 80 billion euros.
These definitions are again clearly represented in Table 9.1.
Even though illicit work violates the law, many people consider it
as legitimate. In a survey of the Allensbach Institute only one in four
regards illicit work as a behaviour that should not be endorsed in any
case (Institut für Demoskopie Allensbach, 2007). Tax evasion is rejected
by every second person, whereas corruption and social fraud are not
accepted by 70 per cent of the population. Another survey conducted by
the TNS-Emnid Institute on behalf of the Cologne Institute for Economic
Research discovered that 25 per cent of the population regard illegal
employment as a peccadillo. Only 3.6 per cent would report illicit workers
to the police.
The fact that the respondents perceive illicit work as less reprehensible
than tax evasion appears paradoxical, since illegal employment is always
accompanied by tax evasion. But a closer look reveals a certain economic
logic behind this, because apart from negative effects on state finances and
competition, illegal employment may also lead to some positive impacts.
Additional incomes can be generated, as illicit work enables production
and supply of goods and services that could not be produced in the official
economy for financial reasons. Two-thirds of this extra income flow back
to the official sector and generate additional tax revenue (see Schneider
and Enste, 2000b for a precise analysis).
Still, it is striking that people rate reneging on a promise as more rep-
rehensible than fooling the state. This leads to the conclusion that social
values and norms remarkably influence individual behaviour. Therefore
it is an essential issue to improve tax morale in order to fight illicit work.


9.2.2 Measuring Shadow Economy and Illicit Work
There are several studies that provide information about illicit work at
irregular intervals (Pedersen, 2003; Feld and Larsen 2006). Additionally
annual estimations measure the size of the shadow economy based
on macroeconomic methods (Schneider, 2005). However, the results of
these studies differ strongly due to the appliance of different estimation
techniques and variations in definitions and assumptions. Consequently,
international comparability can not be assured. It is difficult to compare
national shadow economies since the estimations of variables (such as
tax burden, unemployment rates) differ significantly from state to state.
Moreover, illicit work is carried out secretly and can therefore only be
detected indirectly (for example in interviews and so forth). Macro-
economic methods are, in addition, criticised on various grounds, for
example, the currency demand method. For a discussion of the shortcom-
ings of the methods see Kazemier (2006). Regardless of these limitations,
some international studies have been conducted. Comparable trends can
indeed be observed, even though different estimation methods have been
applied. The emerging differences in size, that have been determined empir-
ically, can be explained to a certain extent with theoretical approaches.
Here, a microeconomic approach has been chosen to find out more
about illicit work in Germany (see for a similar approach Feld and Larsen,
2005, 2006). Since international comparison using surveys is difficult, as
the EU-Commission (2007) stated after a large European Survey failed,
the analyses focus on Germany only.
9.3 ILLICIT WORK IN GERMANY: THE SURVEY
9.3.1 Method
Based on experiences in the field of empirical social research it is known

that people sometimes tend to answer dishonestly and rather give ‘socially
desired’ answers instead (see, for example, Friedrichs, 1998). If people are
asked to indicate their income, the ratio of those who refuse to answer is
fairly high. Similarly, it is understandable that people also are reluctant to
answer questions about illegal activities and income derived from them.
In order to bypass this problem, researchers ask indirect questions such as
‘What do you think, how many of your friends and neighbours do work illic-
itly?’ But results also differ significantly depending on what terms are used
(for example ‘not declared work’ instead of ‘illicit work’), which methods are
used (for example, face-to-face interview or on the phone) and the context


in which the interview is taken (for example, does the survey focus on illegal
employment or are there just few questions in a more general context).
This representative survey has been conducted by TNS Emnid for the
Cologne Institute for Economic Research. For this purpose 1018 people
(minimum age: 18 years) were asked in January 2007 for their attitudes
towards illicit work. A modified interview technique was used for sensitive
questions to tackle the above-mentioned problems. The monitor screen
of the computer faced the respondent allowing secrecy in answering.
Influences posing social pressure on the interviewee were excluded as far
as possible. The questionnaire followed a structure that also allowed inves-
tigation of the relevance of the social environment. An all-day context
was created by appending questions about illegal employment to a more
general questionnaire.
9.3.2 Main Findings
The main findings of the survey are:
● Supply of illicit work: 20.7 per cent of the respondents had worked
illegally at least once during the last 12 months. The medium
working time of these people accounts for 6.5 hours per week and
equates to a wage rate of 10 euro/hour. Gender differences can be
depicted: 25 per cent of men worked illicitly, whereas the proportion
of women accounts only for 16 per cent.
● Demand for illicit work: 30.8 per cent of the respondents admit-
ted that they employed workers without check during the last 12
months. The proportion of men equates to 35.4 per cent, whereas
the proportion of women accounts for 26.5 per cent.
Table 9.2 gives an overview of further influencing factors. It is obvious

that some of the results differ to what some might have expected. Contrary
to the prevalent assumption it is not the ‘usual suspects’ (people, that are
temporarily unemployed or have never been employed at all), but other
population groups that turn out be most engaged in illicit employment.
One reason for this might be the fact, that there is a lack of ‘customers’ and
therefore few possibilities for clandestine employment. Besides, misuse
of social services is prosecuted by public authorities quite intensely. The
result of former studies, that women act more honestly when it comes
to economic issues, can be backed on basis of these data. Apart from
this, correlation analysis show that employed people (workers and self-
employers to be precise) most commonly work illicitly or demand services
from illicit workers.


Table 9.2 Illicit work in Germany (2007) – descriptive statistics
Variable Specification Supply Demand

Gender total (N 5 1018) 20.7 30.8
male 25.0 35.4
female 16.0 26.5
Age categories 18–19 years 42.9 42.9
20–29 years 23.3 26.5
30–39 years 21.1 38.3
40–49 years 19.1 27.6
50–59 years 20.1 30.2
60–69 years 18.4 29.4
> 70 years 20.4 30.9
Education currently attending school 0 0
secondary general school; no 29.9 26.9
vocational training
secondary general school, plus 20.9 29.1
vocational training
intermediate secondary school 18.1 34.2
(no higher education entrance
qualification)
higher education entrance 24.07 27.5
qualification
polytechnic /university 21.42 33.7
Occupation working full-time 22.1 32.8
working part-time 17.1 32.1
pensioner (previously employed) 19.9 31.2
apprentice, pupil, student, in 29.9 33.3
re-education
not employed (previously 26.5 28.4
employed)
temporary unemployed 14.3 20.3
never been employed 0 7.1
Income (person) no personal income 17.9 17.9
up to 1500 Euro/net 21.3 32
above 1500 Euro/net 30.2 42.9
total (N 5 811) 23.2 33.58
Income (household) up to 1500 Euro/net 25.2 27.1
more than 1500 Euro/net 24.4 38.9
total (N 5 778) 24.7 35.2
Living situation house owner 20.2 31.0
flat owner 22.6 41.0
tenant 20.4 27.2
subtenant 20.7 50.0


Variable Specification Supply Demand

Profession self-employed 21.7 47.7
worker 30.0 32.9
employee 16.8 28.6
public servant 23.3 23.3
Size of city up to 20,000 residents 10.1 19.7
20,000–500,000 residents 23.8 38.3
above 500,000 residents 20.4 23.2
Source: own calculation based on representative survey of TNS Emnid (January 19–29,
2007; 1018 interviewees aged 18 and older)
Correlations between different variables and supply/demand of illicit

employment as well as its significance were determined by means of corre-
lation analysis (Spearman and Pearson coefficients). The derived findings
are summarised in Table 9.2.
On the other hand, variables such as educational background, occup-
ation, living situation, size of city and income seem to have no significant
effect on this deviant behaviour. Positive correlations can only be detected,
when income is split up into three categories. A possible explanation for
the reversed-U-shaped relation between size of city and illicit work might
be that neighbourly help is quite common in smaller communities, while
anonymity in major cities leads to difficulties in finding somebody to work
without check. Consequently, illicit work is prospering in medium-size
towns.
When statements in regard to deviant behaviour and reasons for the
supply and demand of illicit work are analysed, the correlations are con-
siderably stronger. The interviewees were also asked if they agreed with
certain statements (see Table 9.3). In this analysis, a high tax burden
(line 12) as well as possibilities to sidestep directives (line 13) and a low
probability of being caught (line 14) turn out to be prominent reasons for
illicit work. If respondents have negative attitudes towards illegal employ-
ment (lines 15, 16), it is probable that corresponding supply and demand
is rather low. Former experiences do not act as a deterrent, but imply a
high probability that people demand or offer illicit work (line 17). The
perception of quality differences does not show any significant correlation
(line 18). Personal attitudes regarding deviant behaviour (lines 19–22)
strongly influence the demand for and supply of illicit work. Yet, the most
important determinants are social norms and the perception of people’s
behaviour in the social environment.


Table 9.3 Influencing factors for illicit work supply and demand
Variable Supply Demand

1 Gender (1 5 male, 2 5 female) −0.121*** −0.096***
2 Age x x
3 Educational background x x
4 Occupation 0.091*** 0.093***
5 Profession (1 5 public servant, 2 5 employee, 0.098*** 0.103***
3 5 worker, 4 5 self-employed)
6 Size of househould x −0.050*
7 Net-income of household x x
8 Income of respondent x x
9 Income of respondent (15 no income/ 0.093*** 0.132***
25 less than €1500 / 35above €1500)
10 Living situation x
11 Size of city x x
12 Statement: it is the state’s fault, taxes are too high x 0.077**
13 Statement: illicit work is a smart way to 0.126*** 0.175***
sidestep directives
14 Statement: being caught is improbable 0.169*** 0.215***
15 Statement: illicit workers should pay high fines −0.66** −0.121***
16 Statement: tax evaders are fraudsters −0.116*** −0.125***
17 Statement: I employed illicit workers during 0.251*** 0.429***
the last 2–3 years
18 Statement: there are no quality differences x 0.151***
concerning illicit work
19 Statement: a job without check is a peccadillo 0.144*** 0.246***
20 Statement: the honest person is the loser 0.119*** 0.246***
21 Statement: skipping work is ok every once a while x 0.096***
22 Statement: on occasion I would work illicitly 0.186*** 0.163***
23 Statement: almost everybody understates income 0.09*** 0.181***
when it comes to taxes
24 Number of neighbours, who have employed 0.263*** 0.393***
somebody illicitly
25 Number of neighbours, who have worked illicitly 0.345*** 0.414***
Notes:
Spearman Correlation Coefficient
*** 5 correlation, significant on 1 per cent level
** 5 correlation, significant on 5 per cent level
* 5 correlation, significant on 10 per cent level
x 5 no significant correlation


9.3.3 Reasons for Illicit Work
According to this analysis one major conclusion may be drawn: if the

respondent assumes that illicit work is quite common in his circle of
friends and neighbours, the probability that he demands/offers illicit work
is rather high. The higher the assumed number of acquaintances who do
not comply with the rules, the more likely the respondent also violates
social norms.
In order to find out about perceived social norms, the interviewees were
asked the following question: ‘Sometimes it occurs that one works for
somebody or places an order without check. Thus it can be considered
illicit work. This represents a win–win situation for both sides: one does
not have to pay VAT, whereas the other saves income tax. This applies for
example in the following cases: babysitting, housework, renovation, car
repairs). Think of your circle of friends and neighbours. Let’s assume this
is 10 persons.
1. What do you think, how many of these 10 persons have employed

somebody without check, thus illicitly? __ of 10
2. What do you think, how many of these 10 persons have worked
illicitly themselves? __ of 10
When taking a closer look at the results of these questions, it is obvious

that few people know any friends or neighbours who have been employed
or worked illicitly: 22 per cent of the respondents do not know anybody
who has employed workers illegally. A total of 34.4 per cent do not know
any person in their social environment who has worked illicitly. On the
contrary, almost every second interviewed person (45.2 per cent) indi-
cated that he knows that at least half of his neighbours and friends have
employed clandestine workers. One-fifth of the respondents assume that at
least half of their friends and neighbours have worked illicitly themselves
(see Figure 9.1).
The findings on the perceived scope of illegal employment allow some
interesting insights about influence of social norms on one’s own behav-
iour. Own demand for illicit work significantly correlates with perceived
demand in circles of acquaintances, as the regression value of r 5 0.388
indicates. The same is true for the relation between the respondent’s illicit
work supply, on the one hand, and the perceived supply in the circle of
friends and neighbours. The calculated value of r 5 0.355 proves the case
for this correlation. This variable indicates the highest explanatory contri-
bution. This also applies if additional variables are included, as a logistic
regression analysis shows. On the basis of some explicative variables, the


90
80
70
Cumulative distribution
60
50
40
30
20
10
0
10 9 8 7 6 5 4 3 2 1
Belief about number of friends and neighbours who have ...
demand supply
Figure 9.1 Cumulative distribution of beliefs about illicit work in social

environment
calculations may lead to forecasts about the probability that a person

works illicitly or employs illicit workers respectively. Tables 9.4 and 9.5
show the results of stepwise logistic regressions, including relevant factors
explaining the supply and demand of illicit work. These results can be
summarised to describe how well one can predict this deviant behaviour if
the value of some variables is known:
● Supply of illicit work: On the basis of the answers to the question,

how many neighbours and friends are working illicitly, it can be
estimated with a 76.9 per cent probability whether the respondent
worked illicitly during the last 12 months or not. This value varies
according to gender: the percentage for correctly estimated results
equates to 83.3 per cent (women) and 70.2 per cent (men). In compli-
ance with this data, it can be said that social norms affect women’s
behaviour more strongly than men’s. There are further variables,
which may enhance the model’s predictive power up to 79.5 per
cent for both sexes. These variables include: (1) possible positive
experiences with illicit workers in earlier years, (2) profession of the
respondent. Self-employed people and workers are more frequently
illegally employed than public servants or employees. For more
relevant factors see Table 9.4.


Table 9.4 Logistic regression: supply of illicit work. Variables in

equation: Logistic regression (men and women)
Variables Regression Standard Wald Sig. Exp(B)

coefficient B error
Step 1(a) Neighbours’ illicit 2.565 0.352 52.994 0.000 13.007
work supply
Constant −3.291 0.340 93.873 0.000 0.037
Step 2(b) Illicitly employed 1.039 0.212 24.090 0.000 2.827
Neighbours’ illicit 2.369 0.356 44.394 0.000 10.689
work supply
Constant −3.361 0.341 97.200 0.000 0.035
Step 3(c) I cannot afford 0.873 0.217 16.234 0.000 2.393
legal work
Illicitly employed 0.881 0.218 16.329 0.000 2.414
work supply
Constant −3.446 0.343 100.755 0.000 0.032
Step 4(d) Profession 19.937 0.000
(self-employed)
(public servant) 0.838 0.557 2.267 0.132 2.312
(employee) 0.222 0.325 0.470 0.493 1.249
(worker) 1.106 0.346 10.195 0.001 3.021
I cannot afford 0.905 0.222 16.564 0.000 2.471
legal work
work supply
Constant −4.023 0.467 74.192 0.000 0.018
Step 5(e) Profession 16.629 0.001
(self-employed)
(public servant) 0.774 0.561 1.905 0.168 2.168
(employee) 0.254 0.327 0.604 0.437 1.289
(worker) 1.060 0.349 9.231 0.002 2.885
I cannot afford 0.825 0.224 13.604 0.000 2.281
legal work
Tax evaders are −0.740 0.272 7.411 0.006 0.477
criminals (tax
compliance)
work supply
Constant −3.863 0.471 67.324 0.000 0.021
Step 6(f) Professions 18.849 0.000
(self-employed)
(public servant) 0.630 0.562 1.256 0.263 1.878



coefficient B error
(employee) 0.352 0.330 1.135 0.287 1.422
(worker) 1.217 0.356 11.702 0.001 3.378
Personal income 7.449 0.024
Personal income (1) −0.253 0.444 0.325 0.569 0.777
Personal income (2) −0.603 0.223 7.310 0.007 0.547
I cannot afford 0.818 0.225 13.162 0.000 2.265
legal work
Tax evaders are −0.727 0.273 7.071 0.008 0.483
criminals (tax
compliance)
neighbours’ illicit 2.368 0.363 42.452 0.000 10.674
work supply
Constant −3.557 0.481 54.642 0.000 0.029
Step 7(g) Profession 19.406 0.000
(self-employed)
(public servant) 0.652 0.565 1.332 0.248 1.919
(employee) 0.353 0.330 1.142 0.285 1.423
(worker) 1.234 0.356 12.029 0.001 3.435
Personal income 7.184 0.028
Personal income (1) −0.248 0.443 0.314 0.575 0.780
Personal income (2) −0.593 0.223 7.052 0.008 0.553
I cannot afford 0.808 0.226 12.847 0.000 2.244
legal work
Tax evaders are −0.746 0.273 7.472 0.006 0.474
criminals (tax
compliance)
work demand
work supply
Constant −4.290 0.683 39.460 0.000 0.014
Notes:
(a) in step 1 entered variables: Neighbours’ illicit work supply.
(b) in step 2 entered variables: Illicitly employed.
(c) in step 3 entered variables: I cannot afford legal work.
(d) in step 4 entered variables: Professions.
(e) in step 5 entered variables: Tax evaders are criminals (tax compliance).
(f) in step 6 entered variables: Personal income.
(g) in step 7 entered variables: Neighbours’ illicit work demand.


● Demand for illicit work: The demand for illegal employment during
the last 12 months can also be forecast on the basis of the perceived
level of deviant behaviour in the social environment (65.8 per cent).
However, in this case, previously gained positive experiences with
illicit workers in the past play a more important role. If both varia-
bles are taken into account, one can forecast the behaviour correctly
with a probability of 75.7 per cent. If responses about attitudes
towards tax evaders and illicit workers are added to this calculation,
the probability of accurate forecasts rises to 78.7 per cent. Further
relevant factors are excuses like: ‘I cannot afford legal work’, the
occupation and the answers to the statement: ‘Illicit workers should
pay high fines.’ (See Table 9.5.)
9.3.4 Welfare Loss Caused by Illicit Work and Affected Branches
The survey of the Cologne Institute for Economic Research shows, that
almost one-quarter of illicit work could be shifted to the official economy.
This result is based on the question, what respondents would do if there
was no shadow economy:
● 48 per cent indicated that this work would not be carried out any
more;
● 30 per cent indicated that they would do this work themselves;
● 22 per cent indicated that they would place orders officially.
Value creation in the shadow economy is estimated to amount to 150

billion euros, which equates to 2.7 billion full-time jobs with an average
added value of 57 000 euros per capita. If shadow economic activities were
fought efficiently, there would be potential of at least 600 000 additional
jobs in the official economy. This in turn would lead to additional fiscal
income, as the following calculation demonstrates:
1 €11 200 revenue from social insurance contributions (per capita, per
year)
1 €4 800 revenue from income tax (per capita, per year)
5 €16 000 total fiscal effects (per capita, per year)
If these numbers are multiplied by the number of possible jobs, the
overall potential for state revenue can be calculated. The potential
for additional tax revenue amounts to 2.88 billion euros, whereas the
budgetary surplus of the social insurances equates to 6.72 billion euros.
Therefore the total fiscal return adds up to almost 10 billion euros per
year. This value is regarded as a minimum value, since some other


Table 9.5 Logistic regression: demand for illicit work. Variables in

equation (men and women)

coefficient B error
Step 1(a) Neighbours’ illicit 2.780 0.292 90.866 0.000 16.119
work supply
Constant −2.851 0.278 105.172 0.000 0.058
Step 2(b) Illicitly employed 1.974 0.245 64.917 0.000 7.201
work supply
Constant −3.038 0.286 113.073 0.000 0.048
Step 3(c) Illicitly employed 2.088 0.257 66.156 0.000 8.065
Tax evaders are −1.157 0.247 21.930 0.000 0.314
criminals
(tax compliance)
work supply
Constant −2.809 0.287 95.859 0.000 0.060
Step 4(d) I cannot afford 0.807 0.231 12.239 0.000 2.241
legal work
Tax evaders are −1.048 0.252 17.238 0.000 0.351
criminals (tax
compliance)
work supply
Constant −2.918 0.293 99.314 0.000 0.054
Step 5(e) Occupation 8.765 0.012
Occupation (1) 0.863 0.506 2.913 0.088 2.370
Occupation (2) 2.055 0.702 8.578 0.003 7.809
I cannot afford 0.837 0.232 13.048 0.000 2.309
legal work
Tax evaders are −1.058 0.256 17.090 0.000 0.347
criminals (tax
compliance)
work supply
Constant −3.777 0.578 42.672 0.000 0.023
Step 6(f) Occupation 8.539 0.014
Occupation (1) 0.923 0.504 3.357 0.067 2.517
Occupation (2) 2.037 0.699 8.484 0.004 7.670
I cannot afford 0.812 0.233 12.138 0.000 2.252
legal work



coefficient B error
Illicit workers −0.972 0.398 5.971 0.015 0.378
should pay high
fines
Tax evaders are −0.842 0.269 9.795 0.002 0.431
criminals (tax
compliance)
Neighbours’illicit 2.439 0.305 63.863 0.000 11.456
work supply
Constant −3.770 0.576 42.776 0.000 0.023
Notes:
(a) in step 1 entered variables: Neighbours’ illicit work supply.
(b) in step 2 entered variables: Illicitly employed.
(c) in step 3 entered variables: Tax evaders are criminals (tax compliance).
(d) in step 4 entered variables: I can not afford legal work.
(e) in step 5 entered variables: Occupation.
(f) in step 6 entered variables: Illicit workers should pay high fines.
studies assume higher numbers of recreated jobs and therefore higher

fiscal profits.
Figure 9.2 gives an overview on where illicit work is – according to the
survey – most commonly spread.
9.4 MEASURES TO FIGHT ILLICIT WORK
9.4.1 Fighting Symptoms: Hardly Successful Strategy
In order to fight illegal employment, politicians as well as bureaucrats

chose tightened sanctions, more frequent raids, controls and higher penal-
ties. A law to prohibit illicit work was introduced in 1957 in Germany. At
that time, offences were penalised with fines up to 5000 euros. Over time,
the fines have risen and the law was finally changed in 2000. Accordingly,
the fines were set up to a maximum of 200 000 euros and 500 000 euros
respectively, in the case of illegal employment. Since then, illicit work may
not only be categorised as misdemeanour, but also as a criminal offence
in some cases. At the same time, the Federal Employment Office was not
in charge of the prosecution any longer, as the responsibility was shifted


housework, building construction

2.0 1.7 1.1
4.3 cleaning
18.7
5.6
car repairs
hotel business, gastronomy
12.1 hairdressing, beauty care
tutoring, child care

15.8
entertainment branch
nursing, elderly care

12.2
upscale services, consulting
industry
13.7
12.8 retail, grocery
Source: IW Survey, 2007
Figure 9.2 Distribution of illicit work according to different branches
to the customs authority (financial control department for illicit work).

Moreover, this implied an increase of workforce. A report by the German
Federal Court of Auditors provides information about how cost-effective
and successful this financial control department was in carrying out the
combat against illicit work. This report was published at the beginning of
2008 and confirms the criticism, which had already been pointed out by
the Cologne Institute for Economic Research in 2004 (Enste, 2004). The
work of the tax inspectors has been characterised as ineffective and costly.
Uncovered losses in the range of 600 billion euros (2006) were presented to
the public as proof of alleged success. But, in fact, only additional revenue
from taxes and social security contributions (30 billion euros) as well as
fines (10 billion euros) were earned. These earnings correspond to not
more than 5–10 per cent of reported losses.
On the other hand, these earnings stand in contrast to the number of
about 6500 employees, who cause costs of as much as 400 billion euros
(2007). In addition, the financial control department for illicit work
is mainly engaged with small offences: more than 70 per cent of the
cases are minor offences, which can be uncovered easily by comparing
data in different funding agencies. The control department should not
be abolished, but needs to be evaluated carefully. The effectiveness of


anti-illicit-work measures necessitates better surveillance. In order to

improve the efficiency of the department, the Federal Court of Advisors
worked out some advice, which includes: increased presence in order to
prevent illicit work (especially at large building sites), focus on major
offences with high potential losses and data examination in collabora-
tion with external institutions. Since the financial control department
has not uncovered many illicit workers, some authors concluded that
illegal employment is only thinly spread. Yet, some of the authors
misinterpreted the Federal Court of Advisors advice: it is not the low
number of exposed cases, but the small amount of revenue that is criti-
cised primarily, since this suggests a low efficiency of encashment and
organisation.
Data derived from other surveys lead to similar conclusions. Even
though surveillance was intensified, 21 per cent of the respondents
assumed the probability of being detected was low (survey of the Cologne
Institute for Economic Research). The fear of being caught has decreased
compared to previous years (2001: 18.2 per cent; 2004: 15.1 per cent).
The objective probabilities for punishment vary considerably depending
on the kind of offence. Misuse of social benefits (for example, receiv-
ing unemployment pay even though one works illicitly) can be detected
easily by comparison of data in many cases. Therefore, it is likely that
disclosure of social benefits misuse leads to preliminary investigations,
whereas moonlighting is hardly prosecuted. These results match the citi-
zens’ perception of injustice: 71 per cent do not accept misuse of social
services, whereas only 25 per cent take a critical position towards illegal
employment.
Fiscal policy measures are discussed alongside tougher punishment
and more controls. This includes, for example, tax reductions on labour-
intensive services and tax rebates on labour costs for officially assigned
jobs. New forms of tax reduction possibilities for home-related services
have been agreed to recently and came into force at the beginning of 2009.
As illicit work is quite common in this area, the above-mentioned meas-
ures should foster regular jobs in home-related services.
9.4.2 Tax Deductibility
In order to fight illegal employment, tax deductibility on labour-

intensive jobs has been agreed. As moonlighting was quite common in
this field, the parliament chose to allow 20 per cent of expenditures to
be subtracted from the tax load. This gives incentives to the people to
prefer legal work, since the price is as high as for illicit work. The areas
covered are:


● Home-related services: jobs may, for example, include cleaning,

yard work and removal help
● Child care, nurture and other care services
● Handicraft services: fostered jobs include measures for renovation,
modernisation and maintenance.
However, a point of criticism needs to be made, as these special arrange-

ments do not accord with an ordo-liberal point of view: instead of reduc-
ing the tax on special services, the tax rate, in general, should have been
decreased.
9.4.3 Effects of a Minimum Wage
Minimum wages of 7.50 euros per hour are subject to the current public
debate. By contrast the lowest agreed wages may amount to about 3 euros
in some regions and branches.
But even if this may tempt someone to plead for minimum wages, the
potentially negative effects need to be carefully taken into consideration.
With regard to the shadow economy it must be clearly stated, that an intro-
duction of minimum wages will lead to an increase of illicit work because
the incentive to employ someone illicitly increases. Critics therefore stress
the serious threat that a notable number of jobs will be shifted towards the
shadow economy as a consequence. Estimations give the additional value
creation in the nonofficial sector as 25 billion euros (estimation is based on
an assumed minimum wage of 7.50 euros). These values were determined
by two different underlying methods: with macro-economic simulation
models, on the one hand, and by means of the survey of the Cologne
Institute for Economic Research, on the other hand.
A calculation example will clarify this position: if minimum wages were
introduced, up to 4 million wage earners (who earn less than 7.50 euros
nowadays) would be affected. The results of the survey by the Cologne
Institute for Economic Research, which were discussed previously, show
roughly 20 per cent of the population have worked illicitly during the past
year, whereas 40 per cent would do so, if they had the opportunity. And,
every third person employed on illicit worker during the last 12 months.
Consequently, one may assume that there is huge potential for illicit work:
up to 800 000 to 1.6 million jobs would be shifted to the shadow economy,
if the minimum wage were set to 7.50 euros nationwide. This equates to an
increase in additional value of 12–25 billion euros in the shadow economy.
Macro-economic estimation methods reveal similar trends. According
to those models, the extra value is determined to be as high as 16.3–28
billion euros. These calculations are based on the following assumptions:


minimum wages of 7.50 euros and an average wage for this kind of illicit
work of 5 euro/hour (Schneider, 2007).
If a minimum wage was introduced, it would affect adolescents and
job starters (up to the age of 25), in particular, since almost 30 per cent
(in West Germany) and almost 50 per cent (in Eastern Germany) earn
less than 7.50 euros per hour. At the same time, this age group, alongside
less qualified people is most commonly involved with illicit work (every
third person). Therefore, even more people are pushed into the shadow
economy due to minimum wages. This in turn implies negative social con-
sequences: if young people start to work illicitly at an early age, this might
impose some negative influences on their development of values.
9.5 SUMMARY
Illicit work is a widespread phenomenon. The dimension of the nonofficial

sector is estimated to amount to about 3 to 4 per cent of the gross national
product based on survey data. But because respondents tend to give
socially desired answers, this portion is likely to be even higher and add up
to 7 per cent of GDP. According to data derived from the survey by the
Cologne Institute of Economic Research, 20.7 per cent of the respondents
worked illicitly during the last 12 months. The demand for illicit workers is
even higher: 30.8 per cent of the respondents indicated that they employed
illicit workers during the last 12 months. Analysis of possible influencing
factors shows that perceived norms in the social environment play by far
the most important role if one offers or demands illicit work oneself. When
information about friends’ and neighbours’ behaviour is known, conclu-
sions about the respondent’s behaviour may be drawn. If some explana-
tory variables are taken into consideration, the probability for correct
estimations can be as high as 80 per cent.
In the context of the survey, 22 per cent of the respondents indicated
that they would make contracts officially if there was no possibility of
demanding and offering goods and services via the black market. This
equates to a number of approximately 2.7 million full-time jobs. One extra
job in the official economy generates a fiscal surplus of 16 000 euros, mul-
tiplied by the potential of new employment this would correspond to (at
least) 9.6 billion euros total tax and social insurance revenue. Since there is
an enormous potential for new jobs and thus positive economic effects, the
issue of fighting illegal employment needs to be pushed to the top of the
political agenda. But even though fines have risen and controls have tight-
ened, measures have hardly proven to be effective. The financial control
department has uncovered some offences, but the resulting income quote


is far too small and is contradictory to the huge sum of costs. The govern-
ment has passed some new laws to foster regular employment, including
tax deductibility of home-related, nurture, care and handicraft services.
But these have to be criticised from an ordo-liberal point of view, too.
Finally, the possible consequences of a minimum wage were discussed. It
can be clearly stated, that an introduction of fixed wages would lead to an
expansion of the shadow economy and, consequently, to negative effects
on the economic performance.
REFERENCES
Enste, D. (2004), Schattenwirtschaft: Boom trotz härterer Strafen – warum?

Zeitschrift Sozialer Fortschritt, Jg. 53, Heft 2, 29–36.
Enste, D.H. and S. Hardege (2007), Regulation and shadow economy. Working
Paper. Cologne Institute for Economic Research.
Enste, D.H. and F. Schneider (2006a), Jahrbuch Schattenwirtschaft 2006/2007.
Münster.
Enste, D.H. and F. Schneider (2006b), Welchen Umfang haben Schattenwirtschaft
und Schwarzarbeit? Ein Versuch zur Lösung des Rätsels. Wirtschaftsdienst, 3,
185–191.
Enste, D.H. and F. Schneider (2007), Schattenwirtschaft und Schwarzarbeit –
Von Mythen, Missverständnissen und Meinungsmonopolen. List Forum für
Wirtschafts- und Finanzpolitik, 33 (3), 251–286.
EU-Commission (2007), Undeclared work in the European Union. Special
EUROBAROMETER 284, Brussels.
Feld, L.P. and C. Larsen (2005), Black activities in Germany in 2001 and 2004. A
comparison based on survey data, Study. No. 12, The Rockwool Foundation
Feld, L.P. and C. Larsen (2006), Strafen, Kontrollen und Schwarzarbeit: Einige
Anmerkungen auf Basis von Befragungsdaten für Deutschland. Jahrbuch
Schattenwirtschaft 2006/2007. Münster.
Friedrichs, J. (1998), Methoden empirischer Sozialforschung. 15th edn, Wiesbaden.
Graf, G. (2007), Die volkswirtschaftliche Bedeutung von Schwarzarbeit. List
Forum für Wirtschafts- und Finanzpolitik, 33 (2), 106–128.
Graf, G. (2008), Stilisierte Fakten zum Zusammenhang zwischen Bargeld und
Schwarzarbeit. Working paper.
Institut für Demoskopie Allensbach (2007), Moral 2007- Sozialbetrug ist bei fast
allen verpönt, Schwarzarbeit nur bei wenigen. Allensbacher Berichte, No. 9.
Kazemier, B. (2006), Monitoring the underground economy: a survey of methods
and estimates. In D. Enste and F. Schneider (eds), Jahrbuch Schattenwirtschaft.
Münster: Wien, pp. 11–53.
Koch, W.A.S. (2008), Sisyphusarbeiten – Untersuchungen zu Schattenwirtschaft
und Schwarzarbeit. List Forum für Wirtschafts- und Finanzpolitik, 34 (2), 81–101.
Scandinavia: a measurement based on questionnaire surveys. The Rockwool
Foundation Research Unit Study, No. 10.


Schäfer, W. (1984), Schattenökonomie. Theoretische Grundlagen und wirtschaft-

spolitische Konsequenzen, Göttingen.
Schneider, F. (2005), Shadow economies around the world. European Journal of
Political Economy, 21(3), 598–642.
Schneider, F. (2007), Reducing the shadow economy in Germany: a blessing or a
curse? Discussion Paper, Department of Economics, Linz.
Schneider, F. and D.H. Enste (2000a), Shadow economies: size, causes, and conse-
Schneider, F. and D.H. Enste (2000b), Schattenwirtschaft und Schwarzarbeit
– Umfang, Ursachen, Wirkungen und wirtschaftspolitische Empfehlungen,
Munich.
Schneider, F. and D.H. Enste (2002), The shadow economy: an international
survey, Cambridge, MA.


PART IV
Tax morale and the shadow economy



10 Tax morale, tax evasion and the
shadow economy
Gebhard Kirchgässner
10.1 INTRODUCTION
Following a usual distinction, the shadow economy can be divided into

three parts: (1) pure tax evasion, that is, legal activities are legally per-
formed but no taxes are paid; (2) the black economy, where legal activi-
ties are illegally performed; and (3) criminal, illegal activities.1 Taxes are
evaded not only in the first area, but also whenever activities are illegally
performed. Thus, activities in the shadow economy are always connected
with tax evasion, and factors influencing the latter will also always have an
impact on the former.
If we put to one side financial psychology which has been all but for-
gotten today,2 moral factors have scarcely been taken into account when
the behaviour of taxpayers has been analysed.3 Following the economic
theory of criminal behaviour the extent to which citizens pay or evade
taxes depends on a simple cost benefit calculus: the additional income by
not paying taxes is compared with the expected punishment, that is, the
product of the expected punishment and the probability that this punish-
ment will be executed.4 As many investigations show, this model identifies
two important factors, but is unable to explain the behaviour of taxpayers:
given the low probability of detection citizens should pay much less tax
than they actually do.5
In order to correctly pay their taxes, individuals must have a motiva-
tion which includes more than the simple economic calculus; other factors
also play an important role. If we still assume that citizens’ behaviour is
rational, it has to be moral behaviour.6 Thus, it is not by chance that we
speak of tax morale, that is, we assume that citizens consider it – under
certain circumstances – as being their moral duty to pay their taxes even
if – from a purely economic point of view – it would be profitable to try to
evade them.
As McGee (2006) shows with reference to Crowe (1944), there exist
three ethical (moral) positions with respect to tax evasion: (1) it can be
347


judged as being categorically unethical, that is, there is no justification at

all for tax evasion; (2) it is generally allowed, because the government is
viewed as a thief which confiscates assets; and (3) that whether tax evasion
is ethical or unethical depends on the concrete circumstances. To the
latter belong, for example, whether the tax system is considered as fair or
unfair, what is done with the tax revenue and, as first mentioned by Weck-
Hannemann and Pommerehne (1989), how far taxpayers can influence
what is done with their tax money.7
What is, however, ‘tax morale’, and how can it be measured? In the
German-speaking countries, in particular, there is often a misunderstand-
ing because tax morale is (largely) equated with tax compliance. But while
the former is a moral attitude, the latter is concrete behaviour, and both
do not necessarily coincide. In the following, we first consider theoreti-
cal arguments on the role of tax morale, and under which conditions tax
evasion might be considered to be justified by citizens or not (Section
10.2). Then we ask how tax morale can be measured (Section 10.3). In
Section 10.4, we discuss the role of tax morale for the shadow economy.
In Section 10.5, determinants of tax morale and in Section 10.6, empirical
results of the effects of tax morale on tax compliance are discussed. The
results show that for a high tax morale, institutional and cultural factors
are at least as important as economic incentives. We conclude with some
remarks on citizens’ duties (Section 10.7).
As mentioned above, tax morale regards the question to what extent citi-
zens consider it their moral duty to pay taxes. Considering it as a moral
duty supposes that there are justified reasons to pay taxes besides the legal-
istic ones. What these reasons are, that is, under which conditions paying
taxes can be considered as a moral duty of citizens (and under which con-
ditions tax evasion might be justified) is not only a problem discussed in
economics, but has also been discussed for a long time in philosophy and,
in particular, in theology.8
Taxes are prices for goods and services provided by the government.
The relevant question for the moral evaluation of taxes is whether this
price corresponds to the value of these services, that is, whether it is ‘just’
in this sense. This approach of the ‘just price’ also holds for the principles
of taxation developed in the nineteenth century which today are still to
be found in textbooks of public finance, especially the equivalence prin-
ciple, but also the ability-to-pay principle.9 The fact that these services
are not distributed via a market and that, therefore, no market price


Tax morale, tax evasion and the shadow economy 349
exists made a simple transfer of the marginalist principle scarcely pos-

sible. Consequently, the original formulations of the principles of just
taxation refer rather to the average costs of the services which are publicly
provided.10
That with respect to this problem – as is often the case in economic
ethics – the discussion is based on the idea of the just price and, therefore,
on the concept of exchange justice, has first of all a practical reason: simi-
larly to the Pareto principle, which, however, is much less far-reaching,
there is no serious dispute about it. In a specific situation it might be dif-
ficult to determine what the real value of goods or services to be exchanged
is, but that any exchange should follow the principle that taking and giving
should be equivalent is hardly ever disputed.11
With respect to exchange justice both sides have to be considered. The
supplier of a good should get a price which covers at least his (justified)
costs. This also holds for those services which are publicly provided. Thus,
public employees have to earn an acceptable (just) wage, but the price
should not be too high, because public providers are wasting resources
at their disposal, in particular, taxpayers’ money. Those who use publicly
provided services have, first, the right to only pay for those services they
are interested in and, second, not to pay too high a price. The tax price
might, however, be different for different individuals, depending, for
example, on their income. Perhaps this should even be the case for going
along with the principle of vertical justice because the willingness to pay
for these services might – ceteris paribus – be a positive function of their
income.12 This makes it possible that – in principle – all citizens can agree
to the package of services and taxes provided by the government.13 This
leads, in principle, to lower and upper limits of the costs of publicly pro-
vided services and at least also to an upper limit of the extent of publicly
provided services.14
Following these considerations, levying taxes is morally justified and,
therefore, paying taxes a moral duty only as long as those services are pub-
licly provided for which such a demand exists on the part of the citizens
and if these services are provided efficiently. In all other cases tax evasion
might be punishable, but from the perspective of the just exchange paying
taxes is not a moral duty.15
These considerations are mainly based on the equivalence principle.
They are not only relevant for the supply of public goods but also for the
redistribution organised by the government as long as this redistribution
can be interpreted as being voluntary or as an insurance against risks in
cases where no private insurance exists (or private insurance cannot be
supplied without making deficits).16 For such situations, one might take
on the perspective of the original position in Rawls (1971). According to


this approach, publicly provided services and their financing by publicly

forced payments can be justified whenever the individuals behind the veil
of ignorance decide that it has to be done in this manner. How far this
holds for the redistribution actually performed today is, however, open.
As the discussion following the publication of his ‘Theory of Justice’
has shown, there is a large distance between redistribution according to
the difference principle of Rawls (1971) and the utilitarian principle of
expected utility maximisation favoured, for example, by Harsanyi (1975).
This distance is scarcely narrowed down if we employ the concept of the
original position, in particular, if we assume risk neutrality when apply-
ing the expected utility maximisation principle.17 Whichever principle
we use, however, it always holds that using the concept of the original
position taxation can be justified even if there is no exact equivalent for
every individual. Moreover, Rawls (1971) provides sound arguments that
individuals in the original position would choose the (or a variant of the)
ability-to-pay principle with progressive income taxation. Following these
arguments this principle can also be justified by the idea of just exchange.
This is not contradicted by the fact that, when discussing problems of
taxation, Rawls (1971, pp. 245ff.) does not refer to the traditional taxa-
tion principles but only to his difference principle, and that he argues that
– under certain conditions – even a proportional income taxation might
be compatible with this principle. On the other hand, he also mentions:
‘It does not follow that, given the injustice of existing institutions, even
steeply progressive income taxes are not justified when all things are con-
sidered.’ The question of whether the income tax tariff should be progres-
sive or not is for him a question ‘of political judgement and not part of
the theory of justice’ (pp. 246f.). The concept of a regressive income tax
would, on the other hand, violate his ideas of justice and is, therefore, not
considered at all.
A progressive income tax might, on the other hand, also be justified if
citizens do not (only) take their absolute income into account but their
relative income as well, that is, if they compare their own income with that
of their fellow citizens.18 Empirical results show that this usually holds.19
This also holds for Switzerland.20 Moreover, theory of optimal taxation
shows that under this condition higher marginal tax rates and, there-
fore, a progressive income tax schedule, is also appropriate for efficiency
reasons.21
When applying the concept of the original position the two conditions
mentioned above still hold: paying taxes is a moral obligation only as
far as the citizens (in the situation of the original position) demand that
the corresponding services are publicly provided and as long as these are
efficiently provided. Under these conditions evading taxes is breaking a


contract and might morally be qualified as being similar to theft. On the

other hand, if these conditions are violated, levying taxes is an illegitimate
infringement of the property of the citizens concerned and might also be
morally qualified as theft.
Such considerations can also be performed with respect to the concept
of a ‘fair tax system’. Citizens feel themselves more justified, the less they
perceive the tax system as fair. New developments of economic theory
have shown that arguments of fairness which traditionally have been
considered as being unimportant have strong impacts on the behaviour of
people.22 This holds, as shown below, in particular in the area of taxes.23 If
tax compliance is to be high, one of the most important aspects is that the
tax system should be perceived by the citizens as fair (or at least that it can
be perceived as being fair).
It is not only the fairness of the tax system itself but also the trust in
the political and legal system which matters. Citizens have to trust their
political representatives and the tax bureaucrats to use the tax revenue
responsibly. And they should have trust in the legal system that they will
be fairly treated should they ever have a dispute with the tax authori-
ties. Otherwise, they will hardly perceive paying taxes as a moral duty.
Correspondingly, Hammar, Jagers and Nordblom (2009, p. 238) conclude
that ‘it is important for politicians to be perceived as trustworthy in order
to be able to collect taxes’.
How people evaluate these questions also depends on the (religious) tra-
dition or the traditional relation between political and religious authorities
in the countries. It is to be noticed that tax morale is considerably lower
in the Romanic compared to the Nordic countries.24 This apparently
reflects a different attitude towards the government. One reason for this
might lie in the different religious traditions. Since the Middle Ages, in the
Romanic, mostly Catholic countries, there have quite often been disputes
between the government and the church, that is, between political and
religious authorities which, for example, were rather vigorous in France
and Italy at the end of the eighteenth and in the nineteenth centuries.
Partly, as for example in the Investiture Controversy, it was forbidden by
the Church to support the political authorities. In the Nordic countries,
which, on the other hand, had a nearly completely Protestant population,
political and religious authorities had been unified since the Reformation;
the king was not only the supreme political authority but also the head of
the state church, which partly (at least in a formal sense) still holds today.
Thus, an offence against the government was also a religious offence: a sin.
This had a strong impact on the population in these countries, and there
was no abrupt change in this respect once the importance of religion in
public (and private) life declined. Seen from this perspective it becomes


understandable that – ceteris paribus – tax evasion is more strongly

pronounced in southern compared to northern Europe.25
The opinions of citizens on how far these two conditions are fulfilled
and, therefore, whether paying taxes is a moral duty differ also largely
within the different societies. As we are never really in the situation of
the original position but are always in the ongoing political process citi-
zens might be strongly influenced by their social positions as well as their
income. Insofar as this may be the case, it is hardly possible to concede
single individuals an individual right of resistance which allows them to
take their own decision to deny payment of taxes, even if some groups
sometimes claim such a right for themselves.26 That, on the other side, such
considerations of justice play a role in political reality is not only shown
by auditing institutions whose role is to fight the waste of tax revenue but
also by a decision of the German Constitutional Court in 1995 accord-
ing to which a tax rate for capital income of more than 50 per cent is not
compatible with the German Constitution.27
The question of how far citizens are able to influence the use of their
tax money might also be important for their willingness to pay taxes. The
more impact the citizens have on the size of their taxes as well as their
usage, the more they can be convinced that they do not have to pay too
high taxes and that their taxes are not wasted. Thus, in more direct democ-
racies the willingness to pay taxes might – ceteris paribus – be higher and
tax evasion lower compared to purely representative systems.28 The federal
structure of a country might have a similar impact. The more decentralised
fiscal decisions are, the more influence citizens can exert and the more trust
they might have in their political institutions.29
Besides this, the willingness to pay taxes might also depend on the way
citizens are treated by the tax authorities. The stronger their feeling that
they are treated unfairly the more they believe in having a right to evade
taxes.30 Nevertheless, such an attitude is highly problematic because, as
long as it is a morally relevant problem, tax evasion is not a problem
between the single taxpayer and the tax authority, but between him/her
and the community of all citizens. On the other hand, tax bureaucrats have
the (moral) duty to treat taxpayers fairly.
10.3 MEASURING TAX MORALE
Before we can evaluate the impact of tax morale on tax compliance, we

have to measure it. This can be done by surveys. In the International
Social Science Panel (ISSP), for example, the following question is
employed:31 ‘A taxpayer does not report all of his or her income in order


to pay less income taxes. Do you feel it is (1) not wrong, (2) a bit wrong,
(3) wrong, (4) seriously wrong?’
If we allocate values between 1 and 4 to the different answers, 1 indicat-
ing the lowest and 4 the highest tax morale, we get an index with values
between 1 and 4.32 Switzerland, for example has a value of 2.65 which is
considerably below the average of 2.93 (with a standard deviation of 0.24).
The question in the World Value Survey (WVS) is slightly different:33
‘Please tell me for each of the following statements whether you think it
can always be justified, never be justified, or something in between: cheat-
ing on taxes if you have a chance?’
Those surveyed were given a 10 point scale where ‘1’ implied that tax
evasion can never and ‘10’ that it can always be justified. Again, it is pos-
sible to construct an index; its mean is 2.27 with a standard deviation of
0.69. Switzerland, for example, gets a value of 2.65 and, therefore, ranked
60 out of the 80 countries considered.34
10.4 THE ROLE OF TAX MORALE IN THE SHADOW

ECONOMY
A low tax morale should – ceteris paribus – have the effect of making
tax evasion high. Institutional (and other factors) might, however, have
the opposite effect. Therefore, a low tax morale does necessarily coincide
with high tax evasion. Thus, we have to ask for the effective impact of tax
morale on tax compliance, which is an empirical question. To answer this,
tax morale has to be recorded independently from tax compliance, and,
together with other possible impact factors, should be included in a model
which tries to explain the extent of tax evasion.
For the first time, such an investigation was performed by Weck (1983)35
where she used, however, not the extent of tax evasion but the size of
the shadow economy as dependent variable. To estimate the size of the
shadow economy of different countries in order to make an international
comparison, she uses the model approach for non-observable variables. It
is assumed that there are (measurable) impact factors as well as indicator
variables for the non-observable variable ‘size of the shadow economy’.
The impact factors are the input variables of the model, they determine
the size of the shadow economy, while the indicators are output variables,
which allow Weck to measure this size. The model is estimated using the
LISREL approach developed by Jöreskog and Thilo (1973) in order to
estimate ‘linear interdependent structural relationships’. It allows the
derivation of estimates of the size of the shadow economy.36
Weck (1983) uses as impact factors the shares of direct taxes (Tdir), of


indirect taxes (Tindir), and of social security contributions (SSC) in relation

to GDP, the perception of the tax burden, measured by the change of the
share of direct taxes (DTdir), the burden of government regulation (REG),
measured by the share of public employees in the labour force, real dispos-
able income per capita (YDR), a ten-year average of the unemployment
rate (UR) as well as estimates of the tax morale (TM) from survey data.37
Besides a stochastic term (e), these variables determine the size of the
shadow economy (S). Indicators are the age adjusted male participation
rate (MPR), the effective working time per week (TIME), as well as the
growth rate of the – officially reported – GDP (GRGDP), always in addition
to a stochastic term (di, i 5 1, 2, 3). Using data for 17 OECD countries and
for the years 1960, 1965, 1970, 1975 and 1978, she estimates the following
model:38
1 The structural model

0.419 Tdir 1 0.090 Tindir 2 0.113 SSC 2 0.042 DTdir
S5
(2.8) (1.1) (1.2) (0.6)
0.294 REG 1 0.136 YDR 2 0.078 UR 2 0.480 TM

1 1 e.
(2.4) (1.4) (1.1) (4.5)
(10.1)
2 The measurement model
MPR 5 20.080 S 1 d1, R2 5 0.621,

TIME 5 21.000 S 1 d2, R2 5 0.419,
(10.2)
GRGDP 5 20.198 S 1 d3, R2 5 0.976.
c2 5 61.5, DF 5 52.
In the structural model, only the burden of direct taxes, the extent of
regulation and tax morale have a significant impact. It is important to
note that the latter is significantly different from zero at a level less than
0.1 per cent: tax morale has the most significant impact on the size of the
shadow economy. Thus, it should be obvious that its impact should not
be neglected.
How important this impact is might be demonstrated by compar-
ing two countries in this sample, the Netherlands and Italy. Following
Frey and Weck-Hannemann (1984, p. 46), in 1978 Italy had a larger
shadow economy (in relation to GDP) than the Netherlands, although
the latter had a considerably higher burden of direct taxes of 16.1 per
cent and the higher share of public employees of 13.9 per cent. (The cor-
responding figures for Italy are 10.2 and 13.2 per cent, respectively.) The


result is mainly determined by the considerably higher tax morale in the

Netherlands.39 Without including this variable it would not make sense to
estimate this model. And this implies that it is hardly possible to capture
the causes of the shadow economy in an international comparative way
without taking into account this moral factor.40 Thus, also when using
the economic approach, it makes sense and is, in many situations, even
necessary to revert to the moral attitudes of the population if the size of
the shadow economy (and the extent of tax evasion) in a country is to be
explained. On the other hand, tax compliance (as well as abidance by other
laws) retroacts on tax morale.41 But then a more detailed investigation of
tax morale and its determinants seems to be appropriate.
Though its validity is not undisputed, in the last two decades this
approach has widely been used to estimate the size of the shadow
economy, in particular by Schneider and several collaborators.42 As far as
data are available, all these models use tax morale as one of the explana-
tory variables of the shadow economy. Relating to our question of inter-
est, whether tax morale has an effect on tax evasion (and the activities of
the shadow economy) the results of these studies can, however, at best be
judged as weak empirical evidence for such a relation but not as results
of strict tests.43 The estimated values of the size of the shadow economy
are linear combinations of variables including tax morale, that is, the
information included in this indicator is already employed by construct-
ing the dependent variable. It can, therefore, not be used a second time
to test whether such a relation exists. Such tests demand estimates of the
shadow economy (or the amount of evaded taxes) that are constructed
independently of this information.
10.5 DETERMINANTS OF TAX MORALE
What are, however, really the determinants of tax morale, that is, which
of the possible impact factors are not only plausible from a theoretical
point of view (and are, perhaps, morally justified), but are also empirically
relevant? More recently, quite a lot of investigations have been under-
taken, concerning rather different countries as well as societal groups.44
The necessary data have been collected by surveys and experiments; as far
as internationally comparative studies are performed the data mentioned
above collected in the World Value Survey and the International Social
Science Panel have been employed.
Following McGee (2005, p. 27), the fairness of the tax system is one of
the most important determinants of a high tax morale.45 It is, for example,
more important than the tax burden. Tax evasion is judged more to be


morally justified the more unjust and/or corrupt a government is per-

ceived. Those arguments, on the other hand, that do not relate to fairness,
are least acceptable as a moral justification of tax evasion.
‘Fairness’ is, of course, a broad term that can include quite a lot. If we
consider this in more detail, it becomes obvious that reciprocity plays an
important role: tax evasion seems to be more morally justified (objection-
able) the more others evade taxes as well (comply).46 Besides this, it is
important how the taxpayers’ money is used, whether it is beneficially
spent or futilely wasted, and whether the tax burden is considered appro-
priate for the services provided by the government. It is obvious that those
(moral) arguments discussed in the previous section really play a role for
tax morale.
While McGee is concerned with subjective evaluations directly related
to the tax system, Torgler considers, aside from individual factors such
as age, sex, marital status and employment status, which are not to be
discussed here, more general evaluation indicators such as trust in the gov-
ernment and in the legal system. Both have a significantly positive impact
on tax morale.47 The same holds for the degree of decentralisation.48
Besides this, cultural factors49 and, in particular, religious behaviour play
important roles: religious people have a higher tax morale.50 In his inves-
tigation for Canada (2003c, p. 296), Catholics have a lower tax morale
than Protestants, but it is open whether the difference is statistically sig-
nificant.51 On the other hand, the comparative study by Alm and Torgler
(2006) shows that those indicators which represent cultural factors have a
relatively high impact in ‘Catholic’ countries.
We take the results of Torgler (2005a) for Switzerland as an example.
He investigates the impact of direct democratic rights on tax morale.52
Thus, this paper is different from his other studies (as well as the papers
in McGee, 2008) insofar as, besides individual attributes, not only are
subjective evaluations included in the test equations but so is an ‘objective’
indicator of the political system. He uses the ISSP data mentioned above.
For Switzerland, he has 1068 observations surveyed in 1998. Dependent
variables are the coded answers to the question regarding tax morale. With
ordered probit he obtains the following estimates:53
TM 5 20.00004 P 2 0.001 FR 2 0.010 TR 1 0.104 DPR 1 0.093 TRU

(0.05) (0.66) (0.77) (3.41) (2.94)
1 0.085 CA 1 0.021 INC 1 . . . 1 e,

(4.75) (0.96) (10.3)
with:


TM tax morale;
P audit probability, approximated by the number of tax auditors per
1000 taxpayers in each canton;
SST fine rate, approximated by the standard legal fine as a multiple of
the evaded tax amount (in per cent) in the canton;
TR individual tax rate (in per cent);
INC the individual income class of the taxpayer (in 1000 CHF);
DPR index for the extent of direct popular rights in the canton;
TRU measure of confidence in the courts and the legal system (ISSP);
CA degree of church attendance (ISSP).
The index of direct popular rights was developed by Stutzer (1999) and
since then has often been employed.54 It measures the degree of participa-
tion, with ‘1’ reflecting the lowest and ‘6’ the highest. It encompasses four
sub-indices: (1) constitutional initiative, (2) legislative initiative, (3) legisla-
tive referendum, and (4) fiscal referendum. Besides these, other variables
such as age group, marital status and employment status are included in
the regression equation.
Out of these variables, only three are highly significant with a positive
impact on tax morale: confidence in the courts and the legal system, the
extent of direct popular rights and the degree of church attendance. On
the other hand, those variables which are assumed to be crucial for the
economic theory of tax evasion, the audit probability and the fine rate,
seem to have no impact at all on tax morale. The impacts of the other vari-
ables are also not statistically significant, even if the signs are plausible: a
higher tax burden tends to provide more justification not to pay taxes, and
higher income, usually providing more possibilities to evade taxes, which
are exploited at least partially, motivates people – at least according to the
theory of cognitive dissonance – to justify tax evasion by de-emphasising
its moral dimension.
In order to investigate which of the four different instruments that
are combined in the index of direct popular rights has an impact on tax
morale, Torgler (2005a) uses the sub-indices for additional estimations.
For the coefficients of the sub-indices he obtains the following results:
1. Constitutional initiative 0.051

2. Legislative initiative 0.064
3. Legislative referendum 0.088
4. Fiscal referendum 0.090
The coefficient of the constitutional initiative is at the 10 per cent level,

that of the legislative initiative at the 5 per cent level, and those of the two


referenda variables are at the 1 per cent level significantly different from
zero. This indicates that a direct democratic instrument contributes more
to raise tax morale, the more possibilities it provides for the citizens to
control public expenditure as well as the possibilities to raise taxes.
The question arises as to what extent these results might be generalised
and hold for other countries as well. With respect to direct political rights,
it might be argued that except for the United States there is no other
country in the world which has comparable direct rights at the state and
local governmental levels, and for the United States there are no compa-
rable studies available. Thus, generalisations might be problematic. On
the other hand, there is no reason why citizens of other countries should
behave totally differently. Thus, the hypothesis that the introduction and/
or extension of direct popular rights might raise tax morale in other coun-
tries as well can be kept, at least preliminarily. With respect to trust in the
legal system the situation is different. This has been investigated by Torgler
(2003a). Using the World Value Survey data for European countries and
the period from 1990 to 1993, he finds that, independent of the concrete
estimation procedure, trust in the legal system always had a highly signifi-
cant positive impact on tax morale. Using data from the Opinion Taxpayer
Survey 1987, he obtained the same result for trust in the legal system.55
He obtains similar results also for Eastern European countries (2003b),
Canada (2003c), as well as India and Japan (2004). Thus, it can be taken for
granted that trust in the legal system has a positive impact on tax morale.
10.6 TAX MORALE AND TAX EVASION
If we ask for the determinants of tax evasion, we might first think of the
individual attributes of taxpayers such as, for example, age, education,
employment and marital status. These variables are, however, rather unin-
teresting in this respect; they can hardly (or, at best, only in the very long
term) be influenced by policy measures. If we use micro-data, in order not
to bias the results, they are included in the regressions as control variables.
For similar reasons, when using macro-data, it makes sense to use their
means as controls. Besides these, there are individual data of interest, such
as tax morale in particular, but also trust in the political and legal system,
which however play hardly any role in the traditional economic approach
to tax evasion. They depend on institutional conditions which can be
changed and can, therefore, be influenced by policy measures. Thus, these
institutional conditions have also to be represented in the estimation
equations. Finally, there are the core variables of the traditional economic
approach: tax rate, probability of detection, fine rate, as well as income.


If we consider tax morale and other variables as determinants of tax

evasion, we have to distinguish between direct and indirect impacts. There
is, for example, the question of whether trust in the political system has
an impact on tax compliance, independent of the moral qualification of
tax evasion. On the other hand, it is open as to whether tax morale has an
independent impact once we account for all other possible determinants.
This could be tested in empirical analyses, but is hardly ever done.
Moreover, there are only relatively few reliable investigations con-
cerning the determinants of tax evasion.56 Sometimes micro-data can be
received from tax authorities.57 These data have, however, their prob-
lems. Because even strong audits cannot detect all evasions, these data
underestimate the actual amount.58 Aggregate data can be constructed by
comparing data for the national income (expenditure) from the national
account with income data collected from tax declarations.59 They rather
overestimate evasion because not every non-declared income is liable for
taxation.60 A third, but rather rare possibility, is tax amnesties.61 Again the
amount of evasion is underestimated because not all of those who have
evaded taxes so far are willing to reveal their real income in case of such
an amnesty.
To circumvent these problems, survey data62 and, more recently, data
from experiments have been used.63 But these data are also not without
problems. Even under optimal conditions, hardly any of those who evade
taxes will actually admit to this if asked. Moreover, if the effect of tax
morale on tax compliance is to be investigated, it would be preferable to
collect these two data sets independently of each other. Data from experi-
ments do not have minor problems. The subjects can lose money, which
might be interpreted as a fine, but, first, it is money given to them before
and not ‘own’ (earned) income, and, second, the possible loss of a small
amount of money can hardly be compared with a real fine or even impris-
onment. Thus, the results should be taken seriously, but rather carefully
interpreted.
The main result is, however, unambiguous: tax morale is a main
determinant of the willingness to pay taxes.64 As was to be expected, the
perceived fairness of the tax system plays an important role.65 Moreover,
there is also reversed causation from tax morale to the other determinants.
As Scholz and Pinney (1995) have shown, the higher the perceived risk of
being caught and punished when evading taxes, the higher the tax morale
of an individual. Finally, reciprocity is crucial. Citizens are more willing to
pay their taxes, the more they are convinced that their fellow citizens are
doing the same. Wenzel (2005b) shows that a systematic bias might exist:
they believe that others behave, on average, less morally than they do
themselves. If this bias is corrected, tax evasion declines.


Those studies which used real data from the tax authorities first asked
for the impact of those determinants that play a role in the traditional
model, in particular income and the marginal tax rate. The results are
heterogeneous. For the United States, Clotfelder (1983) finds a significant
impact of both variables, but Slemrod (1985) does not. Engel and Hines
(2000) find a significant effect for the audit probability, but not for the
marginal tax rate. Similarly, contradictory results with respect to fines
and the probability of detection have been derived for Switzerland. While
Weck-Hannemann and Pommerehne (1989) do not find a significant
impact of fines, Frey and Feld (2002) find a significant result. On the other
hand, they get a significantly ‘wrong’ sign for the probability of detection.
The two latter papers again employ, besides the variables of the eco-
nomic approach (and further controls), the extent of direct popular rights
as explanatory variable. In the paper by Frey and Feld (2002), which we
use as another example, treatment by the tax authorities is also taken into
account. The proxy to represent this has been constructed on the basis of
a survey among the cantonal tax authorities.66 Thus, they take two aspects
into account which are relevant for tax morale.
With data for the 26 Swiss cantons and five years in the period from
1970 to 1995, they have 130 observations. They obtain the following
result:67
TE 5 0.066 P 2 0.064 F 1 0.709 MTR 1 0.423 Y 2 1.038 TI
(2.74) (2.48) (4.92) (2.20) (0.91)
2 0.002 Pop 2 0.463 A65 2 0.678 SSE 1 0.403 SAS 2 7.432 Lat
(1.94) (1.95) (2.61) (2.03) (3.10)
2 2.291 ID 2 2.908 TypP 2 5.725 RespP 2 6.673 AutP 1 e. (10.4)

(3.14) (3.49) (3.84) (3.31)
R2 5 0.798,
with:
TE tax evasion;
P probability of detection;
F fine for tax evasion;
MTR maximum marginal tax rate;
Y gross effective primary income per capita in 1000 CHF;
TI dummy for tax indexation;
Pop cantonal population (in 1000);
A65 percentage of people over 65 years in the population (in per cent);


SSE share of self-employed from total employment (in per cent);

SAE share of employment in the agricultural sector (in per cent);
Lat dummy for French and Italian speaking cantons;
ID index of direct popular rights;
TypV index for the procedure if no taxes are declared;
RespP dummy variable for a respectful procedure by the tax authorities;
AutP dummy variable for an authoritarian procedure by the tax
authorities.
As already mentioned, both variables of the traditional approach are

significant, but the proxy for the probability of detection has the ‘wrong’
sign. This might indicate reversed causality: the higher the tax evasion is in
a canton, the more it pays to employ additional tax collectors. That a high
fine has a significant negative impact on tax evasion despite the fact that
it has (according to the result of Torgler, 2005a) no impact on tax morale,
reveals the difference between tax morale and tax compliance. Repressive
measures can, as other studies show as well, reduce the amount of tax
evasion,68 but this does not imply that citizens revise their moral convic-
tions. And whether this is the most efficient way to reduce tax evasion is,
at least concerning the results of Kucher and Götte (1998), questionable.
The index of direct democratic rights has a highly significant negative
impact on the amount of tax evasion. Insofar, we see the same result as in
the earlier papers by Weck-Hannemann and Pommerehne (1989) as well
as Pommerehne and Weck-Hannemann (1996). Those variables which
reflect the behaviour of the tax authorities are of particular interest in the
paper by Frey and Feld (2002).69 The more rigorous these authorities are
if a declaration is missing, the more taxes are evaded. If the authorities
detect an error in a declaration, respectful as well as authoritarian behav-
iour leads to less evasion, the former, one might guess, because trust in the
cantonal tax authorities (and, more generally, in the public bureaucracy) is
strengthened, the latter, because citizens fear repressive measures.70
The big advantage of these investigations for Switzerland, be it those
by Pommerehne and Weck-Hannemann or those by Feld and Frey is that
the independent variables are constructed using official data without using
indicators of tax morale; the condition that dependent and explanatory
variables have to be constructed independently holds. Thus, for testing
whether tax morale has an impact on tax compliance and, consequently,
also on the size of the shadow economy, these results are much more reli-
able than those of the model approach. We can, therefore, take for granted
that tax morale has a significantly positive impact on tax compliance.
While we can trust in this qualitative result, we have to be somewhat more
cautious with respect to the quantitative (numerical) results. The reason


Table 10.1 Size of the Swiss shadow economy (amount of undeclared

income) (as percentage of the official economy (total
income))
Year Method Size Source

1978 Model approach 4.3 Weck, Pommerehne and Frey
(1984, p. 67)
1990/93 Model approach 6.9 Schneider (2001, p. 22)
1980 Currency approach 6.3 Weck, Pommerehne and Frey
(1986, p. 32)
1990/93 Currency approach 6.9 Schneider and Enste (2000,
p. 102)
1970/95 Comparison of 23.48 Frey and Feld (2002, p. 36)
income declared and
expenditure
for this is that estimates regarding the amount of tax evasion as well as the
size of the shadow economy differ widely, depending on the measurement
method.71 Taking Switzerland as an example, Table 10.1 shows different
estimates for the period covered by Frey and Feld (2002). While the results
for the model and the currency approaches are rather similar, the com-
parison of income and expenditure leads to rather different figures, which
are about three to four times higher. These differences are too large to be
only a result of the overestimation by the latter method discussed above.
Thus, not the qualitative results, but the quantitative figures derived by
these approaches (and used in further studies) should be interpreted rather
cautiously.
10.7 CONCLUDING REMARKS
Taking all the results together, despite all the problems discussed above,
not only theoretical considerations, but also empirical investigations show
that tax morale has an important impact on the behaviour of taxpayers
and, therefore, also on the size of the shadow economy. Tax morale, on
the other hand, depends on institutional conditions, but also, on how citi-
zens are treated by tax authorities, by their trust in the political and the
legal system, and by the extent of direct popular rights: the more citizens
can take part in fiscal decisions, the more they are willing to make their
individual contributions in order to finance the government.72
Given the qualifications discussed in Section 10.2, paying taxes is
considered being a moral duty of citizens. Thus, this discussion is about


the duties of the individual and not about their rights. Not long ago, at
the beginning of the 1970s, people were still rather sceptical about such
arguments. Then, the emancipation of the individual was the main topic,
and the accent was on rights which – according to liberal conviction – are
inalienable properties of the individual. In contrast to this, when taking
up office on 20 January 1961, John F. Kennedy had already made the
following famous statement to his American co-citizens: ‘Don’t ask what
your country can do for you, ask what you can do for your country!’73 But
the political philosophy of that time also mainly asked for the individuals’
rights and under which conditions these rights might be restricted and
asked much less for the duties of the individuals to their community.74
In the meantime, the situation has changed. Firstly, somewhat more
than ten years ago, in addition to the catalogue of human rights, a cata-
logue of human duties was presented to the international discussion. Of
course, opinions about this enterprise are mixed.75 Secondly, the problem
of citizens’ duties is an important question discussed today within politi-
cal communitarism, a movement which is strongly influenced by Amitai
Etzioni.76 The first version of the ‘Communitarian Platform’ developed by
him contains the statement: ‘Paying one’s taxes and encouraging others to
pay their fair share . . . are fully obligatory’ (1993, p. 18).77 The solution
he strives for, to use moral suasion in order to encourage citizens to have
more civic duty might, however, not be very successful. The same holds for
the hope of Gaertner, that ‘the sense for fairness, the feeling for justice as
well as attention for and the compliance with laws’ might be encouraged
‘by investment in the moral education of all members of the society’ (1988,
p. 127). Most such attempts to influence the preferences have hardly been
successful in the past.
On the other hand, at the beginning of his Theory of Moral Sentiment,
Adam Smith (1759, p. 1) tells us that human beings are also moral beings,
that is, not totally reserved against moral arguments. How open they are
for such arguments might depend on the institutional conditions under
which they act, and recognising this is a precondition for really under-
standing the empirical results presented above. In order to reduce tax
evasion and following the economic approach, the appropriate way is to
set incentives so that citizens have as strong a motivation as possible to pay
taxes. However, as Allingham and Sandmo (1972) have already shown, a
strategy which solely relies on these incentives could only be successful
if the taxpayers were extremely risk averse, something they hardly are
now.78 That citizens pay considerably more taxes than is compatible with
only acting in narrow self-interest indicates that they are more strongly
motivated by moral reasons than is suggested by the (standard) economic
model of behaviour. Because this depends on societal conditions, policy


can still be successful even if it is (correctly) assumed that it is hardly pos-

sible to directly influence the preferences of the individual. If the willing-
ness to pay taxes is to be enforced, a responsible use of tax revenue by the
public authorities is necessary as well as a partnership relation (and not
a magisterial one) between them. Moreover, federal structures might be
strengthened by providing real tax autonomy to the lower governmental
levels as well as giving citizens at least some direct political rights with
respect to the determination of public finances. In all these respects,
however, especially with respect to federal structure and direct popular
rights, much could be done in many countries. The situation could be
improved, that is, tax evasion (and, correspondingly, also the shadow
economy) could be reduced if there were real attempts to do so.
NOTES
1. On the definition of the shadow or underground economy see, for example,

Kirchgässner (1984, p. 379).
2. On financial psychology see, for example, Schmölders (1981).
3. There is, for example, only a rather small section on moral aspects of taxpayer behav-
iour in the survey on tax compliance by Andreoni, Erard and Feinstein (1998). The
substantial literature on financial psychology is totally neglected. For an overview of
earlier contributions to tax morality see also Schöbel (2005).
4. On the (traditional) economic theory of tax evasion see, for example, Tanzi (1993), the
classical contribution by Allingham and Sandmo (1972) as well as Richter (2010).
5. This holds, at least as long as we do not assume extreme risk aversion. See, for the
United States, Alm, McClelland and Schulze (1992) or for Switzerland, Pommerehne
and Weck-Hannemann (1996). Graetz and Wilde (1985) call the statement that tax
evasion is a good example to apply the economic theory of criminal behaviour a ‘myth’.
6. On the definition of moral behaviour, see Kirchgässner (2010).
7. See also Torgler (2005a).
8. See also Kirchgässner (2003, pp. 221ff.).
9. See, for example, Musgrave and Musgrave (1984, pp. 227ff.). The basic considerations
can be traced back to the Nicomachean Ethics (350 BC [2009], p. 106) of Aristotle who,
when discussing his concept of ‘particular justice’, distinguishes between a ‘balancing’
and a ‘distributing’ justice, depending on whether the exchange of goods is discussed or
the distribution of means within a community. (See, for example, Hauser, 1974, p. 330.)
This distinction is taken up by Thomas Aquinas in his Summa Theologica (1265–1275
[1948], pp. 254ff.), in which he distinguishes between a ‘justitia commutativa’ and a
‘justitia distributiva’ (as well as a ‘justitia legalis’). (See Hauser, 1974, p. 333.)
10. On the history of the ability-to-pay principle see, for example, Koch (1981, pp. 225ff.).
Independent of this discussion, however, for the goods of public enterprises marginal
cost prices have been advocated in modern public finance. See for this, for example,
Blankart (1980, pp. 21ff.).
11. See for this Höffe (1990, pp. 92f.), but also Höffe (1987, pp. 382ff.; 1994). He considers
the principle of exchange justice as being a general principle of justice which also might
be transferred to the political area. He complements it, however, with a principle of cor-
rective justice. See for this Höffe (1994).
12. For the distinction between horizontal and vertical justice in relation to the ability-to-
pay principle see, for example, Musgrave and Musgrave (1984, pp. 232f.).


13. This has already been mentioned by Wicksell (1986). Buchanan based his approach of
‘constitutional economics’ on this idea assuming that – in principle – it should be pos-
sible to reach such a general consensus. See for this Buchanan (1967, pp. 114ff.) as well
as Buchanan (1987).
14. There are also arguments in favour of a lower limit of these services. (See, for example,
Galbraith (1958) and his arguments about ‘public poverty and private wealth’.) Taking
into account the strong increase of the government share since the 1960s this might
today, however, be of secondary importance in the developed industrial countries of the
Western world.
15. Insofar as tax laws are pure ‘penal laws’, the government might force citizens to obey,
but it does not bind their conscience. As Schmölders (1951, pp. 18f.) shows, the moral
theory of the Scholastic already knows this conception.
16. Following Sinn (1997) most of the services provided by governments can be interpreted
as being just such insurances, that is, it does not only hold for explicit redistribution,
but also for publicly provided goods and services. Moreover, these public services do
only to a small extent crowd out private insurances because ‘the by far largest part of
the social security system covers risks for which no private insurance is available’ (Sinn,
1996, pp. 263). This does not, however, imply that (all) individuals would voluntarily
take out such an insurance. For the discussion of arguments in favour of voluntary
redistribution see, for example, Kirchgässner and Pommerehne (1992). For the rel-
evance of distributive justice for tax compliance see also Verboon and van Dijke (2007).
17. For the presentation of the two principles see, for example, Frey and Kirchgässner
(2002, pp. 35f., pp. 262ff.).
18. See Balestrino (2009).
19. See, for example, Easterlin (2001).
20. See Dorn et al. (2008).
21. See Boskin and Sheshinski (1978) as well as Ireland (2001).
22. See, for example, Kahnemann, Knetsch and Thaler (1986), Broome (1990), Konow
(2003) or Hooker (2005).
23. See, for example, Bordignon (1993) or Vihanto (2003).
24. According to the estimates of Weck (1983) the index of tax immorality in 1978 was 20.6
in Italy, 17.3 in France and 14.0 in Spain. In Sweden, Denmark and Finland, it was,
however, only 4.4. Germany and the Netherlands had a middle position with 10.9.
25. This ‘north–south gradient of tax morality’ is, as Schmölders (1981, p. 131) shows, also
reflected in the languages of the different countries.
26. In the 1980s, some German citizens claimed to have a right of resistance in the form of a
tax boycott against the use of tax revenue for defence purposes. This claim was rejected
by the Financial Court of Cologne in its decision of 15 November 1984 (V K 223/84).
See Selmer (1986).
27. BVerfG 93, 121; 93, 165. The Constitutional Court states the ‘principle of property
preserving taxation’. See for this Tipke and Lang (1973, pp. 121ff.).– For Switzerland,
Kleinewefers proposed the following article for the constitution: ‘(1) Each individual
has the right to the results of his work and his property. (2) Nobody’s current income
is to be burdened by direct and indirect publicly forced payments without individually
equivalent services in return to more than one half.’ Thus, he assumes that today this
limit is often transgressed. See: Kleinewefers (1999).
28. See Pommerehne, Hart and Frey (1994) as well as Pommerehne, Hart and Feld (1997).
29. See Oates (1985, p. 749).
30. See – in a more general connection – Tyler (1997): The more fair the citizens feel they
are treated, the more they are prepared to fulfil legal obligations.
31. These dates relate to 1998. See: http://www.gesis.org/dienstleistungen/daten/umfrag
edaten/issp/modules-study-overview/religion/1998/ (04/03/10) (question V16 for tax
morale).
32. Here, we consider averages of the countries. The whole sample of the ISSP-data
comprises 39,034 persons.


33. See Inglehart et al. (2004, question F116). These data, which relate to the years 1999 to
2002, comprise 89,678 persons.
34. A more detailed analysis of the data shows that the rank correlation for those 28 coun-
tries which are included in both surveys is only 0.546 and, therefore, relatively small.
This casts some doubt on the validity of these data. Nevertheless, these surveys provide
the only data available for international comparisons.
35. See also Frey and Weck-Hannemann (1984).
36. Schneider calls this the MIMIC (Multiple-Indicators, Multiple-Causes) or DYMIMIC
(Dynamic Multiple-Indicators, Multiple-Causes) approach. See, for example, Dell’
Anno and Schneider (2006) or Schneider (2005), respectively.
37. Actually, she uses an ‘index of tax immorality’. For the construction of this variable see
Weck (1983, p. 91, p. 110) as well as Weck, Pommerehne and Frey (1984, p. 57). With
respect to its content this does not, however, change anything if it is taken into account
that the sign of the coefficient is reversed.
38. See Weck (1983, p. 112) as well as Frey and Weck-Hannemann (1984, p. 40), who use
the same model. The numbers in parentheses are the absolute values of the t-statistics
of the estimated parameters. DF is the number of degrees of freedom. The variables are
standardised.
39. See for this Weck (1983, p. 119, p. 133). To estimate the size of the shadow economies
in the different countries only the significant variables of the structural model are used.
40. The estimated coefficients are biased whenever a relevant explanatory variable is
excluded from the regression as long as this variable is correlated with the included
variables.
41. See Orviska and Hudson (2002) as well as Wenzel (2005).
42. See, for example, Schneider (2005), Schneider and Enste (2000, 2002), Chaudhuri,
Schneider and Chattopadhyay (2006), or Torgler and Schneider (2009). For the debate
as to whether this approach is appropriate see, for example, Breusch (2005) and the
response by Dell’Anno and Schneider (2006).
43. This also holds for those papers where (DY)MIMIC estimates derived in other papers
are used to explain the size of the shadow economy using usual econometric methods,
as long as tax morale is used in the (DY)MIMIC approach to construct the depend-
ent variables employed in these other papers. See, for example, Torgler and Schneider
(2009).
44. See, for example, Scholz and Pinney (1995), Torgler (2003a, 2003b, 2003c, 2004, 2005a),
Torgler and Schneider (2007), Torgler and Schaltegger (2006), Wu and Teng (2005), or
McGee (2005, 2006, 2008).
45. The results of McGee (2005) are based on surveys among members of internationally
active manager societies. He obtains very similar results from students of rather dif-
ferent countries. Many of these studies are published in McGee (2008). On the role of
fairness see also Forest and Sheffrin (2002) as well as Bobeck and Hatfield (2003).
46. See for this also Frey and Torgler (2007).
47. See Wenzel (2004), who shows that tax morale is higher, the more citizens identify with
their political community, as well as Wu and Teng (2005), who point to the impact of
corruption.
48. See Torgler (2005b), Torgler and Werner (2005), but also Güth, Levati and Sausgruber
(2005).
49. See, for example, Alm and Torgler (2006) or Torgler and Schneider (2007).
50. See Torgler (2006) in an internationally comparative study covering 30 countries or
Frey and Torgler (2007) looking at several East- and West-European countries. On
the relation between tax morale and religiosity in Germany see Prinz (2004) as well as
Feld, Torgler and Dong (2008). Prinz shows that religious attitudes strongly influence
tax morale, with a significant difference between West and East Germany. In West
Germany, religious education seems to be most important, while in East Germany, it is
personal conviction. Feld, Torgler and Dong (2008) show that church attendance had a
positive and highly significant impact on tax morale in 1990 and 1997, while its impact


in 1999 was insignificantly negative. This difference casts, however, severe doubts on
the validity of these estimates, because it is hardly plausible that the preferences of the
German population changed that drastically within only two years.
51. The respective coefficients have opposite signs but are not significantly different from
zero. Whether the difference is statistically significant has not been tested.
52. Similar results are presented in Torgler (2007).
53. The numbers in parentheses are the absolute values of the z-statistics of the estimated
parameters. Similar results are provided by Torgler and Schneider (2007). In contrast
to here, they find a highly significant coefficient of the penalty tax.
54. See the description of this index in the appendix of Stutzer (2003).
55. There, trust in public officials is the explanatory variable. On the Taxpayer Opinion
Survey, see Harris and Associates (1988). For a survey about papers which set the
theme ‘trust’ in relation to problems of public finances, see Slemrod (2003).
56. Surveys are given, for example, by Pommerehne (1985), Pommerehne and Weck-
Hannemann (1992), Andreoni, Erard and Feinstein (1998) as well as Torgler (2001).
57. See, for example, Clotfelter (1983), Slemrod (1985) or Wenzel (2005b).
58. For this, see Andreoni, Erard and Feinstein (1998, p. 836).
59. See, for example, the two papers on Switzerland by Weck-Hannemann and
Pommerehne (1989) as well as by Frey and Feld (2002).
60. For this, see Slemrod and Yitzhaki (2000). In Switzerland, such overestimations occur,
for example, because private capital gains are not liable for taxation. They do, however,
hardly count for a very large part of the difference between declared income and actual
expenditure.
61. See, for example, Crane and Nourzad (1990) or Alm and Beck (1992).
62. See, for example, Ovriska and Hudson (2002), Forest and Sheffrin (2002) or Hammar,
Jagers and Nordblom (2009). On the measurement of tax compliance using survey data,
see also Kirchler and Wahl (2010).
63. See, for example, Alm, Jackson and McKee (1992), Bosco and Mittone (1997), Feld
and Tyran (2002), Park and Hyun (2003), Cummings et al. (2009), as well as the survey
by Torgler (2002).
64. See, for example, Reckers, Sanders and Roark (1994), Ovriska and Hudson (2002) or
Bobek and Hatfield (2003).
65. See, for example, Forest and Sheffrin (2002) or Bobek and Hatfield (2003).
66. A problem with this data is, however, that – depending on the procedures in the differ-
ent cantons – most taxpayers have no contact at all with the cantonal tax authorities
but exclusively with the local tax authorities. This also holds for most of those taxpayers
whose declarations are – for various reasons – not processed by local but by cantonal
employees. On the other hand, cantonal tax authorities have an impact on local author-
ities, and some ‘big’ taxpayers might negotiate directly with the cantonal authorities.
67. Equation (4) in Table 2, p. 20, in Frey and Feld (2002). The numbers in parentheses are
the absolute values of the t-statistics of the estimated parameters. The equation does
not include a constant term because dummy variables for the five points in time are
included.
68. See, for example, Ledermann (2003).
69. See Feld and Frey (2001).
70. The estimated impact of respectful behaviour might, however, also reflect reversed cau-
sality. The less taxes are evaded in a canton, the less the tax collectors might feel induced
to be very strict. In a further equation using interaction terms, Frey and Feld (2002, p.
20, equation (5)) try to show that authoritarian behaviour of the tax authorities is more
likely in ‘representative’ and respectful behaviour is more likely in ‘direct’ democracies.
The results are, however, not so clear-cut that this conclusion can be drawn.
71. For a description of the different methods, see, for example, Schneider (2001).
72. See also Bird, Martinez-Vazquez and Torgler (2007) who show for developing countries
that accountability of the government has a positive and corruption a negative impact
on tax effort.


73. See Kennedy (1962, p. 28).

74. This holds, for example, for the basic works of the ‘new contractarians’ Nozick (1974)
and Buchanan (1975). Rawls (1971), on the other hand has a chapter about ‘duty
and obligation’ (pp. 368ff.). For an introduction into these theories see, for example,
Gordon (1976).
75. See Allgemeine Erklärung der Menschenpflichten, Die Zeit No. 41 October 3, 1997;
Schmidt (1997) Zeit von den Pflichten zu sprechen, ibid., pp. 17f.; as well as the discus-
sion following the article in this newspaper.
76. See, for example, Etzioni (1993, 1995).
77. See also http://www.gwu.edu/~ccps/platformtext.html (Accessed 16 March 2010).
78. For more realistic theoretical models including psychological costs or ethical prefer-
ences see, for example, Schnellenbach (2006, 2007) or Eisenhauer (2008).
REFERENCES
Alm, J. and W. Beck (1993), Tax amnesties and compliance in the long-run: a time
series analysis. National Tax Journal, 46, 53–60.
Alm J.and B. Torgler (2006), Cultural differences and tax morale in the United
States and in Europe. Journal of Economic Psychology, 27, 234–246.
Alm, J., B.R. Jackson and M. McKee (1992). Estimating the determinants
of taxpayer compliance with experimental data. National Tax Journal, 45,
107–114.
Alm, J., G.H. McClelland and W.D. Schulze (1992). Why do people pay taxes?
Andreoni, J.B., B. Erard and J. Feinstein (1998), Tax compliance. Journal of
Aquinas, T. (1265–1274 [1948]), Summa Theologica. New York: Benzinger Bros.
Aristotle (350 BC [2009]), Nicomachean Ethics. Oxford, UK: Oxford University
Press.
Balestrino, A. (2009), Tax avoidance, endogenous social norms, and the compari-
son income effect. ChilD Working Paper No. 15/2009. University of Pisa, Centre
for Household, Income, Labour and Demographic Economics.
Bird, R.M., J. Martinez-Vazquez and B. Torgler (2007), Tax effort: the impact of
corruption, voice and accountability. CREMA Working paper No. 2007-13,
Basel.
Blankart, Ch. B. (1980), Ökonomie der öffentlichen Unternehmen. Vahlen,
München.
Bobek, D.D. and R.C. Hatfield (2003), An investigation of the theory of planned
behavior and the role of moral obligation in tax compliance. Behavioral
Research in Accounting, 15, 13–38.
Bordignon, M. (1993), A fairness approach to income tax evasion. Journal of
Public Economics, 52, 345–362.
Bosco, L. and L. Mittone (1997), Tax evasion and moral constraints: some experi-
mental evidence. Kyklos, 50, 297–324.
Boskin, M. and E. Sheshinski (1978), Optimal redistributive taxation when indi-
vidual welfare depends upon relative income. Quarterly Journal of Economics,
92, 589–601.



Mimeo, Australian National University, Canberra.
Broome, D.J. (1990), Fairness. Proceedings of the Aristotelian Society, 91,
87–102.
Buchanan, J.M. (1967), Public Finance in Democratic Process: Fiscal Institutions
and Individual Choice. Chapel Hill: University of North Carolina Press.
Buchanan, J.M. (1975), The Limits of Liberty. Chicago: University of Chicago
Press.
Buchanan, J.M. (1987), The constitution of economic policy. American Economic
Review, 77, 243–250.
Chaudhuri, K., F. Schneider and S. Chattopadhyay (2006), The size and develop-
ment of the shadow economy: an empirical investigation from States of India.
Journal of Development Economics, 80, 428–443.
Clotfelter, Ch.T. (1983), Tax evasion and tax rates: an analysis of individual
returns. Review of Economics and Statistics, 65, 363–373.
Crane, S.E. and F. Nourzad (1990), Tax rates and tax evasion: evidence from
California amnesty data. National Tax Journal, 43, 189–199.
Crowe, M.T. (1944), The moral obligation of paying just taxes. The Catholic
University of America Studies in Sacred Theology No. 84.
Cummings, R.C., J. Martinez-Vasquez, M. McKee and B. Torgler (2009), Tax
morale affects tax compliance: evidence from surveys and an artefactual field
experiment. Journal of Economic Behaviour and Organization, 70, 447–457.
Dell’Anno, R. and F. Schneider (2006), Estimating the underground economy
using MIMIC models: a response to T. Breusch’s critique. Working Paper No.
0607. Johannes Keppler University of Linz, Department of Economics.
Dorn, D., J.A.V. Fischer, G. Kirchgässner and A. Sousa-Poza (2008), Direct
democracy and life satisfaction revisited: new evidence for Switzerland. Journal
of Happiness Studies, 9, 227–255.
Easterlin, R. (2001), Income and happiness: towards a unified theory. Economic
Journal, 111, 465–484.
Eisenhauer, J.E. (2008), Ethical preferences, risk aversion, and taxpayer behavior.
Journal of Socio-Economics, 37, 45–63.
Engel, E.M.R.A. and J.R. Hines (2000), Understanding tax evasion dynamics.
Econometric Society World Congress 2000 Contributed Paper 1117.
Etzioni, A. (1993), The Spirit of Community: Rights, Responsibilities, and the
Communitarian Agenda. New York: Crown Publishers.
Etzioni, A. (1995), On restoring the moral voice. In A. Etzioni (ed.), Rights and the
Common Good: The Communitarian Perspective. New York: St. Martin’s Press,
271–276.
Feld, L.P. and B.S. Frey (2001), Trust breeds trust: how taxpayers are treated.
Economics of Governance, 2, 87–99.
Feld, L.P. and J.-R. Tyran (2002), Tax evasion and voting: an experimental analy-
sis. Kyklos, 55, 197–222.
Feld, L.P., B. Torgler and B. Dong (2008), Coming closer? Tax morale, deterrence
and social learning after German Unification. CREMA Working Paper No.
2008-9, Basel.
Forest, A. and S.M. Sheffrin (2002), Complexity and compliance: an empirical
investigation. National Tax Journal, 55, 75–88.
Frey, B.S. and L.P. Feld (2002), Deterrence and morale in taxation: an empirical
analysis. CESifo Working Paper No. 760, München.


Frey, B.S. and G. Kirchgässner (2002), Demokratische Wirtschaftspolitik: Theorie

und Anwendung, 3rd edn. München: Vahlen.
Frey, B.S. and B. Torgler (2007), Tax morale and conditional cooperation. Journal
of Comparative Economics, 35, 136–159.
Frey, B.S. and H. Weck-Hannemann (1984), The hidden economy as an unob-
served variable. European Economic Review, 26, 33–53.
Gaertner, W. (1988), Untergrundwirtschaft, Steuerhinterziehung und Moral.
In H. Hesse (ed.), Wirtschaftswissenschaft und Ethik. Berlin: Duncker und
Humblot, 109–130.
Galbraith, J.K. (1958), The Affluent Society. Boston: Houghton Mifflin.
Gordon, S. (1976), The new contractarians. Journal of Political Economy, 84,
573–590.
Graetz, M.J. and L.L. Wilde (1985), The economics of tax compliance: facts and
fantasy. National Tax Journal, 38, 355–363.
Güth, W., V. Levati and R. Sausgruber (2005), Tax morale and (de-)centralisation:
an experimental study. Public Choice, 125, 171–188.
Hammar, H., S.C. Jagers and K. Nordblom (2009), Perceived tax evasion and the
importance of trust. Journal of Socio-Economics, 38, 238–245.
Harris, L. and Associates (1988), 1987 taxpayer opinion survey, conducted for
the US internal revenue service. Internal Revenue Service Document No. 7292,
Washington DC.
Harsanyi, J.C. (1975), Can the maximim principle serve as a basis of morality?
A critique of John Rawls’s theory. American Political Science Review, 69,
594–606.
Hauser, R. (1974), Gerechtigkeit. In J. Ritter (ed.), Historisches Wörterbuch der
Philosophie. Basel/Stuttgart: Band 3, Schwabe und Co., 329–335.
Höffe, O. (1987), Politische Gerechtigkeit: Grundlegung einer kritischen Philosophie
von Recht und Staat. Frankfurt: Suhrkamp.
Höffe, O. (1990), Gerechtigkeit als Tausch: Ein ökonomisches Prinzip für die
Ethik. In P. Ulrich (ed.), Auf der Suche nach einer modernen Wirtschaftsethik:
Lernschritte zu einer reflexiven Ökonomie. Bern: Haupt, 91–102.
Höffe, O. (1994), Tauschgerechtigkeit und korrektive Gerechtigkeit:
Legitimationsmuster für Staatsaufgaben. In D. Grimm (ed.), Staatsaufgaben.
Baden-Baden: Nomos, 713–737.
Hooker, B. (2005), Fairness. Ethical Theory and Moral Practice, 8, 329–352.
Inglehart, R., M. Basáñez, J. Diez-Medrano, L. Halman and R. Luikx (2004),
Human Beliefs and Values: A Cross-Cultural Sourcebook Based on the 1999–2002
Values Surveys. Mexico: Siglo XXI Editores.
Ireland, N. (2001), Optimal tax in the presence of status effect. Journal of Public
Economics, 81, 193–212.
Jöreskog, K.G. and M. van Thillo (1973), LISREL: a general computer program
for estimating a linear structural equation system involving multiple indica-
tors of unobserved variables. Research Report 73-5, University of Uppsala,
Department of Statistics.
Kahnemann, D., J.L. Knetsch and R.H. Thaler (1986), Fairness and the assump-
tions of economics. Journal of Business, 59, S285–S300.
Kennedy, J.F. (1962), To Turn the Tide: The Major Addresses and Statements of
President John F. Kennedy. New York: Harper and Brothers.
erarbeiteten Sozialprodukts. Allgemeines Statistisches Archiv, 68, 378–405.


Kirchgässner, G. (2003), Moralische Aspekte der Besteuerung. In M. Rose (ed.),

Integriertes Steuer- und Sozialsystem. Heidelberg: Physica, pp. 215–241.
Kirchgässner, G. (2010), On minimal morals. European Journal of Political
Economy, 26(3), 330–339.
Kirchgässner, G. and W.W. Pommerehne (1992), Approaches to explain fiscal
redistribution in democracy. Mimeo, Hochschule St. Gallen.
Kirchler, E. and I. Wahl (2010), Tax compliance inventory TAX-I: designing an
inventory for surveys of tax compliance. Journal of Economic Psychology, 31(3),
331–346.
Kleinewefers, H. (1999), Das vergessene Grundrecht: Für ein Recht auf den Ertrag
der eigenen Arbeit und des eigenen Vermögens. Neue Zürcher Zeitung, 12, 29.
Koch, W. (1981), Finanzwissenschaft I: Geschichte. In W. Albers et al. (eds),
Handwörterbuch der Wirtschaftswissenschaft. Stuttgart: Band 3, Gustav Fischer
et al., 211–234.
Konow, J. (2003), Which is the fairest one of all? A positive analysis of justice
theory. Journal of Economic Literature, 41, 1186–1237.
Kucher, M. and L. Götte (1998), Trust me: an empirical analysis of taxpayer
honesty. Finanzarchiv, 54, 429–444.
Ledermann, L. (2003), The interplay between norms and enforcement in tax
compliance. Law and Economics Working Paper No. 03-12, Georg Mason
University.
McGee, R.W. (2005), The ethics of tax evasion: a survey of international busi-
ness academics. Paper presented at the 60th International Atlantic Economic
Conference, New York, October 6–9, 2005.
McGee, R.W. (2006), Three views on the ethics of tax evasion. Journal of Business
Ethics, 67, 15–35.
McGee, R.W. (2008), Taxation and Public Finance in Transition and Developing
Countries. New York: Springer.
Musgrave, R.A. and P.B. Musgrave (1984), Public Finance in Theory and Practice,
4th edn. New York: McGraw-Hill.
Nozick, R. (1974), Anarchy, State, and Utopia. Oxford, UK: Basil Blackwell.
Oates, W.E. (1985), Searching for leviathan: an empirical study. American
Economic Review, 75, 748–757.
Orviska, M. and J. Hudson (2002), Tax evasion, civic duty and the law abiding
citizen. European Journal of Political Economy, 19, 83–102.
Park, C.-G. and J.K. Hyun (2003), Examining the determinants of tax compli-
ance by experimental data: a case of Korea. Journal of Policy Modeling, 25,
673–684.
Pommerehne, W.W. (1985), Was wissen wir eigentlich über Steuerhinterziehung?
International Review of Economics and Business, 32, 1155–1186.
Pommerehne, W.W. and H. Weck-Hannemann (1992), Steuerhinterziehung:
Einige romantische, realistische und nicht zuletzt empirische Befunde. Zeitschrift
für Wirtschafts- und Sozialwissenschaften, 112, 433–466.
Pommerehne, W.W. and H. Weck-Hannemann (1996), Tax rates, tax administra-
tion and income tax evasion in Switzerland. Public Choice, 88, 161–170.
Pommerehne, W.W., A. Hart and B.S. Frey (1994), Tax morale, tax evasion and
the choice of instruments in different political systems. Public Finance, 49, 52–69.
Pommerehne, W.W., A. Hart and L.P. Feld (1997), Steuerhinterziehung und ihre
Kontrolle in unterschiedlichen politischen Systemen. Homo Oeconomicus, 14,
469–478.


Prinz, A. (2004), Steuermoral und Religiosität in Ost- und Westdeutschland.

Schmollers Jahrbuch, 124, 511–537.
Rawls, J. (1971), A Theory of Justice. Cambridge, MA: Harvard University Press.
Reckers, P.M.J., D.L. Sanders and S. Roark (1994), The influence of ethical atti-
tudes on taxpayer compliance. National Tax Journal, 47, 825–836.
Richter, W. (2010), Geplante Steuerhinterziehung und ihre effiziente Bestrafung.
In B.-A. Wickström (ed.), Finanzpolitik und Schattenwirtschaft. Berlin: Duncker
und Humblot, 67–92.
Schmidt, M. (1997), Zeit von den Pflichten zu sprechen. Die Zeit, 41 (3 October),
17.
Schmölders, G. (1951), Finanzpsychologie. Finanzarchiv, 13, 1–36.
Schmölders, G. (1981), Finanz- und Steuerpsychologie. In W. Albers et al. (eds),
Handwörterbuch der Wirtschaftswissenschaft. Stuttgart: Band 3, Gustav Fischer
et al., 121–133.
Schneider, F. (2001), What do we know about the shadow economy: evidence from
21 OECD countries, World Economics, 2, 19–32.
Schneider, F. and E. Enste (2000), Shadow economies: size, causes, and conse-
Schneider, F. and E. Enste (2002), The Shadow Economy: An International Survey.
Cambridge, UK: Cambridge University Press.
Schnellenbach, J. (2006), Tax morale and the taming of leviathan. Constitutional
Political Economy, 17, 117–132.
Schnellenbach, J. (2007), Tax morale and the legitimacy of economic policy. Homo
Oeconomicus, 17, 21–45.
Schöbel, E. (2005), Otto Veit and subsequent developments of tax morality. In
J.G. Backhaus (ed.), Essays on Fiscal Sociology. Bern: Peter Lang, 187–210.
Scholz, J.T. and N. Pinney (1995), Duty, fear, and tax compliance: the heuristic
basis of citizenship behavior. American Journal of Political Science, 39, 490–512.
Selmer, P. (1986), Kein Recht zum Steuerboykott wegen Rüstungsausgaben.
Juristische Schulung (JUS), 20, 226.
Sinn, H.-W. (1996), Social insurance, incentives and risk taking. International Tax
and Public Finance, 3, 259–280.
Sinn, H.-W. (1997), Das Selektionsprinzip und der Steuerwettbewerb. In A.
Oberhauser (ed.), Fiskalföderalismus in Europa. Berlin: Duncker und Humblot,
9–60.
Slemrod, J. (1985), An empirical test for tax evasion. Review of Economics and
Statistics, 67, 232–238.
Slemrod, J. (2003), Trust in public finance. In S. Cnossen and H.-W. Sinn (eds),
Public Finance and Public Policy in the New Century. Cambridge, MA: MIT
Press, 49–88.
Slemrod, J. and Y. Yitzhaki (2000), Tax avoidance, evasion, and administration.
NBER Working Paper No. 7473, January 2000.
Smith, A. (1759), Theory of Moral Sentiments. London: Millar.
Stutzer, A. (1999), Demokratieindizes für die Kantone der Schweiz. IEW Working
Paper No. 23, Zürich, October 1999.
Stutzer, A. (2003), Eine ökonomische Analyse menschlichen Wohlbefindens. Aachen:
Shaker.
Tanzi, V. (1993), A primer on tax evasion. IMF Staff Papers, 40, 807–829.


Tipke, K. and J. Lang (1973), Steuerrecht, 15th edn. Köln: Otto Schmid.
Torgler, B. (2001), What do we know about tax morale and tax compliance?
International Review of Economics and Business, 48, 395–419.
Torgler, B. (2002), Speaking to theorists and searching for facts: tax morale and
tax compliance in experiments. Journal of Economic Surveys, 15, 657–683.
Torgler, B. (2003a), Tax morale, rule-governed behaviour and trust. Constitutional
Torgler, B. (2003b), Tax morale in transition countries. Post-Communist
Economies, 15, 357–381.
Torgler, B. (2003c), To evade taxes or not to evade: that is the question. Journal of
Socio-Economics, 32, 283–302.
Torgler, B. (2004), Tax morale in Asian countries. Journal of Asian Economics, 15,
237–266.
Torgler, B. (2005a), Tax morale and direct democracy. European Journal of
Torgler, B. (2005b), Tax morale and fiscal autonomy: evidence from Germany.
Public Finance and Management, 5, 460–485.
Torgler, B. (2006), The importance of faith: tax morale and religiosity. Journal of
Economic Behavior and Organization, 61, 81–109.
Torgler, B. (2007), The impact of direct democracy and local autonomy on
tax morale in Switzerland. Discussion paper, The School of Economics and
Finance, Queensland University of Technology, Institute of Local Public
Finance, Brisbane, December 2007.
Torgler, B. and Ch.A. Schaltegger (2006), Tax morale: a survey with a special focus
on Switzerland. Schweizerische Zeitschrift für Volkswirtschaft und Statistik, 142,
395–425.
Torgler, B. and F. Schneider (2007), What shapes attitudes toward paying taxes?
Evidence from multicultural European countries. Social Science Quarterly, 88,
228–245.
Torgler, B. and W. Werner (2005), Tax morale and fiscal autonomy: evidence from
Germany. Public Finance and Management, 5, 460–485.
Tyler, T.R. (1997), Procedural fairness and compliance with the law. Schweizerische
Verboon, P. and M. van Dijke (2007), A self-interested analysis of justice and tax
compliance: how distributive justice moderates the effect of outcome favorabil-
ity. Journal of Economic Psychology, 28, 704–727.
Vihanto, M. (2003), Tax evasion and the psychology of the social contract. Journal
of Socio-Economics, 32, 111–125.
öffentlichen Verwaltung? Frankfurt: Lang.
Weck-Hannemann, H. and W.W. Pommerehne (1989), Einkommensteuer-
Weck, H., W.W. Pommerehne and B.S. Frey (1984), Schattenwirtschaft. München:
Vahlen.
Weck, H., W.W. Pommerehne and B.S. Frey (1986), Die heimliche Wirtschaft:
Struktur und Entwicklung der Schattenwirtschaft in der Schweiz. Bern: Paul
Haupt.


Wenzel, M. (2004), An analysis of norm processes in tax compliance. Journal of

Economic Psychology, 25, 213–228.
Wenzel, M. (2005a), Motivation or rationalisation? Causal relations between
ethics, norms, and tax compliance. Journal of Economic Psychology, 26, 491–508.
Wenzel, M. (2005b), Misperceptions of social norms about tax compliance: from
theory of intervention. Journal of Economic Psychology, 26, 862–883.
Wicksell, K. (1896), Finanztheoretische Untersuchungen. Jena: Gustav Fischer.
Wu, S.-Y. and M.-J. Teng (2005), Determinants of tax compliance: a cross-country
analysis. Finanzarchiv, 61, 393–417.
Die Zeit (1997), Allgemeine Erklärung der Menschenpflichten. Die Zeit, 41 (3
October).


11 The link between the intrinsic
motivation to comply and
compliance behaviour: a critical
appraisal of existing evidence
Martin Halla
11.1 INTRODUCTION
In recent years, the economic literature on the shadow economy and tax
evasion emphasized the importance of moral considerations (or social
norms) to explain compliance behaviour.1 Likewise, research on public
enforcement of law increasingly considers social norms because of their
role to substitute or to complement formal laws, and because of the
potential impact of laws on social norms (Polinsky and Shavell, 2000).
This trend most likely results from the fact that neo-classical models of
compliance – in the spirit of the economics-of-crime approach – over-
predict real-world compliance. Many scholars therefore conclude that
the explanation for the tendency to comply must be that individuals are
obeying a norm (Posner, 2000).
As a response, theoretical papers incorporated individuals with an
intrinsic motivation to comply (for example, Gordon, 1989; Erard and
Feinstein, 1994; Traxler, 2010). More recently, an increasing number
of empirical papers (to be discussed below) have tried to quantify this
intrinsic motivation with survey data. In most of the cases, scholars study
the case of tax evasion, and analyze the intrinsic motivation to pay taxes,
which is known as tax morale.2 The increasing popularity of theses studies
can be shown by the number of papers indexed in Google Scholar over
time.3 Figure 11.1 shows that the number of published papers on tax
morale was below ten per year throughout the 1990s. However, thereafter
the number sharply increased, and since 2006 we are observing more than
100 papers per year.4
These papers typically aim to identify factors (both on an individual-
and a country-level) that affect the level of tax morale. A substantial
375


Notes:
N 5 24, correlation coefficient 5 0.122, p-value 5 0.571
AT: Austria, BE: Belgium, BG: Bulgaria, BY: Belarus, CZ: Czech Republic, DE: Germany,
DK: Denmark, ES: Spain, FI: Finland, FR: France, GB: United Kingdom, GR: Greece,
HR: Croatia, HU: Hungary, IE: Ireland, IS: Iceland, IT: Italy, KV: Latvia, LT: Lithuania,
LU: Luxembourg, MT: Malta, SE: Sweden, TR: Turkey, UA: Ukraine.
Figure 11.1 Relation between underground production and tax morale

WVS-2
amount of empirical evidence on the association between tax morale and

several socio-demographic characteristics from national and international
samples is available.5 Further, a number of papers identified different insti-
tutional arrangements that are correlated with a high level of tax morale
(for example, Torgler, 2005a).
These papers all – explicitly or implicitly – presume that tax morale
affects actual compliance. However, it is important to bear in mind that an
intrinsic motivation to comply captured by survey data, does not measure
individual behaviour but an individual attitude. That means a high intrinsic
motivation to comply does not necessarily translate into a high level of
compliance. In fact, relatively little empirical evidence on the impact of
an intrinsic motivation to comply on actual compliance behaviour exists.6
Most likely, this is due to the fact that is extremely hard to identify this
causal link. First, one has to obtain and quantify both dimensions. In
the case of compliance behaviour, this is a non-trivial problem, since any
form of non-compliance is difficult to observe. In general, non-compliant


Motivation to comply and compliance behaviour 377
behaviour cannot be measured and has to be estimated. The intrinsic

motivation to comply has to be obtained by adequate survey techniques.
Given that one can observe and link both dimensions, one needs a credible
research design to establish a causal effect.
With respect to tax evasion almost all existing evidence is based on
survey data. Obviously, this research design is problematic, since the
accuracy of self-reported tax evasion information is questionable. A
small number of papers try to solve this limitation by combining tax
evasion data observed in laboratory experiments with information from
post-experiment questionnaires. With respect to the shadow economy
a number of papers present simple correlations between the level of tax
morale and the size of the shadow economy. However, this descriptive
evidence allows several interpretations. It is unclear, whether a causal
effect of tax morale on the size of the shadow economy exists. Most
recently, a small number of papers (Torgler, Schaffner and Macintyre,
2007; Torgler and Schneider, 2007, 2009) address this identification
problem and suggest an instrumental variable approach to disentangle a
causal effect.
We think a good understanding of the relation between the intrinsic
motivation to comply and actual compliance behaviour is very important.
Future research in this area should pursue empirical strategies that are
able to establish a causal link between these two dimensions. The existence
of this causal link determines the significance of the whole strand of litera-
ture that analyzes the determinants of the intrinsic motivation to comply.
The remainder of the chapter is organized as follows. In Section 11.2 we
discuss the necessary steps to establish this causal link, which includes a
definition of the shadow economy that fits the question under considera-
tion. Following that, we summarize the literature that aims to quantify
the related phenomena of tax evasion. Thereby, we will evaluate which
methods produce data points that can be used to answer the question
under consideration. Subsequently, we provide an overview of all publicly
available survey data that are useful to measure the intrinsic motivation to
abstain from underground activities, tax evasion and benefit fraud. Under
the assumption that we can observe and link both the intrinsic motiva-
tion to comply and compliance behaviour, we discuss the econometric
identification problem and highlight the necessary assumptions in order to
establish a causal effect.
Based on these findings, Section 11.3 critically reviews the existing liter-
ature. At this stage, we fully acknowledge that a clear identification of this
causal effect is extremely difficult. However, we believe that only a thor-
ough discussion of the identifying assumptions and their credibility allows
steps forward. We do not provide a solution to solve the identification


problem in this chapter. However, we try to contribute to this literature by

providing a clear discussion of the identification problem.
Before we conclude in Section 11.5, we briefly highlight in Section 11.4
the importance of unobserved heterogeneity for the case of tax morale and
the shadow economy.
11.2 ESTABLISHING A LINK BETWEEN ATTITUDE

AND BEHAVIOUR
In order to identify a causal link between the intrinsic motivation to
comply and actual compliance behaviour one has to overcome a number
of obstacles. First, one has to observe and link both dimensions. In a
second step, a research design, most likely based on an instrumental
variable approach, is needed.
11.2.1 Measuring Compliance Behaviour
To start with, one has to decide which specific form of compliance should
be studied. As usual, this choice (should be guided by the relevance of the
question and) is limited by the availability of data. In the case of compli-
ance, this problem is non-trivial since any form of non-compliance is dif-
ficult to observe. In general, non-compliant agents will try to hide their
behaviour in order to avoid punishment. That means, non-compliant
behavior cannot be observed and has to be estimated.
This complicates an a priori assessment of the significance of the issue
and may obscure an evaluation of the (quality of the) available data.
The choice is further complicated by the fact that different forms of non-
compliance that are usually studied by economists – such as underground
economic activities and tax evasion – may overlap and are not mutually
exclusive. Therefore, the choice over the specific form of compliance will
inevitably be related with the selection of its estimation method.7
Defining the shadow economy

In the economic literature there is no agreement on the definition of the
shadow economy or on the method to estimate its size.8 For instance, the
OECD9 uses a very broad definition that specifies five groups of activities
that are collectively said to comprise the shadow economy10:
1. Underground production, defined as those activities that are productive

and legal but are deliberately concealed from the public authorities to
avoid payment of taxes or complying with regulations;


2. Illegal production, defined as those productive activities that generate

goods and services forbidden by law or that are unlawful when carried
out by unauthorized producers;
3. Informal sector production, defined as those productive activities con-
ducted by unincorporated enterprises in the household sector that are
unregistered and/or are less than a specified size in terms of employ-
ment, and that have some market production;
4. Production of households for own final use, defined as those productive
activities that result in goods or services consumed or capitalized by
the households that produced them; and
5. Production missed due to deficiencies in data collection programme,
defined as all the productive activities that should be accounted for by
the basic data collection programme but are missed due to statistical
deficiencies. It is sometimes referred to as the statistical underground.
For the purpose of the question under consideration a narrower definition

of the shadow economy is reasonable. We suggest restricting the definition
of the shadow economy to activities captured by the underground produc-
tion. All other categories are either not (unambiguously) related to non-
compliance behaviour, or are not within the usual domain of this strand
of literature. The production of households for own final use is clearly not
connected to non-compliance behaviour. In the case of the informal sector
production it is less clear, since there can be some overlap. For instance,
informal sector enterprises may prefer to remain unregistered in order
to avoid compliance with regulations and to minimize production costs
(see OECD, 2002, p. 39). However, since these activities are not neces-
sarily performed with the deliberate intention of non-complying (such as
tax evasion, infringing labour legislation or other regulations), we argue
to exclude this category. Finally, illegal production, such as sale of drugs
or trafficking stolen goods, is by definition non-compliant behaviour.
However, these forms of non-compliance are usually not studied (by
economists) with a reference to an intrinsic motivation to comply.
Our recommended definition of the shadow economy comprises there-
fore as stated by OECD (2002, p. 37) ‘only activities that may be both
productive in an economic sense and also quite legal (provided certain
standards or regulations are complied with) but deliberately concealed
from public authorities for the following kinds of reasons’:
● to avoid payment of income, value added or other taxes;

● to avoid the payment of social security contributions;
● to avoid having to meet certain legal standards such as minimum
wages, maximum hours, safety or health standards, and so forth;


● to avoid complying with certain administrative procedures, such as

completing statistical questionnaires or other administrative forms
(see OECD, 2002, p. 38).
Therefore, this definition of the shadow economy covers well-known

practices such as under-reporting of income in order to avoid taxation,
fraudulent receipt of unemployment benefits, unofficially operating enter-
prises who want to avoid long and costly bureaucratic procedures, or
infringement of employment regulations or immigration laws by firms
who hire labour ‘off the books’. Clearly, this quantity is related to tax
evasion. But it does not cover all forms of tax evasion. It includes some
methods of tax evasion (for example, under-reporting of income), and
excludes others that are non-productive, such as over-claiming deductions
or exemptions.11
The economic literature offers various so-called direct and indirect
approaches to estimate the size of the shadow economy, and a full review
of the existing methods is well beyond the scope of this chapter.12 In
general, the methods that are used to estimate the size of the shadow
economy and the extent of tax evasion largely overlap. In the next section,
we will briefly discuss these methods, and evaluate which methods produce
data points that can be used to answer the question under consideration.
Measuring tax evasion

Compared to the shadow economy, it is probably easier to find a consen-
sus among economic scholars on the definition of tax evasion.13 In reality,
however, there are many grey areas where the distinction between tax
evasion and tax avoidance is not so clear. Both tax evasion and tax avoid-
ance are an attempt to reduce one’s own tax liabilities. They only differ
in legal respects. Tax evasion is an illegal activity, whereas tax avoidance
is consistent with existing law. However, even tax authorities may often
inappropriately characterize particular cases (Slemrod and Yitzhaki,
2002).14 Several approaches to estimate the extent of tax evasion – on
an individual- and a more aggregate-level – have been suggested in the
literature.15
One category is called direct approaches. Traditionally, researchers tried
to collect individual-level data on tax evasion with survey techniques.
This has the advantage that questions on tax morale can be added easily.
However, the accuracy of self-reported tax evasion information is highly
questionable. Elffers, Weigel and Hessing (1987) managed to link tax audit
data for approximately 700 Dutch taxpayers with survey responses. They
show that the correlation between assessed and self-reported tax evasion
is essentially zero.


In principal, the most reliable information about tax evasion should be

obtained by tax audits. However, it is widely recognized that even inten-
sive audits are not able to reveal all kinds of non-compliance (Slemrod
and Yitzhaki, 2002). Moreover, regular tax audits do not constitute a rep-
resentative sample. In general, tax authorities do not randomly select tax-
payers to audit, but use properties of submitted returns which indicate the
likelihood of non-compliance.16 Consequently, the best available source
is data from randomly assigned tax audits, such as the US Taxpayer
Compliance Measurement Program (TCMP).17 However, these data have
the disadvantage that they do not include information on tax morale. In
general, it is challenging to obtain survey responses that can be linked with
tax audit data. It is likely ineffective if the IRS sends a questionnaire to US
taxpayers asking about their tax morale. Moreover, survey data collected
by other institutions may often not be linkable to tax audit data due to
privacy law concerns.18
An alternative direct approach to obtain individual-level data taken in
the literature is based on laboratory experiments. For an early applica-
tion, see, Friedland, Maital and Rutenberg (1978). Clearly, the shortcom-
ing of these data is the artificial setting in which they are generated. For
instance, it is unclear whether individuals would behave differently when
they deal with real tax authorities instead of experimenters. However, it
has the advantage that it can be augmented by tax morale information
form pre- or post-experimental questionnaires (see Bosco and Mittone,
1997; Torgler, Schaffner and Macintyre, 2007). Certainly, it has to be
carefully checked, whether pre-experimental questionnaires affect com-
pliance behaviour and/or compliance behaviour affects answers in post-
experimental questionnaires.
A second category discussed in the literature is indirect approaches.
These methods usually provide more aggregated estimates of tax evasion.
Typically, these approaches rely on inferring the levels or trends in tax
evasion from observable quantities, such as currency demand or national
income and product accounts. Clearly, these approaches are able to
produce tax evasion estimates for a large set of countries and years. These
can be matched with country-averages of tax morale from international
surveys. However, one has to note that these approaches have been
heavily criticized in the literature on tax evasion. For instance, Slemrod
and Yitzhaki (2002) conclude that ‘none of these approaches is likely to be
reliable [. . .] as their accuracy depends either on unverifiable assumptions
or on how well the demand for currency is estimated’.
Most recently, a small number of papers (for example, Gorodnichenko,
Martinez-Vazquez and Sabirianova Peter, 2009) have combined economic
theory and natural experiments to obtain estimates of tax evasion. While


this seems to be in general a very promising direction of research, their

results cannot be used for our question under consideration. Typically,
these methods provide only a few data points, which are hard to link with
tax morale data.
11.2.2 Measuring the Intrinsic Motivation to Comply
The selection of the method to quantify the intrinsic motivation to comply

seems to be rather straightforward. Since one tries to measure an attitude,
the only available choice is survey techniques. The more challenging part
is the design of the survey questions. Among others, the formulation of the
survey question has to target the form of compliance under consideration.
Tables A11.1.1 and A11.1.2 (in the Appendix) provide an overview on
all publicly available international survey data that are useful to measure
the intrinsic motivation to abstain from underground activities, tax
evasion and benefit fraud. Only a few survey programmes include ques-
tions on the intrinsic motivation to comply. Fortunately, these are all very
well organized ongoing academic projects that cover a reasonable number
of respondents from a large set of countries.
Measuring tax morale

The most extensive data are available to measure tax morale. To our best
knowledge, there are four international survey data-sets available: (1)
the European and World Values Surveys (WVS), (2) the European Social
Survey (ESS), (3) the International Social Survey Programme (ISSP), and
(4) the Latinobarometro.19 Each survey is a pooled cross-sectional data.
As Table A11.1.1 shows, each survey differs in the exact formulation
of the question, the possible answer categories, and the available country-
years. The WVS and the ESS offer both two different questions on tax
morale. The first question in the WVS (TMWVS 1 ) refers to tax morale in
a very general way: ‘Please tell me for each of the following statements
whether you think it can always be justified, never be justified, or some-
thing in between: Cheating on taxes if you have a chance.’ Respondents are
asked to evaluate on an ordered scale from ‘never justifiable’ (1) to ‘always
justifiable’ (10). The second question (TMWVS
2 ) is more specific, ‘[. . .] Paying
cash for services to avoid taxes’, and offers the same scale to answer.
Similar, the ESS asks first (TMESS
1 ) ‘How much you agree or disagree with
each of these statements: Citizens should not cheat on their taxes’, and
then (TMESS2 ) ‘How wrong, if at all, do you consider the following ways
of behaving to be? How wrong is someone paying cash with no receipt so
as to avoid paying VAT or other taxes?’ In the first case, respondents can
answer on a five-point scale, in the second case on a four-point scale. The


only available question in the ISSP (TMISSP) is more specific, and refers
to tax cheating via income under-reporting: ‘Consider the following situ-
ations below. Do you feel it is wrong or not wrong if a taxpayer does not
report all of his or her income in order to pay less income tax?’ The scale of
answers ranges from ‘not wrong’ (1) to ‘seriously wrong’ (4).
TMWVS1 offers by far the most observations and is, therefore, the
most widely used variable to study tax morale. It was included in each
of the four survey waves (covering the time period from 1981 to 2003).
Information on respondents from 80 countries was collected and in sum
data from 184 country-years are available. TMWVS 2 was only asked in the
fourth wave and provides data from 33 country-years. TMESS 1 and TMESS
2
were both included in the second wave of the ESS, where fieldwork was
conducted between April 2004 and December 2006. In each case, informa-
tion from respondents from 25 countries is available. TMISSP has been col-
lected in 1991 and 1998. In sum, this provides data from 48 country-years.
It is a priori not clear how the ideal survey question to capture tax
morale should be formulated. However, one could argue that a more
general formulation (that does not only refer to one method of tax
evasion) is preferable. Given this criteria, TMWVS 1 and TMESS1 are supe-
WVS ESS ISSP
rior to TM2 , TM2 and TM . Nevertheless, it would be reassuring
if the correlation between responses from different questions within
one survey was high. This can be checked for the WVS and ESS. The
Spearman’s rank correlation coefficient between TMWVS 1 and TMWVS
2 is in
the full sample on an individual-level (38 560 observations from 33 coun-
tries) equal to 0.52 and the hypothesis of a zero rank correlation can be
rejected at a significance level below 0.001. For the ESS (44 802 observa-
tions from 25 countries), we observe a considerably lower correlation of
0.28 between the two alternative measures of tax morale. Nevertheless,
the hypothesis of a zero rank correlation can again be rejected at a sig-
nificance level below 0.001. Notably, on a country-level the correlation
coefficients are more pronounced (WVS: 0.70 and ESS: 0.58).
Tables 11.1 and 11.2 provide descriptive statistics on the tax morale vari-
ables per country. In the case of the WVS (see column 5 in Table 11.1), the
correlation coefficients vary between 0.29 (Romania) and 0.67 (Germany
and Portugal). In each country the hypothesis of a zero rank correlation
can be rejected. As expected, the ESS (see column 4 in Table 11.2) shows
lower within-country correlations, but still, independence can be rejected in
each case at a significance level below 0.001. Interestingly, in both surveys
the Spearman’s rank correlation coefficients between the two alternative
measures of tax morale are not significantly correlated with the number
of observations, or with the level of the country-mean of either tax morale
variable.20 In the case of the WVS, both variables are measured on the same


Table 11.1 Comparison of the survey questions on tax morale in the

WVSa
N TMWVS
1 TMWVS
2 TMWVS
1 2TMWVS
2 Corr (TMWVS
1 , TMWVS
2 )
Austria 1467 8.91 7.84 1.06 0.47
Belarus 824 6.76 7.05 −0.29 0.62
Belgium 1833 7.37 6.74 0.63 0.45
Bulgaria 923 9.03 9.32 −0.29 0.52
Croatia 987 8.25 8.05 0.20 0.65
Czech 1849 8.98 8.62 0.35 0.55
Republic
Denmark 1007 8.99 6.69 2.29 0.38
Estonia 901 7.83 7.73 0.11 0.60
Finland 1017 8.45 7.37 1.08 0.55
France 1551 7.97 6.86 1.11 0.49
Germany 1935 8.65 8.31 0.34 0.67
Great Britain 977 8.56 7.52 1.04 0.53
Greece 1061 7.81 6.98 0.83 0.35
Hungary 947 8.90 8.37 0.53 0.55
Iceland 957 8.77 8.30 0.46 0.58
Ireland 976 8.72 8.10 0.62 0.49
Italy 1948 8.61 8.51 0.10 0.49
Latvia 970 8.63 7.94 0.69 0.54
Lithuania 860 7.08 6.56 0.52 0.53
Luxembourg 1128 7.64 7.01 0.63 0.47
Malta 1002 9.47 8.94 0.53 0.48
Netherlands 995 8.27 6.74 1.53 0.38
Northern 898 8.63 7.98 0.65 0.55
Ireland
Poland 945 8.86 8.29 0.57 0.36
Portugal 954 8.60 8.72 −0.12 0.67
Romania 979 8.23 8.17 0.07 0.29
Russia 2138 7.97 7.86 0.11 0.51
Slovakia 1271 8.84 7.34 1.50 0.38
Slovenia 987 8.66 7.73 0.93 0.49
Spain 1095 8.63 7.65 0.97 0.43
Sweden 1002 8.57 7.20 1.37 0.42
Turkey 1195 9.82 9.63 0.19 0.34
Ukraine 981 7.56 7.50 0.07 0.62
Mean 1168 8.42 7.81 0.62 0.50
Notes: aColumn 1 shows the available observations per country. Columns 2 to 4 list the
respective country-means. Column 5 shows the correlation coefficient within a country. For
the definition of TMWVS
1 and TMWVS
2 see Table A11.1.1.


Table 11.2 Comparison of the survey questions on tax morale in the ESSa
N TMESS
1 TMESS
2 Corr (TMESS
1 , TM2 )
ESS
Austria 2040 3.91 2.25 0.27

Belgium 1728 3.58 2.28 0.31
Switzerland 2059 4.02 2.65 0.21
Czech Republic 2661 4.09 2.77 0.25
Germany 2706 3.80 2.36 0.32
Denmark 1450 4.10 2.74 0.40
Estonia 1772 4.14 2.75 0.20
Spain 1529 3.92 2.65 0.17
Finland 1988 4.11 2.77 0.34
France 1784 4.04 2.12 0.29
Great Britain 1865 3.94 2.59 0.30
Greece 2321 3.93 3.06 0.25
Hungary 1420 4.17 2.68 0.25
Ireland 2231 4.08 2.65 0.21
Iceland 558 4.10 2.91 0.37
Luxembourg 1505 3.98 2.27 0.11
Netherlands 1853 3.93 2.40 0.24
Norway 1749 3.94 2.75 0.40
Poland 1620 4.08 2.58 0.28
Portugal 1924 4.17 3.00 0.17
Sweden 1916 3.90 2.81 0.32
Slovenia 1382 4.08 2.58 0.25
Slovakia 1416 3.93 2.72 0.16
Turkey 1713 4.47 3.23 0.20
Ukraine 1612 3.90 2.68 0.28
Mean 1792 4.01 2.65 0.26
Notes: aColumn 1 shows the available observations per country. Columns 2 and 3 list the
respective country-means. Column 4 shows the correlation coefficient within a country. For
the definition of TMESS
1 and TMESS2 see Table A11.1.1. Note, we recoded TMESS
1 such that
higher values indicate a higher tax morale.
scale and their means can be compared. As column 3 shows, in most of the
cases the level of tax morale is higher for the more general question.
Ideally, different measures of tax morale are not only highly correlated
within surveys, but also across surveys. Unfortunately, there is little
overlap between the available country-years of the different surveys. In
fact, it is impossible to compare the variables from the ESS with those
from the WVS and the ISSP. However, there is some overlap between the
WVS and the ISSP. In order to gain some more country-years for this


comparison, we extrapolate each country average of TMWVS 1 , TMWVS

2 and
ISSP
TM by one year forward. That means we impute all missing country-
averages of tax morale in year t with the value of the preceding year
(t 2 1). As Table 11.3 shows, there is a positive, but only modest, correla-
tion (0.24, N 5 41) between TMWVS 1 and TMISSP.21 However, if we drop
two observations on Austria – which are certainly outliers in the ISSP
sample – the correlation increases to 0.43 (N 5 39), and independence
can be rejected at a significance level below 0.01. The correlation between
TMWVS2 and TMISSP is practically zero (0.08, N 5 20). This result remains
unchanged, even if the one observation on Austria is excluded.22
In a final step we check whether the interrelation between alternative
measures of tax morale and individual characteristics is similar within and
across surveys. Therefore, we regress TMWVS 1 , TMWVS
2 , TMESS
1 and TMESS
2
on basic socio-economic characteristics. We manage to measure the
dimensions of age, sex, marital status, children, educational attainment,
household income and labour market status (base category is employed)
in both surveys on an almost equal scale.23 After cleaning the data-sets,
about 30 700 observations from the WVS and about 16 600 from the ESS
remain.24 Estimation results are presented in Table 11.4. For all four
measurements of tax morale we find consistent results with respective
to the estimated coefficients’ signs. (The only exception is the size of the
place of residence.) In most of the cases, even the statistical significance
coincides across estimations. As previously found in the literature (see,
for example, Halla and Schneider, 2008) a higher age, being female, being
married, a higher educational attainment and being out of the labour force
(compared to being employed) are associated with a higher level of tax
morale. Whereas, a high household income, self-employment, and unem-
ployment are negatively correlated with the intrinsic motivation to pay
taxes. The only notable difference is that the share of explained variation
is comparably low in the case of TMESS 1 .
Measuring benefit morale

We are aware of two international surveys, the WVS and the ISSP, that
include a question to study the phenomenon of benefit morale, that is the
intrinsic motivation to abstain from cheating on the state via benefit fraud.
Table A11.1.2 shows the exact formulation of the question, the possible
answer categories, and the available country-years.
The question on benefit morale (BMWVS) in the WVS questionnaire is
very similar to TMWVS
1 , and reads as follows: ‘Please tell me for each of the
following statements whether you think it can always be justified, never
be justified, or something in between: Claiming governments benefits to
which you are not entitled.’ Again, respondents are asked to evaluate on


Table 11.3 Correlation among different measurements of tax morale and benefit moralea
TMWVS
1 TMWVS
2 TMESS
1 TMESS
2 TMISSP BMWVS BMISSP
TMWVS
1 1
TMWVS
2 0.692*** 1
(N 5 33)
TMESS
1 – – 1
TMESS
2 – – 0.584*** 1
(N 5 25)
387
TMISSP 0.238 0.080 – – 1
(N 5 41) (N 5 20)
BMWVS 0.403*** 0.537*** – – −0.081 1
(N 5 196) (N 5 34) (N 5 41)
BMISSP 0.029 -0.381* – – 0.561*** 0.200 1
(N 5 41) (N 5 20) (N 5 70) (N 5 41)
Notes: a For the definition of the variables see Table A11.1.1 and A11.1.2. *, ** and *** indicate statistical significance at the 10 per cent level, 5
per cent level and 1 per cent level.

27/07/2011 16:16
Table 11.4 Relation of tax morale with socio-economic characteristicsa
Dependent variable
TMWVS
1 TMWVS
2 TMESS
1 TMESS
2
Age 0.023*** 0.025*** 0.005*** 0.006***

(0.002) (0.002) (0.001) (0.001)
Female 0.338*** 0.360*** 0.104*** 0.040***
(0.037) (0.034) (0.014) (0.014)
Marriedb 0.198*** 0.180*** 0.066*** 0.057***
(0.048) (0.044) (0.014) (0.017)
Childrenc −0.032 0.003 −0.044* −0.033**
(0.048) (0.065) (0.023) (0.013)
School leaving 0.012** 0.005 0.008*** 0.005*
aged (0.005) (0.005) (0.003) (0.003)
Household −0.020** −0.044*** 0.001 −0.012**
incomee (0.009) (0.011) (0.003) (0.005)
Self-employedf −0.304*** −0.150 −0.107*** −0.042
(0.084) (0.105) (0.023) (0.030)
Unemployedf −0.223*** −0.036 −0.021 −0.008
(0.065) (0.052) (0.034) (0.029)
Out of labour 0.061 0.086* 0.036** 0.028**
forcef (0.048) (0.046) (0.016) (0.012)
Size of place of −0.125*** −0.132** 0.010 0.033**
residenceg (0.045) (0.051) (0.013) (0.012)
No. of 30 773 30 773 28 778 28 778
observations
Adjusted 0.112 0.126 0.052 0.105

R-squared
Notes:
a
The dependent variable is in each estimation a measure of tax morale (TM), where
higher values indicate a higher level of TM. In columns 2 and 3 TM is measured on a ten
point-scale; in columns 4 and 5 on a four-point scale. Each estimation includes country
and year fixed effects. Method of estimation is ordinary least squares. Standard errors
(allowing for clustering by countries) are in parentheses below. *, ** and *** indicate
statistical significance at the 10 per cent level, 5 per cent level, and 1 per cent level.
b
This is a binary variable equal to one if the individual is married, and zero otherwise.
c
This is a binary variable equal to one if the individual is parent, and zero otherwise.
d
In the case of the columns 2 and 3 (data from the WVS) some observations have been
imputed base on the highest educational level. For detailed information please refer to
the Data Appendix in Halla and Schneider (2008).
e
In the case of columns 2 and 3 this ordinal variable is measured on a ten-point scale; in
columns 4 and 5 on a twelve-point scale.
f
The base group is equal to employed individuals.
g
This ordinal variable is measured on a three-point scale.


an ordered scale from ‘never justifiable’ (1) to ‘always justifiable’ (10). The
available question in the ISSP (BMISSP) is a little bit more specific and
refers to providing incorrect information: ‘Consider the following situa-
tions below. Do you feel it is wrong or not wrong if a person gives the gov-
ernment incorrect information about himself to get government benefits
that he is not entitled to?’ The scale of answers ranges from ‘not wrong’ (1)
to 4 ‘seriously wrong’ (4).25
BMWVS was included in each of the four survey waves. Information on
respondents from 80 countries covering data from 186 country-years is
available. BMISSP has been collected in 1991 and 1998. In sum, this pro-
vides data for only 48 country-years.
In order to check the correlation between the two available measure-
ments of benefit morale, we apply the same imputation procedure as in
the case of tax morale. The correlation between BMWVS and BMISSP is
positive (0.20, N 5 41), however, not statistically significant from zero at
conventional significance levels (p-value is equal to 0.20). If we restrict our
sample to observations without imputations only five observations are
left, however, the correlation increases to 0.85.
Finally, one may also be interested in the correlation between tax
morale and benefit morale.26 Table 11.3 shows that tax morale and benefit
morale are highly correlated within a survey. The correlation between tax
morale in one survey, and benefit morale in another survey is, however,
practically zero.
To sum up this section, we have shown that available measurements of
tax morale and benefit morale are quite consistent within a given survey
and to a lesser extent across surveys. While we think that all discussed
variables are suitable to study the question under consideration, we have
a slight preference for variables based on more general formulated survey
questions. Clearly, in terms of available data points the WVS is superior
to ISSP and the ESS. Unfortunately, no survey question that explicitly
(or comprehensively) refers to the shadow economy is available. A major
shortcoming is that no data source offers individual-level panel data on the
intrinsic motivation to comply.
11.2.3 The Identification Problem
In the ideal case, the researcher has access to information on both vari-
ables for a random sample of people on an individual-level, combined with
a large set of covariates Xit, over time t. While it is practically impossible to
obtain a data-set that fulfils all these criteria, we will for a moment assume
that it exists.27 This data-set would allow us to estimate an equation of the
following form:


behaviourit 5 a 1 b # attitudeit 1 GXit 1 eit. (11.1)
Assuming that attitudeit and eit are uncorrelated, an ordinary least squares
regression (OLS) would give us a bOLS that is equal to an unbiased estimate
of the causal impact of the intrinsic motivation to comply on actual com-
pliance behaviour. Whether this necessary assumption is fulfilled, depends
crucially on the set of covariates Xit. Since OLS is a control strategy, we
can increase the likelihood to observe an unbiased estimate by controlling
for a large set of covariates.28 In particular, Xit has to include all variables
that affect behaviourit and that are correlated with attitudeit. However,
many determinants of behaviourit (such as an innate disposition to comply
or aspects of socialization) are most likely unobservable. If these factors
are correlated with attitudeit, the OLS-estimate is biased.
One way to mitigate this problem is controlling for individual fixed
effects di. These account for unobserved time-invariant individual het-
erogeneity. Adding in addition year fixed effects zi we get the following
equation:
behaviourit 5 a 1 b # attitudeit 1 GXit 1 di 1 zt 1 eit. (11.2)
The fixed effects model in (11.2) gives an unbiased estimate of the causal
effect, as long as the relevant attitudeit is not correlated with time-varying
unobservables that affect behaviourit, and reversed causality can be ruled
out. Are these reasonable assumptions? It is hard to evaluate the case
of time-varying omitted variables on a general basis. However, reversed
causality (or simultaneity) cannot be ruled out, or seems almost highly
likely.29 While it is plausible to assume that the intrinsic motivation to
comply (attitudeit) affects actual compliance behaviour (behaviourit), it
is also reasonable that actual behaviour has an impact on individuals’
attitude. That means, individuals justify or confirm their own (self-
interested) behaviour.30 If this hypothesis is true, then the fixed effects
estimate from (11.2) is inconsistent.31
A potential estimation strategy to solve endogeneity problems, such as
this simultaneity bias, is an instrumental variable (IV) approach.32 An IV
approach can give a consistent estimate when OLS cannot. Therefore, a
valid IV, let’s call it zit, has to be available. The IV has to be correlated
with attitudeit, but uncorrelated with any other determinant of behaviourit.
The second requirement can be stated as follows: Cov (zit, eit) 5 0. Since eit
is unobserved, there is no way to prove that an IV is actually valid. The
researcher must rely on theoretical justifications in order to persuade criti-
cal readers.33 As usual, the hardest part is to find such a credible IV. In
fact, specific advice does not exist. Angrist and Krueger (2001) recommend


that ‘progress comes from detailed institutional knowledge and the careful
investigation of the forces at work’.
To sum up, we argue in this section that it seems impossible to iden-
tify the causal effect of the intrinsic motivation to comply on actual
compliance behaviour, without an IV approach.
11.3 EXISTING EVIDENCE: A SHORT BUT

CRITICAL REVIEW
11.3.1 Tax Morale and Tax Evasion
With respect to tax evasion almost all evidence is based on survey data. A
number of papers contrast self-reported tax evasion with different meas-
urements of an intrinsic motivation to comply.34 For instance, Torgler et
al. (2008) examine the relation between self-reported tax evasion and tax
morale based on survey data from the US and Turkey. Their regression
analysis shows that a high level of tax morale is associated with a low level
of tax evasion. One obvious critique of such a research design is the ques-
tionable accuracy of self-reported tax evasion data.
In order to solve this problem, some papers (for example, Bosco and
Mittone, 1997; Torgler, Schaffner and Macintyre, 2007) combine tax
evasion data observed in laboratory experiments with tax morale infor-
mation from post-experiment questionnaires, and confirm the findings
obtained with survey data. However, it is not clear whether the answers
in the questionnaire are independent from the behaviour in the experi-
ment. Therefore, it is not clear whether the intrinsic motivation to comply
with the tax law causally affects compliance behaviour. The correlation
between these two variables can be explained by simultaneity or reversed
causality. If individuals justify or confirm their own self-interested behav-
iour, then actual behaviour has an impact on individuals’ moral consid-
erations (Wenzel, 2005). In line with this argument, Halla and Schneider
(2008) point out that tax morale deteriorates with rising income, while
benefit morale improves with rising income. The authors conclude that
individuals who have comparably more opportunities and low cost to
commit a certain offence develop the attitude that it is a minor offence.
Rich people have comparably more opportunities to commit tax evasion;
they self-servingly adjust their attitude that cheating on taxes is more or
less justifiable. For poor people it is easer to fraudulently collect benefits;
they report that benefit fraud is not a big deal.
In sum, we are not aware of any convincing empirical evidence that tax
morale has a causal impact on tax compliance.


11.3.2 Tax Morale and the Shadow Economy
With respect to the shadow economy, some papers report a negative cor-
relation between average levels of tax morale and the size of the shadow
economy: Weck (1983), Torgler (2005b) for Latin America, Alm and
Torgler (2006) for the US and Europe, Alm, Martinez-Vazquez and
Torgler (2006) for several transition countries, and Barone and Mocetti
(2009) for Italy. As in the case of tax evasion, these descriptive results
allow different interpretations. First, a low level of tax morale may caus-
ally impact peoples’ behaviour, which results in a bigger shadow economy.
Second, a pronounced shadow economy may undermine peoples’ tax
morale. Or third, the correlation may just be driven by an unobserved
factor, such as complex tax legislation.
Most recently, a small number of papers (Torgler, Schaffner and
Macintyre, 2007; Torgler and Schneider, 2007, 2009) have tried to disen-
tangle the causal effect of tax morale on the size of the shadow economy
based on an IV approach. In each case the authors use a definition of the
shadow economy as suggested in Section 11.2.1. The estimates of the size
of the shadow economy are based on a combination of the DYMIMIC-
method and the currency demand method.35 Tax morale is captured as
country-averages based on a re-scaled variable from the WVS (and the
Latinobarometro). Since each paper has a different focus, the exact sample,
the set of control variables, and the suggested IV’s vary. In essence, Torgler
and Schneider (2009) present a cross-sectional analysis of the effect of tax
morale and institutional quality on the size of the shadow economy,
where the authors try to account for the endogeneity of tax morale and
institutional quality with a set of IVs, such as legal origins of commer-
cial laws. Torgler, Schaffner and Macintyre (2007) include a panel data
analysis of the impact of tax morale on the size of the shadow economy,
where weather conditions (a measure for cloudiness) serves as an IV for
tax morale. Finally, Torgler and Schneider (2007) employ a panel data
analysis of study on the effect of tax morale, institutional quality and
governance on the size of the shadow economy. To instrument for tax
morale, a measure of cloudiness and an index for moral values based on
data from the WVS is used. All papers use a Two-Stage Least Squares
estimation (2SLS) and find a statistically significant negative effect of tax
morale on the size of the shadow economy.36 In turn, we will discuss each
IV approach in more detail.
In Torgler and Schneider (2009) tax morale and institutional quality is
instrumented by the following variables: legal origin (English, German
and French), latitude, fractionalization (language), religion (protestant,
catholic) and the legal system (political rights). These IVs (or subsets of


them) are widely used in the literature as a source of exogenous variation

in institutions and their quality (La Porta et al., 1999). The origin of single
IVs is different. For instance, the idea to use legal origin as an IV has been
suggested by La Porta et al. (1997, 1998). The authors use legal origins
of commercial laws to uncover the causal effect of legal rules on financial
development. It is argued that legal origin is largely exogenous, since it
was typically introduced into various countries through conquest and col-
onization. Subsequently, a large body of research has shown that the influ-
ence of legal origins on legal rules is not restricted to finance.37 However,
the usage of these variables as IVs is not without critique.38 Moreover, one
disadvantage of these IVs is that most of them do not vary over time, and
cannot be combined with a fixed effects model.
In any case, the application of these IVs for tax morale requires further
discussion and should not be used ad hoc. A discussion of the expected
and actual sign of each IV in the first stage would be informative. Further,
the assumption that none of these IVs influences the size of the shadow
economy through channels other than tax morale needs some support.
For instance, it is unclear whether some legal origins are correlated with
higher penalties for non-compliance (or higher enforcement effort), which
would have an independent effect on the size of the shadow economy.
Torgler, Schaffner and Macintyre (2007) instrument tax morale with a
measure for cloudiness. The authors cite literature showing that cloudi-
ness has a negative impact on individuals’ well-being and they find that it
has also a statistically significant negative impact on tax morale (in their
first stage regression). However, it is hard to rule out that weather condi-
tions do not have a separate effect on the size of the shadow economy. For
instance, the construction industry, a sector which is difficult to tax and
known for high underground activity, is affected by weather conditions.
In general, bad weather (such as rain, snow or wind) slows down construc-
tion activity, and may also reduce the size of the shadow economy.
Torgler and Schneider (2007) suggest a (in addition to a measure of
cloudiness as discussed above) an index for moral values based on data
from the WVS as an IV for tax morale. This index is based on benefit
morale and on the justifiability of avoiding a fare on public transport. As
expected, this index enters statistically significant in the first stage regres-
sion and the first requirement of the IV is clearly fulfilled. However, under
the assumption that the intrinsic motivation to comply has an impact on
compliance behaviour, the second assumption could possibly fail. For
instance, if a low level of benefit morale translates into higher benefit
fraud, then the IV is correlated with the error term in the second stage.
We think that the papers using an IV approach to disentangle the
causal effect of tax morale on the size of the shadow economy constitute a


very promising direction of research. In fact, these papers are some of the
rare exceptions that aim to address the causality issue between compliance
attitude and compliance behaviour. We believe that a waterproof identi-
fication of this causal effect is almost impossible, since it is hard to think
of a natural experiment that provides an ‘as-if’ randomly assigned high
intrinsic motivation to comply. Nevertheless, future research in this area
should pursue along the lines of existing empirical strategies; however,
provide a thorough discussion on the validity of the used IVs. Murray
(2006) states strikingly ‘Indeed, all instruments arrive on the scene with a
dark cloud of invalidity hanging overhead. This cloud never goes entirely
away, but researchers should chase away as much of the cloud as they
can.’
11.4 THE IMPORTANCE OF UNOBSERVED

HETEROGENEITY
In this section we want to evaluate the importance of time-invariant unob-

served heterogeneity in the relation between the intrinsic motivation to
comply and actual compliance behaviour. Therefore, we study the case of
tax morale and the shadow economy, or more precisely the underground
production.
We use estimates of the size of the shadow economy from Schneider,
Buehn and Montenegro (2010). This paper uses a narrow definition of
the shadow economy (as suggested in Section 11.2.1 1) that coincides
with the underground production. Based on a multiple indicators multiple
causes (MIMIC) model estimates on the size of the underground pro-
duction for 162 countries over the period 1999 to 2007 are available.39
If we match this data with country-averages of TMWVS 1 we get a sample
of 52 country-years.40 In order to gain more observations (in particular
observations for countries over time), we amend our data-set with esti-
mates of the underground production for the available country-years in
the WVS before 1999.41 In sum, we have an unbalanced panel data-set
with 75 observations from 53 countries. In line with the literature we
find a negative correlation between tax morale and underground pro-
duction in our sample, see Figure A11.1.2. However, the correlation
coefficient of minus 0.07 is not statistically significant different from zero
(p-value 5 0.55).42
In Table 11.5 we analyze this relationship based on a series of regres-
sions. Column 1 shows that when controlling for year fixed effects, tax
morale and the size of the underground production are statistically sig-
nificantly correlated. This association is lower in OECD-member states


Notes:
AT: Austria, BE: Belgium, CH: Switzerland, CZ: Czech Republic, DE: Germany, DK:
Denmark, EE: Estonia, ES: Spain, FI: Finland, FR: France, GB: United Kingdom, GR:
Greece, HU: Hungary, IE: Ireland, IS: Iceland, LU: Luxembourg, SE: Sweden, TR:
Turkey, UA: Ukraine.

ESS-1
(see column 3) compared to non-member states (see column 4).43 Most

importantly, column 5 shows that once we control for country fixed effects,
the estimated coefficient on tax morale remains statistically significant, but
switches sign.44 If we control for a small set of control variables (see column
5) the coefficient turns statistically insignificant, however, it stays positive.
It seems that tax morale is correlated with unobserved country-specific
time-invariant heterogeneity in a way that disregarding country fixed
effects can diametrically reverse results. The suggested positive relation-
ship between tax morale and the size of the underground production is
counter-intuitive. That means, either our measurements of tax morale
and/or the underground production have fundamental problems, or
unobserved time-varying unobservables are crucial. In the latter case an
IV-strategy (as described in Section 11.2.3) can be used to fix the problem.
An ideal IV would vary over time, such that the IV strategy complements
the fixed effects model.


Table 11.5 Tax morale and the size of the underground productiona
VARIABLES (1) (2) (3) (4) (5) (6)

Full sample Full sample OECD Non- OECD OECD
sample OECD sample sample
sample
TMWVS
1 −4.639** −4.225** −3.544* −5.183* 2.829** 2.152
(1.870) (1.621) (1.779) (2.712) (1.284) (1.708)
GDP p. c. 0.064
(in $1000)b (0.329)
GDP deflatorb 0.014
(0.221)
Population size −0.053
(in mill.)b (0.052)
OECDc −13.517***
(2.854)
Constant 46.722** 76.857*** 43.402*** 84.792*** 2.863 −9.046
(19.064) (16.036) (15.299) (25.423) (8.613) (21.243)
Year fixed effects yes yes yes yes yes yes

Country fixed effects no no no no yes yes
No. of observations 75 75 46 29 46 46
Adjusted R-squared 0.366 0.525 0.316 0.009 0.979 0.977
Notes:
a
The dependent variable is in each case an estimate of the underground production
measured as percentage of GDP. Schneider, Buehn and Montenegro (2010) is the
source of the following country-years: Albania (2002), Algeria (2002), Argentina
(1999), Austria (1999), Bangladesh (2002), Belgium (1999), Bosnia and Herzegovina
(2001), Bulgaria (1999), Belarus (2000), Canada (2000), Chile (2000), China (2001),
Croatia (1999), Czech Republic (1999), Denmark (1999), El Salvador (1999), Finland
(2000), France (1999), Germany (1999), Greece (1999), Hungary (1999), Iceland (1999),
India (2001), Indonesia (2001), Iran (2000), Ireland (1999), Italy (1999), Japan (2000),
Jordan (2001), Republic of Korea (2001), Kyrgyzstan (2003), Latvia (1999), Lithuania
(1999), Luxembourg (1999), Malta (1999), Mexico (2000), Republic of Moldova (2002),
Morocco (2001), Vietnam (2001), Zimbabwe (2001), Spain (1999), Spain (2000), Sweden
(1999), Turkey (2001), Uganda (2001), Ukraine (1999), Republic of Macedonia (2001),
Egypt (2000), United Kingdom (1999), United Republic of Tanzania (2001), United
States (1999) and Venezuela (2000). The remaining country-years – Australia (1995),
Austria (1990), Belgium (1990), Canada (1990), Denmark (1990), Finland (1990),
Finland (1996), France 1990), Germany (1990, 1997), Ireland (1990), Italy (1990), Japan
(1990, 1995), Spain (1990, 1995), Sweden (1990, 1996), Switzerland (1989, 1996), United
Kingdom (1990) and United States (1990, 1995) – are imputed by estimates kindly
provided by Prof. Schneider. Method of estimation is ordinary least squares. Standards
are in parentheses below. *, ** and *** indicate statistical significance at the 10 per cent
level, 5 per cent level, and 1 per cent level.
b
This variable is derived from the OECD Factbook 2007.
c
This is a binary variable equal to one if the country is an OECD-member state, and zero
otherwise.


Notes:
AT: Austria, BE: Belgium, CH: Switzerland, CZ: Czech Republic, DE: Germany, DK:
Denmark, EE: Estonia, ES: Spain, FI: Finland, FR: France, GB: United Kingdom, GR:
Greece, HU: Hungary, IE: Ireland, IS: Iceland, LU: Luxembourg, SE: Sweden, TR:
Turkey, UA: Ukraine.

ESS-2
11.5 CONCLUSIONS
Why should economists be interested in the (determinants of the) intrinsic
motivation to comply, if this variable has no causal impact on actual com-
pliance behaviour? In this chapter we argued that a good understanding of
this relationship is very important for this strand of literature and further
empirical evidence is needed. We think future research should pursue
empirical strategies that are able to establish a causal link between these
two dimensions.
We see three (potentially) feasible endeavours that could help to uncover
a causal effect. First, we suggest trying to create a link between randomized
in-depth audits (such as the TCMP) and panel survey data. Second, we
recommend laboratory experiments augmented with well designed survey
techniques. The effect of survey participation on compliance behaviour
could be checked with random assignment to different groups with a


Notes:
AT: Austria, AU: Australia, CA: Canada, CH: Switzerland, DE: Germany, DK: Denmark,
ES: Spain, FR: France, GB: United Kingdom, IE: Ireland, IT: Italy, JP: Japan SE: Sweden,
US: United States.

ISSP
pre-experimental survey, a post-experimental survey and no survey. Third,

scholars should pursue along the lines of existing IV approaches, however,
provide a thorough discussion of the validity of the used IVs.
NOTES
1. For a general discussion of moral considerations and social norms within economics
see, for instance, Elster (1989), Posner (1997) and Posner (2002).
2. A smaller number of papers deal with benefit morale, that is, with the individual reluc-
tance to exploit the state via benefit fraud. See, for instance, Halla and Schneider (2008),
Heinemann (2008) and Halla, Lackner and Schneider (2010).
3. Google Scholar, a freely-accessible Web search engine, indexes the full text of schol-
arly literature across an array of publishing formats and disciplines. It includes most
peer-reviewed journals of large scholarly publishers. For more information see http://
scholar.google.com/intl/en/scholar/about.html.
4. To be precise, German scholars around Günter Schmölders known as the ‘Cologne
school of tax psychology’ already tried in the 1950s and 1960s to build a bridge between


economics and social psychology. They emphasized the importance of tax morale
to explain tax compliance behaviour. See, for instance, Schmölders (1951/52, 1960,
1969).
5. For a survey of this literature, see Torgler (2007).
6. We will discuss the existing evidence in detail below.
7. A possible way to avoid this problem is laboratory experiments.
8. See, for instance, Frey and Schneider (2001).
9. See OECD (2002); this is a joint publication of the Organisation for Economic Co-
operation and Development, the International Monetary Fund, the International Labour
Organisation, and the Statistical Committee of the Commonwealth of Independent States.
10. To be precise, OECD (2002) uses the term non-observed economy.
11. Tanzi (1999) provides a detailed discussion of the nexus between the shadow economy
and tax evasion.
12. For a comprehensive review of the literature on the shadow economy, see Schneider
and Enste (2000). Some recent applications are Lemieux, Fortin and Fréchette (1994);
Lyssiotou, Pashardes and Stengos (2004) and Breusch (2005).
13. The development of the literature through the 1980s is surveyed by Cowell (1990).
More recent literature surveys are provided by Andreoni, Erard and Feinstein (1998);
Slemrod and Yitzhaki (2002); and Slemrod (2007).
14. Denis Healey, a former UK Chancellor of the Exchequer, phrased it strikingly:
‘The difference between tax avoidance and tax evasion is the thickness of a prison
wall.’
15. As mentioned before, most of these approaches are also to infer on the size of the
shadow economy.
16. Similarly, data from tax amnesties have a sample selection problem (Andreoni, Erard
and Feinstein, 1998).
17. Under the TCMP, the Examination Branch of the Internal Revenues Service (IRS)
periodically conducted (until the late 1980s) random in-depth audits to estimate com-
pliance and revenue lost from non-compliance. The resulting data consisted of detailed
information about what the taxpayer reported, and what the examiner concluded was
correct.
18. The aforementioned paper by Elffers, Weigel and Hessing (1987) provides a notable
exception.
19. Due to the rather specific geographic restriction we will not cover the Latinobarometro
in this chapter. For further information, see http://www.latinobarometro.org/.
20. Results are not shown, but are available upon request.
21. Without imputation the correlation between TMWVS 1 and TMISSP is equal to 0.34
(N 5 5).
22. Without imputation there are no observations for a comparison of TMWVS 2 and TMISSP
available.
23. Details on the definition of the variables are provided in the notes to Table 11.4.
24. In the case of the WVS (compare Table 11.1), we had to exclude all 954 observa-
tions from Portugal from the regression analysis, since household income is only
available on a 6-point scale. Another 6833 observations (from various countries) are
excluded due to missing information on one or more covariates. In the case of the EES
(compare Table 11.2), we had to exclude all observations from France (1784), Estonia
(1772), Hungary (1420) and Ukraine (1612). For France and Hungary, information
on self-employment is missing. For Estonia and Ukraine, no information on the
household income is provided. Another 9436 observations (from various countries)
are excluded due to missing information on one or more covariates. In most of these
cases, non-response on the household income question is responsible. The Spearman’s
rank correlation coefficients between the two alternative tax morale variables in the
reduced samples (WVS: 0.52, ESS: 0.30) are very similar to those from the full samples
discussed above.
25. It should be noted that the ESS asks, ‘Suppose you planned to get benefits or services


you were not entitled to. How many of your friends or relatives do you think you could
ask for support?’ This question is not perfectly suited to capture benefit morale, but
rather measures the (perceived) benefit morale among the respondent’s circle of friends
and acquaintances.
26. Halla and Schneider (2008) provide a more elaborate discussion.
27. Alternatively, one may think of repeated observations of countries c over time t.
28. It should be noted that more control variables is not necessary better. If variables are
themselves outcome variables, then they should not be included in the regression. For a
detailed discussion see Chapter 3 in Angrist and Pischke (2009). Therefore, in our case,
we should not control for factors that are determined by the intrinsic motivation to
comply.
29. Another source of contemporaneous correlation between attitudeit and eitis measure-
ment error in the intrinsic motivation to comply.
30. We will discuss this hypothesis in more detail in the next section.
31. See, for instance, Wooldridge (2002) for a formal discussion.
32. An IV approach will be especially important, if one cannot control for time-invariant
unobserved heterogeneity.
33. For a further discussion and aspects of estimation, see, for instance, Angrist and
Pischke (2009).
34. Other papers, such as Kaplan and Reckers (1985) and Webley, Cole and Eidjar (2001),
study the relation between the perceived prevalence of tax evasion among others, and
the respondent’s own self-reported compliance behaviour.
35. For further details of all papers refer to Schneider (2005a, b).
36. The estimated quantitative impact of tax morale on the size of the shadow economy
from OLS and 2SLS can unfortunately not be directly compared. In the former case
standardized coefficients are listed, while in the latter case only conventional (or
unstandardized) coefficients are presented.
37. For a survey of this literature see, La Porta, Lopez-de-Silanes and Shleifer (2008).
38. Respectively, La Porta et al. (2008) comment on page 326 ‘[that their] interpretation of
the meaning of legal origins has evolved considerably over time [. . .] and that is the idea
that legal origins – broadly interpreted as highly persistent systems of social control of
economic life – have significant consequences for the legal and regulatory framework
of the society, as well as for economic outcomes.’ On page 291, they admit ‘[that] legal
origins influence many spheres of law making and regulation, which makes it dangerous
to use them as instruments.’
39. Details on the estimation method are provided in Section 3.1 of Schneider, Buehn and
Montenegro (2010).
40. The notes to Table 11.5 list these country-years.
41. Professor Schneider kindly provided data on the estimated size of the underground pro-
duction based on a MIMIC model for the following country-years: Australia (1995),
Austria (1990), Belgium (1990), Canada (1990), Denmark (1990), Finland (1990),
Finland (1996), France 1990), Germany (1990, 1997), Ireland (1990), Italy (1990),
Japan (1990, 1995), Spain (1990, 1995), Sweden (1990, 1996), Switzerland (1989, 1996),
United Kingdom (1990) and United States (1990, 1995).
42. Notably, as Figures A11.1.1 to A11.1.2 (in the Appendix) show, the correlation between
the other available tax morale variables and the estimated underground production is in
each case positive. For TMESS2 and TM2
ISSP
the correlation coefficient is even statistically
significant at conventional levels.
43. Column 2 shows, as expected, that the underground production is on average lower in
OECD-member states (about minus 14 per cent of the GDP).
44. We do not have observations over time for non-OECD member states.


REFERENCES
Alm, J. and B. Torgler (2006), Culture differences and tax morale in the United
States and Europe. Journal of Economic Psychology, 27(2), 224–246.
Alm, J., J. Martinez-Vazquez and B. Torgler (2006), Russian attitudes toward
paying taxes: before, during, and after the transition. International Journal of
Social Economics, 33(12), 832–857.
Economic Literature, 36(2), 818–860.
Angrist, J.D. and A. B. Krueger (2001), Instrumental variables and the search
for identification: from supply and demand to natural experiments. Journal of
Economic Perspectives, 15(4), 69–87.
Angrist, J.D. and J.-S. Pischke (2009), Mostly Harmless Econometrics: An
Empiricist’s Companion. Princeton, NJ: Princeton University Press.
Barone, G. and S. Mocetti (2009), Tax morale and public spending inefficiency.
Economic Working Paper 732, Economic Research Department, Bank of
Italy.
Bosco, L. and L. Mittone (1997), Tax evasion and moral constraints: some experi-
mental evidence. Kyklos, 50(3), 297–324.
Unpublished manuscript, Australian National University.
Cowell, F.A. (1990), Cheating the Government. Cambridge, MA: MIT Press.
Elffers, H., R.H. Weigel and D.J. Hessing (1987), The consequences of different
strategies for measuring tax evasion behavior. Journal of Economic Psychology,
8(3), 311– 337.
Elster, J. (1989), Social norms and economic theory. Journal of Economic
Perspectives, 3(4), 99–117.
Erard, B. and J.S. Feinstein (1994), Honesty and evasion in the tax compliance
game. RAND Journal of Economics, 25(1), 1–19.
Frey, B.S. and F. Schneider (2001), Informal and underground economy. In
N.J. Smelser and P.B. Baltes (eds), International Encyclopedia of Social and
Behavioral Science. Oxford, UK: Elsevier, pp. 7441–7446.
Friedland, N., S. Maital and A. Rutenberg (1978), A simultation study of income
tax evasion. Journal of Public Economics, 10(1), 107–116.
Gordon, J.P. (1989), Individual morality and reputation costs as deterrents to tax
evasion. European Economic Review, 33(4), 797–805.
Gorodnichenko, Y., J. Martinez-Vazquez and K.S. Peter (2009), Myth and reality
of flat tax reform: micro estimates of tax evasion response and welfare effects in
Russia. Journal of Political Economy, 117(3), 504–554.
Halla, M. and F.G. Schneider (2008), Taxes and benefits: two distinct options to
cheat on the state?, IZA Discussion Paper 3536, Institute for the Study of Labor
(IZA), Bonn.
Halla, M., M. Lackner and F.G. Schneider (2010), An empirical analysis of
the dynamics of the welfare state: the case of benefit morale. Kyklos, 62(1),
55–74.
Heinemann, F. (2008), Is the welfare state self-destructive? A study of government
benefit morale. Kyklos, 61(2), 237–257.
Kaplan, S.E. and P.M.J. Reckers (1985), A study of tax evasion judgments.
National Tax Journal, 38(1), 97–102.


La Porta, R., F. Lopez-de-Silanes and A. Shleifer (2008), The economic conse-

quences of legal origins. Journal of Economic Literature, 46(2), 285–332.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R.W. Vishny (1997), Legal
determinants of external finance. Journal of Finance, 52(3), 1131–1150.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R.W. Vishny (1998), Law and
Finance. Journal of Political Economy, 106(6), 1113–1155.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R.W. Vishny (1999), The quality
of government. Journal of Law, Economics and Organization, 15(1), 222–279.
Lemieux, T., B. Fortin and P. Fréchette (1994), The effect of taxes on labor supply
in the underground economy. American Economic Review, 84(1), 231–254.
Lyssiotou, P., P. Pashardes and T. Stengos (2004), Estimates of the black
economy based on consumer demand approaches. Economic Journal, 114(497),
622–640.
Murray, M.P. (2006), Avoiding invalid instruments and coping with weak instru-
ments. Journal of Economic Perspectives, 20(4), 111–132.
OECD (2002), Measuring the Non-Observed Economy: A Handbook. Technical
report, Organisation for Economic Co-operation and Development, Paris.
Polinsky, M.A. and S. Shavell (2000), The economic theory of public enforcement
of law. Journal of Economic Literature, 38(1), 45–76.
Posner, E.A. (2000), Law and social norms: the case of tax compliance. Virginia
Law Review, 86(8), 1781–1819.
Posner, E.A. (2002), Law and Social Norms. Cambridge, MA: Harvard University
Press.
Posner, R.A. (1997), Social norms and the law: an economic approach. American
Economic Review, 87(2), 365–369.
Schmölders, G. (1951/52), Finanzpsychologie. Finanzarchiv, 13(1), 1–36.
Schmölders, G. (1960), Das Irrationale in der öffentlichen Finanzwirtschaft.
Hamburg: Rowohlt.
Schmölders, G. (1969), Steuermentalität und Steuermoral in Großbritannien,
Frankreich, Italien und Spanien. In B. Beichelt, B. Biervert, J. Daviter, G.
Schmölders and B. Strümpel (eds), Steuernorm und Steuerwirklichkeit Band II.
Köln: Westdeutscher Verlag.
Schneider, F. (2005a), Shadow economies around the world: what do we really
Schneider, F. (2005b), Shadow Economies of 145 countries all over the world: what
do we really know? CREMA Working Paper Series 13, Center for Research in
Economics, Management and the Arts (CREMA).
Schneider, F., A. Buehn and C.E. Montenegro (2010), Shadow economies all
over the world: new estimates for 162 countries from 1999 to 2007. Background
paper for the World Bank study of the informal sector in Central, Southern
Europe and the Baltic countries (Task number P112988).
Schneider, F. and D. Enste (2000), Shadow economies: sizes, causes, and conse-
quences. Journal of Economic Literature, 38(1), 77–114.
Slemrod, J. (2007), Cheating ourselves: the economics of tax evasion. Journal of
Economic Perspectives 21(1), 25–48.
Slemrod, J. and S. Yitzhaki (2002), Tax avoidance, evasion, and administration.
In A. J. Auerbach and M. Feldstein (eds), Handbook of Public Economics, Vol.
3. Amsterdam: Elsevier, chapter 22, pp. 1423–1470.
Economic Journal, 109(456), F338–F347.


Torgler, B. (2005a), Tax morale and direct democracy. European Journal of

Political Economy, 21(2), 521–531.
Torgler, B. (2005b), Tax morale in Latin America. Public Choice, 122(1–2),
133–157.
Torgler, B. (2007), Tax Compliance and Tax Morale: A Theoretical and Empirical
Analysis. Cheltenham, UK: Edward Elgar.
Torgler, B. and F. Schneider (2007), Shadow economy, tax morale, governance
and institutional quality: a panel analysis. IZA Discussion Paper 2563, Institute
for the Study of Labor (IZA).
quality on the shadow economy. Journal of Economic Psychology, 30(2),
228–245.
Torgler, B., I.C. Demir, A. Macintyre and M. Schaffner (2008), Causes and conse-
quences of tax morale: an empirical investigation. Economic Analysis and Policy
(EAP), 38(2), 313–339.
Torgler, B., M. Schaffner and A. Macintyre (2007), Tax compliance, tax morale,
and governance quality. International Studies Program Working Paper, Andrew
Young School of Policy Studies, Georgia State University.
Traxler, C. (2010), Social norms and conditional cooperative taxpayers. European
Journal of Political Economy, 26(1), 89–103.
Webley, P., M. Cole and O.-P. Eidjar (2001), The prediction of self-reported and
hypothetical tax-evasion: evidence from England, France and Norway. Journal
of Economic Psychology, 22(2), 141–155.
öffentlichen Verwaltung? Eine ökonomische Analyse. In Finanzwissenschaftliche
Schriften Nummer 22. Bern-Frankfurt: Lang.
Wenzel, M. (2005), Motivation or rationalisation? Causal relations between ethics,
norms and tax compliance. Journal of Economic Psychology, 26(4), 491–508.
Wooldridge, J.M. (2002), Econometric Analysis on Cross Section and Panel Data.
Cambridge, MA: MIT Press.
403


APPENDIX 11.1

Table A11.1.1 Available survey data on tax morale
Survey Question Answers Years covered Total no. of Total no.

(no. of countries) countries of country-years
WVS TMWVS
1 : Please tell me for each Respondents are asked to 1981 (12) 80 184
of the following statements evaluate on an ordered scale 1982 (5)
whether you think it can always from never justifiable (1) to 1983 (1)
be justified, never be justified, or always justifiable (10). 1984 (2)
something in between: Cheating 1989 (2)
on taxes if you have a chance. 1990 (34)
404
1991 (6)
1992 (1)
1993 (1)
1994 (1)
1995 (11)
1996 (22)
1997 (8)
1998 (9)
1999 (33)
2000 (11)
2001 (19)
2002 (5)

2003 (1)
27/07/2011 16:16
WVS TMWVS 2 : Please tell me for each Respondents are asked to 1999 (30) 33 33
of the following statements evaluate on an ordered scale 2000 (2)

whether you think it can always from never justifiable (1) to 2001 (1)
be justified, never always justifiable (10).
be justified, or something in
between: Paying cash for
services to avoid taxes.
ESS TMESS 1 : How much do you agree or Respondents can answer: agree 2004 (16) 25 25
disagree with each of these strongly (1), agree (2), neither 2005 (8)
statements: Citizens should not agree nor disagree (3), disagree 2006 (1)
cheat on their taxes. (4) or disagree strongly (5).
ESS TMESS 2 : How wrong, if at all, do Respondents can answer: not 2004 (16) 25 25
405
you consider the following ways wrong at all (1), a bit wrong (2), 2005 (8)
of behaving to be? How wrong wrong (3) or seriously wrong (4). 2006 (1)
is someone paying cash with no
receipt so as to avoid paying
VAT or other taxes?
ISSP TMISSP: Consider the following Respondents can answer: not 1991 (17) 31 48
situations below. Do you feel wrong (1), a bit wrong (2), wrong 1998 (31)
it is wrong or not wrong if a (3) or seriously wrong (4).
taxpayer does not report all of
his or her income in order to pay
less income tax?

27/07/2011 16:16
Table A11.1.2 Available survey data on benefit morale
Survey Question Answers Years covered Total no. of Total no.

(no. of countries) countries of country-years
WVS BMWVS: Please tell me for each of the Respondents are asked 1981 (12) 81 186
following statements whether you think it to evaluate on an 1982 (6)
can always be justified, never be justified, ordered scale from never 1983 (1)
or something in between: Claiming justifiable (1) to always 1984 (2)
governments benefits to which you are not justifiable (10). 1989 (2)
entitled. 1990 (36)
1991 (6)
1992 (1)
1993 (1)
1994 (1)
406
1995 (11)
1996 (21)
1997 (8)
1998 (9)
1999 (33)
2000 (11)
2001 (18)
2002 (5)
2003 (2)
ISSP BMISSP: Consider the following situations Respondents can 1991 (17) 31 48
below. Do you feel it is wrong or not wrong answer: not wrong (1), a 1998 (31)
if a person gives the government incorrect bit wrong (2), wrong (3)
information about himself to get government or seriously wrong (4).

benefits that he is not entitled to.
27/07/2011 16:16
Notes: Query (excluding citations) was done on February 19, 2010.
Figure A11.1.1 Hits for ‘tax morale’ on Google Scholar over time


Notes: N 5 75, correlation coefficient 5 20.070, p-value 5 0.553

AL: Albania, AR: Argentina, AT: Austria, AU: Australia, BA: Bosnia and Herzegovina,
BD: Bangladesh, BE: Belgium, BG: Bulgaria, BY: Belarus, CA: Canada, CH: Switzerland,
CN: China, CZ: Czech Republic, DE: Germany, DK: Denmark, DZ: Algeria, EG: Egypt,
ES: Spain, FI: Finland, FR: France, GB: United Kingdom, GR: Greece, HR: Croatia, HU:
Hungary, ID: Indonesia, IE: Ireland, IN: India, IR: Iran, IS: Iceland, IT: Italy, JO: Jordan,
JP: Japan, KG: Kyrgyzstan, KR: Republic of Korea, KV: Latvia, LT: Lithuania, LU:
Luxembourg, MA: Morocco, MD: Republic of Moldova, MK: Republic of Macedonia,
MT: Malta, MX: Mexico, SE: Sweden, SV: El Salvador, TR: Turkey, TZ: United Republic
of Tanzania, UA: Ukraine, UG: Uganda, US: United States, VE: Venezuela, VN: Vietnam,
ZW: Zimbabwe.
Figure A11.1.2 Relation between underground production and tax morale

WVS-1


12 Deterrence policy and the size of
the shadow economy in Germany:
an institutional and empirical
analysis
Lars P. Feld, Andreas J. Schmidt and
Friedrich Schneider
12.1 INTRODUCTION
As recently spotlighted by the cases of tax evasion by German taxpayers

at Liechtenstein banks, the most important policy practice to fight tax
evasion and undeclared work in OECD countries is to increase fines or
investigation efforts of tax authorities. In these recent cases, investigation
efforts and hence expectations of detection were emphasized. The German
tax authorities (even more mysteriously, actually the German Secret
Service) paid 5 million euros to an informant in order to get their hands on
confidential bank customer data.
In economic theory, such deterrence measures as remedy against tax
evasion and undeclared work are founded in the economics of crime
pioneered by Becker (1968). He argued theoretically that costs and ben-
efits determine criminal behaviour. According to this theory, expected
punishment is the cost each criminal faces for committing a crime or, as
Gneezy and Rusticchini (2000) coined it, a fine is a price. The higher the
expectations of detection, the higher the real detection efforts and the
higher the fine, the more it deters from committing a crime. In contrast,
on the benefit side, the incentives to commit a crime are stronger the
higher the expected gains from crime. Allingham and Sandmo (1972)
applied this approach to tax evasion, and enhanced it by taking individual
risk preferences and risk aversion into account. In this frame, it turns out
to be theoretically ambiguous how the (true) personal income and the
marginal tax rate as the most important determinants of the benefits of
tax evasion affect tax compliance. Thus, the cost components of crime,
fines, auditing and investigations remain the most important explanation
409


for tax compliance and the one policy implications are most easily to be
connected to.
German criminal tax law development in the last years supports this
attitude: in 2004 the ‘Black Activities’ Act’ or more exactly the ‘Law to
intensify the fight against black activities and accompanying tax evasion’
(SchwarzArbG, Bundesrats-Drucksache 155/04a) passed both houses
of parliament by comfortable majorities. This new legislation allows for
more intensive tax auditing by the finance control unit ‘black activities’
(Finanzkontrolle Schwarzarbeit, FKS), but also imposes higher fines in
order to raise tax compliance in Germany.
In the empirical literature, deterrence measures as instruments to fight
tax evasion have not remained uncontested (for surveys see Andreoni,
Erard and Feinstein, 1998; Slemrod and Yitzhaki, 2002; Torgler, 2003a;
Braithwaite and Wenzel, 2006). From a bird’s-eye perspective fines and
tax auditing are unable to explain the actual level of tax compliance as
they are too low to provide effective deterrence in most OECD countries
(Graetz and Wilde, 1985; Pommerehne and Frey, 1992). Alm, McClelland
and Schulze (1992) even contend that, given the rather low level of fines
and intensity of control, the real enigma of tax evasion consists in the
question why taxpayers honestly pay taxes (the ‘tax compliance puzzle’).
Moreover, the empirical evidence on the specific partial impact of tax
auditing and fines on tax evasion and the shadow economy, which mainly
comes from experimental studies and field studies in the US, is not unam-
biguous. Dubin, Graetz and Wilde (1987), Dubin and Wilde (1988),
Beron, Tauchen and Witte (1992) and Slemrod, Blumenthal and Christian
(2001) report a significantly positive effect of tax auditing on tax evasion
at least for a part of the income groups investigated. While Schwartz and
Orleans (1967), Friedland, Maital and Rutenberg (1978), Klepper and
Nagin (1989), De Juan, Lasheras and Mayo (1994), Alm, Sanchez and De
Juan (1995) and Blackwell (2010) report a positive influence of fines on tax
compliance, the results offered by Spicer and Lundstedt (1976), Friedland
(1982), Elffers, Weigel and Hessing (1987) and Varma and Doob (1998)
are mixed. In a study for Switzerland, Frey and Feld (2002) find a posi-
tive impact of fines, but a negative impact of tax auditing on tax compli-
ance. Martinez-Vazquez and Rider (2005) report evidence for the US that
enforcement efforts affect the mode of tax evasion targeted by these efforts
negatively, but the untargeted mode positively (as a substitution effect).
While they find an overall positive effect of enforcement on tax compli-
ance, it remains generally open whether the unintended side effect on the
untargeted mode over-compensates the intended effect.
Taken this empirical ambiguity, doubts on the exclusive reliance on
deterrence to reduce tax evasion and undeclared work emerge. Frey


Deterrence policy and the shadow economy in Germany 411
(1997a) argues that tax compliance results from an intrinsic motivation

to pay taxes and hence from tax morale. Such an intrinsic motivation can
however be crowded out by auditing and fines, for example, if taxpayers
perceive a higher control intensity as an unjustified intrusion into their
private lives. Tax compliance can thus be enhanced if tax authorities treat
taxpayers in a friendly way (Feld and Frey 2002a, 2002b). All in all, the
relationship between tax authorities and taxpayers is much more complex
than the economic theory of crime suggests (Feld and Frey 2007).1 The
mixed and even contradictory results obtained in the literature up to now
seem to call for additional evidence.
In this chapter, we will present some additional evidence about the
shadow economy and deterrence policy by presenting time series data
from Germany. This approach provides interesting insights: first, as there
are not many studies using European and especially German data, and
second, as expected punishment is not an explanatory variable in the most
prominent studies for Germany (Schneider and Enste, 2000; Schneider,
2004; Kanniainen, Pääkkönen and Schneider, 2004). Our aim is to provide
some new data and basic insights into the following research question: do
policies that raise deterrence measures work to fight the shadow economy?
As a working hypothesis, we consider three answers into causality direc-
tions to be convincing: first, more severe deterrence measures cause lower
undeclared work and tax evasion, or second, on the contrary, deterrence
may lead to a crowding out effect of tax morale and thus an increased
shadow economy, and third, a higher shadow economy may cause a
policy of more severe deterrence. The approach used is an explorative and
mainly descriptive analysis focusing on German macro-data which were
not systematically collected before. We survey research into the German
shadow economy and its measurement and combine that with a portrayal
of the institutional environment and practice of the deterrence policy from
1970 to 2005. To derive some first insights into the mentioned causality
directions we finally link both parts with a first Granger causality test.
The remainder of the chapter is structured as follows: in Section 12.2,
we first discuss the state, methods and data of empirical research in the
field of shadow economy and tax evasion and, second, assess the size and
development of the shadow economy in Germany as a first proxy for
further analyses. Section 12.3 characterizes the legal foundations as well
as the practice of fines and sentences and its development. In Section 12.4,
we provide time series about the probability of detection in terms of audit-
ing as the second determinant of expected punishment. In Section 12.5, we
link the preceding descriptive empirics with Granger causality tests on the
impact of deterrence on the size of the shadow economy, while the results
are summarized in Section 12.6.


12.2 EMPIRICAL RESEARCH INTO TAX EVASION

AND SHADOW ECONOMY
12.2.1 Shadow Economy, Undeclared Work, Tax Evasion and Tax
Morale
The terms and scope of this inquiry in the interconnected fields of research
into shadow economy and tax evasion needs to be rendered more pre-
cisely (for broader discussions, see Schneider and Enste, 2000; Feld and
Larsen, 2005, 2011). We do so to shed some light on the possibilities and
restrictions for empirical studies in the field. The term ‘shadow economy’
mainly refers to its property of being hidden. It includes economically
legal but hidden activities in the sense of black work as well as some
illegal hidden activities like trade of illicit drugs or prostitution. The first
part, legal undeclared work in the shadow economy usually involves tax
evasion, but taxes could also be evaded pursuing different activities than
those in the shadow economy. This is, for example, the case when capital
income earned officially is not truthfully reported. Tax compliance can
be understood, in contrast to the tax gap, as the amount of the projected
total tax base that tax authorities actually collect (Andreoni, Erard and
Feinstein, 1998). Finally, tax morale traditionally refers to the residuum
of tax compliance which cannot be explained by standard portfolio choice
determinants and the deterrence measures.
These terms and activities are overlapping to a certain extent and are of
a very clandestine nature. As we will see, it will not be a question of taxo-
nomical precision, but of measurability which activity we use as approxi-
mation for the further empirical analysis. Hence, we will further focus on
economically legal but illegally hidden activities and leave other criminal
activities aside. Also we focus on the criminal tax code and audits rather
than general criminal law.
12.2.2 Studies of Tax Compliance and Tax Morale
Nowadays’ citations give the impression that economic studies on the

shadow economy or undeclared work, tax compliance and tax morale has
started only recently. Nevertheless, analyses of tax compliance and tax
evasion have a long tradition in German public finance.
The empirical examination of tax evasion in Germany started with
the works of Schmölders (1932, 1960, 1964, 1970, 1978). In his inaugural
lecture about ‘tax morale and tax mentality’ at Humboldt University
in 1932, he adopted an interdisciplinary approach using both insights
from psychology and economics. He conceptualized basic ideas like the


subjectively perceived tax burden in different groups of taxpayers or

stressed the importance of taxpayers’ attitude towards the state or their
satisfaction with policies for tax compliance. Long before the ‘tax compli-
ance puzzle’ following the Allingham/Sandmo-type rational portfolio-
choice theory was recognized, he explored why taxpayers contribute to
the ‘bonum commune’ deliberately. For him, the key explanation can be
found in the psychological sphere of values and attitudes. He understood
‘tax mentality’ as a combination of general attitudes, norms and beliefs
about the state, its provision of public goods and its financial basis. On
this background, ‘tax morale’ is the attitude towards (1) the duty to pay
his taxes correctly and (2) the criminal act of evasion compared to other
offences. He conducted interviews to measure tax mentalities and morale
and added the tax authorities’ behaviour and the revenue situation to
his data, comparing different countries’ tax morale, tax mentalities and
institutions in their impact on tax compliance.
However, neither hard measures of tax evasion nor theoretical models
are constructed in this work, and no hypotheses are econometrically
tested. Yet, the conclusion of five key factors that shape ‘tax compli-
ance’ fit into today’s insights: first, the importance of the appropriate-
ness of the overall level of the tax burden; second, a fair distribution
of tax burdens among taxpayers; third, the treatment of the taxpayer
by the tax authorities; fourth, an efficient government spending; and,
finally, the importance of the benefit principle of taxation, that is, an
equivalence between taxes and public goods in each group of taxpayers
(Mackscheidt, 1994, 2004). Schmölders and Strümpel (1969) compared
tax mentality and tax evasion in Great Britain, France, Italy and Spain.
Schmölders Institute for Empirical Socio-Economics (Forschungsinstitut
für empirische Sozialökonomik, abbr. FORES) continued his work and
published several continuous surveys on ‘The Tax Mentality of German
Taxpayers’.2 While these studies captured many aspects of modern tax
morale research, they neither measured the shadow economy or the levels
of tax evasion, nor focused on the impact of deterrence on tax compliance.
12.2.3 Measuring the Shadow Economy, Tax Evasion and Tax Morale
Data on the size of the shadow economy, its partial activities and, even
more, on the extent of tax evasion are not easily available for Germany
because of their very clandestine nature and the fiscal secrecy laws in
Germany. Thus, several estimation methods have been developed to
‘measure the unmeasurable’ which are usually linked to the one or the
other aspect of tax evasion (Thomas, 1999; Schneider and Enste, 2000;
Pedersen, 2003; Lyssiotou, Pashardes and Stengos, 2004; Feld and Larsen,


2005, 2011). Some of the methods rather capture the shadow economy or
undeclared work by focusing on the labour market, physical production
or particular economic transactions. Others aim at an assessment of tax
compliance.
There are indirect and direct methods of measurement (see Table 12.1).
The first indirect method is called the income gap approach. It uses the
basic definition in national accounts that the income measure should be
the same as the expenditure measure of the domestic product. If there are
statistical discrepancies, they might occur because the quality of the data is
insufficient. However, it is highly implausible that these statistical discrep-
ancies increase substantially over time. Thus, tax evasion explains why
people in an economy buy more products and services than they officially
have money for, given their earned income according to income tax decla-
rations. In Europe, Larsen (2002) applies this method to Denmark, while
Weck-Hannemann and Pommerehne (1989), Pommerehne and Weck-
Hannemann (1996), Pommerehne and Frey (1992), Frey (1997a), Feld and
Frey (2002) and Frey and Feld (2002) use it to measure Swiss tax evasion.
In a similar fashion, the official participation rate in the labour market can
be compared with actual employment (Pedersen, 2003).
The second indirect measurement method is based on monetary
approaches. On the one hand, the transactions approach, starting from
the Fisher equation of the quantity theory of money, relates total nominal
GNP to total transactions. The GNP of the shadow economy is obtained
by subtracting official GNP from total nominal GNP, assuming a base
year in which the ratio of total transactions to total nominal GNP was
normal, that is, no shadow economy existed (Feige, 1989). On the other
hand, the currency demand approach assumes that transactions in the
shadow economy are more strongly done in cash than transactions in the
official economy in order to leave no accounting traces (Kirchgässner,
1983; Schneider, 2004). The size of the shadow economy is then inferred by
simulating currency demand with and without tax variables.
The third indirect method is the electricity consumption method
(Schneider and Enste, 2000). It assumes that electricity serves as a good
indicator of overall economic activity and that the electricity to GDP
elasticity is close to one. Then, a calculation can be made of how large
the actual total GDP of a country is. The difference from official GDP
provides an estimate of the shadow economy. Schneider and Enste (2000)
describe the more sophisticated econometric method developed by Lackó
(1998), who uses household electricity consumption.
The fourth indirect method is the hidden variable approach (Frey
and Weck-Hannemann, 1984). Three or four macroeconomic indica-
tors, usually the labour participation rate, real GDP growth, currency


Table 12.1 The size of the shadow economy and the extent of undeclared
work in Germany (relative to official GDP) according to
different methods and definitions
Methods 1970 1975 1980 1990 2000 2001 2003 2004 2005 2006 2007
and sources
Survey
IfD Allensbach – 3.6 – – – – –
(1975)
Lamnek et al. – – – – 29/16 – – – – – –
(2000)a
Mummert and – – – – 25/13 – – – – – –
Schneider (2001)b
Feld and Larsen – – – – – 4.1 – 3.1 3.6 1.8 3.2
(2011)c
West Germany – – – – – 4.1 – 2.8 3.2 1.8 2.8
East Germany – – – – – 4.3 – 4.4 4.8 2.5 4.4
Enste et al. – – – – – – – – – 6-7 –
(2009)d
Income Gape
Lippert and 11.0 10.2 13.4 – – –
Walker (1997)
Physical Input
Schneider and – – – 14.6 – – – – – – –
Enste (2000)f
Transactions
Approach
Schneider and 17.2 22.3 29.3 – – – – – – – –
Enste (2000)g
Currency Demand
Kirchgässner 3.1 6.0 10.3 – – – – – – – –
(1983)
Langfeldt 12.1 11.8 12.6 – – – – – – – –
(1984a, 1984b)
Schneider and 4.5 7.8 9.2 11.8 14.7 – – – – – –
Enste (2000)h
Hidden Variable
Approachi
Frey and Weck- 5.8 6.1 8.2 – – – – – – – –
Hanneman
(1984)
Pickhardt and – – 9.4 11.4 16.3 – – – – – –
Sardà Pons
(2006)
Schneider (2010) 4.2 5.8 10.8 12.2 16.0 16.0 17.1 16.1 15.4 15.0 14.7
Soft modeling
Weck (1983) – 8.3 – – – – – – – – –


Notes:
1970–2000:
a
Refers to 1997; 18 years and older.
b
Refers to 1998; 14 years and older. Proportion of the interviewed which has ever carried
out undeclared work.
c
The shadow economy (‘Schwarzarbeit’) relative to official GDP calculated on the
basis of hours worked by the 18–66-year-olds and the wages pertaining in the official
economy. Data for 2001 were first published in Pedersen (2003).
d
The shadow economy (‘Schwarzarbeit’) relative to official GDP calculated on the basis
of hours worked and the wages pertaining in the official economy. 18 years and older.
e
Discrepancy between expenditure and income.
f
Physical input: electricity consumption. Refer to work by Mariá Lackó.
g
Refer to work by Edgar L. Feige.
h
Refer to work by Vito Tanzi.
i
(DY)MIMIC estimates, a combination of the currency demand approach and the
Multiple Indicators Multiple Causes (MIMIC) method.
demand and working hours, are used as indicator variables for the shadow
economy and linked to explanatory variables such as tax rates or the regu-
latory burden using LISREL techniques (structural causal modelling tech-
niques, or DYMIMIC approach; Schneider and Enste, 2000; IAW, 2006).
With the hidden variable approach, only a relative assessment of the size of
the shadow economy is possible such that analyses using this method often
relate their estimates to the currency demand approach (Pickhardt and
Sardà Pons, 2006). In contrast to the income gap method, the latter three
approaches capture activities in the shadow economy, but not overall tax
evasion, as they are not able to capture undeclared income from capital.
There are three main direct methods. The first focuses on undeclared
work, as a part of the shadow economy, by using surveys in which indi-
viduals are directly asked whether they have carried out undeclared work,
either for cash payments or payments in kind (Pedersen, 2003; Feld and
Larsen, 2005). The second direct method, applied by the US Internal
Revenue Service (IRS), is based on actual tax auditing and other compli-
ance methods (Engel and Hines, 1999). In 1963, the IRS started to conduct
periodic tax audits (Taxpayer Compliance Measurement Program TCMP,
later followed by the National Research Program, NRP) measuring
understatement of income, overstatement of deductions and exemptions,
and so forth, for a random sample of individual income taxpayers. The
data are used to calculate tax evasion for the whole population. The IRS
also applies an income gap method for non-filers by calculating the dis-
crepancy between the declared income and actual income of randomly
audited individuals (Andreoni, Erard and Feinstein, 1998). The third


direct method aims at measuring tax morale instead of tax evasion in

surveys. For instance, the World Values Survey elicits tax morale for a
representative sample of individuals (Torgler 2003a). Torgler (2003b) and
Feld and Torgler (2007) analyze the tax morale data for Germany.
Any of these indirect and direct methods has disadvantages. The income
gap method has to cope with the unreliability of statistical mistakes. The
monetary methods may overestimate the rationality of the money market.
In addition, many transactions in the shadow economy take place without
cash payments. The electricity approach heavily depends on the assump-
tion that undeclared work involves the use of electricity. As indirect
methods minimize strategic problems that emerge if individuals are directly
confronted with questions about tax honesty, it could be argued that the
indirect methods provide for an upper boundary of tax evasion or the
shadow economy. The survey approach is sensitive to the formulation of
the questions, and participants in the survey may behave strategically and
simply not tell the truth. Even in face-to-face interviews, which promote
the greatest degree of participation in the survey, a respondent may simply
lie. The survey method may thus measure a lower limit of undeclared work
in the economy. The tax auditing method is prone to sample selection bias,
because the selection for audit is based on the properties of the tax returns
submitted to the tax office and thus not independent of the probability of
evading taxes. Those taxpayers identified as tax cheaters could be the tip of
the iceberg only, because it is highly improbable that tax authorities would
detect all tax cheaters even if they wanted to. The survey of individual tax
morale only measures hypothetical tax morale and not real tax compli-
ance. Nevertheless, all methods taken together describe recent possibilities
to measure the phenomenon. Thus, in Table 12.1, we provide an overview
of the different estimates for Germany obtained from these methods.
If we look at the size of the shadow economy as measured by the trans-
actions approach, it seems that this approach provides implausibly large
estimates of the size of the German shadow economy. A size of the shadow
economy of almost one third of the official economy in 1980, given that by
nearly all other estimates the shadow economy grew further in the 1990s,
would seem to be an overestimate. It thus appears to be more realistic to
think of the income gap method as providing for the upper boundaries of
tax non-compliance. This may indeed be reasonable, because the income
gap method supposedly includes capital income tax evasion, so that it
could produce a higher figure than measures that mainly focus on labour
income. For the 1970s, the figures from the income gap method are larger
than those from the currency demand or the hidden variable approaches.
This qualitative evidence for Germany is corroborated to some extent by
evidence from the United States and Switzerland. A comparison of the


16
12
0
1970 1975 1980 1985 1990 1995 2000 2005
Pickhardt and Sardà Pons Schneider
Source: Pickhardt and Sardà Pons (2006) and Schneider (2006)
Figure 12.1 The size of the German shadow economy, 1970 to 2006 (in
percent of the official economy)
estimates of the extent of tax evasion produced by the auditing approach

of the IRS (Engel and Hines, 1999) with the figures of the shadow
economy reported by Schneider and Enste (2000) using the currency
demand approach shows that the former produces larger estimates of
the tax evasion figures in the United States. Although based on a micro-
approach, it also uses an income gap measure. Similarly, the comparison
of the extent of tax evasion according to the income gap and the shadow
economy according to currency demand for Switzerland shows higher
figures when using the income gap method (Feld and Frey, 2005).
Independent of which authors have conducted the analysis, the currency
demand figures and those obtained from the hidden variable approach are
relatively close together and slightly lower than those from the income
gap method. The closeness of the outcomes of these two approaches is not
really surprising, given the fact that the estimates of the shadow economy
from the hidden variable approach are derived by taking point estimates
from the currency demand approach. The greatest deviation results for
the year 2000 with 1.3 to 1.6 percentage points. Both approaches show
an increase of the size of the shadow economy during the 1980s and
1990s. This is illustrated by Figure 12.1 that plots the joint model data of
Pickhardt and Sardà Pons (2006), which combine the currency demand
and hidden variable estimates, and the currency demand data by Schneider


(2006). According to the figures by Pickhardt and Sardà Pons, the German
shadow economy shows a steadily increasing trend with a peak in 1999
and only modest stagnation during the observed time period. From 1999,
a decrease can be observed. The data by Schneider (2006) show a peak in
2003 and a decline afterwards. Both series are relatively close together.
Each explanation of the size of the shadow economy must be able to cope
with that stylized fact of a steady increase with a first observable decline
since 1999 or 2003.
The figures from the surveys reported in Table 12.1 do not give such
a uniform picture. Admittedly, they are less systematically provided on
a yearly basis and lack consistency across questionnaires and survey
approaches. While the estimates reported by IfD Allensbach (1975) for
1975 cannot be evaluated due to the lack of further information, the
results reported by Mummert and Schneider (2001) for 1998 and by
Lamnek, Olbrich and Schäfer (2000) for 1997 can be compared to those
reported by Pedersen (2003) for 2001 and Feld and Larsen (2011) for 2004
to 2007. The latter two studies use almost the same questionnaire and sam-
pling methods for their surveys. While the former two studies found con-
siderably more difference between West and East Germany than Pedersen,
these differences are again remarkable in the Feld and Larsen study.
Moreover, the proportion of respondents conceding that they had carried
out undeclared work in the survey by Mummert and Schneider (2001) is
double the proportion of people admitting that they conducted undeclared
work for payments in cash or in kind in the survey by Pedersen (2003) and
Feld and Larsen (2005). According to the information given in the survey
on the number of black hours and the actual wages paid for undeclared
work, this amounts to only 1.3 per cent of GDP for Germany as a whole in
2001 and 1 per cent in 2007. When the size of the shadow economy is meas-
ured assuming that labour productivity in the official and in the shadow
economy is the same, the shadow economy decreased from 4.1 per cent in
2001 to 3.2 per cent in 2007. In line with the indirect methods, the survey
method thus indicates a decline of the shadow economy in recent years
like the estimates by Schneider (2006). The strong differences in the size of
the shadow economy across approaches can be attributed to differences in
the methods, but also to the fact that the surveys conducted up to now are
focused on a smaller part of the shadow economy in the first place. They
only ask households and dismiss any undeclared work between firms.
Moreover, other illegal behaviour or transactions than black work are not
captured.
A similar development across time can be found for the assessment of
tax morale using the WVS. Torgler (2003b) compared tax morale between
East and West Germans after reunification. In his analysis for 1990, 67.2


per cent of the East Germans stated that tax evasion is never justifiable,
while at the same time only 40 per cent of West Germans were of that
opinion. In a subsequent study, Feld and Torgler (2007) use the most
recent WVS data and find that tax morale in East and West Germany in
1999 is still significantly different from each other. They further provide
evidence that the convergence in tax morale of both parts of Germany
cannot be attributed to deterrence, but is the result of the East Germans’
willingness to support the West German welfare state. The share of
respondents replying that tax evasion is not acceptable at all declined
considerably during the 1980s, by more than 20 per cent. Tax morale
continued to increase again remarkably until the end of the 1990s. If
this translated into a smaller amount of undeclared work or a higher tax
compliance, perhaps after a time lag, as might be cautiously guessed from
the slight decline in the shadow economy suggested by Schneider’s (2004,
2006) recent estimates, it adds to the main stylized fact that needs to be
explained: steady decreases in tax compliance until the end of the 1990s or
beginning of the new millennium that appear to turn around afterwards.
If we try to sum up our survey about the investigation into the German
shadow economy and tax compliance, we state a first stylized fact regard-
ing the underground economic activity:
Fact 1: The shadow economy and undeclared work in Germany steadily

increased until the beginning of the new millennium and decreased after-
wards, while tax morale in Germany declined and rose in about the same
period.
12.3 FINES AND PUNISHMENT
According to the prominent theoretical models in the literature, a reduc-

tion in the size of the shadow economy, if it indicates a trend reversal,
could be the result of increased deterrence, tax or social policy reforms, or
additional pressure from social norms. As the last of these develops slowly
over time and is influenced by more important institutional changes, such
as a shift in the tax authorities’ treatment of taxpayers (Feld and Frey,
2002b) or the introduction of direct democracy (Weck-Hannemann and
Pommerehne, 1989), we leave this possible explanation aside. The second
explanation is difficult and complex to capture empirically. Here, we
will centre on changes in deterrence. This is, however, rather difficult to
do as well, because the administration of the German tax system is not
uniform across the country, due both to the strong role that the state
(Länder) administrations have in auditing and tax investigations and to


the differences in the sentences imposed by the courts in different states.

Nevertheless, we will try to put things into perspective, by first portray-
ing the criminal tax code and the involved institutions and second by
delivering facts and data about its developments over time.
12.3.1 German Criminal Tax Code in a Nutshell
The prosecution of tax evasion is legally founded in the eighth chapter

(§§369–412) of the Abgabenordnung (general fiscal code; abbr. AO).
There, tax offences are distinguished into tax crimes and tax misdemean-
ors.3 Tax crimes are distinct from tax misdemeanours by the deliberate
act of tax evasion. Additionally, if not ruled differently in tax laws, the
general criminal code (Strafgesetzbuch, abbr. StGB), the code of criminal
procedure (Strafprozessordnung, abbr. StPO) and the regulatory offences
act (Ordnungswidrigkeitengesetz, abbr. OWiG) apply. Underneath this
level of statute law the administrative instructions for the crimes depart-
ments and the case law based on decisions of the Federal Finance Court
(Bundesfinanzhof, abbr. BFH), which functions as appellate court, are of
importance.
The main offence within the category of tax crimes is tax evasion (§370
AO). Tax evasion is committed by (1) a misrepresentation or concealing of
relevant information for taxation to tax authorities, by (2) neglect of tax
disclosure duties or by (3) refraining from compulsory use of tax stamps.
Tax evasion must be committed deliberately and the attempt is liable for
prosecution also. The statutory limitation period for prosecution of tax
crimes is 5 years (§78 StGB). But the limitation period for back duties is 10
years, and for back duties 6 per cent interest per year is added.
Possible sentences for tax evasion range from a penalty to a prison
sentence up to 5 years. In serious cases of tax evasion in combination with
abuse of an evader’s official authority or with fraudulent counterfeit the
possible sentence ranges from a minimum of 6 months up to 10 years of
imprisonment. If tax evasion is committed professionally or as an organ-
ized crime (§370a AO) the possible sentence is a minimum of 1 year up to
10 years of imprisonment. The sum of the penalty depends on the amount
of taxes evaded, the cooperation in the proceedings and the individual
daily net income of the tax evader. Imposable penalties start at the equiva-
lent of 6-times the daily net income of the tax evader and can be imposed
up to 360-times the daily net income (while the accountable part of daily
net income ranges from €1 to €5000). For a more severe sentence the judge
must impose a prison sentence.4
The Regional Tax Offices (Oberfinanzdirektion, abbr. OFD) have
developed sentence tables for standard cases of tax evasion. They serve


as the basis for the demand for penalties by the crime departments of the
tax authorities in the proceedings. While the final sentence is decided by a
judge and based on individual guilt and circumstances, the demand of the
crimes department provides the framework for penalty sentences in cases
of conviction. In Figure 12.2, we show six examples of mild, average and
severe sentencing demands.
The main offence within the category of tax misdemeanors could be liter-
ally translated as tax shortening (§378 AO). In comparison to tax evasion,
tax shortening is not enacted deliberately but is grossly negligent. Gross
negligence is, for example, presumed, if the taxpayer does not hand in a tax
return, does not inform himself about his tax duties or does not scrutinize
the tax statement of his tax advisor. Tax shortening can also be conducted
by tax advisors or accountants if they do not pay the necessary profes-
sional attention. In contrast to the Anglo-Saxon system, tax misdemeanors
in Germany can only be fined. For tax shortening, a possible fine up to
€50 000 can be imposed. The statutory limitation period for prosecution
of tax misdemeanors is 5 years (§384 AO). Other tax misdemeanors are the
different acts of preparation of or assistance to tax evasion (§372–382 AO).
Liable for prosecution are acts like fraud of documents adequate to achieve
tax allowable expenses, violation of legal obligations to keep records,
violation of obligations to notify foreign business transactions or opening
accounts under the wrong identity to camouflage transactions. For those
preparation acts a possible fine up to €5000 can be imposed. It is important
to note that the fined person does not need not to be the tax evader. Finally,
illegal trade with tax refunds (§373 AO) can be fined up to €50 000.
12.3.2 Involved Authorities, Investigation and Criminal Proceedings
In general, in the case of criminal proceedings, the prosecution authority is

responsible for the investigation procedure of criminal offences (§§160, 161
StPO). Notwithstanding, in cases of pure tax offences the tax authorities
are in charge of the investigation. Regularly, responsibility is only reas-
signed to prosecutors if (1) other non-tax-offences are connected or become
revealed or (2) a warrant of arrest has to be decreed. In addition, pros-
ecutors are entitled to retract the investigation from tax authorities or to
delegate back to them at any time. Within the tax authorities, professional
tax crimes and fines departments take the lead on the investigation and tax
crime investigators perform them. The starting point of criminal proceed-
ings is the opening of an investigation procedure. Such an investigation is
mandatory if an authority gains knowledge of an offence. The investigation
may have three possible outcomes: first, the authorities can decide to close
the proceedings (§§386, 389 AO); second, they can enact a penalty order


400
350

300
250
Berlin
Chemnitz
200 Magdeburg
Muenchen
150 Saarbruecken
Penalty demand –
Stuttgart
100
423
no. of daily net income equivalents
50
0
M M M M M M M M M M M M M M M M M M M M M M M M M M
D D D D D D D D D D D D D D D D D D D D D D D D D D
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
2 3 5 10 15 20 25 30 40 45 50 60 70 75 80 90 00 10 44 50 71 86 00 50 85 60
1 1 1 1 1 1 2 2 2 3
Sum of taxes evaded
Source: INF. Informationen fuer Steuern und Wirtschaft, 1998, 11, pp. Vf., own calculations
Figure 12.2 Penalty demands for tax evasion according to administrative instruction tables of different regional tax

offices
27/07/2011 16:16
(§§400, 407 StPO); and third, if sufficient evidence is collected, the prosecu-
tor can go to court and charge the suspect with tax evasion (§170 StPO).
A closure of the proceedings is chosen if suspicions cannot be proved. In
the case of weak evidence or minor severity, the authorities are also enti-
tled to offer a closure of the proceedings. In this case, the closure is com-
bined with a payment of the accused to avoid prosecution (§153a StPO).
Penalty orders are also used to cut short the proceedings, and are possible
for penalties and for sentences up to 1 year of imprisonment. They are
requested by the prosecutors and enacted by a judge. If a sentence per
penalty order is accepted by the accused, no court trial will take place and
the sentence becomes legally binding after 2 weeks. In cases of sufficient
and severe evidence for tax evasion or if the accused rejects a sentence per
penalty order, a court trial will take place.
The proceedings for misdemeanour remain under the competence of the
tax authorities. In contrast to criminal proceedings, the authorities have
discretion to decide whether they pursue an offence or not. The investi-
gation procedure is organized similarly to criminal investigations. Here,
the completion of investigation may have the following results: First, tax
authorities can close the investigation in case of insufficient proof. Second,
even in the case of probable cause tax authorities can close because of neg-
ligibility (normally if the sum of evaded taxes is less than €1533 or the sum
of forged expense vouchers does not exceed €2556). Third, if evidence of
criminal offence or connected criminal offences is secured, tax authorities
may transfer the proceedings to the prosecution authorities. Fourth, in
the case of verification of the allegations, the tax authorities can impose a
fine on the tax evader. If so, the tax evader is entitled to submit an appeal
within 2 weeks. Then the department has to decide whether the reason for
the appeal is acceptable or not. If not, the procedure is transferred to the
prosecution authority. If the prosecution authority decides to pursue the
issue, it will be negotiated at a municipal court.
12.3.3 Amendments to the Criminal Tax Code
Germany’s general criminal tax code, the sixth chapter of the

Abgabenordnung, remained relatively constant after its predecessor,
the Reichsabgabenordnung, was revised fundamentally and enacted as
AO on 1 January 1969. Besides editorial revision since then, only four
changes seem relevant to be noted: First, in 1993 taxes and custom duties
of the European Union (EU) fiscal code became taxes in the sense of the
German criminal tax code. Second, in 2001, the maximum fine for evasion
of withholding taxes was increased from €5000 to €25 000. Third, in 2002,
the possibility of handing in amended returns was applied to tax evasion


committed professionally or as an organized crime (§370a StGB). This

change is intended to support whistle-blowing in criminal organizations.
As a result, a person handing in amended returns can only by sentenced
to half of the usual sentence (a maximum of 5 years instead of 10 years).
Fourth, the Bundesgerichtshof, as the highest federal court below the
federal constitutional court, decided in 2008 to establish minimum sen-
tences for particular severities of tax evasion. Tax evasion of €100 000 and
above must be punished by prison sentences. In severe cases of more than
1 million euros of concealed taxes, no agreement between the state and the
taxpayers outside the court can be concluded anymore.
For 1982–2002, the developments of case law due to decisions of
the Finance Courts, especially the BFH, have been collected by Peter
Bilsdorfer (for example, 2003). The basic principle of sentencing must be
the individual extent of guilt and the amount of taxes evaded. Besides,
conviction effects on the evader’s status in society should be taken into
account. The evader’s actions after the commitment of the offence have to
be considered. Hereby, disclaims of the evasion in defence at court cannot
result in a severe sentencing. In contrast, sentencing should be milder if
the evasion was stopped deliberately, encouraged by authority officials
or police informers or if the proceedings are delayed by the actions of the
prosecution authorities. Furthermore, for prison sentences of 1 up to 2
years, a general release on licence should be generously considered and
additional legal obligation to keep records like a business can be consid-
ered. While some norms have been interpreted more precisely, general
amendments of the statute law were not made by the courts.
Finally, the administrative instructions for the crimes departments of
tax authorities have undergone three important revisions. First published
in 1983, new editions were enacted in 1991, 1995 and finally in 2004.
Nevertheless, Part 6 of the instructions, which deals with the sentencing,
remained the same, reflecting the court decisions described above and
binding the administration to them.
12.3.4 Developments in Sentencing Practice
While statutory law has not changed dramatically during the last few
decades, sentencing practice reveals stronger changes in deterrence. Thus,
we take a look at the development of a long time-series of central
indicators of sentencing practice.
Considering the developments in the overall number of sentences and
fines imposed for tax evasion in criminal proceedings in long time-series
(Figure 12.3), that is the number of cases that were actually prosecuted,
the figures show peaks in severe punishment at the beginning of the 1980s


45 000

40 000
35 000
Total prison sentences,
penalties and fines in
30 000 criminal proceedings
25 000
Prison sentences or
penalties for major
20 000 offences against tax
laws
No. of sentences/fines
15 000
Fines in criminal
426
proceedings for minor
10 000 offences against tax
laws
5 000
0
70 72 74 76 78 80 82 84 86 88 990 92 94 96 98 00 02 04
19 19 19 19 19 19 19 19 19 19 1 19 19 19 19 20 20 20
Year
Source: German Tax Offences Statistic, BMF (1970–2005)
Figure 12.3 Sentences and fines in criminal proceedings for tax evasion (all tax types)

27/07/2011 16:16
as well as in the mid and late 1990s. The changes in the 1990s mainly result
from offences in VAT, customs duties and excise duties (Figure 12.4), and
thus reflect the change in the law in 1993. With regard to the punishment
of minor and major offences, it seems evident that the number of cases
with fines for minor tax offences decreased a lot more than the number of
those with prison sentences or penalties for major tax offences.
Since the beginning of the 1970s, a steadily increasing trend can be
observed for the sum of nominal penalties imposed for tax evasion (Figure
12.5). For the sum of prison sentences (Figure 12.6), that is, in the more
serious cases of tax evasion, there is a decline which is first observable in
the beginning of the 1980s and which became steeper towards the end of
the 1980s. When offences in cases of indirect taxation are excluded, the
figures show a steady increase. This could, of course, reflect the fact that
the extent of tax evasion and the shadow economy has increased over time
as well. But taking these figures together with those shown in Figures 12.3
and 12.4, it becomes clear that the lower number of offences punished with
prison and fines was more than compensated for by more severe sentences
(higher fines and longer imprisonment).
12.3.5 Amendments at the Recent Margin: The ‘Black Activities’ Act’
In 2004 the ‘Law to intensify the fight against black activities and accom-
panying tax evasion’ (SchwarzArbG, Bundesrats-Drucksache 155/04a)
passed legislation to intensify deterrence. This law has provided for a
uniform framework to investigate and sanction undeclared work and the
shadow economy. It linked several different legal aspects of undeclared
work from tax laws regarding tax evasion and fraud to social security
or social assistance regarding social benefit fraud. Moreover, it mainly
comprises paragraphs which punish acts of concealment rather than
acts of evasion and focuses on employers. Misdemeanours for offences
against declarations and duties which favour ostensible self-employment
are defined and fined up to €300 000 by law. A fine for resistance against
or neglect of audit duties up to €50 000 is enacted and illegal employment
of foreigners can be sentenced by penalty or prison from 1 up to 6 years
depending on the severity of the offence and circumstances.
12.4 DETECTION BY TAX AUDITING AND

INVESTIGATIONS
While in the United States, differentiation between auditing and investiga-

tion is not common, in Germany, a distinction is made between regular


14 000

12 000
10 000
8 000 Prison sentences or

penalties for major offences
against tax laws
6 000
Fines in criminal
No. of sentences/fines
proceedings for minor
428
4 000 offences against tax laws
2 000
0
70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04
19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20
Year

Figure 12.4 Sentences and fines in criminal proceedings for tax evasion (excluding VAT, excise and custom duties)
27/07/2011 16:16
60€
50€

40€
Sum of
30€ penalties for tax
evasion - all tax
types
All tax types,
429
20€
excl. VAT,
Sum of penalties [in million €]

custom and
excise duties
10€
–€
70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04
19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20
Year

Figure 12.5 Sum of penalties imposed for tax evasion (all tax types, and excluding VAT, custom and excise duties)
27/07/2011 16:16
3500

3000
2500
2000
Sum of prison
sentences - all
tax types
1500
All tax types,
excl. VAT,
430
custom and
Sum of imprisonment [in years]

1000 excise duties
500
0
70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04
19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20
Year

Figure 12.6 Sum of prison sentences (all tax types, and excluding VAT, custom and excise duties)
27/07/2011 16:16
auditing without suspicion of tax evasion and precise investigations in

case of suspicion. Contrary to punishment, the intensity of control cannot
easily be inferred from the statistics. Figure 12.7 contains information on
the average number of firms per auditor. The figures reveal that tax audi-
tors have had to audit more and more firms on average since the begin-
ning of the 1990s. This results in a decreasing probability of detection.
However, this trend stopped at the end of the 1990s, due to the allocation
of more and more customs officials to the investigation of undeclared
work as indicated by Figure 12.8. The number of tax investigations thus
considerably increased at the end of the 1990s.
The ‘Black Activities Act’ put this practice on a new legal basis. The
financial control unit for undeclared work (Finanzkontrolle Schwarzarbeit,
FKS) as part of the customs regularly informs about its activities. These
statistics presented in Table 12.2 reveal an increased intensity of control
at the individual level across time while the control intensity of employ-
ers decreased. The number of concluded investigations first went up until
2007 and then slightly decreased. Overall, the sum of penalties and fines
in million euros and the sum of prison sentences in years rose steadily
indicating the considerable increase in deterrence the new law and the new
investigation unit brought. The FKS mainly concentrates on professional
forms of undeclared work, for example, in the construction sector where
the phenomenon is not infrequently accompanied by illegal immigration.
Thus, the statistics in Table 12.2 also give a flavour as to how widespread
undeclared work is in Germany. Overall, deterrence in Germany can be
characterized as having increased mainly due to the Black Activities Act.
Taking the two variables punishment and audit capabilities together, it
becomes obvious that deterrence as the product of these two variables gen-
erally increased from the mid-1980s until recently in Germany. This leads
us to state a second stylized fact regarding deterrence policy:
Fact 2: Deterrence has increased in Germany from the mid-1980s until 2001.
12.5 A TIME-SERIES ANALYSIS
As our two stylized facts indicate, there is first an increasing trend in the
size of the shadow economy from the 1970s until the beginning of the new
millennium and, second, deterrence has increased, though less steadily and
from the mid-1980s only. This could imply first that the shadow economy
increased despite deterrence efforts; second, that increases in deterrence
led to a crowding effect, as proposed by Frey (1997b), and thus increased
the shadow economy; or third, that increases in the shadow economy


800
700

600
Big enterprises
500
Medium enterprises
400 Small enterprises
Smallest enterprises
No. of enterprises
and self-employed
300
Sum
432
200
100
0
69 71 973 975 977 979 981 983 985 987 989 991 993 995 997 999 001 003 005
19 19 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2
Year
Source: German Tax Audits Statistic, BMF (1970–2005)

Figure 12.7 Audit capabilities (on-site inspections): average no. of enterprises per auditor
27/07/2011 16:16
60 000
50 000

40 000
Aggregate
Western Germany
30 000
Eastern Germany
No. of cases
20 000
433
10 000
0
92 93 94 95 96 97 98 99 00 01 02 03 04 05
19 19 19 19 19 19 19 19 20 20 20 20 20 20
Year
Source: German Tax Investigation Statistic, BMF (1970–2005)
Figure 12.8 Tax investigations and administrative assistance of tax investigators: Eastern, Western Germany and

Aggregate
27/07/2011 16:16
Table 12.2 Results of the work of the federal financial investigation unit
(Finanzkontrolle Schwarzarbeit FKS), 2005–2008
2005 2006 2007 2008

Controls of individuals at the 355 876 423 175 477 035 488 996
workplace
Control of employers 78 316 83 258 62 256 46 058
Concluded investigations in 81 290 91 820 117 441 106 960
criminal proceedings
Concluded investigations of 53 852 54 087 72 969 63 274
misdemeanors
Sum of penalties in million Euros 67.1 46.4 51.9 56.7
Sum of damages according to 562.8a 603.6 561.8 549.7
investigations in criminal
proceedings by FKS in million Euros
Sum of damages according to 37.0 39.1
investigations in criminal
proceedings by tax investigators of
the Laender induced by intelligence
of the FKS in million Eurosb
Sum of fines in million Euros 21.2 19.8 25.4 33.9
Sum of prison sentences in years 995 1123 1398 1556
Notes:
a
37 million Euros of it by special commissions.
b
In 2005 and 2006 not separately recorded.
Source: BMF (2009), Entwicklung der Bekämpfung der Schwarzarbeit und der illegalen
Beschäftigung, 11. Bericht der Bundesregierung, Monatsbericht, September 2009, p. 68
induced an increase in deterrence measures. In order to investigate these

relationships, we propose to test Granger causality using the data of tax
compliance and deterrence that are available as a time series. As outlined
above, the only measure of tax compliance that is available on a time-series
basis is the size of the shadow economy as measured by the MIMIC cum
currency demand approach (Schneider, 2006). These data are available
from 1970 until 2005. Deterrence is measured by four different variables.
Penalties per investigation, prison sentences per investigation and fines
per investigation are used as measures of punishment, while the number
of firms per audit supposedly captures the probability of detection. Due
to data availability problems, we have to restrict the period of analysis to
the years 1974 to 2001. We could thus unfortunately not capture the turn
in the trend of the German shadow economy. The results of our tests are
reported in Table 12.3.


Table 12.3 Tests of Granger Causality for deterrence and the size of the
shadow economy, Germany, 1974–2001, 4 lags
y x F (y d x) F (y S x)
Shadow Economy Penalties per Investigation 1.047 4.027**
Shadow Economy Prison Sentences per 1.156 0.555
Investigation
Shadow Economy Fines per Investigation 3.359** 0.501
Shadow Economy Firms per Auditor 1.207 1.374
Notes: ‘***’, ‘**’, ‘*’ or ‘(*)’ indicate that the null hypothesis of no causal relationship can
be rejected at the 0.1, 1, 5 or 10 per cent level, respectively.
The results of the Granger causality tests are at best mixed. Only
twice can the null hypothesis that no causal relationship exists be
rejected at the 5 per cent significance level. Both results contradict each
other, however. First, the hypothesis that penalties per investigation
do not Granger-cause the shadow economy cannot be rejected on any
conventional significance level, while the hypothesis that the shadow
economy does not Granger-cause penalties per investigation is rejected
at the 5 per cent level. Prison sentences per investigation and the shadow
economy are not Granger-causing each other, according to our results.
However, in the case of fines per investigations, the hypothesis that
they do not Granger-cause the shadow economy can be rejected at the
5 per cent significance level, while the reverse causality is not supported
by the test statistics. Finally, firms per audit and the shadow economy
are not causing each other according to the test statistics. Thus, there is
one precedent relationship from the shadow economy to penalties per
investigation and another from fines per investigation to the shadow
economy. Moreover, it should be noted that the time series of fines per
investigation follow a more cyclical pattern with local maxima in 1980,
1986, 1997 and 1999.
In sum, these results imply that the impact of punishment measures
on the size of the shadow economy is ambiguous, while the probability
of detection as measured by the number of firms per audit does not have
any impact on the size of the shadow economy. Unfortunately, our data
set does not cover the turning trend in the shadow economy numbers,
nor are data on the number of investigations available for a longer
time period than the 1990s. We can thus only present this evidence as
preliminary.


12.6 CONCLUSION
In this chapter, we have investigated whether tax compliance as measured
by the size of the shadow economy, undeclared work or tax morale is actu-
ally negatively affected by deterrence as supposed in the seminal analysis
of Allingham and Sandmo (1972) and a lot of the follow-up literature as
well as in practice. The unique descriptive data presented in this chapter
indicate that the size of the shadow economy follows a steadily increas-
ing trend from the beginning of the 1970s until the beginning of the new
millennium and drops afterwards. Similar inferences can be drawn from
survey data on undeclared work or tax morale though no continuous time
series are available. The more diverse picture of deterrence in Germany
since the beginning of the 1970s also allows for the conclusion that deter-
rence has increased in Germany from the mid-1980s to the beginning of
the new millennium. With two steadily increasing time series, simple OLS
regressions are not useful. With Granger causality tests, we find that the
shadow economy Granger-causes penalties per investigation, while fines
per investigation Granger-cause the shadow economy. Investigation in
terms of number of firms per auditor does not have any impact on the
shadow economy.
In a further step with first simple OLS regressions (not reported here),
no significant effect of deterrence on the size of the shadow economy could
be found. Our analysis has an additional drawback: the data period for
the more systematic analysis only covers the years 1974 to 2001 such that
the most interesting turning point in the trend of the shadow economy
may not be captured sufficiently by the data yet due to time lag properties.
Nevertheless, our analysis casts some doubts as to the usefulness of a pure
deterrence policy to fight the shadow economy.
ACKNOWLEDGEMENTS
Lars P. Feld gratefully acknowledges a grant from the German Science

Foundation (DFG SPP 1142).
NOTES
1. Veit (1927) and Schmölders (1932) have already argued that tax compliance is shaped by
the relationship between citizens as taxpayers and the state. See Schöbel (2005).
2. See FORES (1977, 1987, 1997). Additionally two studies about the taxpayers’ reactions
in small and medium sized companies were conducted (FORES 1963, 1982). Another
study about tax simplification completes their efforts (1994).


3. By definition, in the United States, misdemeanours are punished by fines, penalties or

less than a year of imprisonment. In Germany, only tax crimes can be punished by penal-
ties and imprisonment.
4. Furthermore the breach of import and export ban (§372 AO), organized contraband
crime (§373 AO) and fencing contraband or non-taxed goods (§374 AO) are liable to
prosecution.
REFERENCES
Alm, J., G.H. McClelland and W.D. Schulze (1992), Why do people pay taxes?
Alm, J., I. Sanchez and A. De Juan (1995), Economic and non-economic factors in
tax compliance. Kyklos, 48, 3–18.
Becker, G.S. (1968), Crime and punishment: an economic approach. Journal of
Beron, K.J., H.V. Tauchen and A.D. Witte (1992), The effect of audits and socio-
economic variables on compliance. In J. Slemrod (ed.), Why People Pay Taxes.
Ann Arbor, MI: University of Michigan Press, pp. 67–89.
Bilsdorfer, P. (2003), Die Entwicklung des Steuerstraf- und Steueror-
dnungswidrigkeitenrechts. Neue Juristische Wochenschrift (NJW), 32, 2281–2288.
Blackwell C. (2010), A meta-analysis of incentive effects in tax compliance experi-
ments. In J. Alm, J. Martinez-Vazquez and B. Torgler (eds), Tax Compliance
and Tax Evasion. London: Routledge, pp. 97–112.
Braithwaite, V. and M. Wenzel (2006), Integrating explanations of tax evasion and
avoidance. In A. Lewis (ed.), Cambridge Handbook of Psychology and Economic
Behaviour, Cambridge, UK: Cambridge University Press.
De Juan, A., M.A. Lasheras and R. Mayo (1994), Voluntary tax compliant behav-
ior of Spanish income taxpayers, Public Finance, 49, 90–105.
Dubin, J.A. and L.L. Wilde (1988), An empirical analysis of federal income tax
auditing and compliance. National Tax Journal, 41, 61–74.
Dubin, J.A., M.J. Graetz and L.L. Wilde (1987), Are we a nation of tax cheaters?
New econometric evidence on tax compliance. American Economic Review, 77,
240–245.
Elffers, H., R.H. Weigel and D.J. Hessing (1987), The consequences of different
strategies for measuring tax evasion behavior. Journal of Economic Psychology,
8, 311–337.
Engel, E. and J.R. Hines (1999), Understanding tax evasion dynamics. NBER
Working Paper No. 6903, NBER, Cambridge.
Distortion, Cambridge, UK: Cambridge University Press.
Feld, L.P. and B.S. Frey (2002a), The tax authority and the taxpayer: an explora-
tory analysis. Unpublished Manuscript, University of St. Gallen.
Feld, L.P. and B.S. Frey (2002b), Trust breeds trust: how taxpayers are treated.
Economics of Governance, 3, 87–99.


Feld, L.P. and B.S. Frey (2005), Illegal, immoral, fattening or what? How
deterrence and responsive regulation shape tax morale. In C. Bajada and F.
Schneider (eds), Size, Cause and Consequences of the Underground Economy.
Aldershot, UK: Ashgate, pp. 15–37.
Feld, L.P. and B.S. Frey (2007), Tax compliance as the result of a psychological
tax contract: the role of incentives and responsive regulation. Law and Policy,
29, 102–120.
Feld, L.P. and C. Larsen (2005), Black activities in Germany in 2001 and in 2004.
A comparison based on survey data. Study No. 12, The Rockwool Foundation
Feld, L.P. and C. Larsen (2011), Undeclared Work, Deterrence and Social Norms:
The Case of Germany. Berlin: Springer.
Feld, L.P. and B. Torgler (2007), Tax morale after the re-unification of Germany:
results from a quasi-natural experiment. CESifo Working Paper No. 1921,
February.
FORES (Forschungsstelle für empirische Sozialökonomik) (1963), Steuerbelastung
und Steuerbelastungsgefühl des selbstständigen Mittelstandes. Cologne.
FORES (Forschungsstelle für empirische Sozialökonomik) (1977), Steuermentalität
der bundesdeutschen Steuerzahler und deren Vorstellungsbild vom Bund der
Steuerzahler. Cologne.
FORES (Forschungsstelle fuer empirische Sozialöekonomik) (1982), Die
Besteuerung mittelständischer Unternehmen: Eine finanzpsychologische Unter-
suchung des Zensitenverhaltens kleiner und mittelständiger Unternehmen.
Cologne.
FORES (Forschungsstelle für empirische Sozialökonomik) (1987), Die
Steuermentalität der bundesdeutschen Steuerzahler und deren Vorstellungsbild
von einer Interessenvertretung der Steuerzahler. Cologne.
FORES (Forschungsstelle für empirische Sozialökonomik) (1994), Steuerver-
einfachung und Steuergerechtigkeit. Cologne.
FORES (Forschungsstelle für empirische Sozialökonomik) (1997), Steuermen-
talität und Steuermoral der bundesdeutschen Bevölkerung und deren Einstellung
zur Steuerreform. Cologne.
Frey, B.S. (1997a), A constitution for knaves crowds out civic virtues. Economic
Journal, 107, 1043–1053.
Frey, B.S. (1997b), Not Just for The Money. An Economic Theory of Personal
Motivation. Cheltenham, UK: Edward Elgar.
Frey, B.S. and L.P. Feld (2002), Deterrence and morale in taxation: an empirical
analysis. CESifo Working Paper No. 760, August 2002.
served’ variable. European Economic Review, 26, S. 333–353.
Friedland, N. (1982), A note on tax evasion as a function of the quality of informa-
tion about the magnitude and credibility of threatened fines: some preliminary
research. Journal of Applied Social Psychology, 12, 54–59.
Friedland, N., S. Maital and A. Rutenberg (1978), A simulation study of income
tax evasion. Journal of Public Economics, 10, 107–116.
Gneezy, U. and A. Rustichini (2000), A fine is a price. Journal of Legal Studies,
29, 1–17.
Graetz, M.J. and L.L. Wilde (1985), The economics of tax compliance: facts and
fantasy. National Tax Journal, 38, 355–363.
IAW (Institut für angewandte Wirtschaftsforschung) (2006), Steuermoral – Das


Spannungsfeld von Freiwilligkeit der Steuerhinterzahlung und Regelverstoß durch

Steuerhinterziehung. IAW Forschungsbericht Nr. 64, Tübingen.
Institut für Demoskopie (IfD) Allensbach (1975), Freizeitarbeit 1974. Studie
im Auftrag der Kommission für wirtschaftlichen und sozialen Wandel,
Allensbach.
Kanniainen, V., J. Pääkönen and F. Schneider (2004), Fiscal and ethical deter-
minants of shadow economy: theory and evidence. Mimeo, Johannes-Kepler
University Linz.
economy, 1955–1980. Zeitschrift für die gesamte Staatswissenschaft (ZgS)/
Journal of Institutional and Theoretical Economics (JITE), 139, 197–214.
Klepper, S. and D. Nagin (1989), Tax compliance and perceptions of the risks of
detection and criminal prosecution. Law and Society Review, 23, 209–240.
Lackó, M. (1998), The hidden economies of Visegrad countries in international
comparison: a household electricity approach. In L. Halpern and Ch. Wyplosz
(Eds), Hungary: Towards a Market Economy. Cambridge, UK: Cambridge
University Press.
Lamnek, S., G. Olbrich and W.J. Schäfer (2000), Tatort Sozialstaat: Schwarzarbeit.
Leistungsmissbrauch, Steuerhinterziehung und ihre (Hinter)Gründe. Leske and
Budrich, Opladen.
Larsen, C. (2002), Underdeklaration af personlig indkomst måt ved sammenligning
af dansk skatte- og nationalregnskabsstatistik (Unreported Personal Income
Estimated by Examining Discrepancies between Danish National Accounts
and Tax Statistics). Mimeo, Institute of Economics, University of Copenhagen,
Copenhagen.
Lyssiotou, P., P. Pashardes and T. Stengos (2004), Estimates of the black economy
based on consumer demand approaches. Economic Journal, 114, 622–640.
Mackscheidt, K. (1994), Die Entwicklung der Kölner Schule der Finanzpsycho-
logie. In C. Smekal and E. Theurl (eds), Stand und Entwicklung der
Finanzpsychologie. Baden-Baden: Nomos.
Mackscheidt, K. (2004), Die Entwicklung der Steuermoralforschung. In K.
Bizer, A. Falk and J. Lange (eds), Am Staat vorbei: Transparenz, Fairness und
Partizipation kontra Steuerhinterziehung. Berlin: Duncker & Humbolt.
Martinez-Vazquez, J. and M. Rider (2005), Multiple modes of tax evasion: theory
and evidence. National Tax Journal, 58, 51–76.
united Germany? Finanzarchiv, 58, 286–316.
Scandinavia: a measurement based on questionnaire surveys, Study No. 10, The
Pickhardt, M. and J. Sardà Pons (2006), Size and scope of the underground
economy in Germany. Applied Economics, 38, 1707–1713.
Pommerehne, W.W. and B.S. Frey (1992), The effects of tax administration on tax
morale. Unpublished Manuscript, University of Zurich.
Pommerehne, W.W. and H. Weck-Hannemann (1996), Tax rates, tax administra-
tion and income tax evasion in Switzerland. Public Choice, 88, 161–170.
Schmölders, G. (1932), Steuermoral und Steuerbelastung. Vierteljahresschrift für
Steuer- und Finanzrecht, 6, 151–161.
Schmölders, G. (1960), Das Irrationale in der öffentlichen Finanzwirtschaft.
Probleme der Finanzpsychologie. Hamburg: Rowohlt.


Schmölders, G. (1964), Staatsbürgerliche Gesinnung und Steuermoral. In F.

Schäfer (eds), Finanzwissenschaft und Finanzpolitik. Tübingen: Mohr Siebeck.
Schmölders, G. (1970), Survey research in public finance: a behavioural approach
to fiscal theory. Public Finance, 25, 300–306.
Schmölders, G. (1978), Verhaltensforschung im Wirtschaftsleben: Theorie und
Wirklichkeit. Munich: Wirtschaftsverlag Langen-Mueller/Herbig.
Schmölders, G. and B. Strümpel (1969), Vergleichende Finanzpsychologie. In
B. Beichelt and J. Daviter (eds), Steuernorm und Steuerwirklichkeit. Cologne:
Westdeutscher Verlag.
Schneider, F. (2004), Arbeit im Schatten: Wo Deutschlands Wirtschaft wirklich
wächst. Wiesbaden: Gabler.
Schneider, F. (2006), Shadow economies of 145 countries around the world.
What do we know? In B.-A. Wickström (ed.), Schattenwirtschaft. Schriften des
Vereins für Socialpolitik, Berlin: Duncker & Humblot.
Schneider, F. and D. Enste (2000), Shadow economies: size, causes, consequences.
Journal of Economic Literature, 38, 77–114.
Schöbel, E. (2005), Otto Veit and subsequent developments of tax morality. In J.G.
Backhaus (ed.), Essays on Fiscal Sociology. Frankfurt: Peter Lang, pp. 187–210.
Schwartz, R.D. and S. Orleans (1967), On legal sanctions. University of Chicago
Law Review, 34, 282–300.
Slemrod, J. and S. Yitzhaki (2002), Tax avoidance, evasion and administration. In
A.J. Auerbach and M. Feldstein (eds), Handbook of Public Economics, Vol. 3.
Amsterdam: North-Holland, pp. 1423–1470.
Slemrod, J., M. Blumenthal and C.W. Christian (2001), Taxpayer response to
an increased probability of audit: evidence from a controlled experiment in
Minnesota. Journal of Public Economics, 79, 455–483.
Spicer, M.W. and S.B. Lundstedt (1976), Understanding tax evasion. Public
Finance, 31, 295–305.
Thomas, J. (1999), Quantifying the black economy: measurement without theory
yet again. Economic Journal, 109, F381–F387.
Torgler, B. (2003a), Tax morale: theory and empirical analysis of tax compliance.
Dissertation, University of Basel.
Torgler, B. (2003b), Does culture matter? Tax morale in an East-West-German
comparison. Finanzarchiv, 59, 504–528.
Varma K.N. and A.N. Doob (1998), Deterring economic crimes: the case of tax
evasion. Canadian Journal of Criminology, 40, 165–184.
Veit, O. (1927), Grundlagen der Steuermoral – Eine finanzsoziologische Studie,
Zeitschrift für die gesamte Staatswissenschaft, 83, 317–349.
Weck, H. (1983), Schattenwirtschaft: EineMöglichkeit zur Einschränkung der
öffentlichen Verwaltung? Eine ökonomischeAnalyse. Bern: Peter Lang.
Weck-Hannemann, H. and W.W. Pommerehne (1989), Einkommensteuer-
.


PART V
Corruption and the shadow economy



13 The impact of institutions on the
shadow economy and corruption:
a latent variables approach
Axel Dreher, Christos Kotsogiannis and
Steve McCorriston
13.1 INTRODUCTION
Corruption has always been present in societies, in one form or another.1

It is, in general, a significant contributor to low economic growth, to dis-
tortionary investment and provision of public services. As such it increases
inequality to such an extent that international organizations have identi-
fied it as ‘the single greatest obstacle to economic and social development’
(World Bank, 2001).2 This view is not, however, supported unanimously.
Routine corruption may be efficiency enhancing. As Leff (1964, p. 11) puts
it: ‘If the government has erred in its decision, the course made possible
by corruption may well be the better one’. Corruption may also ‘grease
the wheels’ in the rigid public administration, a view that, recently, has
been supported empirically in the contributions of Dreher and Gassebner
(2007) and Méon and Weill (2008), who provide evidence in favour of the
‘grease the wheels’ hypothesis.3
An important and, up until recently, relatively unexplored element of
corruption is its relationship with the shadow economy. Unquestionably,
corruption and shadow economy share a common characteristic: they are
both, in general, illegal.4 In understanding the relationship of corruption
with the shadow economy, it is important to understand what causes the
shadow economy. With the risk of over-simplification, two schools of
thought can be identified. One school of thought identifies high tax and
social security burdens as the principal causes.5 Economic agents, the story
goes, are not willing to pay high taxes and so are driven out of the official
economy.6 The second school of thought identifies institutional quality7
as the main cause of driving economic agents underground. This view is
based on the presumption that governments are not sufficiently constitu-
tionally constrained and, hence, use their coercive powers to exploit the
443


citizenry. The natural response of economic agents to this government

behaviour is to go underground losing all publicly provided benefits. It
is, therefore, intuitive that there is a potential relationship between cor-
ruption (and the misbehaviour of government in general) and the shadow
economy. But what is the precise relationship? Is it that a high corruption
leads to high unofficial economy and so they are complements, or to a low
unofficial economy? Theoretically, both types of relationship may stand.
Indeed,8 Choi and Thum (2005) show that the shadow economy mitigates
government-induced distortions leading to enhanced economic activities
in the official sector, corruption and the shadow economy then being
substitutes.9 This is a view that seems to be in line with Rose-Ackerman
(1997) who notes that ‘going underground is a substitute for bribery,
although sometimes firms bribe officials in order to avoid official taxes’
(p. 21). Alternatively, Friedman et al. (2000) show that corruption and the
shadow economy are positively related. When faced with weak economic
institutions entrepreneurs go underground hiding their activities. As a
consequence, tax revenues fall as well as the quality of public adminis-
tration further reducing a firm’s incentive to remain official. Using an
inspector–tax payer model, Hindriks, Muthoo and Keen (1999) also show
that the shadow economy is a complement to corruption. This is because,
in this case, the tax payer colludes with the inspector so the inspector
underreports the tax liability of the tax payer in exchange for a bribe.
According to Dreher, Kotsogiannis and McCorriston (2009) – a frame-
work which we re-visit in Section 13.2 – improvements in institutional
quality reduce the size of the shadow economy and affect the corruption
market. The exact relationship between corruption and institutional
quality is, however, ambiguous and depends on the relative effectiveness
of institutional quality in the shadow and corruption markets.10 We turn
to this shortly.
The empirical evidence, though limited, is in favour of corruption
and the shadow economy being complements. Johnson, Kaufmann and
Zoido-Lobatón’s (1998) investigation of 49 countries in Latin America,
the OECD, and the former Soviet Union bloc find a statistically significant
positive relationship between the various measures of corruption and the
shadow economy. Using data for 69 countries, Friedman et al. (2000) find
evidence that ‘more bureaucracy, greater corruption, and a weaker legal
environment are all associated with a larger unofficial economy’ (p. 460).11
They conclude that ‘poor institutions and a large unofficial economy go
hand in hand’ (p. 460).12 The relationship between the shadow economy
and corruption may depend on the level of income, an issue taken up
by Dreher and Schneider (2010). Dreher and Schneider (2010) argue
that corruption and the shadow economy are substitutes in high income


The impact of institutions on the shadow economy 445
countries while they are complements in low income countries. The

hypotheses are tested for a cross-section of 98 countries. The results show
that there is no robust relationship between corruption and the size of the
shadow economy, when perceptions-based indices of corruption are used.
Employing an index of corruption (estimated in Dreher, Kotsogiannis and
McCorriston (2007)) based on a structural model, however, corruption
and the shadow economy are complements in countries with low income,
but not in high income countries.
Aside from the theoretical aspects of the relationship between corrup-
tion and the shadow economy, an attempt to establish empirically the
relationship faces a difficulty; that corruption and the shadow economy
do not lend themselves to measurement and so they are inherently latent
variables.13 Most of the studies quoted above deal with this by using
indices of corruption and indicators for the shadow economy derived from
different sources, rather then deriving them jointly. The indices on per-
ceived corruption, however, have recently been criticized. The reason for
such criticism stems from the fact that perceived corruption is completely
unrelated to actual corruption once other relevant factors are controlled
for (Mocan, 2004). Corruption might also not be related to other forms
of ‘illegal activities’ (Weber Abramo, 2007). Analyzing the relationship
between corruption and the shadow economy using a measure of corrup-
tion that is not based on perceptions is, thus, clearly warranted.14
A fruitful avenue to deal with this difficulty is the use of structural equa-
tion modelling.15 Though the fundamental importance of the quality of
institutions for combating corruption and the shadow economy is well rec-
ognized, to the best of our knowledge, very few studies exist that attempt
to articulate, and empirically estimate, the relationship treating the vari-
ables in question as inherently latent. This is the objective of this chapter.
To capture the relationship between institutional quality, the shadow
economy and corruption in a stark way, we follow Dreher, Kotsogiannis
and McCorriston (2009).16 The theoretical model shows that corruption
and the shadow economy are substitutes in the sense that the existence
of the shadow economy reduces the propensity of officials to demand
rents from firms. It also shows that institutional quality, under a certain
condition, may reduce both the magnitude of corruption and the size of
the shadow economy. The predictions of the model are then tested and
confirmed in a sample of 18 OECD countries.17 To the best of our knowl-
edge, this is the first study that attempts to confirm a specific relationship –
treating these variables as inherently latent – between institutional quality,
corruption and shadow economy.
The organization of the chapter is as follows. In Section 13.2, we intro-
duce a simple model that investigates the relationship between institutional


quality, corruption and the shadow economy. In Section 13.3, we present

the structural equation model used to confirm the hypothesized relation-
ship between institutional quality, corruption and the shadow economy.
We also derive the scores of the latent endogenous variables and so derive
an index of corruption and shadow economy in the 18 OECD countries.
In Section 13.4, we present and discuss the results. Finally, in Section 13.5
we conclude.
13.2 DESCRIPTION OF THE MODEL
Since the model is essentially that of Dreher, Kotsogiannis and McCorriston

(2009), it is only briefly described here.18 There is a continuum of firms, each
of which is characterized by a parameter q distributed with cumulative prob-
ability F(q) with F9(q) $ 0. q is known to the firms but not to the officials.
There are two markets in this economy: an official, within which corrupt
officials operate, and the shadow market. What distinguishes the former
from the latter market is the cost of operating in the corruption market.
Public officials are corrupt and (attempt to) expropriate both the govern-
ment property and firms. Firms, in order to operate in the official market,
must purchase a permit, which is government property, from public officials
at price denoted by m. To avoid, however, expropriation firms can enter the
shadow market.19 Entering the shadow market is, however, not cost free.
The (expected) cost firms face is the fine associated with operating in the
shadow economy. To minimize the possibility of getting caught, firms scale
down their degree of operation. We turn to this shortly below. Independent
of their earning ability and market of operation, firms also incur fixed oper-
ating costs, such as the purchase of capital, denoted by r > 0.
Institutions impact differently on the corruption and shadow markets.
To capture this, and after denoting the effectiveness of institutions by e, it
is assumed that both the quality of the official market, denoted by q(e),20
and the quality of the shadow market, denoted by s(e), improve with more
effective institutions. It is, thus, the case that q9(e), s9(e) > 0, but with
q9(e) ? s9(e).21 As noted earlier, the effectiveness of institutions in the two
markets should be interpreted broadly to include elements such as, among
others, the rule of law, and the complexity of the tax system and regulatory
discretion. For a study on the effect of the complexity of the tax system on
the size of the shadow economy, see Schneider and Neck (1993).
In the framework considered here, events unfold as follows. Firstly,
and for given institutional quality, public officials must decide what level
of rent m to set in order to maximize rent revenues. Secondly, and for
given level of rents, firms decide to either enter the official market, the


shadow economy market or not to enter the markets at all.22 The underly-
ing assumption here is that there is commitment on the part of corrupt
public officials. When public officials announce the level of rent demanded
they cannot renege. This seems a natural assumption on the working of
institutions in OECD countries (that comprise the empirical study of
Section 13.3). To investigate the precise relationship between the corrupt
and shadow markets and their relationship with the quality of the institu-
tions, the strategy is to consider the behaviour of public officials when the
shadow market is not present and when it is. We start with the former.
13.2.1 Corruption in the Absence of the Shadow Market
A firm of type q that operates in the official economy realizes profits
p 5 q 2 r 2 mq (e) , (13.1)
where mq(e) > 0 reflects the monetary cost of institutional quality to a firm
of type q with, in particular, mq9(e) > 0.23 Underlying this is the idea that
an improvement of institutional quality is costly for firms being caught
engaging in corrupt activities. It then follows that the marginal firm will
enter the official market if and only if q $ r 1 mq(e) ; q; which defines a
cutoff level of q, denoted by q, such that all firms with earning ability above
this will enter the official market by purchasing the permit at the cost of m.
The proportion of firms, then, that will enter the market is 1 2 F(q). For
notational convenience, denote G(q) ; 1 2 F(q), with G9 (q) # 0.
In the absence of a shadow economy, the public official maximizes
R 5 mG (q) , (13.2)
with the appropriate choice of m, giving necessary condition
Rm 5 G 1 mGrq (e) 5 0. (13.3)
Equation (13.3) implicitly defines the equilibrium level of rents m*(e, r)

with, in particular,
m*r (e) 5 2m*qr (e) /q (e) , 0. (13.4)
Total equilibrium rents, denoted by R*, decrease with institutional quality.

To see this, evaluate (13.2), using (13.3), and perturb to find
Rr (e) 5 (m* (e)) 2Grqr (e) , 0. (13.5)


That total equilibrium rents decrease with institutional quality is not

surprising. The reason for this is that institutional quality, e, does not, as
an envelope property, affect rents through m*(e) but only through q(e);
for given m*(e) an increase in e increases the cutoff level of q reducing the
official market and thereby rents. It is also interesting to observe that an
improvement in institutional quality does not affect the equilibrium size
of the official market, denoted by H. To see this, notice that the size of the
official market is given by H(e) 5 G(q(e)). Differentiating H, and evaluat-
ing using (13.3) and (13.4), gives, as an envelope property, H9(e) 5 0. That
an increase in institutional quality reduces rents (but leaves the official
market unchanged) is perhaps not surprising. But it does serve as a useful
benchmark to compare against the equilibrium outcomes in the presence
of the shadow economy. This is what we turn to next.
13.2.2 Corruption in the Presence of the Shadow Market
Consider now an economy in which there is an underground sector. As

noted previously, if a firm enters the shadow market it pays no rents but
it requires to scale down the level of the economic activity so as not to be
detected. To capture this in a simple way, it is assumed that for a given
s(e), the cost of entering the shadow economy is qs(e). Denoting vari-
ables pertaining to the shadow economy with a superscript s, profits are
given by
ps 5 q 2 r 2 qs (e) . (13.6)
It is also natural to assume that the cost of entering the shadow market
is not too high that is, s(e) < 1. Consider now the choice that a typical
firm, with characteristic q, faces. A simple comparison between (13.1) and
(13.6) reveals that if q $ mq(e)/s(e) then this firm participates in corrup-
tion, whereas if r / (1 2 s(e)) # q < mq(e)/s(e) then it enters the shadow
economy. With a sufficiently low q < r / (1 2 s(e)) the firm makes, follow-
ing (13.6), negative profits and hence ceases activity altogether.
Denoting total rents by Rs officials, choosing m, and anticipating the
market equilibrium, maximize
mq (e)
Rs 5 mGa b, (13.7)
s (e)
with necessary condition
Rsm 5 G 1 mGrq (e) /s (e) 5 0, (13.8)


which implicitly defines the optimal rents ms*(e) with, for later use,
ms*r 5 2ms*e21 (eq 2 es) , (13.9)
where eh ; h9(e)e/h > 0 denotes the elasticity of h 5 q, s with respect to

institutional quality e. Equation (13.9) reveals that the change in equilib-
rium rents, due to a change in institutional quality e, takes the opposite
sign of eq 2 es. Consider, for instance, eq > es. In this case, corruption is
easier to detect, relative to the illegal activities taking place in the shadow
market, and so the public officials respond to an increase in institu-
tional quality by reducing the level of rents demanded, that is m s*9 < 0.
Analogous reasoning applies to eq < es.
It is now straightforward to see that the equilibrium level of rents,
denoted by Rs*, when the shadow economy exists is less than the level of
rents when the shadow economy does not exist that is, ms*(e) < m*(e). To
see this, evaluate (13.7) at the optimal level of rents m*(e) and then, start-
ing from m*(e), consider a small change dm to find
s (e) 2 1
dRs* 5 Ga bdm , 0, (13.10)
s (e)
with the inequality following upon s(e) < 1. Clearly, then, since Rs* is
strictly concave in m, ms*(e) < m*(e) and so optimal rents under the exist-
ence of the shadow economy is smaller than without. This is intuitive. The
shadow economy, for given institutional quality, imposes a constraint on
the officials. When firms face high rents, they have the option of going
underground. The lower ms*(e), the more firms enter the official economy.
Hence, when the shadow economy exists, corruption is lower and so the
official economy is larger implying that the shadow market and the official
economy are complements.24
What about total rent revenues? One would expect that these depend on
the relative magnitudes of eq and es. Indeed, perturbing optimal rents Rs*,
and evaluating at ms* using (13.8), one obtains
(ms*) 2qGr
Rs*r 5 (eq 2 es) , (13.11)
se
and so, with G9 < 0, total rent revenues decrease (increase) in institutional
quality when eq > es (eq < es). The intuition of this is similar to the one
given when the shadow economy was not present. An increase in institu-
tional quality does not affect equilibrium revenues through m s* but only


via the cutoff level of q thereby reducing the proportion of firms enter-
ing the official market.25 Interestingly, however, improved institutional
quality does not affect the size (number of firms) of the official market but
it does affect the magnitude of rent revenues. This is for the same reason
as that given above, but modified here to incorporate the shadow market.
Specifically, with eq > es (eq < es), the cutoff level of q increases (decreases)
thereby reducing (increasing) the number of firms entering in the official
market. But this is not the end of the story. Following (13.9), to maintain
maximum rent revenues, officials reduce (increase) the equilibrium level of
rents undoing the change in the cutoff level of q. Thus, overall, the size of
the official market remains the same.
An increase in institutional quality, e, however, unambiguously reduces
the size of the shadow economy. To see this, notice first that, following
the discussion in the preceding paragraph, with the equilibrium size of the
official market being unaffected by institutional quality, there are no firms
contemplating entering the official market after an improvement in institu-
tional quality has taken place. What institutional quality affects, however,
is the marginal firm that decides to stay in the shadow market or of exiting
the market completely. It is straightforward to show that the size of the
shadow economy decreases with changes in institutional quality, e. To see
this, denoting the proportion of firms entering the shadow economy by
S(e) 5 F(r / (1 2 s(e))), we thus have S9(e) 5 F9rs9(e) / (1 2 s(e))2 < 0.
Summarizing:
Proposition 1 Ceteris paribus, when eq > es (eq < es), countries with better
institutional quality are characterized by an official market with a low (high)
magnitude of corruption. Moreover, while the size of the official market is
independent of institutional quality, the size of the shadow economy reduces
with an improvement in institutional quality.
Proposition 1 emphasizes that the size of the shadow market unambigu-

ously decreases with institutional quality, whereas the size of the corrup-
tion market remains unchanged. The exact relationship, however, between
the magnitude of corruption and institutional quality is ambiguous as it
depends on the relative effectiveness of the institutional quality in the two
markets. If, for instance, institutions are more effective in combating cor-
ruption (and so more costly for the firms engaged in corrupt activities)
relative to the shadow economy, in the sense that eq > es, then a further
improvement in institutional quality reduces the magnitude of corruption.
Officials, anticipating that the official market is less profitable for firms,
reduce the level of rents demanded. But the opposite is also true. If the
effectiveness of the institutions is biased towards combating the shadow


economy, eq < es, then institutional quality may exacerbate corruption.

The preceding analysis has emphasized that:
1. The corruption and shadow markets are substitutes in the sense that
the existence of the shadow market is associated with smaller levels of
rents;
2. The effect of institutional quality on the shadow market is unambigu-
ously negative; whereas
3. The effect of institutional quality on the magnitude of corruption is
ambiguous and depends on the relative effectiveness of institutional
quality on the two markets.
We now take the model to the data. We start with a description of the
methodology used. As briefly touched upon in Section 13.1, the diffi-
culty with the variables of corruption, the shadow economy and insti-
tutional quality is that they are inherently latent since they do not lend
themselves easily to measurement. A fruitful and promising approach
to estimating latent variables is to use structural equation modelling.26
This methodology is ‘becoming a powerful approach’ to these problems,
Giles (1999, p. 372) and, applied to the present context, is outlined in
the next section.
13.3 A STRUCTURAL EQUATION MODEL
Structural equation modelling allows a set of relationships between one

or more independent variables and one or more dependent variables to
be examined.27 Both independent and dependent variables can be either
latent variables (factors) or measured variables (indicators). The underly-
ing assumption is that the observed variables are perfectly correlated (or at
least nearly so) with the latent variables that they measure.28
13.3.1 General Model Specification
It is instructive to provide the general specification of the model. A struc-

tural equation model has the following form:
x 5 Lxx 1 d, (13.12)
y 5 Lyh 1 e, (13.13)
h 5 Bh 1 Gx 1 z, (13.14)


where x 5 (x1, . . ., xq)9 and y 5 (y1, . . ., yp)9 are the observed indicators
of the latent factors x 5 (x1, . . ., xn)9 and h 5 (h1, . . ., hm)9 respectively. d
(a q 3 1 vector) and e (a p 3 1 vector) are the measurement errors for x
and y respectively. Lx is a q 3 n matrix of coefficients (‘loadings’) relating
manifest exogenous variables x to exogenous latent variables x whereas
Ly is a p 3 m matrix of coefficients relating manifest endogenous vari-
ables y to endogenous latent variables h. B is the m 3 m coefficient matrix
showing the influence of the latent endogenous variables on each other. G
is the m 3 n coefficient matrix for the effects of x on h. The model (13.12)
is called the exogenous measurement model, whereas the model (13.13) is
called the endogenous measurement model. The errors of measurement
are assumed to be uncorrelated with x and z and with each other. Also
E(d) 5 0q31 and E(e) 5 0p31. To simplify matters x, y, x and h are written as
deviations from their means (Bollen, 1989). To incorporate causal indica-
tors into the current model structure (see below) one needs to define each
indicator, xi, as equal to a latent variable x that is x 5 Ix,29 where x is a
vector of causal indicators.
The hypothesis of the model is that S 5 S(␾). Estimation is performed
by choosing ␾ (the vector that contains the model parameters) minimizing
the maximum likelihood
F 5 log 0 S (␾) 0 1 tr { S S21 (␾) } 2 log 0 S 0 2 (p 1 q) , (13.15)
where p 1 q is the number of measured variables.

Following Equations (13.12) and (13.13), the implied covariance matrix
is
Syy (␾) Syx (␾)
S (␾) 5 c d, (13.6)
Sxy (␾) Sxx (␾)
where,
Syy (␾) 5 LyE (hhr) Lry 1 Qe, (13.17)
5 Ly (I 2 B) 21 (GFGr 1 Y) [ (I 2 B) 21 ] rLry 1 Qe, (13.18)
Syx (␾) 5 LyE (hxr) Lrx, (13.19)
5 Ly (I 2 B) 21GFLrx, (13.20)
Sxy (␾) 5 Sryx (␾) , (13.21)
Sxx (␾) 5 LxFLrx 1 Qd (13.22)


where the equalities in (13.18) and (13.20) follow from substituting the
reduced form of Equation (13.14) that is, h 5 (I 2 B)–1(Gx 1 z). Qe is the
p 3 p covariance matrix of e, Qd is the q 3 q covariance matrix of d, and F
is the n 3 n covariance matrix of the latent factors x. (I 2 B) is required to
be non-singular and so invertible.
Analysis of the covariance matrix of observed variables leads to
unstandardized coefficients that depend upon the units in which the vari-
ables are scaled. To compare the effects of two or more variables on the
same dependent variable when they have different units of measurement,
the coefficients are standardized as follows
s^ jj
l^ sij 5 l^ ij a b,
Å sjj
(13.23)
s^ jj
b^ sij 5 b^ ij a b,
Å sjj
(13.24)
s^ jj
g^ sij 5 g^ ij a b,
Å sjj
(13.25)
where the superscript s represents a standardized coefficient, i is the

dependent variable, j is the independent, and s^ ii, s^ jj are the model pre-
dicted variances of the ith and jth variables. Once the hypothesized relation-
ship between the variables has been identified and estimated, the latent
variable scores hj for each country j 5 1, . . . J can be obtained following
the procedure suggested by Jöreskog (2000).
The analysis outlined above allows us to decompose the effects of one
variable on another into direct, indirect, and total effects.30 Direct effects
are the influences of one variable on another that are not mediated by any
other variable. For non-recursive models as the present one, the direct,
indirect and total effects of x on h (the variables of interest) are given by G,
(I 2 B)–1 G 2 G, and (I 2 B)–1 G, respectively.31
13.3.2 Model Specification
Figure 13.1 presents a representation of the system of simultaneous equa-

tions (path diagram) of the hypothesized set of relationships that link
the variables of interest, institutional quality, the shadow economy and
corruption.
Notice the direction of the arrows connecting the two constructs
(factors), the shadow market32 (‘SHADOW’-h1) and magnitude of corrup-
tion (‘CORRUPTION’-h2), to their indicators. These constructs predict


TAX ε1
δ1 DURATION λ1 y1
x1 ζ1 λ5

ξ1 γ 11
λ6 LABOUR
SHADOW y2 ε2
γ 12 η1
δ2 FLEXIBILITY λ7
x2 λ2
ξ2 γ 13 β 21 CURRENCY ε3
y3
LAW λ3 γ 23 λ8
δ3 CORRUPTION
454
x3
ξ3 QUALITY η2
TI ε4
λ4 λ9 y4
ζ2
EFFECTIVENESS PROCEDURES ε5
δ4 x4 y5
Notes: It is common practice to represent the measured variables (indicators) by squares and the latent variables (factors) by circles. The
hypothesized relationship between the variables are indicated by lines. Straight single-headed arrows represent one-way influences from the
variable at the arrow base to the variable to which the arrow points.

Figure 13.1 Hypothesized relationships between variables of interest, institutional quality, the shadow economy and
corruption.
27/07/2011 16:16
the measured variables. The implication is that SHADOW is assessed

using a country’s tax revenue as a percentage of GDP (‘TAX’-y1), women
labour participation rate (‘LABOUR’-y2) and the national currency in
circulation relative to GDP (‘CURRENCY’-y3). The latent variable
SHADOW is also predicated by the duration (in days) of starting up a
business (‘DURATION’-x1), and an index indicating rigid regulation
in the labour market (‘FLEXIBILITY’-x2). The magnitude of corrup-
tion, CORRUPTION is indicated by the Transparency International
Corruption Perception Index (‘TI’-y4), and an index capturing proce-
dural costs with meeting governments’ requirements to start operating
a business legally (‘PROCEDURES’-y5). These two endogenous latent
variables are both predicated by an exogenous latent variable, captur-
ing institutional quality (‘QUALITY’-x3). The exogenous latent variable
is indicated by a rule of law index (‘LAW’-x3) and an index of govern-
ment effectiveness (‘EFFECTIVENESS’-x4).33 Finally, the directional link
between CORRUPTION and SHADOW tests the relationship between
these two endogenous latent variables.34 As predicted by the theoretical
model, we expect a direct relation between QUALITY, SHADOW and
CORRUPTION. We turn now to the results.
13.4 RESULTS
13.4.1 Structural Equation Model Estimates
To increase the number of observations the data have been averaged over
the period 1998–2002. The sample, driven by data availability, covers 18
OECD countries. As noted in Subsection 13.3.1, prior to estimation the
data are standardized.35 Figure 13.2 presents the estimated coefficients. To
derive the t-ratios for the indicator variables, one of the coefficients of the
indicators must be normalized to 1. We have chosen to normalize the esti-
mated parameters with respect to CURRENCY; this should be taken to
mean higher currency circulation relative to GDP reflects a higher unofficial
economy. The indicator variables are statistically significant at the 10 per
cent level at least, with higher TAX indicating a smaller shadow economy.
This is in line with the evidence found by Friedman et al. (2000). The varia-
ble LABOUR has also the expected sign implying that a low participation
rate, assuming constant labour force, indicates a high shadow economy.
For the latent variable of corruption, CORRUPTION, we have normal-
ized the estimated parameters with respect to TI (this should be taken
to mean that high values of the Transparency International index reflect
high corruption). The indicator variable of the latent CORRUPTION


TAX
DURATION 1 y1

x1 −0.59***
ζ1
ξ1 −0.36
−0.39*** LABOUR
SHADOW y2
0.37***
η1
FLEXIBILITY 1 1
x2
ξ2 CURRENCY
−0.65** −0.25* y3
456
LAW 1
x3 CORRUPTION −1
−1.11*
ξ 3 QUALITY η2
TI
0.43** y4
0.97*
ζ2
EFFECTIVENESS PROCEDURES
x4 y5
Notes: Levels of significance: 1 per cent (*), 5 per cent (**), 10 per cent (***).

Figure 13.2 Structural equation model estimated coefficients
27/07/2011 16:16
– PROCEDURES – is significant at the 5 per cent level. Turning now to

the indicator of QUALITY – EFFECTIVENESS – is positive – following
the normalization of the estimated coefficient with respect to the variable
LAW – and significant at the 1 per cent level.
13.4.2 Structural Equation Model Evaluation
To evaluate the model we make use of a number of statistics. The c2 statis-

tic for testing the model against the alternative that the covariance matrix
of the observed variables is unconstrained, where smaller values indicate
a better fit is c2 5 14.75 (with 16 degrees of freedom). In other words, a
small c2 does not reject the null hypothesis that the model reproduces
the covariance matrix. The c2 test of exact fit accepts the model at least
at the five per cent level of significance. The Root Mean Square Error
of Approximation (RMSEA) accounts for the error of approximation
in the population and has recently been recognized as one of the most
informative criteria in covariance structure modelling (Steiger, 1990).
Expressed differently, the RMSEA measures how well the model fits
based on the difference between the estimated and the actual covariance
matrix (and degrees of freedom). The value of RMSEA is almost zero
indicating a good fit.36 Other indices providing evidence of an acceptable
fit are the Goodness of Fit Index (GFI 5 0.84), the Adjusted Goodness
of Fit Index (AGFI 5 0.55) and the Normed Fit Index (NFI 5 0.87).
These indices range from zero to one, with values close to one indicating a
better fit.37 The squared multiple correlations for the structural equations
are 0.51, and so moderately high, for SHADOW and, very high, 0.94 for
CORRUPTION. Based on these goodness-of-fit statistics, we conclude
that the model fits the data fairly well.
13.4.3 Direct and Indirect Effects
The path analysis in Figure 13.1, as noted in Section 13.3.1, allows us

to distinguish three types of effects: direct, indirect and total. The direct
effect is the influence of one variable on another that is unmediated by any
other variable in the path diagram. The indirect effects of a variable are
mediated by at least one intervening variable. The total effect of institu-
tional quality captured by LAW on the latent variable SHADOW is g13 5
20.65, and is significant at the 1 per cent level. The interpretation of this
is that a marginal improvement in institutional quality reduces the latent
score of the shadow economy by 0.65. The direct effect of QUALITY on
the latent CORRUPTION (h2) is g23 5 21.11, and is significant at the 1
per cent level, implying that a marginal increase in the rule of law reduces


the magnitude of corruption by 1.11, as measured by the latent score. The

estimated model, therefore, shows that an increase in institutional quality
affects negatively both the shadow economy and corruption markets
in OECD countries. The interpretation of this is that, in the sample of
OECD countries, institutions seem to work better in combating corrup-
tion relative to the underground economy. Turning now to the total effect
of institutional quality, denoted by Th x , on CORRUPTION is given by
2 3
Th x 5 (1 2 b21b12) 21 (b21g13 1 g23)

2 3
5 20.950, (13.26)
and is significant at the 1 per cent level. Thus, a marginal increase in

QUALITY reduces CORRUPTION by 20.950.38
The coefficient (b21 5 20.25) of the link between the two endogenous
latent (CORRUPTION and SHADOW) variables is significant at the 1
per cent level implying that the existence of SHADOW reduces the mag-
nitude of CORRUPTION. This, therefore, is consistent with the shadow
economy and corruption being substitutes.
We turn now to deriving the scores for the latent variables for the coun-
tries in our sample. These latent scores allow us to provide a ranking of
the shadow economy and the extent of corruption for the countries in our
sample.
13.4.4 Latent Scores (Indices)
To derive the latent scores, as noted in Section 13.3.1, we adopt the pro-
cedure suggested by Jöreskog (2000). The results are presented in Table
13.1.39
As can be seen in Table 13.1, the country with the smallest shadow
economy is Canada (with normalized index value 0), followed by Hungary
(0.139) and Belgium (0.144). Among the countries in our sample, Mexico
and the Slovak Republic have the largest unofficial sectors, with 1 and
0.440, respectively. In terms of corruption, Finland is the least corrupt
country (with normalized index value 0) followed by New Zealand (0.081),
and the UK (0.204). The most corrupt country is Mexico (with normal-
ized index of 1), followed by the Slovak Republic (0.970), Poland (0.867),
Czech Republic (0.853) and Greece (0.810).
One of course may still be not convinced of the accuracy of the indices
and in particular with the index of CORRUPTION.40 A natural additional
test one could perform is the derivation of the correlation of the derived
index of CORRUPTION with other existing indices. These correlations


Table 13.1 Indices of SHADOW and CORRUPTION
Country SHADOW CORRUPTION

Index Rank Index Rank
Mexico 1 1 1 1
Slovak Rep. 0.440 2 0.970 2
Germany 0.243 3 0.374 10
Australia 0.235 4 0.208 15
Poland 0.223 5 0.867 3
Czech Rep. 0.209 6 0.853 4
Switzerland 0.207 7 0.216 14
New Zealand 0.201 8 0.081 17
Finland 0.200 9 0 18
United Kingdom 0.199 10 0.204 16
Italy 0.198 11 0.746 7
Greece 0.195 12 0.810 5
Portugal 0.178 13 0.570 8
Norway 0.163 14 0.222 13
Austria 0.160 15 0.342 11
Belgium 0.144 16 0.542 9
Hungary 0.139 17 0.748 6
Canada 0 18 0.300 12
Table 13.2 Correlation of index of CORRUPTION with other indices
Model TI ICRG WVS EBI CI DKM CC

Model 1.0000
TI 0.9888 1.0000
ICRG 0.7177 0.7542 1.0000
WVS 0.5953 0.5741 0.3356 1.0000
EBI 0.8587 0.8590 0.5590 0.4201 1.0000
CI 0.9719 0.9817 0.8028 0.6059 0.8410 1.0000
DKM 0.7634 0.7434 0.3128 0.4570 0.6839 0.5683 1.0000
CC 0.9734 0.9744 0.7361 0.5928 0.8727 0.6774 0.9576 1.0000
Notes: TI is the Corruption Perception Index of Transparency International; ICRG is

the International Country Risk Guide Corruption index; WVS is the World Value Survey
(asking the question whether accepting a bribe is justifiable); EBI is the Exporter Bribery
Index of Friedman et al. (2000); CI is the Corruption Index of Friedman et al. (2000);
DKM is the Corruption Index based on the full model of 1997 of Dreher, Kotsogiannis
and McCorriston (2007); and CC is the Control of Corruption of Kaufmann, Kraay and
Mastruzzi (2003).


are reported in Table 13.2. This table reveals that the ranking of the index
of CORRUPTION across the 18 OECD countries is highly correlated
with the majority of the existing indices of CORRUPTION. It is worth
noting that the index of corruption has the highest correlation (with
correlation 0.9888) with the TI index, a widely used index.41
Taken together, the estimates from the structural equation model
are consistent with the theoretical framework, that institutional quality
affects negatively the shadow economy and corruption both directly and
indirectly and that corruption and the shadow economy are substitutes.
13.5 CONCLUDING REMARKS
This chapter has taken a further step towards understanding the relation-
ship between institutional quality, corruption and the shadow market. The
model described confirmed existing results, that corruption and shadow
markets are substitutes in the sense that the existence of the shadow
market is associated with smaller levels of rents and that the effect of
institutional quality on the shadow market is unambiguously negative
whereas the effect of institutional quality on the magnitude of corruption
is ambiguous and depends on the relative effectiveness of institutional
quality on the two markets. These results have been tested using data from
18 OECD countries. The empirical estimation confirmed that institutional
quality reduces the shadow economy and corruption. The total effect of
institutional quality on corruption was estimated to be negative and sig-
nificant. While these results are in line with those of Dreher, Kotsogiannis
and McCorriston (2009), the key innovation is that we modelled corrup-
tion and the shadow economy as latent variables, rather than considering
them as exogenous estimates.
ACKNOWLEDGEMENTS
We thank Christian Conrad, Michael Devereux, David de Meza, Bruno

Frey, Thomas Herzfeld, Johann Graf Lambsdorff, Paolo Mauro, Michael
Pickhardt, Jordi Sardà, Michael Smart, and Gary Stasser, for many con-
structive comments that have considerably improved this chapter. This
work has also benefited from comments from seminar participants at
the Centre for Economic and Business Research (Copenhagen) and the
Autonomous University of Barcelona, and conference participants at the
European Public Choice Society Annual Meeting 2005, the Public Choice
Society Meeting, New Orleans, 2005, and the conference on ‘The Shadow


Economy, Tax Evasion and Social Norms’ at the University of Muenster,

July 23–26, 2009, Muenster. Kotsogiannis acknowledges support from
the Catalan Government Science Network (Project No. SGR2005-177)
and the Spanish Ministry of Education and Science Research Project
(SEJ2006-4444). The usual caveat applies.
NOTES
1. Corruption, in the common usage of the word, can mean different things in different
contexts. For a discussion of some of the alternative denotations of the problem of cor-
ruption and its damaging consequences see the insightful survey by Bardhan (1997). See
also Klitgaard (1988) and Rose-Ackerman (1999).
2. Tackling corruption and improving governance and the rule of law could increase per
capita incomes by a staggering 400 per cent (World Bank, 2004).
3. Additional justification is also provided by Huntington (1968, p. 386) who notes: ‘In
terms of economic growth, the only thing worse than a society with a rigid, over-
centralized, dishonest bureaucracy is one with a rigid, over-centralized, honest bureauc-
racy’. Another efficiency argument in favour of corruption is to consider it as the ‘speed
of money’ which considerably reduces the slow-moving queues in public offices. These,
however, have been met with criticism, see Tanzi (1998), Kaufmann and Wei (1999),
and Rose-Ackerman (1999). Bardhan (1997), Jain (2001) and Aidt (2003) provide com-
prehensive accounts of the latest developments on corruption. See also Rose-Ackerman
(1978, 1999) and Lambsdorff (1999). Kaufmann, Kraay and Zoido-Lobatón (1999)
provide a comprehensive account of empirical work on corruption.
4. The shadow economy is widely believed, and existing estimates also confirm, to be
both pervasive and significant. For the OECD countries – the subject of this chapter
– it is estimated, for instance, that for countries such as Greece and Italy, the under-
ground economies are almost one-third as large as the officially measured GNP. The
smallest underground economies are estimated to be in Japan, the United States and
Switzerland (countries that have traditionally relatively small public sectors and high
tax morale), Schneider and Enste (2000, 2002). These estimates, however, need to be
interpreted with some caution.
5. See, for instance, among others, Tanzi (1982, 1999), Frey and Pommerehne (1984),
Schneider (1994a, b, 1997) and Giles (1999). For a recent contribution that explores
the extent to which moral sentiments may control shadow activities see Kanniainen,
Pääkkönen and Schneider (2005).
6. This view, it has to be said, is closely associated with the literature on tax evasion
(a subset of underground economic activities). The seminal work on tax evasion is
Allinghmam and Sandmo (1972). A thorough account of theoretical contributions can
be found in Cowell (1990).
7. Taken to mean bureaucracy, regulatory discretion, rule of law, corruption and a weak
legal system.
8. Building on the work of Bliss and Di Tella (1997) and Shleifer and Vishny (1993) who
analyze a bureaucracy issuing permits to perform some economic activity in exchange
for bribes.
9. The official market and the shadow economy are, however, complements. We address
this in Section 13.2.
10. The focus, more recently, has been on the relationship between corruption, centraliza-
tion and the shadow economy (Echazu and Bose, 2008).
11. They also find evidence that higher taxes are associated with a smaller underground
sector.


12. Addressing the issue of causality Friedman et al. (2000) show that the causal link
runs from weak economic institutions to the size of the shadow economy. As they
put it: ‘[the] results show there is an exogenous component of “institutions” that is
significantly correlated with the size of the unofficial economy’ (p. 460). Focusing on
regulation, Johnson, Kaufmann and Schleifer (1997) show that countries with more
regulation of their economies tend to have a higher share of the unofficial economy in
GDP.
13. This measurement difficulty is to a great extent exacerbated by the lack and dif-
ficulty of a precise definition of the shadow economy. A working – and widely used
– definition is ‘all economic activities that contribute to the officially calculated (or
observed) gross national product but are currently unregistered’ (Schneider and
Enste, 2000. p. 78). Clearly, this definition involves legal as well as illegal and corrupt
activities.
14. A view also taken by Dreher and Schneider (2010). The difference, however, between
what we do here and the analysis in Dreher and Schneider (2010), is they use already
estimated data – taken from Dreher, Kotsogiannis and McCorriston (2007) – whereas,
in the present paper, the indices are estimated simultaneously.
15. The structural equation model was introduced to economics by Weck (1983) and Frey
and Weck-Hannemann (1984) and latterly explored by Loayza (1996) and Giles (1999),
among others, to measure the size of the shadow economy, Raiser, Tommaso and
Weeks (2000) to investigate the institutional change in Eastern Europe, and Kuklys
(2004) to measure welfare. Schneider and Enste (2000, 2002) offer a comprehensive
account of studies on the hidden economy that have employed this approach. This
approach has recently been used in Dreher, Kotsogiannis and McCorriston (2007) to
derive an index of corruption based on around 100 countries.
16. A word of clarification is in order here. This chapter does not attempt to address the
recent criticism regarding the extensive use of variables capturing the institutional
quality that are based on surveys and suffer from ‘artificial inertia’ (Rodrik, 2004).
Nevertheless, by treating the variables as latent we, to some extent, circumvent this
problem.
17. The small number of countries reflects, of course, data availability.
18. Corruption is defined, following Shleifer and Vishny (1993, p. 599), ‘as the sale by gov-
ernment officials of government property for personal gain’.
19. It is implicitly assumed that there is no rent extraction in the shadow economy. If firms
are caught operating in the shadow economy they are fined but the revenues are not
observed, and so cannot be expropriated, by the corrupt officials.
20. The effectiveness of the institution in the two markets should be interpreted broadly to
include, for instance, elements such as, among others, the complexity of the tax system
and regulatory discretion. For a study on the effect of the complexity of the tax system
on the size of the shadow economy see Schneider and Neck (1993).
21. It is feasible to endogenize the quality of the institution e, but this will add no further
insights.
22. There is, therefore, commitment on the part of corrupt public officials: once announced,
public officials cannot renege on the level of graft. This seems a natural assumption on
the working of institutions in OECD countries.
23. At first sight this specification might look peculiar. What we have in mind though is the
fact that (an improvement of) institutional quality does not only affect corrupt officials
but also affects the policing of firms caught engaging in corrupt activities.
24. This conclusion is also reached by Choi and Thum (2005), though they do not treat the
role of institutional quality.
25. It can be easily seen that with, as an envelope property, ms*(e) being unaffected by insti-
tutional quality, the threshold q is increasing (decreasing) in e if and only if eq > es (eq <
es).
26. Latent random variables represent unit-dimensional concepts. The observed variables
or indicators of a latent variable contain random or systematic measurement errors but


the latent variables are free of these errors. Latent variables correspond to hypothetical
variables (constructs) that may vary in their degree of abstractness. We turn to this in
the next section.
27. For an application, see Jöreskog and Goldberger (1975).
28. Of course, all measures or abstract factors have far from perfect associations with the
factor.
29. The observed x variables, so, contain no measurement error.
30. Total and indirect effects are only defined under certain conditions. A sufficient condi-
tion for the total effects to exist is that the largest eigenvalue of the matrix B is less than
one (Bollen, 1989).
31. For the effects of x on y and x as well as of h on h, y and x see Bollen (1989, Table 8.9).
32. To avoid confusion, and where appropriate, we provide both the name of the variables
and their symbolic representation in Subsection 13.3.1.
33. For the choice of the indicators of the latent variables, we have been guided by the
existing literature on corruption and shadow economy. The indicators for SHADOW
– TAX and CURRENCY – are all variables that have been widely used (Schneider and
Enste, 2000). The indicator LABOUR follows from the fact that a decline in women
labour force participation in the official economy can be seen, assuming constant
labour force participation, as an indication of increased activity in the shadow economy
(Schneider and Enste, 2000). For CORRUPTION, TI is an obvious (perception based)
indicator. Following Djankov et al. (2002), PROCEDURES (that cover direct costs of
time associated with meeting government requirements that a start-up must bear before
it can operate legally) should be correlated with CORRUPTION, too. The indicators
of QUALITY seem to be obvious choices.
34. From a theoretical point of view other variables could measure the underlying con-
structs equally well. In the empirical estimation, other variables have been tried out.
The final set of variables is the one that produces the best fit. Correlations between the
error terms, where this increases model fit, have also been accounted for.
35. All variables with their definitions and sources can be found in Appendix 13.1. All esti-
mations have been performed with LISREL® V. 8.5.4.
36. Values of the RMSEA less than 0.05 indicate good fit, values as high as 0.08 represent
reasonable fit, values from 0.08 to 0.10 indicate mediocre fit, and those greater than 0.10
indicate poor fit (MacCallum et al., 1996).
37. These indices are based on comparison with a null model predicting all covariances to
be zero. The NFI relates the c2 of this null model to the c2 of the actual model. The GFI
and the AGFI compare the loss function of the null model with the loss function of the
actual model, with AGFI adjusting for the complexity of the model.
38. Notice that, as noted in Subsection 13.3.1, the total effects of x on h are defined if and
only if the largest eigenvalue of the matrix of the estimated coefficients of the endog-
enous latent variables, B, is less than 1. In the estimated model the largest eigenvalue is
0.062 and so the total effects are defined.
39. The results, of course, depend on the choice of the observed variables.
40. Since indices comparable to the index for SHADOW do not exist one should exercise
caution in providing correlations between those indices and the index for SHADOW
derived in this chapter. This is because studies that are based on single latent models
potentially are picking up aspects of corruption in their shadow economy measure.
41. Two observations are in order here. Firstly, one might argue that the correlation of the
derived index for CORRUPTION with TI is high because TI appears as an indicator
of the latent variable of CORRUPTION. Yet, other indicators are a priori of equal
importance. What the result here indicates is that perceived corruption – at least in the
OECD countries – is highly correlated with actual corruption. This corroborates using
the TI index as index for corruption in those countries. Secondly, one might be tempted
to benchmark the indices SHADOW and CORRUPTION to existing estimates of the
shadow economy and corruption. Though feasible, this benchmarking exercise suffers
from a choice-bias, and is, in the present context, of limited use.


REFERENCES
Aidt, T.S. (2003), Economic analysis of corruption: a survey. Economic Journal,
113, 632–652.
Allingham, M. and A. Sandmo (1972), Income tax evasion: a theoretical analysis.
Bardhan, P. (1997), Corruption and development: a review of issues. Journal of
Bliss, C. and R. Di Tella (1997), Does competition kill corruption? Journal of
Bollen, K. (1989), Structural Equations With Latent Variables. New York: Wiley.
Choi, J. and M. Thum (2005), Corruption and the shadow economy. International
Economic Review, 46, 817–836.
Cowell, F. (1990), Cheating the Government. Cambridge, MA: MIT Press.
Djankov, S., R. La Porta, F. Lopez de Silanes and A. Schleifer (2002), The regula-
tion of entry. Quarterly Journal of Economics, 117, 1–37.
Dreher, A. and A. Gassebner (2007), Greasing the wheels of entrepreneurship?
The impact of regulations and corruption on firm entry. KOF Working Paper
166. ETH Zurich.
empirical analysis. Public Choice, 144, 215–238.
Dreher, A., C. Kotsogiannis and S. McCorriston (2007), Corruption around the world:
evidence from a structural model. Journal of Comparative Economics, 35, 443–466.
Finance, 16, 773–796.
Echazu, L. and P. Bose (2008), Corruption, centralization, and the shadow
economy. Southern Economic Journal, 75, 524–537.
Frey, B.S. and H. Weck-Hannemann (1984), The hidden economy as an unob-
served variable. European Economic Review, 26, 33–53.
for measurement. Review of Income and Wealth, 30, 1–23.
Giles, D.E. (1999), Measuring the hidden economy: implications for econometric
modeling. Economic Journal, 109, F370–F380.
Hindriks, J., A. Muthoo and M. Keen (1999), Corruption, extortion and evasion.
Huntington, S. (1968), Political Order in Changing Societies. New Haven, CT: Yale
University Press.
Jain, A.K. (2001), Corruption: a review. Journal of Economic Surveys, 15, 71–121.
and corruption. American Economic Review, 88, 387–392.
sition. Brookings Papers on Economic Activity, Washington DC.
Statistical Association, 70, 631–639.


Jöreskog, K.G. (2000), Latent variable scores and their uses. Available at http://
www.ssicentral.com/lisrel/column6.htm.
Kanniainen, V., J. Pääkkönen and F. Schneider (2005), Determinants of shadow
economy: theory and evidence. Mimeo, Johannes Kepler University Linz.
Kaufmann, D., A. Kraay and M. Mastruzzi (2003), Governance matters III. Draft
Paper, World Bank.
Kaufmann, D., A. Kraay and P. Zoido-Lobatón (1999), Governance matters.
Policy Research Working Paper 2196, World Bank.
Kaufmann, D. and S.J. Wei (1999), Does ‘grease money’ speed up the wheels of
commerce. NBER Working Paper 7093, Cambridge, MA.
Klitgaard, R. (1988). Controlling Corruption. Berkeley: University of California
Press.
Kuklys, W. (2004), Measuring standard of living in the UK: an application
of Sen’s functioning approach using structural equation models. Mimeo,
University of Cambridge and Max Planck Institute of Research Into Economic
Systems.
Lambsdorff, J. (1999), Corruption in empirical research: a review. Transparency
International Working Paper, November 1999.
Leff, N.H. (1964), Economic development through bureaucratic corruption.
American Behavioural Scientist, 8, 8–14.
MacCallum, R.C., M.W. Browne and H.M. Sugawara (1996), Power analysis and
determination of sample size for covariance structure modelling. Psychological
Methods, 1, 130–149.
Meon, P.-G. and L. Weill (2008), Is corruption an efficient grease? BOFIT
Discussion Paper No. 20/2008.
Mocan, H. (2004), What determines corruption? International evidence from
micro data. NBER Working Paper No. W10460.
Raiser, M., M. DiTommaso and M. Weeks (2000), The measurement and determi-
nation of institutional change: evidence from transition economics. Cambridge
Working Papers in Economics 29.
Rodrik, D. (2004), Getting institutions right. Mimeo, Harvard University.
Rose-Ackerman, S. (1978), Corruption: A Study in Political Economy. New York:
Academic Press.
Rose-Ackerman, S. (1997), Corruption and development. Annual Bank
Conference on Development Economics. The World Bank, Washington DC.
Rose-Ackerman, S. (1999) Corruption and Government: Causes, Consequences and
Reform. Cambridge, UK: Cambridge University Press.
Shleifer, A. and R. Vishny (1993), Corruption. Quarterly Journal of Economics,
108, 599–617.
Schneider, F. (1994a), Measuring the size and development of the shadow
economy: can the causes be found and the obstacles be overcome? In H.
Brandstaetter and W. Güth (eds), Essays on Economic Psychology. Berlin:
Schneider, F. (1994b), Can the shadow economy be reduced through major tax
reforms? An empirical investigation for Austria. Public Finance, 49, 137–152.
Schneider, F. (1997), The shadow economies of Western Europe. Journal of
Institute of Economic Affairs, 17, 42–48.


quences. Journal of Economic Literature, XXXVIII, 77–114.
Survey. Cambridge, UK: Cambridge University Press.
Schneider, F. and R. Neck (1993), The development of the shadow economy under
changing tax systems and structures. Finanzarchiv N.F, 50, 344–369.
Shleifer, A. and R. Vishny (1993). Corruption. Quarterly Journal of Economics,
108, 599–617.
Steiger, J.H. (1990), Structural model evaluation and modification: an interval
estimation approach. Multivariate Behavioral Research, 25, 173–180.
Tanzi, V. (1982), The Underground Economy in the United States and Abroad.
Lexington: D.C. Heath.
Tanzi, V. (1998), Corruption around the world-causes, consequences, scope and
cures. International Monetary Fund Staff Papers, 45, 559–594.
Transparency International (2002), Available at http://www.transparency.org/
policy¡research/surveys. . ./cpi/2002.
öffentlichen Verwaltung? Eine ökonomische Analyse. Finanzwissenschaftliche
Schriften 22, Bern/Frankfurt: Lang.
Weber Abramo, C. (2007), How Much Do Perceptions of Corruption Really Tell
Us? Economics Discussion Papers No. 2007-19.
World Bank (2001), Available at http://www.worldbank.org/publicsector/
anticorrupt/index.ctm.
World Bank (2003), World Development Indicators. CD-ROM, Washington DC.
World Bank (2004), Available at http://www.worldbank.org/WBSITE/EXTER
NAL/NEWS.


APPENDIX 13.1 DATA DEFINITIONS

A13.1.1 Indicators of the Endogenous Latent Variables
TAX: Tax revenue in percent of GDP. Source: World Bank (2003), World
Development Indicators.
LABOUR: Women activity rate. Source: International Labour

Organization, LABORSTA Labour Statistics Database, extracted on 30
May 2004.
CURRENCY: Central Bank-Currency in Circulation in US$, divided by

GDP. Source: Thompson Data Stream.
TI: Corruption Perception Index, Transparency International (2002).

Scaling implies higher values of the index implies more CORRUPTION.
PROCEDURES: The costs associated with starting to operate legally a

firm. Source: Djankov et al. (2002).
A13.1.2 Indicators of the Exogenous Latent Variables
DURATION: Starting a business, duration (days). Source: Djankov et

al. (2002).
FLEXIBILITY: Flexibility of hiring index. Higher values represent

more rigid regulation. Source: Doing Business, World Bank online
Database.
LAW: Rule of law index. This index measures the extent to which agents
have confidence in and abide by the rules of society. It refers to perceptions
of the incidence of both violent and non-violent crime, the effectiveness
and predictability of the judiciary, and the enforceability of contracts.
Measures the success of a society in developing an environment in which
fair and predictable rules form the basis for social and economic interac-
tions. Source: Kaufmann, Kraay and Zoido-Lobatón (1999); Kraay and
Mastruzzi (2003).
EFFECTIVENESS: This variable combines perceptions of the quality of

public service provision, the quality of the bureaucracy, the competence
of civil servants, the independence of the civil service from political pres-
sures and the credibility of the government’s commitment to policies into a


single grouping. The main focus is on ‘inputs’ required for the government
to be able to produce and implement good policies. Source: Kaufmann,
Kraay and Zoido-Lobatón (1999); Kaufmann, Kraay and Mastruzzi
(2003).


14 Shadow economy, voice and
accountability, and corruption
Benno Torgler, Friedrich Schneider and
Alison Macintyre
14.1 INTRODUCTION
While economics has devoted extensive resources to understanding the

legitimate side of market operations, there is not yet a commensurate body
of work on the ‘shadow economy’. Tanzi (2002) reflects on this situation
and states, ‘it seems that the economic profession, immersed as it was in
its theories, could not cope or was unwilling to cope with the messy world
of the underground economy’ (p. xiii). Encouragingly, in the past decades
economists have ceased to ignore the topic, and interest in this phenom-
enon has increased (see, for example, Schneider and Enste, 2000, 2002).
The transformation of socialist economies ignited the current concern
regarding governance quality, as institutional weaknesses and corrup-
tion surfaced as major obstacles to market reforms (Abed and Gupta,
2002). However, even now, investigation into the causes of the shadow
economy is still an undeveloped yet critical area of research. There is
‘universal recognition of the importance of the unofficial economy’ (Choi
and Thum, 2005, p. 818, see also Virta, 2009), as the informal sector plays
a large and important role both in transition countries and in developing
countries. Furthermore, employment in the informal sector seems to be
a significant source of income for many people. The lack of reliable data
may be responsible for the relative lack of work on this topic (Dreher and
Schneider, 2010), and as noted by Choi and Thum (2005, p. 817): ‘By defi-
nition, the unofficial economy constitutes activities that are not recorded
by government statistics’.
This study investigates the extent to which governance and institutional
quality affect the shadow economy. The day-to-day reality of life in any
country is to some extent determined in the political arena, and countries
vary enormously with respect to the nature and effectiveness of their politi-
cal systems. We use measures of corruption and voice and accountability to
469


determine whether these might be key driving forces for the level of the
shadow economy. Being active in the shadow economy can be seen as an
‘exit’ option, a signal through which taxpayers can express their disagree-
ment. Hence, we also measure the extent of tax effort to discover whether
institutional quality (measured by corruption and voice and accountabil-
ity) affects that behaviour. This contribution stresses the importance of
investigating not only traditional variables such as tax burden, the sectoral
composition, the richness of a country or the labour market conditions,
but also institutional and governance quality.
In Section 14.2, we present our theoretical approach. Section 14.3
describes the data set and Section 14.4 contains the empirical results using
a cross-sectional and a panel analysis exploring the shadow economy as a
dependent variable. Section 14.5 then uses a large panel data set to analyse
the impact of institutional/governance factors on tax performance. Finally,
Section 14.6 concludes with a summary and discussion of the main results.
Taxation and public finance matters are, in democratic states, resolved

through political channels. History suggests that the need to secure an ade-
quate degree of consensus is one of the principal ways in which (over the
centuries) democratic institutions have spread. In an age of information
and mobility, it is not possible for a non-dictatorial government to stay in
power without securing a certain degree of consent from the populace in
the area of taxation and government activities (Bird, Martinez-Vazquez
and Torgler, 2008). State legitimacy thus rests to a considerable extent
on citizens’ ‘quasi-voluntary compliance’ (Levi, 1988). To secure such
compliance, the government systems must, over time, represent the basic
values of at least a minimum supporting coalition of the population. In
other words, it is not only the economic but also the political system that
affects formal and informal economic activities. The general performance
in many countries may be explained by underlying political conditions.
The political equilibrium position reflects the balance of political forces
and institutions (Bird, Martinez-Vazquez and Torgler, 2006). To examine
the effect on the shadow economy, we implement two measurements of
institutional quality, namely corruption, and voice and accountability.
14.2.1 The Effect of Voice and Accountability on the Shadow Economy
If citizens perceive that their interests (preferences) are properly repre-

sented in political institutions and they perceive to receive an adequate


Shadow economy, voice and accountability, and corruption 471
supply of public goods (high voice and accountability), their trust in the
government and their identification with the state increases, increasing
also their willingness to contribute. If the government is not benevolent,
the citizens’ voice has the potential to control politicians’ discretionary
power. An effective voice can help limit the abuse of political power by
selfish politicians and allows citizens to express their preferences in the
political process.
Levi (1988) points out one way a government can create or maintain
compliance is to provide reassurance by giving citizens that effective voice.
A government that precommits itself to providing citizens a voice declares
self-imposed restraints on its own power and thus sends a signal that
citizens are seen as responsible persons. In turn, this signals that citizens
are not perceived as ignorant, an indication of trust that could potentially
create or maintain a certain social capital stock, and the government
demonstrates that citizens’ preferences are taken into account during the
political process. In other words, the social contract between citizens and
the government is based on trust and this trust in turn will add to the
moral costs of behaving illegally.
Having a voice also produces a kind of procedural utility as the oppor-
tunity set increases. It leads to a more favourable outcome compared to
the situation where no such possibilities exist. If voice and accountability
is lacking, citizens might feel less satisfied with the system as well as feeling
powerless, and thus might be less inclined to comply (Alm, Jackson and
McKee, 1993). Rules developed through active involvement of the citizens
enhances rule obedience and the willingness to cooperate and act in line
with those rules. The more involved people are in establishing rules, the
stronger is their sense of obligation to comply (Lempert, 1972; McEwen
and Maiman, 1986; Cialdini, 1989; Kidder and McEwen, 1989).
Tyler’s research (1990a, 1990b, 1997) also provides support for the
importance of legitimacy and allegiance to authority in compliance deci-
sions. The way people are treated by the authorities affects their evaluation
of these authorities and their willingness to cooperate (see, for example,
Tyler, Casper and Fisher, 1989). Tyler (1997) argues that understanding
what people want in a legal procedure helps to explain public dissatisfac-
tion with the law and provides direction for building public support for the
law in the future.
In addition to the relationship between voice and accountability and
the shadow economy, it could be suggested that there is a relationship
between voice and accountability and corruption. Buehn and Schneider
(2009, p. 15) suggest that ‘increasing transparency and accountability can
reduce the scope for bribery’. Aidt, Dutta and Sena (2008) investigate
the effect of political accountability on the development of corruption,


suggesting that the ability to hold political leaders accountable may be ‘a

source of nonlinearlity in the mapping between corruption and growth’
(Aidt, Dutta and Sena, 2008, p. 195), that is, accountability affects cor-
ruption, which in turn affects economic growth. Wagener (2004) widens
the understanding of accountability to compare the relationship between
governance and economic welfare, and order and economic welfare, and
finds that good governance is related to economic welfare, more so than a
strong state. Wagener provides a broader understanding of good govern-
ance as something not restricted only to the responsibility of the state, but
also incorporating the attitudes of citizens toward the state and the law, in
a kind of collaboration between the state and the citizens.
14.2.2 The Effect of Corruption on the Shadow Economy
On the other hand, in an inefficient state where corruption is rampant, the

citizens will have little trust in authority and thus a low incentive to coop-
erate with societies’ rules. In other words, a more encompassing and legiti-
mate state increases citizens’ willingness to contribute to public goods.
Levin and Satarov (2000) explore how the collapse of the socialist regime
in Russia and the inability of the state to protect private property or
fulfilment of contracts resulted in a thriving system of bribes and private
protection offered by corrupt officials. One argument regarding the rela-
tionship between corruption and the shadow economy suggests that if the
government and the administration have significant discretionary power
over the allocation of resources, the level of corruption will be higher than
if they do not. In order to maintain a sustainable tax system, the govern-
ment must be responsive and the tax system must be fair, a condition that
is achieved with a strong connection between tax payments and the supply
of public goods (Bird, Martinez-Vazquez and Torgler, 2006).
The relationship between corruption and the shadow economy is
‘ambiguous from a theoretical point of view’ (Buehn and Schneider, 2009,
p. 4), and the two may be either complements or substitutes (Dreher and
Schneider, 2010). If distribution of resources is not fair, or the government
is not responsive to the needs of citizens, an entrepreneur has less to gain
through participation in the official economy. They may move their activi-
ties to the shadow economy, which restricts the ability of corrupt officials
to ask for bribes. For this reason, the shadow economy may in fact reduce
the level of bribes (Dreher, Kotsogiannis and McCorriston, 2008). Choi
and Thum (2005) also model the reduction in bribery offered by the exit
option of the shadow economy, and find that the presence of the shadow
economy may be a complement to the official economy, but a substitute
for corruption and bribes. Friedman et al. (2000) show empirically that


countries with more corruption have a higher share of unofficial economy.

Dreher and Schneider (2010) investigate the correlation between shadow
economy and corruption using an index of corruption based on a struc-
tural model. They observe the tendency that shadow economy and cor-
ruption are substitutes in high-income countries, but complements in
low-income countries. Citizens who do not have the right connections
to the corrupt officials will instead participate in the shadow economy,
meaning that corruption and the shadow economy are complements in
Greece. Furthermore, Virta (2009) discusses the way in which the type
of corruption might be related to country characteristics, and explores
whether geography affects ‘the relationship between corruption and the
shadow economy through its effects on technological diffusion, disease
burden and economic policy choices’ (Virta, 2009, p. 7). The author finds
that corruption seems to enlarge the shadow economy within the tropics,
but does not have an effect on the shadow economy outside the tropics
(2009, p. 4).
Agents such as the political elite, administration staff and legislators
have discretionary power if institutions are neither credible nor working
well. One negative consequence of this situation is that citizens lose their
trust in the authority. In countries where corruption is systemic and the
government budget lacks transparency and accountability, it cannot
be assumed that the obligation to pay taxes is an accepted social norm.
Levin and Satarov (2000, p. 114) describe the process as ‘trust of author-
ity declines and ordinary people become more and more alienated from
society’. Institutional instability, lack of transparency and rule of law
undermine the willingness of frustrated citizens to be active in the formal
economy. Tax administrators might also experience a crowding-out effect
of morality when there are a great number of corrupt colleagues.
Moreover, regulatory restraints and bureaucratic procedures not only
limit competition and the operation of markets (Choi and Thum, 2005),
but also provide a favourable environment for corrupt activities (Schneider
and Buehn, 2009). Examining conditions in Greece, there is a framework
resembling transition countries where high levels of regulations create
incentives for bribes and participation in the shadow economy. Levin and
Satarov (2000, p. 114) find that in Russia, ‘institutions and norms of politi-
cal behavior’ have provided the conditions whereby ‘corruption is a prin-
cipal means of acquiring wealth at all levels’. Strict regulatory conditions
are also among the causes of corruption discovered by Karymshakov and
Abdykaparov (2008) in their study on corruption in Kyrgyzstan. Yet, con-
trasting results are discovered by the econometric investigation of corrup-
tion in developing countries by Rei and Bhattacharya (2008). The authors
find that the effect of higher taxation and regulation might be overstated,


as they find no generalized evidence of this relationship between stricter

regulations and corruption. Perhaps most importantly, the authors find
that it is stricter regulation combined with effective governance that helps
to reduce the size of the shadow economy. De Soto (1989) and his research
team conducted an experiment, setting up a small garment factory in
Lima, with the intention to comply with the bureaucratic procedures and
thus to act in accordance with the law. He reports that ten times they were
asked for a bribe to speed up the process and twice it was the only possibil-
ity to continue the experiment. It took 10 months in total to start the busi-
ness. Similarly, de Soto (2000) tested the seriousness of barriers to entry by
creating a new and perfectly legal small business in Lima. His team spent
6 hours a day at it and was able to register the business 289 days later. The
cost of the legal registration was $1231, or 31 times the monthly minimum
wage. To obtain the authorization to build a house on state-owned land
took 6 years and 11 months, with 207 administrative steps in 52 govern-
ment offices and to obtain legal title to that piece of land took 728 steps.1
Similar experiences have been described in other countries, for example,
Philippines, Egypt and Haiti. In such cases where transaction costs of
behaving honestly are too high, lawbreaking helps people to survive. If
citizens perceive that their interests (preferences) are properly represented
in political institutions and consider government to be helpful rather
than wasteful, their willingness to opt for staying in the official sector and
comply with their obligations will increase.
If individuals and businesses believe that contracts will not be enforced
and productive efforts will not be protected, their incentive to be active in
the shadow economy increases. If investment in productive assets is either
postponed or abandoned, economic growth does not take place (Levin
and Satarov, 2000, p. 116). Citizens will feel cheated if they believe that
corruption is widespread, their tax burden is not well spent, their govern-
ment lacks accountability and that they are not protected by the rules of
law. This increases the incentive to enter the informal sector. The costs
of corruption are widespread and include a loss of democracy and a loss
of efficiency in distributing public resources. Sructure of society is affected,
and corrupt bureaucrats seek private gain from public activities, and the
most ‘honest and able’ citizens will often emigrate (Buehn and Schneider,
2009, p. 7).
14.2.3 The Shadow Economy and Tax Performance
The level of tax performance might also be related to the availability of

what may be called the ‘exit option’ of the so-called shadow economy.
In general, the larger the shadow economy, the lower we would expect


tax effort to be. The more taxpayers believe that others work in the
shadow economy, the lower the moral costs of behaving dishonestly and
evading taxes by moving their own activities to the shadow economy.
In this way, the potential intrinsic motivation to comply and contribute
to public sector activities gets crowded out. This relationship has been
shown empirically by Bird, Martinez-Vazquez and Torgler (2006). Public
finance matters are usually resolved through political channels, hence, as
mentioned previously, history suggests that the need to secure an adequate
degree of consensus from the taxed is one of the principal ways in which
democratic institutions have spread. In this age of information and mobil-
ity, no non-dictatorial government can stay in power without securing a
certain degree of consent from the populace. A better political system is
more interested in providing citizens what they want, and it transmutes
individuals’ and firms’ preferences into policy decisions in a more efficient
manner (Bird et al., 2008). Moreover, Kaufmann, Kraay and Mastruzzi
(2003) stress that
presence of corruption is often a manifestation of a lack of respect of both the

corrupter (typically a private citizen) and the corrupted (typically a public offi-
cial) for the rules which govern their interactions, and hence represents a failure
of governance according to our definition (p. 8).
If the formal economy does not represent the preferences of the citizens,
the resulting loss of respect may increase participation in underground
activities. Johnson, Kaufmann, and Shleifer (1997) model the shadow
economy as a substitute for the official economy from the assumption
that individuals are either employed in the official economy or the shadow
economy. If more people are in the shadow economy, this reduces the
tax revenue and reduces the money available for public goods or institu-
tional reform, which increases the returns to participation in the shadow
economy. The erosion of the tax base by the size of the shadow economy
might reduce the ability of the government to make effective policy (Dreher
et al., 2009). This effect is observed in Greece, where the inability to tax the
underground economy weakens the ability of the government to stabilize
and manage the economy. Furthermore, Levin and Satarov (2000, p. 115)
report the estimated costs to the Russian economy of corrupt activities,
and find it is more than ‘the combined expenditures on science, education,
health care, culture and art allocated in the government budget’. Such a
large cost means there is less money available for public goods and greater
returns from the shadow economy, and that some criminal groups spend
‘up to 50% of their revenues (actual, not declared) on bribing officials at
various levels’. This increase in the shadow economy then means that


governments are not able to collect taxes and further reduces the size of the
government budget, which further reduces the ability of the government
to provide public goods. This can increase social problems which might
decrease social capital and trust in the economy, which can lead to further
deterioration of institutional conditions.
Dreher and Schneider (2010, p. 6) state that ‘better institutional quality
[. . .] increases the benefits entrepreneurs can derive from operating in the
official sector’; reduces the shadow economy, and ‘should thus reduce
corruption and the size of the shadow economy alike’. Once operations
are transferred to the shadow economy, the entrepreneur can no longer
benefit from the public goods available in the formal economy (Choi
and Thum, 2005, p. 829). However, the value of those public goods will
depend on the level of corruption, as the greater the level of corruption,
the lower the tax effort, and hence the lower the resources available for
public goods.
14.3 DATA
14.3.1 Shadow Economy
The shadow economy by its nature is not readily observable or quantifia-

ble, but a working definition includes all market-based legal production of
goods and services that are deliberately concealed from public authorities
for the following reasons (Schneider 2005a):

minimum wages, maximum working hours, safety standards, and
4. to avoid complying with certain administrative procedures, such as
Hence, in this study, we will not deal with those economic activities
typically classified as underground, that is, all illegal actions with
the characteristics of classical crimes like burglary, robbery and drug
dealing. We also do not include the informal household economy
which consists of all household services and production. To measure
the shadow economy as a percentage of the official GDP we will use
the DYMIMIC method to estimate the parameters for determining the
size of the shadow economy and with the help of the Currency Demand
Method, we calibrate the estimated coefficients of the DYMIMIC


procedure into absolute ones. For the cross-sectional analysis we build

averages for 1990, 1995 and 1999. In the panel analysis we explore all
the years. The fundament of the database has been elaborated in previ-
ous studies and is therefore not further discussed in this chapter (see
Schneider, 2005a, b).
14.3.2 Measuring Voice and Accountability and Corruption
Several data sources are used to investigate the relationships discussed in

the theoretical section.
Aggregate governance indicators

We use the Quality of Governance Index as a key proxy for institutional
quality (see Kaufmann, Kraay and Mastruzzi, 2003). Our index values
report the mean value of six governance dimensions for the periods 1996,
1998 and 2000 (first three rounds). It is based on several hundred variables
measuring perceptions of governance and derived from 25 different data
sources. The disadvantage is that no data are available for the year 1990.
Thus, for these variables only two time periods are available in the panel
analysis. In the cross-sectional analysis we use the mean value for these
three time periods. All scores lie between –2.5 and 2.5, with higher scores
corresponding to better institutions (outcomes). The variable voice and
accountability index includes in it a number of indicators that measure
various aspects of the political process, civil liberties and political rights.
The variable measures the extent to which citizens of a country are able
to participate in the selection of governments. The index also includes
three indicators that measure the independence of the media as a proxy
for monitoring the authority and holding them accountable for their
actions. An overview of the measurement is provided in Figure 14.1. The
variable corruption measures perceptions of corruption using the conven-
tional definition of corruption; namely ‘abuse of public power for private
gains’ (Buehn and Schneider, 2009, p. 6). The index is developed from
various sources covering different aspects that range from the frequency
of ‘additional payments to get things done’ to the effects of corruption on
the business environment (Kaufmann, Kraay and Mastruzzi, 2003, p. 8).
Further explanations of the control of corruption variable is provided in
Figure 14.2.
To check the robustness of our results we will also use the ICRG. The
ICRG has a special emphasis on aspects affecting private foreign invest-
ment decisions. The rating comprises 22 variables in three subcategories
of risk: political, financial and economic. We will mainly focus on the
political risk component. We will investigate CORRUPTION.2 A higher


Source: Kaufmann, Kraay and Mastruzzi (2003, p. 91)
Figure 14.1 Measuring voice and accountability
number of points indicates a lower potential risk and therefore higher

scores are in line with a higher institutional and governance quality. The
use of the ICRG will afford an increased number of observations in the
panel regression.


Figure 14.2 Measuring control of corruption
International country risk guide (ICRG) (see also Knack, 1999)

The variables of the data sets ICRG and Aggregate Governance
Indicators are highly correlated. For example, the correlation between the
POLITICAL RISK RATING and the average of all six variables in the
Aggregate Governance Indicators is 0.88. As mentioned, we will use these
two sets of variables in alternative estimations to check the robustness
of our first two core hypotheses. Moreover, due to the high correlation
between our proxy for voice and accountability and corruption, we use
these two sets of indexes in alternate estimations.


14.4 EMPIRICAL RESULTS ON THE SHADOW

ECONOMY
To test our hypotheses we propose the following baseline equation in the
cross-sectional analysis:
SHADOWi 5 a 1 b1 CTRLi 1 b2 GOVINSTi 1 b3 TAXMi
1 REGIONi 1 ei (14.1)
where i indexes the countries in the sample, SHADOWi denotes the

country’s level of shadow economy as a percentage of official GDP,
GOVINSTi are our two indicators for institutional quality as described
in the previous section. TAXMi represents the level of tax morale.3 The
regression also contains several control variables, CTRLi, including
factors such as government interventions, fiscal burden, wage and prices
controls, log GDP per capita, the share of GDP due to agriculture,
the unemployment rate and the share of urban population. REGIONi
are dummy variables that differentiate between developed, Asian, and
developing or transition countries. ei denotes the error term.4 The model
is estimated using cross-section data with mean values for the years 1990
to 1999.5
The panel model has the following specification:
SHADOWit 5 a 1 b1 CTRLit 1 b2 CORRit 1 TDt 1 REGIONi 1 eit

(14.2)
where i indexes the countries in the sample, SHADOWit denotes coun-

tries’ size of the shadow economy as a percentage of the official GDP
over the three time periods t. CORRit are our indicators for governance
and institutional quality; one regression uses the Kaufmann, Kraay and
Mastruzzi (2003) data set and the other regression uses the ICRG data.
We do not control for tax morale in order to maximize the number of
observations. The regressions also contain several control variables,
CTRLi, including factors such as GDP per capita, the share of agriculture
in GDP, the share of urban population, the size of the population, and the
labour force, and the level of trade. To control for time as well as regional
invariant factors, we include fixed time, TDt, and fixed regional effects,
REGIONi.6 eit denotes the error term.7 A detailed discussion of the impact
of further institutional variables are provided in Torgler and Schneider
(2007, 2009).


14.4.1 Empirical Results
Table 14.1 presents the first results focusing on a cross-sectional analysis.

The relative role played by our main variables vis-à-vis other factors is
investigated by estimating beta or standardized regression coefficients.
All estimations use regional dummy variables.8 In the first regression, we
explore the impact of voice and accountability on the size of the shadow
economy. The results indicate that a higher level of voice and account-
ability reduces the size of the shadow economy. Looking at the relative
strength we observe that only GDP per capita has a stronger impact on the
shadow economy. Thus, we find support for the idea that there is a strong
negative correlation between voice and accountability and the size of the
shadow economy. In specification (2) we explore the impact of corruption,
and again we observe a strong correlation between corruption and the size
of the shadow economy. A lower level of corruption is correlated with
higher size of shadow economy. The coefficient is statistically significant
at the 1 per cent level. The effect of corruption seems to be stronger than
the effect of voice and accountability (comparable to GDP per capita
effect). To check the robustness of the results, in specifications (3) and
(4) we present variables from other sources that measure corruption,
namely (as previously discussed) the International Country Risk Guide
(3) and the Transparency International corruption index (CPI) (4). The
CPI attributes a single (CPI) score to each nation. This score ranges from
1 to 10, and a higher value means a lower level of corruption. Published
annually by Transparency International, it relies on the perception of cor-
ruption by business people and country analysts (for a recent discussion
of the methodology, see, Lambsdorff, 2005). Table 14.1 indicates that the
results are robust. In both cases the coefficient is statistically significant at
the 1 per cent level and reports a large effect.
In addition, looking at the control variables we observe not only that eco-
nomic performance (measured via GDP per capita) reduces the size of the
shadow economy, but it also reduces tax morale. Thus, social norms have
an impact on the size of the shadow economy (for a detailed discussion see
Torgler and Schneider, 2009). The beta coefficients also show that its quanti-
tative impact is comparable to other determinants. Table 14.1 indicates that
GOVERNMENT INTERVENTIONS9 have a positive impact on the size
of the shadow economy. More government interventions crowd out private
initiative and investments in the private sector. The economic freedom to
engage in business activities suffers, and frustration arising from too many
interventions by the government might increase the inclination to engage
in illegal activities. We also observe the tendency that a higher level of
UNEMPLOYMENT is positively correlated with a higher level of shadow


Table 14.1 Cross-sectional analysis
OLS
Dependent Variable: Shadow Economy
Independent Variables (1) (2) (3) (4)
(A) INSTITUTIONS (KKM)
VOICE AND −0.307**
ACCOUNTABILITY (−2.14)
CONTROL OF −0.596***
CORRUPTION (−5.68)
TRANSPARENCY
INTERNATIONAL
CORRUPTION −0.429***
(−3.26)
ICRG −0.438***
CORRUPTION (−3.23)
(B) SOCIAL NORMS
TAX MORALE −0.221*** −0.145** −0.156** −0.131**
(−3.01) (−2.16) (−2.11) (−2.09)
(C) GOVERNMENT
GOVERNMENT 0.252** 0.303*** 0.318*** 0.310***
INTERVENTIONS (2.18) (3.16) (2.92) (2.76)
FISCAL BURDEN −0.132 −0.156* −0.158 −0.078
(−1.20) (−1.78) (−1.63) (−0.75)
WAGE AND PRICES −0.240** −0.232*** −0.187** −0.356***
(−2.25) (−2.80) (−2.15) (−3.93)
(D) CONTROL VARIABLES
LOG (GDP PER CAPITA) −0.677** −0.513** −0.568* −0.600*
(−2.50) (−2.09) (−1.86) (−1.98)
AGRICULTURE/GDP −0.038 −0.033 0.048 0.089
(−0.24) (−0.20) (0.26) (0.46)
UNEMPLOYMENT 0.134** 0.052 0.014 0.176**
(2.10) (0.83) (0.26) (2.46)
Regional Fixed Effects YES YES YES YES
Observations 55 55 53 49
R-squared 0.767 0.798 0.814 0.812
Prob > F 0.000 0.000 0.000 0.000
Notes: t-statistics in parentheses. Significance levels: * 0.05 < p < 0.10, ** 0.01< p < 0.05,
*** p < 0.01.


economy. However, the coefficient is not always statistically significant.

Time does not act as a restriction to being active in the shadow economy.
Moreover, these people have an incentive not to report their additional work
hours as otherwise they would lose their financial support. If the wage of
illicit work and the financial aid together yield more income than regular and
overtime work, taking into account the costs of detection and punishment
and assuming risk neutrality, full-time illicit work as an unemployed person
yields, ceteris paribus, a higher utility. In such a situation, the danger that a
person remains in the shadow economy and turns down job offers increases
(Schneider and Enste, 2002). Table 14.1 also shows a negative coefficient
of FISCAL BURDEN10 (but only in one specification is the coefficient
statistically significant). The first result is in line with previous findings such
as Friedman et al. (2000) and Dreher and Schneider (2010). It seems that a
higher fiscal burden does not per se drive firms into the unofficial economy.
As Friedman et al. (2000) stress, such proxies do not measure how the tax
system is administered, which might explain such a result. On the other
hand, we observe a somewhat surprising result that price and wage regula-
tions are not reasons for firms to move into the unofficial economy. As a
proxy we use the variable WAGE AND PRICES developed by the Index of
Economic Freedom, and provided by Heritage.11
Evaluating the direct effect of institutional/governance quality on the
size of the shadow economy requires an investigation of any potential
causality problems and therefore an instrumental variable technique.
Similarly, one may also raise the criticism that tax morale is endogenous
(Torgler and Schneider, 2009). Recent studies highlight the importance of
considering countries’ historical and geographic features as instrumental
variables. These characteristics can influence countries’ performance
through their impact on the institutional and political environment.12
Studies such as those by Alesina et al. (2003) or La Porta et al. (1999)
implement a broad data set to consider factors such as latitude, ethnic
fractionalization, language, religion or legal origin. In our case we take
the following instruments for institutional quality and tax morale: legal
origin (English, German, French dummies), latitude, fractionalization
(language), religion (protestant, catholic dummies) and the legal system
(political rights). Table 14.2 shows three 2SLS estimations with several
diagnostic tests. The results indicate that the coefficients of institutional/
governance quality are statistically significant in all cases, which supports
the previous findings. Similarly, tax morale also remains statistically
significant. The F-tests for the instrument exclusion set in the first-stage
regressions are in all cases statistically significant at the 5 per cent level
for tax morale. In addition, Table 14.2 also reports a test for instrument
relevance using the Anderson canonical correlations LR for whether


Table 14.2 2SLS regressions
2SLS Regressions
Dependent Variable: Shadow Economy (5) (6)
(A) INSTITUTIONS (KKM)
VOICE AND ACCOUNT. −5.826*
(−1.82)
CONTROL OF CORRUPTION −9.935***

(−3.05)
(B) SOCIAL NORMS
TAX MORALE −12.018** −8.332*
(−2.36) (−1.88)
(C) GOVERNMENT
GOVERNMENT INTERVENTIONS 4.149** 5.312***
(2.29) (3.03)
FISCAL BURDEN −3.769 −5.005*
(−1.40) (−1.96)
WAGE AND PRICES −5.278** −5.344***
(−2.35) (−2.68)
(D) CONTROL VARIABLES
LOG (GDP PER CAPITA) −12.024*** −7.483*
(−3.00) (−1.77)
AGRICULTURE/GDP −0.104 −0.080
(−0.42) (−0.34)
UNEMPLOYMENT 0.493 0.185
(0.80) (0.78)
Regional Fixed Effects YES YES
Prob > F 0.000 0.000
Centered R2 0.761 0.787
First Stage Regressions
Tax Morale:
F–Test of excluded instruments 2.99** 2.99**
Institutions:
F-Test of excluded instruments 9.20*** 3.09***
Anderson canon. corr. LR statistic 21.353*** 22.082***
Sargan statistic 8.230 3.706
*** p < 0.01.


Table 14.3 Governance and institutional quality and the size of shadow
economy
Dependent Variable: Shadow Economy FE FE

(7) (8)
(A) CORRUPTION
CORRUPTION (ICRG) −3.018***
(−4.82)
CONTROL OF CORRUPTION −7.881***
(KAUFMANN ET AL.) (−5.62)
(B) CONTROL VARIABLES
LOG (GDP PER CAPITA) −5.032*** −3.532**
(−4.63) (−2.48)
AGRICULTURE (% OF GDP) –0.196** −0.122
(−2.07) (1.13)
URBANIZATION 0.029 0.070
(0.58) (1.24)
LOG (POPULATION) −12.255*** −14.400***
(−3.13) (−3.03)
LOG (LABOUR FORCE) 10.507*** 12.029**
(2.71) (2.55)
TRADE (% GDP) −0.011 −0.015
(−0.64) (−0.75)
Time Fixed Effects YES YES
Observations 274 204
R-squared 0.524 0.558
Prob > F 0.000 0.000
*** p < 0.01.
the equation is identified. The test shows that the null hypothesis can be
rejected. We also present the Sargan’s test for over-identification for all
2SLS to examine the validity of the exclusion restrictions. This test fails to
reject the null hypothesis that our instruments are valid.
Next, we explore the impact of corruption on the shadow economy
using the panel data set using regional and time fixed effects, and both
estimations reported in Table 14.3 indicate that a reduction in corruption
leads to a decrease in the size of the shadow economy.
Moreover, in line with our previous results in Tables 14.1 and 14.2, we


Table 14.4 2SLS estimations focusing on corruption
Dependent Variable: Shadow Economy

(9) (10)
(A) CORRUPTION
ICRG
CORRUPTION −9.540***
(−3.13)
CONTR. OF CORRUPTION −12.245***
(−3.19)
(B) CONTROL VARIABLES INCL. INCL.
FIRST STAGE REGRESSIONS

INSTR. INST./GOV. Q.
TEMPERATURE −0.037*** −0.021***
(−3.44) (−3.32)
SOCIOECON. CONDITIONS 0.091** 0.107***
(2.31) (−4.58)
Test of excluded instruments 7.99*** 15.07***
Time Fixed Effects YES YES
Anderson canon. corr. LR statistic 16.130*** 29.240***
Anderson Rubin test 6.71*** 4.78***
Sargan statistic 0.017 0.295
Prob > F 0.000 0.000
Observations 219 150
*** p < 0.01. Control variables added.
find that economic performance matters. We observe that an increase in

the GDP per capita is associated with a decrease in the shadow economy.
We use fewer control variables in these estimations to retain a higher
number of observations. Looking at other factors we observe that the
population size and the labour force participation changes are correlated
with changes in the shadow economy.
Table 14.4 reports two 2SLS estimations. An increasing number of
studies stress that climatic conditions have an impact on countries’
or regions’ institutions and their development and individuals’ atti-
tudes and their behaviour (see, for example, Diamond, 1997; Engerman


and Sokoloff, 1997; Landes, 1998; La Porta et al., 1999; Sachs, 2000;
Hirshleifer and Shumway, 2003; Coyle, 2004). Such external situations
may affect the character of inhabitants and hence their culture and insti-
tutional arrangements. According to Diamond (1997), geography and
climate help to explain different nations’ economic destinies. The studies
of Engerman and Sokoloff (1997), Landes (1998) and Sachs (2000) inves-
tigate the connection between climate and economic development. Sachs
(2000), for example, presents evidence that production technology in
the tropics has lagged behind temperate zone technology in the areas of
agriculture and health which opened a considerable income gap between
climate zones. Roll (1992) stresses that the unambiguously observable
weather is a genuinely exogenous identifying variable. Schaltegger and
Torgler (2007), for example, have shown that weather conditions are
valid instruments for government accountability. An advantage of using
temperature as an instrument is that we observe a certain variety over
time and therefore it can be implemented in a panel analysis. Coyle (2004)
claims that a higher temperature is related to a lower performance and
productivity, and even now, many countries in Europe do not have air-
conditioning. Hence, we implement a nation’s yearly mean TEMPERA-
TURE in Celsius13 as an instrument for governance/ institutional quality.
In addition, we also use the SOCIOECONOMIC CONDITIONS as a
second instrument of governance and institutional quality. It measures
general public satisfaction or dissatisfaction covering a broad spectrum
of factors ranging from infant mortality and medical provision to housing
and interest rates. The data are provided by the EFW. Table 14.4 also
reports several diagnostic tests. In both specifications we observe that
corruption remains statistically significant, which supports our previous
results. The instruments used are effective in explaining corruption (see
first stage regression results). The F-tests for the instrument exclusion set
in the first-stage regressions are also statistically significant at the 1 per
cent level. In addition, a test for instrument relevance using the Anderson
canonical correlations LR for whether the equation is identified shows
that the null hypothesis can be rejected in both cases. The Anderson–
Rubin test suggests that the endogenous variables are jointly statistically
significant. Such a test is robust to the presence of weak instruments.
We also present the Sargan’s test for over-identification for those 2SLS
regressions in which we have more than two instruments to examine
the validity of the exclusion restrictions. The test fails to reject the null
hypothesis that our instruments are valid.
In sum, the empirical results provided in this section suggest that our
key hypotheses cannot be rejected. Corruption and voice and account-
ability play a significant role in the determination of the size of the shadow


economy. In the next section we are going to explore whether these factors
also influence countries’ tax performance.
14.5 EMPIRICAL RESULTS ON TAX

PERFORMANCE
To test whether government quality fosters tax performance or tax efforts,

we propose the following baseline equation:
TEit 5 a 1 b1 Yit 1 b2 POPit 1 b3 XMit 1 b4 AGRit 1 b5 GOVQit
1 TDt 1 REGIONi 1 eit (14.3)
where i indexes the countries in the sample and t the time period, TEi
denotes the country’s level of tax effort measured by the tax revenue as a
share of gross domestic product (GDP), Yi the GDP per capita (measured
in US$), POPi the rate of population growth, XMi the average of exports
plus imports as a share of the GDP, AGRi the agriculture level as a share
of GDP and GOVQi are our indicators for voice and accountability and
corruption. To control for time as well as regional invariant factors, we
include fixed time, TDt, and fixed regional effects, REGIONi.14 eit denotes
the error term. We report FE regressions. We have data for a time period
of 16 years when using the ICRG risk guide data (1990–2005). When
implementing the Kaufmann, Kraay and Mastruzzi (2003) data set we
work with the years 1996, 1998, 2000, 2002 and 2005. Such an analysis
goes beyond the previous studies that have explored this framework
mainly in a cross-sectional environment (for example, Bird, Martinez-
Vazquez and Torgler, 2006, 2008).
The explanatory variables employed in the model follow those used
in the conventional tax effort literature (traditional supply factors). Per
capita GDP is a proxy for the level of development of a country. A higher
level of development goes together with a higher capacity to pay and
collect taxes, as well as a higher relative demand for income elastic public
goods and services (Bahl, 1971; Chelliah, 1971). In general, we would
expect a positive relationship between the level of per capita income and
the level of tax effort. Demographic characteristics may also be an impor-
tant determinant of tax effort. As Bahl (2003, p. 13) points out, in coun-
tries with faster growing populations, tax systems may lag behind in the
ability to capture new taxpayers. This suggests that the rate of population
growth is negatively related to the level of tax effort. The most traditional
explanatory variables in the conventional tax effort literature are those


controlling for a country’s economic structure. These variables reflect the

idea that the availability of ‘tax handles’ should influence the level of tax
effort. For example, trade taxes are often a major source of government
revenues in less developed countries because they are easier to collect than
income taxes. We measure the availability of this tax handle by openness,
defined as the sum of exports and imports as a share of GDP. The tax
ratio is expected to be positively related to the degree of openness of the
economy. Moreover, the sectoral composition of domestic product may
also affect the ability to tax. A traditional measure signalling the difficulty
in taxing domestic output is the share of agriculture in GDP. Some argue
that the agricultural sector is not much more difficult to tax (Bahl, 2003),
but the larger its relative importance in a country’s economy the lower
the need to spend on governmental activities and services, as many public
sector activities are city-based (Tanzi, 1992). In addition, for political
reasons some countries exempt a large share of agricultural activities from
taxes. A higher agriculture share in GDP should thus produce a lower tax
ratio.
Table 14.5 presents the results. Both corruption proxies are statisti-
cally significant at the 1 per cent level (see specifications 11 and 12).
Similarly, specification (13) shows that voice and accountability is also
statistically significant at the 1 per cent level. Thus, an improvement in
the governance/institutional conditions leads to a reduction of the size
of the shadow economy. These results give support to the hypothesis
that societies’ willingness to tax themselves depends on the perception
that government institutions are honest and responsive and that there is
a fair and predictable public sector environment. Looking at the control
variables we observe that a faster rate of population growth leads to a
lower tax ratio. A higher share of agricultural sector is correlated with a
lower tax effort. On the other hand, the coefficient for GDP per capita
is not statistically significant, but the results are in line with previous
studies. However, in our results, openness of the economy is associ-
ated with a higher tax effort. Thus, the results imply that conventional
supply factors continue to play a robust role in influencing tax effort.
However, demand factors such as governance/institutional conditions
clearly matter.
14.6 CONCLUSIONS
Our study shows that policies improving voice and accountability and
reducing corruption can help to reduce the incentive to take economic
activities underground. The institutional architecture and governance


Table 14.5 Tax effort and institutions/governance
FE REGRESSIONS
Dependent Variable: (11) (12) (13)
Tax Effort
Independent Variables Coeff. t-Stat. Coeff. t-Stat. Coeff. t-Stat.
(A) INSTITUTIONS/GOVERNANCE
CORRUPTION ICRG 1.296*** 6.84
CONTROL OF 2.069*** 3.96
CORRUPTION
VOICE/ 2.331*** 4.86
ACCOUNTABILITY
(B) ECONOMIC STRUCTURE
AGRICULTURE/GDP −0.127*** −6.16 −0.118*** −3.66 −0.102*** −3.18
(C) DEVELOPMENT
GDP PER CAPITA −0.00003 −1.27 −0.00003 −0.61 0.00004 0.87
POPULATION −0.434** −2.23 −0.842*** −2.65 −0.728** −2.33
GROWTH
(D) OPENNESS
(EXPORT/GDP 1 0.061*** 7.04 0.075*** 5.70 0.080*** 6.13
IMPORT/GDP)/2
REGIONS YES YES YES
YEARS YES YES YES
Observations 941 388 395

Prob > F 0.000 0.000 0.000
R-squared 0.414 0.501 0.507
*** p < 0.01.
quality seem to be a key component in the size of the shadow economy. A

more legitimate and responsive state appears to be an essential precondi-
tion to reduce participation in the shadow economy. Citizens feel cheated
if corruption is widespread and react through more activity in the shadow
economy. A high level of voice and accountability allows expression of
one’s own preferences and participation in the political process enhances
identification with a state’s institutions; this counteracts the inclination to
be active in the shadow economy. Participation and identification through
voice and accountability therefore reduces the size of free-rider problems.


If citizens and authorities interact with a sense of collective responsibility

thanks to the institutional structures, the system may be better governed
and the policies more effective, as accountability promotes effectiveness
through its impact on government behaviour (Schaltegger and Torgler,
2007).
Furthermore, our results indicate that improving institutions through
channels such as enhancing voice or accountability and reducing cor-
ruption generates a better tax performance. Thus, a demand policy that
improves such factors should be quite successful in improving tax efforts.
Such changes should not take longer and should not be more difficult than
the opportunities for tax handles and economic structure, such as the rela-
tive share of the agriculture sector in the economy or the weight of imports
and exports in GDP.
NOTES
1. Furthermore, de Soto argues that it is nearly as difficult to stay legal, as it is to become

legal. In Venezuela, the share of employees working in legal enterprises decreased from
two-thirds in 1976 to less than half at the end of the century as people have created new
business illegally to fill the gaps in the legal economy.
2. Assessment of corruption within the political system. Lower scores indicate ‘high
government officials are likely to demand special payments’ and that ‘illegal payments
are generally expected throughout lower levels of government’ in the form of ‘bribes
connected with import and export licences, exchange controls, tax assessment, police
protection or loans.’
3. For a description of the variables, see Torgler and Schneider (2009).
4. For summary statistics see Appendix Table A14.1.1.
5. The use of average values over a period allows maximizing the number of observations.
6. We differentiate between developed, Asian, and developing or transition countries.
7. For summary statistics and an overview of the countries see Appendix Table A14.1.1.
8. For an overview of the countries see Table A14.1.3 in the appendix.
9. As a proxy for government interventions we use the Index of Economic Freedom
provided by Heritage. According to Beach and Miles (2005, p. 65) this factor measures
‘government’s direct use of scarce resources for its own purposes and government’s
control over resources through ownership’. Five factors are included in this variable
(1) government consumption as a percentage of the economy, (2) government owner-
ship of businesses and industries, (3) share of government revenues from state-owned
enterprises, (4) government ownership of property of property and (5) economic output
produced by the government. The scale goes from 1 to 5 (the more interventions, the
higher the score).
10. We use the fiscal burden variable of the Index of Economic Freedom provided by
Heritage as a proxy. The variable measures the marginal tax rates (top marginal income
and corporate tax rate) and the year-to-year change in the level of government expendi-
tures as a percent of GDP. The scale lists scores from 1 through 5: the higher the fiscal
burden, the higher the score.
11. It measures the extent to which the government allows the market to set wages and
prices, and evaluates the following factors: minimum wage laws, freedom to set
prices privately without government influence, government price controls, extent of
government price controls and price affecting subsidies to businesses. The higher the


value in a scale from 1 to 5, the more strict the governmental regulations of wages and
prices.
12. See Hall and Jones (1999) and Acemoglu, Johnson and Robinson (2001).
13. See Mitchell et al. (2003).
14. We differentiate between Europe, Latin America, North America, North Africa, Sub
Saharan Africa, the Pacific, Asia, the Caribbean and Australia.
REFERENCES
Abed, G.T., and S. Gupta (eds) (2002), Governance, Corruption and Economic
Performance. Washington: International Monetary Fund.
Acemoglu, D., S. Johnson and J. Robinson (2001). The colonial origins of com-
parative development: an empirical investigation. American Economic Review,
91, 1369–1401.
Aidt, T., J. Dutta and V. Sena (2008), Governance regimes, corruption and
growth: theory and evidence. Journal of Comparative Economics, 36, 195–220.
Alesina, A., A. Devleeschauwer, W. Easterly, S. Kurlat and R. Wacziarg (2003),
Fractionalization. Journal of Economic Growth, 8, 155–194.
Alm, J., B.R. Jackson and M. McKee (1993), Fiscal exchange, collective deci-
sion institutions, and tax compliance. Journal of Economic Behavior and
Organization, 22, 285–303.
Bahl, R.W. (1971), A regression approach to tax effort and tax ratio analysis.
International Monetary Fund Staff Paper, 18, 570–612.
Bahl, R.W. (2003), Reaching the hardest to tax: consequences and possibilities.
Paper presented at the ‘Hard to Tax: An International Perspective’ conference,
Andrew Young School of Policy Studies, Georgia State University, 15–16 May.
Bird, R., J. Martinez-Vazquez and B. Torgler (2006). Societal institutions and
tax effort in developing countries. In J. Alm, J. Martinez-Vazquez and M.
Rider (eds), The Challenges of Tax Reform in the Global Economy. New York:
Bird, R.M., J. Martinez-Vazquez and B. Torgler (2008), Tax effort in develop-
ing countries and high income countries: the impact of corruption, voice and
accountability. Economic Analysis and Policy, 38, 55–71.
Beach, W.W. and M.A. Miles (2005), Explaining the factors of the index of eco-
nomic freedom. In 2005 Index of Economic Freedom. Washington, The Heritage
Foundation, Chapter 5.
Buehn, A. and F. Schneider (2009), Corruption and the shadow economy: a struc-
tural equation model approach. Institute for the Study of Labor, IZA DP No.
4182.
Chelliah, R.J. (1971), Trends in taxation in developing countries. International
Monetary Fund Staff Papers, 18, 254–331.
Choi, J.P. and M. Thum (2005), Corruption and the shadow economy.
International Economic Review, 46, 817–836.
Cialdini, R.B. (1989), Social motivations to comply: norms, values and prin-
ciples. In J.A. Roth and J.T. Scholz (eds), Taxpayer Compliance, Volume 2.
Philadelphia: University of Pennsylvania Press, pp. 200–227.
Coyle, D. (2004), Sex, Drugs and Economics: An Unconventional Introduction to
Economics. Australia: Thomson Texere.


de Soto, H. (1989), The Other Path: The Invisible Revolution in the Third World.
New York: Harper & Row.
de Soto, H. (2000), The Mystery of Capital: Why Capitalism Triumphs in the West
and Fails Everywhere Else. New York: Basic Books.
Diamond, J. (1997), Guns, Germs, and Steel. New York: W.W. Northon & Co.,
Inc.
empirical analysis. Public Choice, 144(1–2), 215–238.
Finance, 16, 773–796.
Engerman, S. and K. Sokoloff (1997), Factor endowments, institutions and differ-
ential paths of growth among the New World economies. In S. Haber (ed.), How
Latin America Fell Behind. Stanford: Stanford University Press.
the grabbing hand: the determinants of unofficial activities in 69 countries.
Hall, R. and C. Jones (1999), Why do some countries produce so much more per
worker than others? Quarterly Journal of Economics, 114, 83–116.
Hirshleifer, D. and T. Shumway (2003), Good day sunshine: stock returns and the
weather. Journal of Finance, 58, 1009–1032.
Johnson, A., Kaufman, D. and Schleifer, A. (1997), The unofficial economy in
transition. Brookings Papers on Economic Activity, 28, 157–240.
Karymshakov, K. and Y. Abdykaparov (2008), Corruption in Kyrgyzstan. Sosyo
Ekonomi, 2008-1.
Kaufmann, D., A. Kraay and M. Mastruzzi (2003), Governance matters III: gov-
ernance indicators for 1996–2002. World Bank, 30 June.
Kidder, R. and C. McEwen (1989), Taxpaying behavior in social context: a ten-
tative typology of tax compliance and noncompliance. In J.A. Roth and J.T.
Scholz (eds), Taxpayer Compliance, Volume 2. Philadelphia: University of
Pennsylvania Press, pp. 46–75.
Knack, S. (1999), Aid dependence and the quality of governance. Working Paper,
IRIS Center, University of Maryland.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R. Vishny (1999), The quality of
government. Journal of Law, Economics, and Organization, 15, 222–278.
Landes, D. (1998), The Wealth and Poverty of Nations: Why Some Are So Rich and
Some So Poor. New York: Norton.
Lambsdorff, J.G. (2005), Methodology of the 2005 corruption perceptions index.
Transparency International and University of Passau, September.
Lempert, R.O. (1972), Norm-making in social exchange: a contract law model.
Law and Society Review, 1, 1–32.
Levi, M. (1988), Rules and Revenue. Berkeley: University of California Press.
Levin, M. and G. Satarov (2000), Corruption and institutions in Russia. European
Journal of Political Economy, 16, 113–132.
McEwen, C.A. and R.J. Maiman (1986), In search of legitimacy: toward an empir-
ical response analysis. Law and Policy, 8, 257–273.
Mitchell, T.D., T.R. Carter, P.D. Jones, M. Hulme and M. New (2003), A com-
prehensive set of high-resolution grids of monthly climate for Europe and
the Globe: the observed record (1901–2000) and 16 scenarios (2001–2100).
Unpublished manuscript, Tyndall Centre.


Rei, D. and M. Bhattacharya (2008), The impact of institutions and policy

on informal economy in developing countries: an econometric explora-
tion. Working Paper No. 84, Policy Integration and Statistics Department,
International Labour Office, Geneva.
Roll, R.W. (1992), Weather. In P. Newman, M. Milgate and J. Eatwell (eds),
The New Palgrave Dictionary of Money and Finance. London: Macmillan
Press.
Sachs, J. (2000), Tropical underdevelopment. NBER Working Paper, No. 8119.
Schaltegger, C.A. and B. Torgler (2007), Government accountability and fiscal
discipline: a panel analysis with Swiss data. Journal of Public Economics, 91,
117–140.
Schneider, F. (2005a), Shadow economies of 145 countries all over the world: what
do we really know? CREMA Working Paper 2006-01, Center for Research in
Economics, Management and the Arts, Basel.
Schneider, F. (2005b), Shadow economies around the world: what do we really
Schneider, F. and A. Buehn (2007), Shadow economies and corruption all over the
world: revised estimates for 120 countries. Economics E-Journal, 2007-9.
Survey. Cambridge: Cambridge University Press.
Tanzi, V. (1992). Structural factors and tax revenue in developing countries: a
decade of evidence. In I. Goldin and L.A. Winters (eds), Open Economies:
Structural Adjustment and Agriculture. Cambridge, UK: Cambridge University
Press, pp. 267–281.
Tanzi, V. (2002), Foreword. In F. Schneider and D.H. Enste (eds), The Shadow
Economy. An International Survey. Cambridge, UK: Cambridge University
Press.
Torgler, B. and F. Schneider (2007), Shadow economy, tax morale, governance
and institutional quality: a panel analysis. IZA Discussion Papers 2563, Institute
for the Study of Labor (IZA).
Tyler, T.R. (1990a), Justice, self-interest, and the legitimacy of legal and political
authority. In J.J. Mansbridge (ed.). Beyond Self-Interest. Chicago: University of
Chicago Press, pp. 171–179.
Tyler, T.R. (1990b), Why People Obey the Law. New Haven, CT: Yale.
Tyler, T.R. (1997), Procedural fairness and compliance with the law. Swiss Journal
of Economics and Statistics, 133, 219–240.
Tyler, T.R., J.D. Casper and B. Fisher (1989), Maintaining allegiance toward
political authorities: the role of prior attitudes and the use of fair procedures.
American Journal of Political Science, 33, 629–652.
Virta, H. (2009), The linkage between corruption and shadow economy size: does
geography matter? International Journal of Development Issues, 9, 4–24.
Wagener, H.J. (2004), Good governance, welfare, and transformation. The
European Journal of Comparative Economics, 1, 127–143.


APPENDIX 14.1
Table A14.1.1 Descriptive statistics of the cross-sectional analysis
VARIABLES Mean Std. Min Max Source

Dev.
DEPENDENT VARIABLE
SHADOW ECONOMY 29.413 12.944 7.670 62.500 Schneider (2005)
INDEPENDENT VARIABLES
TAX MORALE 2.103 0.355 1.370 3.014 WVS/
Latinobarometro
VOICE AND ACCOUNT. 0.092 0.941 −1.890 1.610 Kaufmann et al.
(2003)
CONTROL OF CORRUP. 0.130 0.955 −1.610 2.390 Kaufmann et al.
(KAUFMANN ET AL.) (2003)
CORRUPTION (TI) 4.603 2.320 1.600 10.000 Transparency
International
CORRUPTION (ICRG) 3.565 1.204 0.338 6.000 ICRG
FISCAL BURDEN 3.693 0.613 1.750 4.960 Heritage
WAGE PRICES 2.716 0.761 1.000 4.750 Heritage
LOG (GDP PER CAPITA) 8.470 1.021 6.209 10.224 World Development
Indicators
AGRICULTURE/GDP 17.123 13.843 0.210 59.970 World Development
Indicators
UNEMPLOYMENT 9.308 6.170 0.720 39.300 World Development
Indicators


Table A14.1.2 Descriptive statistics and a summary of the results panel

analysis
VARIABLES Mean Std. Min Max Source

Dev.
DEPENDENT VARIABLE
SHADOW ECONOMY 29.594 13.193 6.90 67.30 Schneider (2005a, b)
CORRUPTION 3.473 1.273 0.08 6.00 ICRG
CONTROL OF 0.156 1.040 −1.98 2.56 Kaufmann et al.
CORRUP. (2003)
CONTROL VARIABLES
LOG (GDP PER 7.654 1.586 4.71 10.53 World Development
CAPITA) Indicators
AGRICULTURE 16.640 13.442 0.07 57.65 World Development
(% of GDP) Indicators
URBANIZATION 55.715 22.131 8.90 100.00 World Development
Indicators
LOG (POPULATION) 16.550 1.306 14.17 20.95 World Development
Indicators
LOG (LABOUR 15.705 1.315 13.15 20.42 World Development
FORCE) Indicators
TRADE (% GDP) 71.811 39.133 14.41 290.85 World Development
Indicators
INSTRUMENTS
ANNUAL 16.789 8.194 −5.50 29.00 Mitchell et al. (2003)
TEMPERATURE
SOCIO-ECONOMIC 5.693 1.943 1.00 11.00 ICRG
CONDITIONS


Table A14.1.3 Overview of the countries of the cross-country analysis
Argentina Italy
Australia Japan
Austria Korea, Rep.
Azerbaijan Latvia
Bangladesh Mexico
Belarus Moldova
Belgium Netherlands
Bolivia Nicaragua
Brazil Norway
Bulgaria Panama
Canada Peru
Chile Philippines
China Poland
Colombia Portugal
Costa Rica Romania
Croatia Russian Federation
Czech Republic Slovak Republic
Denmark Slovenia
Dominican Republic South Africa
Ecuador Spain
Egypt, Arab Rep. Sweden
Finland Switzerland
France Turkey
Georgia Ukraine
Germany United Kingdom
Greece United States
Hungary Uruguay
India Venezuela
Ireland


Table A14.1.4 Overview of the countries in the panel analysis on the shadow economy
1990 1995 2000

Countries Countries Countries Countries Countries Countries
Albania Madagascar Albania Malawi Albania Lebanon
Algeria Malawi Algeria Malaysia Algeria Lithuania
Argentina Malaysia Argentina Mali Argentina Madagascar
Australia Mali Australia Mexico Armenia Malawi
Austria Mexico Austria Mongolia Australia Malaysia
498
Bangladesh Mongolia Bangladesh Morocco Austria Mali
Belgium Morocco Belgium Mozambique Azerbaijan Mexico
Bolivia Mozambique Bolivia Netherlands Bangladesh Moldova
Botswana Netherlands Botswana New Zealand Belarus Mongolia
Brazil New Zealand Brazil Nicaragua Belgium Morocco
Bulgaria Nicaragua Burkina Faso Niger Bolivia Mozambique
Burkina Faso Niger Cameroon Nigeria Botswana Netherlands
Cameroon Nigeria Canada Norway Brazil New Zealand
Canada Norway Chile Pakistan Bulgaria Nicaragua
Chile Pakistan China Panama Burkina Faso Niger
China Panama Colombia Peru Cameroon Nigeria
Colombia Peru Costa Rica Philippines Canada Norway

Costa Rica Philippines Côte d’Ivoire Poland Chile Pakistan
Côte d’Ivoire Poland Czech Republic Portugal China Panama
27/07/2011 16:16
Denmark Portugal Denmark Romania Colombia Peru
Dominican Romania Dominican Russian Federation Costa Rica Philippines
Republic Republic
Ecuador Saudi Arabia Ecuador Saudi Arabia Côte d’Ivoire Poland

Egypt, Arab Rep. Senegal Egypt, Arab Rep. Senegal Croatia Portugal
Ethiopia South Africa Ethiopia Slovak Republic Czech Republic Romania
Finland Spain Finland South Africa Denmark Russian Federation
France Sri Lanka France Spain Dominican Saudi Arabia
Republic
Germany Sweden Germany Sri Lanka Ecuador Senegal
Ghana Switzerland Ghana Sweden Egypt, Arab Rep. Slovak Republic
Greece Syrian Arab Greece Switzerland Ethiopia Slovenia
Republic
Guatemala Tanzania Guatemala Syrian Arab Finland South Africa
499
Republic
Honduras Thailand Honduras Tanzania France Spain
Hong Kong, China Tunisia Hong Kong, China Thailand Germany Sri Lanka
Hungary Turkey Hungary Tunisia Ghana Sweden
India Uganda India Turkey Greece Switzerland
Indonesia United Arab Indonesia Uganda Guatemala Syrian Arab Republic
Emirates

27/07/2011 16:16

1990 1995 2000
Countries Countries Countries Countries Countries Countries
Iran, Islamic Rep. United Kingdom Iran, Islamic Rep. United Arab Honduras Tanzania
Emirates
Ireland United States Ireland United Kingdom Hong Kong, China Thailand
Italy Uruguay Italy United States Hungary Tunisia
Jamaica Venezuela, RB Jamaica Uruguay India Turkey
Japan Vietnam Japan Venezuela, RB Indonesia Uganda
500
Jordan Yemen, Rep. Jordan Vietnam Iran, Islamic Rep. Ukraine
Kenya Zambia Korea, Rep. Yemen, Rep. Ireland United Arab Emirates
Korea, Rep. Zimbabwe Lebanon Zambia Italy United Kingdom
Madagascar Zimbabwe Jamaica United States
Japan Uruguay
Jordan Venezuela, RB
Kazakhstan Vietnam
Kenya Yemen, Rep.
Korea, Rep. Zambia
Latvia Zimbabwe
TOTAL 86 88 100

27/07/2011 16:16
Table A14.1.5 Descriptive statistics tax performance analysis
VARIABLES Mean Std. Dev. Min Max Source

DEPENDENT VARIABLE
TAX EFFORT 16.3981 7.1185 0.0800 44.3400 WDI
(tax revenue as a share of
gross domestic product
(GDP))
INDEPENDENT VARIABLES
CORRUPTION ICRG 3.0799 1.3354 0.0000 6.0000 ICRG
CONTROL OF 0.0009 0.9987 −2.0499 2.5830 Kaufmann
CORRUPTION et al. (2005)
VOICE/ −0.0005 0.9989 −2.3223 1.7551 Kaufmann
ACCOUNTABILITY et al. (2005)
ACCOUNTABILITY
AGRICULTURE/GDP 17.5938 15.2093 0.0000 93.9700 WDI
GDP PER CAPITA 5877.158 8699.657 56.4600 51590.170 WDI
POPULATION 1.4888 1.6142−44.4000 11.1800 WDI
GROWTH
OPENNESS 42.4389 23.9190 0.7600 229.5550 WDI
average (export and
import as a share of the
gross domestic product)



Index
Abdykaparov, Y. 473 auditing (Germany) 411, 432
ability-to-pay principle 348, 350, 364 Augmented Dickey Fuller tests 135–6
A.C. Nielsen Company 279 Australia 72, 106, 155, 267, 317, 459
accountability see corruption, voice Bureau of Statistics 294
and accountability Centrelink 314
Acharya, S.N. 132, 134 Office of Fair Trading 314, 322
Adjusted Goodness of Fit Index 457 ranking according to size of shadow
administrative relief 162 economy 35, 36, 40
Africa 30, 156, 267 Taxation Office (ATO) 294, 300,
age 269, 329, 386, 388 314, 322–3
AGFI 71 WorkCover Authority of NSW 314
aggregate efficiency 112–13 see also residential construction
Ahrend, R. 141 sector in Australia
Ahumada, H. 62 Austria 72, 97–100, 117, 155, 193, 459
Aidt, T. 471–2 coefficients of variation 234–5
Aigner, D. 133 construction and craftsmen sector 99
Alañon-Pardo, A. 212, 214 do-it-yourself activities 267, 273
Albania 33, 34, 38, 42, 72 labour market and productivity
Alesina, A. 483 109–10
Alexeev, M. 53, 267 moonlighting 167
Algeria 28, 31, 37, 42, 44, 72 national average of shadow
Allingham, M.G. 83–4, 363, 409 economy 225
Alm, J. 356 ranking according to size of shadow
alphabetical listing of 162 countries economy 35, 36, 39, 40
72–6 size and development of shadow
anchoring procedures 222–4, 226, 227, economy 105, 107, 156, 160–61
229, 231, 233, 242 tax gaps 236, 239
Anderson canonical correlations LR tax morale and compliance 384–5,
483, 487 386
Anderson–Rubin test 487 tax and social security contribution
Andreoni, J. 85 burdens 12, 87
Angola 33, 43, 72 undeclared and under-declared work
Angrist, J.D. 390–91 187
Anselin, L. 208 average hours worked per week 277,
Aquinas, T. 364 279
Argentina 28, 31, 37, 41, 51, 72 average informality 46, 47
Aristotle 364 average shadow economy measure by
Armenia 43, 72 region 47
Asia 30, 46–7, 109, 156, 188, 267 average working time in OECD (high-
see also Central Asia income) countries 101–102
attitude (identification problem) 390 Azerbaijan 44, 72
503


‘bad’ equilibrium 14, 88 Botswana 28, 32, 37, 42, 72

Bagachwa, M.S.D. 132 Brazil 29, 32, 38, 43, 51, 72
Bahamas, the 41, 72 Breusch, T. 129, 204–205
Bahrain 28, 31, 36, 40, 72 bribery 472, 473–4, 475
Bajada, C. 113–14 Brodersen, S. 270, 288
Baltic countries 226 Brunei Darussalam 41, 72
see also Estonia; Latvia; Lithuania Buchanan, J.M. 365
Bangladesh 27, 29, 32, 38, 42, 72 Buehn, A. 261, 266–8, 278–9, 394, 471
Becker, G.S. 268, 289, 409 Bühn, A. 114–15
behaviour (identification problem) 390 Bulgaria 72, 186–7, 192–3, 384
Belarus 43, 72, 384 do-it-yourself activities 270, 271, 272
Belev, B. 30, 267 size of shadow economy
Belgium 72, 106, 155, 192–3, 234, (1999–2006/07) 34, 38, 42
458–9 bureaucratic procedures 473
do-it-yourself activities 267, 271, 272 Burgess, R. 141
estimates of shadow economy 233 Burkina Faso 29, 32, 38, 43, 72
national average of shadow Burundi 29, 32, 43, 72
economy 225 business freedom index 14, 22–5, 63,
ranking according to size of shadow 67–71
economy 35, 37, 41
tax gaps 236, 237, 239 Cagan, P. 60
tax morale and compliance 384–5 Cambodia 44, 72
undeclared and under-declared work Cameroon 28, 31, 37, 42, 72
187 Canada 61, 72, 106, 155, 315–17, 458–9
Belize 33, 43, 72 Atlantic Home Building and
benchmark method 129, 139 Renovation Sector Council 294,
benefit morale 386–9, 393, 398 304, 314, 317
benefit principle of taxation 413 and comparison with residential
Benelux countries 133 construction sector in Australia
see also Belgium; Luxembourg; 314–17, 318
Netherlands labour market and productivity 108
Benin 30, 33, 39, 44, 72 MIMIC (multiple indicators
Besley, T. 141 multiple causes) model 133
beta or standardized regression Nova Scotia Department of Labour
coefficients 481 317
Bhattacharya, M. 473–4 Novia Scotia Worker’s
Bhattacharyya, D.K. 132, 134, 135, Compensation Board and
139 Building Inspectors 317
Bhutan 41, 72 Ontario Construction Secretariat
Bilsdorfer, P. 425 294
Bird, R.M. 475 ranking according to size of shadow
‘Black Activities’ Act 78–9, 410, 431 economy 35, 36, 40
Blackwell, C. 85 residential construction sector 294,
Blades, D. 61 295
Bolivia 72, 267 Statistics Canada 294
size of shadow economy tax morale and tax evasion 356, 358
(1999–2006/07) 27, 30, 33, 39, tax and social security contribution
44 burdens 87
Bollen, K.A. 129 Cape Verde 27, 29, 32, 38, 42, 72
Bosnia & Herzegovina 42, 72 CAPI 182


Index 505
Caragata, P.J. 133 Cohesion Fund 202

Caribbean 45–7 collective action (to tackle corruption)
cash demand approach 52, 134, 139 139
see also currency demand approach Cologne school of tax psychology
Cassar, A. 128 398–9
causality 462 Colombia 29, 32, 38, 42, 51, 72
causal variables 11, 16–17, 22, 70, 71 Comoros 43, 72
European Union 204–206, 212–18, confirmatory factor analysis 51–2
220–21 Congo, Democratic Republic of 30, 33,
India 134–7 43, 72
OECD (high-income) countries Congo, Republic of 30, 33, 39, 43,
101–102 72
Central African Republic 30, 33, 39, constitutional initiative 357–8
43, 72 Continental Europe 186–7, 188–9,
Central America 156 191–3, 195
Central Asia 21, 52, 69, 70, 266–7 contingent valuation method 82
size of shadow economy corporate tax to GDP ratio in India
(1999–2006/07) 30, 33–4, 45–8 138
Central and Eastern Europe 9, 10, 52, corruption, voice and accountability
69, 70, 156, 213 89, 469–92, 495–501
do-it-yourself activities 261, 266–7 2SLS regressions 483, 484–6, 487
econometric findings 21 aggregate governance indicators
estimates of shadow economy 228, 477–9
230 countries in panel analysis 498–500
geographical variations 188–9, cross-country analysis 497
191–3 descriptive statistics of panel
public sector services 14, 88 analysis 496
regional variations 180–81, 186–7, descriptive statistics for tax
195 performance analysis 501
size of shadow economies effect of corruption on shadow
(1999–2006/07) 30, 33–4 economy 472–4
tax morale and tax evasion 358 effect of voice and accountability on
Chad 29, 33, 38, 43, 72 shadow economy 470–72
Chattopadhyay, S. 202 empirical results on shadow
Chaudhuri, K. 202 economy 481–8, 495
Chi-square (p-value) 71 empirical results on tax performance
children (socio-economic 488–9, 490
characteristic) 386, 388 governance and institutional quality
Chile 28, 31, 37, 41, 51, 72 485, 490
China 52, 72, 267, 289 international country risk guide
size of shadow economy (ICRG) 479
(1999–2006/07) 27, 28, 31, 36, shadow economy 476–7
40 tax performance 474–6
Choi, J. 444, 469, 472 see also latent (hidden) variables
Chopra, O.P. 134 approach and corruption
city size 330 Costa Rica 28, 31, 37, 41, 72
closest birthday rule 182 Côte D’Ivoire 29, 33, 38, 43, 73
Clotfelder, Ch.T. 360 covariance matrix 18–20, 126–7, 129,
coefficient matrix 452 134
coefficients of variation 233–4 do-it-yourself activities 281–2


European Union 206–207 size of shadow economy

latent (hidden) variables approach (1999–2006/07) 35, 36, 40
and corruption 452–3, 457 tax gaps 236, 239
Coyle, D. 487 tax morale and compliance 384–5
criminal activity 178 tax morale and tax evasion 365
criminality 263–4 tax and social security contribution
Croatia 33, 34, 37, 42, 73, 384 burdens 87
crowding-out effect 411, 431, 473, 475 undeclared and under-declared work
Crowe, M.T. 347 187
currency: dependent variables 451
in circulation 283 descriptive statistics 67–8
definition 63 deterrence 78, 84, 85–6
demand approach 26, 52, 60–62, 71, see also Germany
82, 154 developed countries and do-it-yourself
European Union 205 activities 261–2
Germany 95, 97, 103–104, 414–16, developing countries 21, 22, 24, 52,
418–19, 434 473–4
Switzerland 154, 163 do-it-yourself activities 261–2,
tax morale and compliance 392 266–7, 268, 287–8
voice, accountability and size of shadow economy
corruption 476 (1999–2006/07) 27–30, 31–3
M0 over M1 15, 21–4, 63, 67–8, 69 development of shadow economy in
M0 over M2 69, 70 OECD (high-income) countries
to M3 ratio in India 137–8 102–105
cut-off distance 249–50 development of shadow economy in
Cyprus 28, 31, 41, 73, 187, 193 Switzerland 154–6, 160–65
Czech Republic 73, 186–7, 192–3, Diamond, J. 487
234–5, 267, 458–9 difference principle 350
national average of shadow direct democratic rights index 361
economy 225 direct effects:
size of shadow economy latent (hidden) variables approach
(1999–2006/07) 33, 34, 36, 40 and corruption 453, 457–8
tax gaps 236, 239 tax morale and tax evasion 359
tax morale and compliance 384–5 voice, accountability and corruption
483
de Soto, H. 474 direct household surveys in European
defining the shadow economy 11–12, Union 204–205
80–81, 152, 325–6, 378–80 direct measurement approach in
degrees of freedom 71, 219 Switzerland 152–3
Dell’Anno, R. 126, 129, 212, 214, 222 direct methods 179–80, 380–81, 414,
demographic characteristics 488 416–17
Denmark 12, 73, 106, 155, 193, 225 direct popular rights index 357
coefficients of variation 234 direct survey in European Union 194
deterrence 86, 414 direct tax 27, 67–8
do-it-yourself activities 269–70, 272, burden 108
273 European Union 212, 215
estimates of shadow economy 226 OECD (high-income) countries
labour market and productivity 101–102
109–10, 112 share of 13, 21–2, 65
Rockwool Foundation 153 to GDP ratio in India 138


Index 507
discrepancy method 104 regulation intensity 14

distance-based weight matrix 209 tax and social security contribution
DIY stores turnover 279 burdens 13
Djankov, S. 141 economy, state of 69
do-it-yourself activities 261–89 Ecuador 29, 32, 37, 42, 73
criteria of informal economic educational attainment 269, 329–30,
activities 264 386, 388
data 278–9 EFW 487
empirical model and hypotheses Egypt, Arab Rep. 73, 474
281–4 size of shadow economy
Germany 279–80 (1999–2006/07) 29, 32, 38, 42
indicators 276–7 El Salvador 30, 33, 39, 43, 73
labour market 275–6 electricity (physical input) method 30,
OECD (high-income) countries 134, 267, 303
114–15 European Union 205
regulations intensity 274–5 Germany 103, 104, 414, 417
size and distribution 268–73 Elffers, H. 380
size and trends of shadow economy employment rate in OECD
266–8 (high-income) countries
substitutes or complements 284–7 101–102
tax burden 273–4 endogeneity problem 213–14
Dominican Republic 28, 31, 37, 42, 44, endogenous variables 452, 455, 458,
73 467, 487
Dreher, A. 44–5, 443, 444–5, 446, 460, Engel, E.M.R.A. 360
473, 476, 483 Engerman, S. 487
Dutta, J. 471–2 Enste, D. 104, 127, 128, 151, 153, 157,
DYMIMIC (dynamic multiple 204, 206, 266, 414
indicators multiple causes) ‘envelope’ work see under-declared
approach 133 work
Germany 103, 416 Equatorial Guinea 31, 41
Switzerland 154, 162 equivalence principle 348–9
tax morale and compliance 392 Erard, B. 85
tax morale and tax evasion 366 Eritrea 73
voice, accountability and corruption estimation specifications for 162
476–7 countries 20–1
Estonia 73, 186–7, 192–3, 225, 234
Eastern Europe see Central and do-it-yourself activities 271, 272
Eastern Europe estimates of shadow economy 226
Eck, R. van 86, 90 size of shadow economy
econometric estimation in OECD (1999–2006/07) 34, 37, 41, 44
(high-income) countries 100–102 tax gaps 236, 237, 239
econometric methodology for 162 tax morale and compliance 384–5,
countries 17–20 399
econometric results for 162 countries Ethiopia 29, 32, 38, 42, 73
20–6 Etzioni, A. 363
economic causal variables 25 Europe 45–8, 392
economic freedom index 24, 25, 63, 69, see also Central and Eastern Europe;
70, 71 Continental Europe; European;
corruption, voice and accountability Southern Europe
483, 491 European Barometer survey 180


European Commission 179, 327 F-tests 483, 487

DG Employment and Social Affairs: fairness of tax system 351, 355–6
Employment Analysis Division Fakhruddin Ali Ahmed, President 142
182 Feige, E.L. 61, 104, 152, 154, 235, 263
European Social Survey (ESS) 382–3, Feinstein, J. 85
385–6, 389, 399, 405 Feld, L.P. 82, 84, 85–6, 87, 89–90,
European Union 177–96, 201–44, 93–5, 104, 167, 266, 360–62, 414,
248–57 419–20
anchoring technique based on Fiess, N. 51
sample moments 222–4 Fiji 28, 31, 37, 42, 44, 73
coefficients of variation of shadow fines 338, 409, 410–11, 420–28,
economy 234 434–5
former estimates of shadow Finland 73, 106, 155, 193, 458–9
economy quota 226 coefficients of variation 234
geographical variations in do-it-yourself activities 270, 271
undeclared work 183–7, 188–90, estimates of shadow economy 226,
194–5 233
geographical variations in under- national average of shadow
declared work 183–7, 190–93, economy 225
194–5 size of shadow economy
labour market and productivity 111, (1999–2006/07) 35, 36, 40
112 tax gaps 236, 239
MIMIC (multiple indicators tax morale and compliance 384–5
multiple causes) model tax morale and tax evasion 365
201–203, 206–208, 212–14, 242, undeclared and under-declared work
244 187
data 214–17 VAT rate 216
deriving estimates 221–2, 224 fiscal freedom index 13, 21–5, 63,
parameter estimates 217–21 67–71
regional estimation considerations fiscal referendum 357–8
204–205 fixed effects model 390, 395
spatial effects 211 former Soviet Union 14, 44, 88, 156,
national averages of shadow 261
economy 225 Fortin, B. 108
NUTS 2 regions 227–32, 237–8, fractionalization 392, 483
240–41, 242, 251–7 France 73, 106, 155, 192–3, 212
previous research 179–81 coefficients of variation 234
regional estimation considerations deterrence 413
204–205 do-it-yourself activities 271, 272
spatial effects 208–12 Federal Finance Court (BFH) 421,
structural equations model (SEM) 425
249–50 labour market and productivity 110,
survey method 83 112
tax gaps 235–41 national average of shadow
weight matrix 248–9 economy 225
Eurostat regional database 214 size and development of shadow
exit option 474 economy 156
exogenous variables 452, 455, size of shadow economy
467–8 (1999–2006/07) 35, 36, 40
exploratory factor analysis 51–2 tax gaps 236, 239


Index 509
tax morale and compliance 384–5, Germany 73, 155, 202, 324–43
399 Allensbach Institute 326
tax morale and tax evasion 351, 365 average hourly wages by form of
undeclared and under-declared work payment 96
187 ‘Black Activities’ Act 78–9, 410, 431
Fréchette, P. 108 Bundesgerichtshof 425
Frey, B.S. 61, 76, 84, 89–90, 104, 133, code of criminal procedure (StPO)
151, 154, 162–3, 206, 354, 360–62, 421
410–11, 414, 431 coefficients of variation 234
Friedman, M. 13, 88, 150, 274–5, 444, Cologne Institute for Economic
455, 472–3, 483 Research survey 326, 328, 336,
339, 340, 341, 342
Gabon 33, 44, 73 comparison using survey and
Gaertner, W. 363 MIMIC method 96
Galbraith, J.K. 365 Constitution 352
Gambia, The 33, 43, 73 Constitutional Court 352
Gassebner, A. 443 cumulative distribution of beliefs in
GDP 476 social environment 333
annual growth rate in OECD definitions 325–6
(high-income) countries demand for illicit work 336, 337–8
101–102 descriptive statistics 329–30
deflator 69 deterrence policy 85–6, 409–37
European Union 35, 213, 223–4, audit capabilities 432
234, 236–8, 240, 243 ‘Black Activities’ Act amendments
geographical variations in 427
under-declared work 192–3 Criminal Tax Code 421–2, 424–5
Germany 279 detection by tax auditing 427, 431
India 139 federal financial investigation unit
official and do-it-yourself activities 434
266–7 involved authorities, investigation
real: and criminal proceedings
Germany 414 422–4, 427
growth rate 138, 279 penalties 429
total country 46–7 prison sentences 430
see also GDP per capita sentences and fines in criminal
GDP per capita 15–17, 21–5, 67–8, 69, proceedings 428
70, 71 sentencing practice developments
definition 63–4 425–7
European Union 202, 215, 216–17, tax compliance and tax morale
219, 220, 244 412–13
growth rate 67–8, 69, 70, 71 tax evasion and tax morale 413–20
labour market and productivity 112 tax investigations and
OECD (high-income) countries administrative assistance 433
101–102 time-series analysis 431, 434–5
voice, accountability and corruption undeclared work, tax evasion and
481, 486, 488–9 tax morale 412
gender 329, 386, 388 do-it-yourself activities 114–15, 269,
Generalized Least Squares (GLS) 271–2, 274, 278–80, 288–9
207–208, 217, 224, 226, 227, 250 East 187, 233, 278, 280, 420
Georgia 34, 39, 44, 73, 267 estimates of shadow economy 226


Federal Court of Advisors 340 survey findings 328, 330

Federal Court of Auditors 339 survey method 83, 327–8
Federal Employment Office 338–9 tax deductibility 340–41
Federal Statistical Office 278, 279, tax gap 236, 237, 239
284 tax morale and compliance 383–5
Finance Courts 425 tax morale and tax evasion 365,
financial control unit for undeclared 366–7
work (FKS) 431 tax and social security contribution
GDP 153, 414, 419 burdens 12, 87
general criminal code (StGB) 421 TNS-Emnid Institute 152–3, 326,
general fiscal code (AO) 421, 424 328
geographical variations in undeclared work 73, 78–9, 91–7, 98,
under-declared work 193 105, 106, 108, 116
GNP 414 underground economy 325–6
hourly wage rates of activities (2004) West 187, 233, 279, 420
93 work in shadow economy (2007) 92
illicit work and shadow economy Ghana 29, 32, 38, 43, 73
327 Ghose, S. 134
illicit work, supply of 333–5 Ghosh, D. 133
illicit work, welfare loss caused by Giles, D.E. 128, 133, 157, 206, 451
336–8 Gneezy, U. 409
indirect measurement approach 154 Goldberger, A.S. 133, 206
influencing factors supply and Gómez-Antonio, M. 212, 214
demand 331 ‘good’ equilibrium 14
labour market and productivity 110, Goodness of Fit Index 127, 457
112 Google Scholar 398, 407
latent (hidden) variables approach Götte, L. 361
and corruption 459 governance:
measures to prevent illicit work indicators, aggregate 477–9
338–40 quality 483
minimum wage effects 341–2 voice, accountability and corruption
moonlighting 166–7 485, 487, 489, 490
national average of shadow government:
economy 225 effectiveness 65, 67
participation in black activities regulation 60–61, 69
(2001–06) 94 size 21–2, 25, 27, 64–5, 67–71, 114
proportion of activities within last Granger causality 85, 411, 434–5,
12 months by gender, age and 436
occupation 95 ‘grease the wheels’ hypothesis 443
reasons for shadow economy Great Britain 106, 187, 269
activities (2007) 93 deterrence 86, 413
Regional Tax Offices (OFD) 421–2 tax morale and compliance 384–5
regulation intensity 13, 87 Greece 73, 106, 155, 212, 458–9, 461
Regulatory Offences Act (OWiG) coefficients of variation 234
421 do-it-yourself activities 267, 268
self-sufficient economy 325–6 estimates of shadow economy 233
shadow economy 325–6 former estimates of shadow
size and development of shadow economy 226
economy 35, 36, 40, 103, 156, geographical variations in
160–61 under-declared work 193


Index 511
national average of shadow tax morale and compliance 384–5,

economy 225 399
size of shadow economy Huntington, S. 461
(1999–2006/07) 35, 37, 41
tax gaps 236, 239 Iceland 35, 36, 40, 73, 384–5
tax morale and compliance 384–5 illegal production 379
undeclared and under-declared work illegality 263–4
187 income 269, 329–30
voice, accountability and corruption gap approach 414, 415, 416–18
473, 475 household 386, 388
Guatemala 30, 33, 39, 44, 73 personal 158
Guinea 29, 32, 38, 43, 45, 73 real disposable per capita 278
Guinea-Bissau 32, 43, 73 tax, progressive 350
Gupta, P. 134 independent variables 451
Gutmann, P.M. 60 India 73, 131–46, 149, 202
Guyana 31, 42, 73 coalition governments 142–4
elections 142–3, 146
Haiti 33, 44, 73, 474 empirical results 142–5
Halla, M. 391 Gross National Product 132
Hammar, H. 351 media/press freedom 139, 141, 142,
Harding, P. 263 144, 145–6
Harmonized European Time Use MIMIC (multiple indicators
Survey 270, 272 multiple causes) model
Harsanyi, J.C. 350 133–9
Hayduk, L.A. 130 National Congress Party 144
Helberger, C. 129 political and institutional actors,
Heritage Foundation 25, 274, 275, role of 139–42
483 size of shadow economy
see also Economic Freedom (1999–2006/07) 28, 31, 37, 41
Index tax morale and tax evasion 358
Herwartz, H. 201, 235 indicator approach see currency
Hessing, D.J. 380 demand approach
HETUS 273 indicator variables 16–17, 23, 69, 70,
Hill, R. 61, 76 71, 455
Hindriks, J. 444 do-it-yourself activities 276–7
Hines, J.R. 360 econometric findings 22–4
Höffe, O. 364 European Union 204–205, 210–11,
Honduras 30, 33, 39, 44, 73 213–17, 219–21
Hong Kong 36, 40, 73 India 134–7
hours worked per week, average OECD (high-income) countries
101–102, 277, 279 101–102
human capital 268 indirect household surveys in
Hungary 73, 187, 192–3, 458–9 European Union 204–205
coefficients of variation 234–5 indirect impacts and tax morale and
do-it-yourself activities 267 tax evasion 359
national average of shadow indirect measurement approach 82,
economy 225 380–81, 453, 457–8
size of shadow economy European Union 179
(1999–2006/07) 33, 34, 37, 41 Germany 414, 417
tax gaps 236, 239 Switzerland 152, 154


indirect measures: Ireland 73, 106, 155, 187, 193

Germany 415 coefficients of variation 234
indirect tax 25 estimates of shadow economy 226
burden 108 national average of shadow
European Union 212, 215 economy 225
OECD (high-income) countries size of shadow economy
101–102 (1999–2006/07) 35, 36, 40
to GDP ratio in India 137–8 tax gaps 236, 237, 239
Indonesia 28, 31, 36, 40, 73 tax morale and compliance 384–5
inflation: Isachsen, A.J. 61
India 138 Israel 28, 31, 37, 41, 73
rate definition 65, 69, 70, 71 Italy 73, 105, 106, 155, 459, 461
informal sector production 379 coefficients of variation 234
informality: deterrence 413
average 46, 47 do-it-yourself activities 270, 271, 272
measurement 48 estimates of shadow economy 233
world view 49 former estimates of shadow
information (to tackle corruption) 139 economy 226
input variables 353 geographical variations in
inspector–tax payer model 444 under-declared work 192–3
institutional actors, role of in India labour market and productivity 111,
139–42 112
institutional conditions and corruption national average of shadow
489 economy 225
institutional instability 473 size of shadow economy
institutional quality 445–6, 460, 483, (1999–2006/07) 35, 37, 41
485, 487 tax gaps 236, 237, 239
corruption in absence of shadow tax morale and compliance 384, 392
economy 447–8 tax morale and tax evasion 351, 354,
corruption in presence of shadow 365
economy 449–51 undeclared and under-declared work
direct and indirect effects 458 187
OECD (high-income) countries
101–102 Jagers, S.C. 351
structural equation model 454 Jamaica 27, 29, 32, 38, 42, 73
tax performance 476, 490 Japan 73, 106, 133, 155, 461
instrumental variable approach 390 size of shadow economy
intercellular weighting 182 (1999–2006/07) 35, 36, 40
international country risk guide tax morale and tax evasion 358
(ICRG) 477–8, 479, 480, 481, 488 Jenkins, R. 263
International Labour Organisation Johnson, S. 13, 14, 30, 87, 88, 157, 267,
(ILO) 188 273, 274, 276, 444, 475
International Social Science Panel Jordan 28, 31, 36, 40, 74
(ISSP) 352–3, 355–6, 382–3, Jöreskog, K.G. 126, 133, 206, 353, 453,
385–6, 389, 398, 405–406 458
intrinsic motivation to pay taxes see just exchange 349–50
tax morale ‘just price’ 348
Investiture Controversy 351
Iran, Islamic Republic 28, 31, 36, 40, Kabir, M. 61, 76
73 Kaldor, N. 134


Index 513
Kaliberda, A. 30, 267 latent (hidden) variables approach 133,

Kanniainen, V. 84 282–4
Karmann, A. 114–15, 274, 278–9 European Union 205, 206–207, 210,
Karymshakov, K. 473 222
Kaufmann, D. 13, 14, 30, 87, 88, 139, Germany 104, 414, 415, 416, 418
157, 267, 273, 274, 276, 444, 475, see also latent (hidden) variables
480, 488 approach and corruption
Kazakhstan 34, 38, 43, 74 latent (hidden) variables approach and
Kazemier, B. 86, 90 corruption 443–63, 467–8
Keen, M. 444 corruption in absence of shadow
Kennedy, J.F. 363 economy 447–8
Kenya 29, 32, 37, 42, 74 corruption in presence of shadow
Kirchgässner, G. 76, 87, 104, 274 economy 448–51
Kirchler, E. 90 direct and indirect effects 457–8
Klovland, J.T. 61, 87, 274 endogenous latent variable
Knepel, H. 129 indicators 467
knowledge (to tackle corruption) 139 exogenous latent variable indicators
Komogorov–Smirnov test 45 467–8
Korea, Republic of 35, 37, 41, 74, latent scores (indices) 458–60
268 structural equations model (SEM)
Körner, M. 90 451–5
Kotsogiannis, C. 444–5, 446, 460 structural equations model (SEM)
Kraay, A. 475, 480, 488 estimates 455–7
Krueger, A.B. 390–1 structural equations model (SEM)
Kucera, D. 109 evaluation 457
Kucher, M. 361 Latin America 14, 30, 45–7, 444
Kuwait 28, 31, 36, 41, 74 do-it-yourself activities 267
Kyrgyz Republic 34, 38, 43, 74, 473 geographical variations 188
public sector services 88
La Porta, R. 393, 400, 483 tax morale and compliance 392
labour force participation 16, 24, 63, Latinobarometro 382, 392
67–8, 70–71, 486 latitude (institutional quality and tax
European Union 213, 215, 216–17, morale) 392, 483
219, 220, 244 Latvia 74, 181, 186–7, 192–3
Germany 414 coefficients of variation 234
labour force, total 66 do-it-yourself activities 271
growth rate 16, 69, 70, 71 former estimates of shadow
labour market: economy 226
do-it-yourself activities 275–6 national average of shadow
indicators 16 economy 225
and productivity in OECD (high- size of shadow economy
income) countries 108–12 (1999–2006/07) 34, 37, 41
regulations 158 tax gaps 236, 237, 239
status 386 tax morale and compliance 384
Lackó, M. 30, 267, 414 leadership (to tackle corruption) 139
Lamnek, S. 419 Lebanon 28, 32, 37, 42, 74
Landes, D. 487 Leff, N.H. 443
Lao PDR 28, 31, 37, 41, 74 legal origin 392–3, 483
Larsen, C. 82, 85–6, 87, 90, 93–5, 104, legal system 392
167, 414, 419 legislative initiative 357–8


legislative referendum 357–8 marital status 269, 386, 388

Lemieux, T. 108 marriage penalty 158
LeSage, J. 208 Martinez-Vazquez, J. 475
Lesotho 28, 31, 37, 41, 74 Mastruzzi, M. 475, 480, 488
Leuthard, D. 150 Mauritania 29, 32, 45, 74
Levi, M. 471 Mauritius 28, 31, 37, 41, 74
Levin, M. 472, 473, 475 Maximum Likelihood Estimation
Liberia 43, 74 (MLE) procedure 127, 128, 134,
Libyan Arab Jamahiria 42, 74 284
Liechtenstein: Zumwinkel Affair 78 Maximum Likelihood (ML) 20,
LISREL (linear independent structural 207–208, 217
relationships) model 51, 126–7, May, J.W. 158
133, 353, 416 measured variables (indicators) 451,
Lithuania 74, 181, 187, 192–3 452
coefficients of variation 234 measurement equations in European
do-it-yourself activities 271 Union 206
former estimates of shadow measurement model 18, 127, 281–2,
economy 226 354
national average of shadow measurement of shadow economy:
economy 225 OECD (high-income) countries 81–3
size of shadow economy Switzerland 152–4
(1999–2006/07) 34, 37, 42 Mehta, R. 134
tax gaps 236, 239 Méon, P.-G. 112, 443
tax morale and compliance 384 Mexico 51, 74, 268, 458–9
living situation 329–30 size of shadow economy
local currency per capita, change of in (1999–2006/07) 28, 31, 35, 37,
OECD (high-income) countries 41
101–102 Middle East 46–7
Long, J.S. 129 MIMIC (multiple indicators multiple
Luxembourg 74, 187, 192–3, 384–5 causes) model 10, 51, 52
size of shadow economy advantages 127–8
(1999–2006/07) 35, 36, 40, 44 application 127
Austria 98
Macao, China 40, 74 deterrence 85
McCorriston, S. 444–5, 446, 460 developing countries 27
Macedonia 34, 42, 74 disadvantages 128–30
McGee, R.W. 347, 355–6 do-it-yourself activities 114
Macintyre, A. 392–3 econometric findings 21, 24
McMillen, D.P. 248 econometric methodology 17–20
Madagascar 29, 32, 38, 43, 74 estimation results 23, 70, 71
Malawi 29, 32, 38, 43, 74 Germany 94–5, 97, 104, 434
Malaysia 28, 31, 37, 41, 44, 74 India 131, 133–9
Maldives 41, 74 indicator versus cause variables 16
Mali 29, 32, 38, 43, 74 OECD (high-income) countries 100,
Maloney, W.F. 51 101, 105
Malta 74, 187, 192–3, 384 residential construction sector 303
size of shadow economy size of shadow economy
(1999–2006/07) 28, 31, 37, 41 (1999–2006/07) 26, 39, 44, 48
marginal weighting 182 tax morale 90
marginalist principle 349 tax morale and compliance 394


Index 515
tax morale and tax evasion 366 New Zealand 74, 107, 133, 155, 458–9
undeclared work in OECD (high- size of shadow economy
income) countries 82, 115, (1999–2006/07) 35, 36, 40
126–30 Nicaragua 30, 33, 38, 43, 74
weight matrix 294 Niger 29, 32, 45, 74
see also DYMIMIC (dynamic Nigeria 33, 45, 74
multiple indicators multiple Nordblom, K. 351
causes) approach; European Nordic countries 186, 187, 195–6
Union geographical variations 188–9,
minimum wage effects in Germany 191–3
341–2 survey method 83
Moldova 34, 45, 52, 74 tax morale and tax evasion 351
monetary approach see transaction see also Denmark; Finland; Iceland;
approach Norway; Sweden
monetary indicators 15, 277 Normed Fit Index 457
Mongolia 28, 31, 36, 40, 74 North Africa 46–7, 188
Montenegro, C. 261, 266–8, 394 Norway 12, 61, 74, 107, 155
moonlighting, public perception of in deterrence 86
Switzerland 165–9 do-it-yourself activities 269, 271,
Morocco 29, 32, 38, 42, 74 272, 273–4, 289
Mozambique 29, 32, 45, 74 latent (hidden) variables approach
Mróz, B. 202 and corruption 459
multiple indicators 179 size of shadow economy
see also MIMIC (multiple indicators (1999–2006/07) 35, 36, 40
multiple causes) model tax and social security contribution
Mummert, A. 202, 419 burdens 87
Murray, M.P. 394 NUTS (Nomenclature of Territorial
Muthoo, A. 444 Units for Statistics) 244
Myanmar 30, 33, 45, 74 NUTS 1 regions 215
NUTS 2 regions 202, 205, 237–8,
Naho, A. 132 242, 244, 251–7
Namibia 28, 31, 37, 41, 74 MIMIC (multiple indicators
Neck, R. 157 multiple causes) model
Nepal 27, 29, 32, 38, 42, 74 estimation 214–15
Netherlands 35, 36, 40, 74, 155 shadow economy estimates
coefficients of variation 234 227–32, 235
deterrence 86 spatial effects adjustments 248
estimates of shadow economy 232–3 tax gaps 237–8, 240–41
geographical variations in under- NUTS 3 regions 248
declared work 193
national average of shadow observations, number of 71
economy 225 observed indicators 452–3
survey method 83 observed variables 207, 282
tax gaps 236, 239, 243 occupation 269, 329–30
tax morale and compliance 380, OECD (high-income) countries 9–10,
384–5 50–52, 69–70, 78–118, 126–30
tax morale and tax evasion 354–5, aggregate efficiency 112–13
365 Austria 97–100
undeclared and under-declared work currency demand approach 61
187 defining shadow economy 80–81


deterrence 85–6, 409, 410 Panama 75, 267

development and size of shadow size of shadow economy
economy 102–105 (1999–2006/07) 27, 30, 33, 45
do-it-yourself activities 114–15, panel data 485
266–8, 275, 278 panel model 480
econometric estimation 100–102 Papua New Guinea 27, 29, 32, 38, 42,
econometric findings 21, 24, 25 75
indicator versus cause variables 17 Paraguay 29, 32, 38, 43, 75
labour market and productivity Pareto principle 349
108–12 Pedersen, S. 85–6, 419
latent (hidden) variables approach penalties 429, 434–5
and corruption 444–7, 455, 458, penalty clauses 170
460, 461, 462 penalty orders 423–4
measuring shadow economy 81–3 Peru 75, 267, 474
MIMIC (multiple indicators size of shadow economy
multiple causes) model 71, (1999–2006/07) 27, 30, 33, 39,
126–30, 133 44
public institutions 89 Petersen, H.-G. 154
public sector services 14, 88 Philippines 29, 32, 38, 43, 75, 474
regional variations 177, 179 physical indicator approach in India
regulations intensity 87–8 139
size and development of shadow physical input methods see electricity
economy 105–108, 156 (physical input) method
size of shadow economy Pickhardt, M. 418–19
(1999–2006/07) 34–5, 45–8 Pinney, N. 359
size of shadow economy as Poland 75, 187, 192–3, 202, 458–9
percentage of official GDP 155 coefficients of variation 234
tax morale 89–90 do-it-yourself activities 271
tax morale and compliance 378–9, estimates of shadow economy 226,
394 229, 232
tax and social security contribution national average of shadow
burdens 86–7 economy 225
unemployment 113–14 size of shadow economy
see also Germany (1999–2006/07) 34, 37, 41
official economy 15, 16 tax gaps 236, 237, 239
Olbrich, G. 419 tax morale and compliance 384–5
Oman 28, 31, 36, 40, 74 policy causal variables 25
openness variable 65, 69, 70, 71 political actors, role of in India 139–42
Opinion Taxpayer Survey 358 Pommerehne, W.W. 61, 76, 151, 154,
optimal taxation theory 350 162–3, 348, 360–61, 414
ordinary least squares (OLS) 390, 436 population size 65–6, 486
out of labour force 388 Portugal 75, 107, 155, 212, 459
output variables 353 coefficients of variation 234
owner-occupier 269 geographical variations in under-
declared work 193
‘p’ indicators 133–4 national average of shadow
Pääkönen, J. 84 economy 225
Pace, R.K. 208 size of shadow economy
Pacific 46–7 (1999–2006/07) 35, 37, 41
Pakistan 29, 32, 38, 42, 75 tax gaps 236, 239


Index 517
tax morale and compliance 383–5, Rei, D. 473–4

399 religious tradition 351, 356, 366–7,
undeclared and under-declared work 392, 483
187 Renooy, P.H. 243
poverty trap 157 RESET approach 303
prison sentences 425, 427, 428, 430, residence, size of 388
434–5 residential construction sector in
production of households for own final Australia 293–319, 321–3
use 379 characteristics of respondents
production missed due to deficiencies 304–306
in data collection programme 379 comparison with Canada 314–17,
profession 330 318
proportionality factor 223 employment 300–303
proxy or surrogate indicators 179 factors affecting size of shadow
public control 212 economy 321
public employment: perceptions on behaviour of those
European Union 214, 218 employed 308–10
India 138 perceptions on characteristics and
public goods, quality of 158 consequences of shadow
public revenue data 278 economy 306–308
public sector 14–15 perceptions on size of shadow
European Union 215 economy 310–14
OECD (high-income) countries 88 perceptions on strategies to reduce
public services 212, 278 shadow economy 314
punishment in Germany 420–27 size and mix of building
purchasing power parity (definition) construction activity 296–300
63–4 suggestions for reducing shadow
Pyle, W. 30, 158, 267 economy 322–3
target population and methodology
Q–Q plot 45–6 304
Qatar 40, 75 reversed causality 391
Quality of Governance Index 477 Rockwool Foundation Research Unit
quantitative modelling 208 269, 272
quantity equation 154 Rogoff, K. 76
quasi-voluntary compliance 470 Roll, R.W. 487
questionnaire-based surveys 269, 288 Romania 75, 181, 186–7, 192–3, 383–4
size of shadow economy
random route procedures 182 (1999–2006/07) 33, 34, 37, 42,
Rawls, J. 349–50 44
reciprocity 359 Roncolato, L. 109
regression equation 60 Root Mean Square Error of
regulation 24, 70, 287–8 Approximation (RMSEA) 71,
intensity 13–14, 21–2, 274–5 457
OECD (high-income) countries Rose-Ackerman, S. 444
87–8 rule of law 446, 455, 457, 473
regulatory burden 25, 70, 473–4 running water principle 132
Germany 416 Russia 181, 472, 473, 475
regulatory discretion 446 see also former Soviet Union
regulatory framework 157, 170 Russian Federation 34, 38, 43, 75, 267,
regulatory quality index 65, 69, 70, 71 384


Rusticchini, A. 409 average informality weighted by

Rwanda 29, 32, 45, 75 total GDP in 2005 47
average shadow economy measure
Sachs, J. 487 by region 47
Sandmo, A. 83–4, 363, 409 developing countries 27–30, 31–3
Sardà Pons, J. 418–19 Eastern European and Central
Sargan’s test 485, 487 Asian (transition) countries 30,
Satarov, G. 472, 473, 475 33–4
Saudi Arabia 28, 31, 36, 40, 75 OECD (high-income) countries 34–5
scaling procedures 204–205 Q–Q plot of informality measure 46
Scandinavian countries 61, 105, 133, ranking of 120 countries 36–9
156 ranking of 151 countries 40–44
see also Denmark; Norway; Sweden world view of informality 49
Schäfer, W.J. 419 size of shadow economy 266–8
Schaffner, M. 392–3 OECD (high-income) countries
Schaltegger, C.A. 162–3, 487 102–105
Schmid, H. 151 Switzerland 154–6, 160–65
Schmidt, A. 85 Slemrod, J. 360, 381
Schmölders, G. 412–13 Slovak Republic 75, 187, 192–3, 384–5,
Schneider, F. 9, 12, 15, 26, 50–53, 76, 458–9
83–5, 87, 90–91, 93, 96–7, coefficients of variation 234
103–105, 110–14, 116–17, 127–8, do-it-yourself activities 267
130, 132–3, 139, 146, 150–64, 166, national average of shadow
167, 177–9, 202, 204–206, 221–2, economy 225
224–6, 235, 244, 261, 265–8, 273, size of shadow economy
275, 278–80, 288–9, 319, 324, (1999–2006/07) 33, 34, 36, 40
325–7, 342, 355, 362, 366–7, 377, tax gaps 236, 239
386, 388, 391–4, 396, 398–400, Slovenia 34, 37, 41, 75, 186–7
411, 412–16, 418–20, 434, 444, coefficients of variation 234
446, 461–3, 469, 471–4, 476–7, do-it-yourself activities 270, 271, 272
480–81, 483, 491, 495–6 former estimates of shadow
Scholz, J.T. 359 economy 226
Sedlenieks, K. 181 geographical variations in
self-employment 51, 388 under-declared work 192–3
European Union 213, 215, 218, 220 national average of shadow
Sena, V. 471–2 economy 225
Senegal 29, 33, 38, 43, 75 tax gaps 236, 239
Shankar, R. 51 tax morale and compliance 384–5
Shleifer, A. 13, 30, 87–8, 267, 274, 475 Smith, A. 363
Sierra Leone 29, 33, 38, 43, 75 Smith, J.D. 154
simultaneity 390, 391 Smith, P. 11, 80, 152
Singapore 75, 267 snowball effect 158
size of shadow economy social relations 188–9
(1999–2006/07) 27, 28, 31, 36, social security contribution burdens:
40 do-it-yourself activities 278
size of shadow economies for 162 OECD (high-income) countries 86–7
countries from 1999–2006/7 social security contributions 12–13,
26–48 101–102, 157
average informality by World social security system 288
Bank’s regions 46 ‘soft modelling’ 51, 103, 415


Index 519
Sokoloff, K. 487 structural equations model (SEM)

Solomon Islands 42, 75 17–18, 20, 26, 51–2
Solomon, O.H. 126 do-it-yourself activities 278, 280–82,
Sommers, P.M. 158 283, 284–7, 288
Sörbom, D. 126 European Union 206, 210, 224,
South Africa 28, 31, 37, 41, 75 249–50
South America 156 India 134
South Asia 46–7, 109 latent (hidden) variables approach
Southern Europe 156, 186–7, 195–6 and corruption 445, 451–7, 462
estimates of shadow economy 228, OECD (high-income) countries
230 126–7, 129
geographical variations 188–9, Switzerland 154
191–3 see also MIMIC (multiple indicators
undeclared and under-declared work multiple causes) model
187, 191–3 Structural Funds 202
Spain 75, 107, 155, 193, 212 structural model and tax morale and
coefficients of variation 234 tax evasion 354
deterrence 413 Stutzer, A. 357
do-it-yourself activities 270, 271, 272 Sub-Saharan Africa 45–8, 50, 109, 188
labour market and productivity 111, substitution effect 113–14
112 Sudan 29, 32, 45, 75
national average of shadow Suriname 32, 42, 75
economy 225 survey method 82–3
size of shadow economy Austria 98
(1999–2006/07) 35, 37, 41 Germany 94–5, 97, 103, 104–105,
tax gaps 236, 237, 239 415, 416–17, 419
tax morale and compliance 384–5 Switzerland 152–3
tax morale and tax evasion 365 Swaziland 29, 32, 45, 75
undeclared and under-declared work Sweden 12, 75, 105, 107, 155
187 coefficients of variation 234–5
spatial autocorrelation 208, 210 deterrence 86
spatial autoregression 209, 210–11 do-it-yourself activities 269, 271,
spatial effects 208–12, 221 272, 273–4, 289
spatial error model 209 estimates of shadow economy 226
spatial heterogeneity 208 geographical variations in under-
spatial lag model 208–209 declared work 193
spatial weight matrix 209 labour market and productivity 111,
Special Eurobarometer Survey No. 284 112
182 national average of shadow
specifications, additional 69–70 economy 225
Spiro, P.S. 87 size of shadow economy
Sri Lanka 29, 33, 38, 43, 75 (1999–2006/07) 35, 36, 40
Stapenhurst, R. 139, 141 Statistics Sweden 270
state regulation burden in OECD tax gaps 236, 237, 239
(high-income) countries 101–102 tax morale and compliance 384–5
statistical models 82 tax morale and tax evasion 365
Statistical Package for Social Scientists tax and social security contribution
182 burdens 87
statistical tests 23, 71 undeclared and under-declared work
Strom, S. 61 187


Switzerland 75, 78–9, 105, 107–108, do-it-yourself activities 278

150–71, 174 European Union 214, 216, 218
causes of shadow economy 156–9 evasion 83–4, 263–4, 377, 391
definition of shadow economy 152 Germany 409, 411, 412, 413–20,
deterrence 410, 414, 418 421, 424–5
do-it-yourself activities 267 penalties 429
Federal law, effect of 162 sentences and fines 428
GDP 151 fairness of tax system 351, 355–6
GFS Bern 165–6, 167–8 gaps 170, 203, 235–41, 243
latent (hidden) variables approach Germany 416
and corruption 459, 461 handles 489
measurement of shadow economy immorality 76
152–4 investigations and administrative
MIMIC (multiple indicators assistance in Germany 433
multiple causes) model 133 mentality 412–13
moonlighting, public perception of misdemeanour 421
165–9 OECD (high-income) countries 86–7
public campaign activity 174 performance 474–6
recent developments 159 empirical results 488–9, 490
Science Foundation 151 shortening 422
Secretariat of State for Economic system complexity 60, 446
Affairs 159 value added (VAT) 218, 220
size and development of shadow variable 76
economy 154–6, 160–65 wedge 218, 220
size of shadow economy willingness to pay 352
(1999–2006/07) 35, 36, 39, 40, see also direct tax; indirect tax; tax
44 morale
tax morale and tax evasion 350, 353, tax morale 84, 86, 158, 347–58
356, 360–62, 364, 365, 367 determinants 355–8
Syrian Arab Republic 28, 31, 36, 41, 75 European Union 212–13
Germany 92, 412–20
Tafenau, E. 235 measurement 352–3
Taiwan 41, 75 OECD (high-income) countries
Tajikistan 34, 38, 43, 75 89–90, 101–102
Tanzania 30, 33, 39, 44, 75 role of in shadow economy 353–5
Tanzi, V. 60, 61, 76, 112, 113, 132, 151, and tax evasion 358–62
154, 157, 263, 469 theoretical considerations 348–52
tax: voice, accountability and corruption
auditing 381, 416, 418, 427, 431 483
avoidance 380 see also tax morale and compliance
benefit principle 413 behaviour
burden 12–13, 25, 27, 70, 71, 157 tax morale and compliance behaviour
direct 60–61 375–400, 404–408
do-it-yourself activities 273–4, 287 benefit morale measurement 386–9
indirect 60–61 defining shadow economy 378–80
total 69 Google Scholar 407
variables 24 identification problem 389–91
compliance 410, 412–13 and shadow economy 392–4
crimes 421 survey data 404–406
deductibility in Germany 340–41 tax evasion 391


Index 521
tax evasion measurement 380–82 Two-Stage Least Squares (2SLS)

tax morale measurement 382, 385–6, estimation 392
387 Tyler, T.R. 471
unobserved heterogeneity,
importance of 394–7 Uganda 29, 32, 38, 43, 75
Taylor, M.P. 51 Ukraine 34, 44, 75, 181
Tedds, L.M. 128, 133 do-it-yourself activities 267
test-statistics 102 tax morale and compliance 384–5,
Thailand 30, 33, 39, 44, 75 399
Theilen, B. 201 undeclared work 178–9, 180–81, 263–4
theoretical considerations 11–17 deterrence 437
defining the shadow economy 11–12 geographical variations in European
indicator and cause variables 16–17 Union 183–7, 188–90, 194–5
labour market indicators 16 Germany 409, 411, 412, 416
monetary indicators 15 under-declared work 178–9, 181, 183
official economy 15, 16 geographical variations in European
public sector services 14–15 Union 183–7, 190–93, 194–5
regulations intensity 13–14 underground production 378, 379, 394,
tax and social security contribution 396
burdens 12–13 unemployment rate 15, 21–5, 67–71,
Thilo, M. van 353 388
Thomas, J.J. 61, 128 definition 64
Thum, M. 444, 469, 472 estimated 64
TI index 460 European Union 213, 215, 218, 220
time-series analysis in Germany 431, OECD (high-income) countries
434–5 101–102, 113–14
Togo 29, 32, 45, 75 United Arab Emirates 28, 31, 37, 41,
Torgler, B. 90, 162–3, 356–8, 361, 391, 76
392–3, 419–20, 475, 480, 487 United Kingdom 35, 36, 40, 76, 192–3,
total effects 453 458–9
transaction (monetary) approach: coefficients of variation 234
Germany 414, 415, 417 do-it-yourself activities 270, 271, 272
OECD (high-income) countries 103, England 155
104 estimates of shadow economy 226,
residential construction sector 293, 233
303 indirect measurement approach
Switzerland 154 154
transition countries 17, 24, 25, 71, 158 national average of shadow
do-it-yourself activities 261, 268, 287 economy 225
public sector services 88 residential construction sector 294
Transparency International survey method 83
Corruption Perception Index 455, tax gaps 236, 237, 239
481 see also Great Britain; Ireland
transparency, lack of 473 United States 76, 105, 107, 108, 151,
trends of shadow economy 266–8 155
tri-cube function 248–9 currency demand approach 60, 61
Trinidad and Tobago 29, 32, 38, 42, 75 deterrence 410, 418, 427
trust 351, 352, 358 do-it-yourself activities 267
Tunisia 29, 32, 38, 42, 75 Internal Revenue Service 83, 416,
Turkey 34, 37, 41, 75, 384–5, 391 418


Taxpayer Compliance Wagener, H.J. 472

Measurement Program 83 weather conditions (cloudiness
latent (hidden) variables approach measure) 392–3, 486–7
and corruption 461 Weck, H. 51, 353–4
MIMIC (multiple indicators Weck-Hannemann, H. 76, 104, 133,
multiple causes) model 133 154, 162–3, 206, 348, 354, 360–61,
National Research Program 416 414
size of shadow economy Weigel, R.H. 380
(1999–2006/07) 35, 36, 39, 40, weight matrix 248–9
44 weighted averages 224
tax morale and compliance 391, 392 weighted least squares (WLS) 127
tax morale and tax evasion 358, 360, weighting procedure 182
364 Weill, L. 112, 443
Taxpayer Compliance Measurement welfare loss in Germany 336, 338
Program (TCMP) 381, 416 Wenzel, M. 359
unobserved heterogeneity, importance willingness to pay taxes 352
of 39, 394–7 Woolfson, C. 181
unregulated or unlicensed enterprises working hours:
263–4 average per week 101–102, 277, 279
Unweighted Least Squares (ULS) in Germany 416
207–208, 217, 224, 226, 244, 250 World Bank 46, 47, 69, 275
Urban, I. 235 Worldwide Governance Indicators
urban-rural variations 189–90 15
Uruguay 30, 33, 39, 44, 76 World Value Survey (WVS):
utilitarian principle 350 European Union 213
utility maximisation principle 350 Germany 417, 419–20
tax morale and compliance
value added tax (VAT) 218, 220 behaviour 382–6, 389, 392–4,
variable definitions and data sources 399, 404–406, 408
63–6 tax morale and tax evasion 353, 355,
vector error correction model 279 358
Venezuela, RB 28, 32, 38, 42, 76, 491
Vietnam 52, 76, 267, 289 Yemen, Republic of 28, 31, 37, 41, 76
size of shadow economy Yitzhaki, S. 381
(1999–2006/07) 27, 28, 31, 36,
40 Zambia 30, 33, 39, 45, 76
Villiger, K. 151 Zellner, A. 133, 206
Virta, H. 473 Zimbabwe 44, 76
voice see corruption, voice and Zoido-Lobatón, P. 13, 14, 87, 88, 157,
accountability 273, 274, 276, 444










Handbook On The Shadow Economy

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Handbook On The Shadow Economy

Uploaded by

Copyright:

Available Formats

Handbook on the Shadow Economy

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd i 27/07/2011 16:16

M2678 - SCHNEIDER TEXT.indd ii 27/07/2011 16:16

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd iii 27/07/2011 16:16

All rights reserved. No part of this publication may be reproduced, stored in a

Edward Elgar Publishing, Inc.

A catalogue record for this book

Library of Congress Control Number: 2011924162

ISBN 978 1 84844 335 8

Typeset by Servis Filmsetting Ltd, Stockport, Cheshire

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd iv 27/07/2011 16:16

PART I SIZE AND DEVELOPMENT OF THE SHADOW

1 Shadow economies all over the world: new estimates for

PART II REGIONAL VARIATION IN THE SIZE AND

5 Regional variations in the nature of the shadow economy:

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd v 27/07/2011 16:16

PART III SHADOW ECONOMY, ILLICIT WORK AND

7 The shadow economy and do-it-yourself activities: what do

PART IV TAX MORALE AND THE SHADOW ECONOMY

10 Tax morale, tax evasion and the shadow economy 347

PART V CORRUPTION AND THE SHADOW ECONOMY

13 The impact of institutions on the shadow economy and

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd vi 27/07/2011 16:16

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd vii 27/07/2011 16:16

Development Studies, Social Indicators Research, Applied Economics,

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd viii 27/07/2011 16:16

His research interests include public economics, public choice, economic

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd ix 27/07/2011 16:16

2003–2005; and President of the Economic Advisory Board of the Swiss

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd x 27/07/2011 16:16

Christoph A. Schaltegger has been Professor of Economics at the

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd xi 27/07/2011 16:16

Benno Torgler is a Professor of Economics in the School of Economics

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd xii 27/07/2011 16:16

The goal of this handbook on the shadow economy is to provide the

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd 1 27/07/2011 16:16

from a theoretical point of view, the empirical evidence on its success is

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd 2 27/07/2011 16:16

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd 3 27/07/2011 16:16

Chapter 8, the contribution of Christopher Bajada, deals with the

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd 4 27/07/2011 16:16

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd 5 27/07/2011 16:16

empirically analyse how governance or institutional quality affects par-

Friedrich Schneider - 9781848443358

M2678 - SCHNEIDER TEXT.indd 6 27/07/2011 16:16

Size and development of the shadow

Friedrich Schneider - 9781848443358