Thesesdissertations 2633 OBJ

INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality of this reproduction is dependent upon the quality of the

copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by

sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced

xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UM I directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/321-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ECONOMIC STRUCTURE AND INVESTMENT UNDER UNCERTAINTY
by
Bernhard G. Gunter
submitted to the
Faculty o f the College of Arts and Sciences of
American University
in Partial Fulfillment of
the Requirements for the Degree of
Doctor of Philosophy
Economics
Chair:
Prof. Daniel M. SclWdlowsky
Prof. Robert A. Blecker
Cd/Aa*~
Prof. William H. Branson
Dean of the College
American U niversity^' •. ST034

Washington, D.C. 20016
THE AMERICAN UV" STY LIBRARY
UMI Number: 9826672
Copyright 1998 by
Gunter, Bernhard Georg
All rights reserved.
UMI Microform 9826672

Copyright 1998, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized

copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
° COPYRIGHT
by
BERNHARD G. GUNTER
1998
ALL RIGHTS RESERVED
For My Parents
"A good mathematical theorem dealing with economic
hypothesis is very unlikely to be good economics."
Alfred Marshall (1906)
ECONOMIC STRUCTURE AND INVESTMENT UNDER UNCERTAINTY
by
Bernhard G. Gunter
ABSTRACT
This dissertation combines the structural approach of development economics with the
more recent real option theory of investment under uncertainty. First, the main patterns of
development are established. Based on data for 93 countries from 1970-94, it is shown that there
exist robust relationships between development and 34 structural indicators. Second, using a
similar cross-country analysis, it is shown that economic structures of less developed countries
increase macroeconomic uncertainty. One whole chapter is used to define macroeconomic
uncertainty and to illustrate the crucial difference between volatility and uncertainty. Third, using
panel data, it is demonstrated that macroeconomic uncertainty has a significantly negative impact
on gross domestic investment. Fourth, based on the recent empirical literature on the relationship
between investment and development, it is concluded that investment usually increases
development. Finally, the results of the four parts are connected with each other. Conclusions
are drawn on implications for economic policy and future research.
ii
ACKNOWLEDGEMENTS
I am indebted to Prof. William H. Branson at Princeton University and Isabel
Guerrero at the World Bank for providing me with the core idea that uncertainty is
related to economic structure and many other suggestions. I am also indebted to
Professors Daniel M. Schydlowsky and Robert A. Blecker, who have not only provided
me with many critical comments and suggestions to this dissertation, but who also have
supported me throughout my graduate studies at American University. Last not least, I
am thankful for comments received to an earlier version presented at the World Bank in
June 1997, especially from Pierre-Richard Agenor, Farrukh Iqbal, Sandeep Mahajan, and
Quentin Wodon; and for editorial suggestions from Jo Marie Griesgraber, Jeannie
O’Donnell, and Eileen Pereira.
iii
TABLE OF CONTENTS
ABSTRACT........................................................................................................................ u
ACKNOWLEDGEMENTS............................................................................................. iii
LIST OF TABLES..........................................................................................................viii
LIST OF FIGURES.........................................................................................................xu
CHAPTERS:
I. INTRODUCTION.............................................................................................. 1
1. Objectives...................................................................................................... 1
2. Context.......................................................................................................... 4
H. HOW TO RELATE DEVELOPMENT TO ECONOMIC STRUCTURE 10
1. Outline of Some Major Contributions.................................................. II
2. Tools of Analysis................................................................................... 22
3. Defining Economic Structure.................................................................... 39
HI. EMPIRICAL RESULTS OF PATTERNS OF DEVELOPMENT............. 58
1. Sectoral Composition of O utput........................................................... 59
2. Investment Shares in G D P ................................................................... 66
3. Savings and Consumption...................................................................... 69
iv
4. Government Expenditures and Revenues................................................ 72
5. Inflation and Money Supply..................................................................... 75
6. Overall Trade and Import V ariables...................................................... 78
7. Export Shares........................................................................................... 83
8. Export Product Concentration (E P C ).................................................... 87
9. Market Power in World Export M arkets............................................. 89
10. Financial Market Development.............................................................. 90
11. Summary of Results: Stylized F a c ts..................................................... 92
IV. HOW TO MEASURE MACROECONOMIC UNCERTAINTY.............. 95
1. Uncertainty versus R isk .......................................................................... 96
2. The Different Concepts of Volatility and Uncertainty........................ 100
3. Measures of Volatility and Uncertainty................................................ 106
4. A Preliminary Measure of Uncertainty................................................. 119
5. Areas of Macroeconomic Uncertainty.................................................. 122
6. How to Obtain an Unbiased Uncertainty M easure............................. 123
7. Checking for Consistency...................................................................... 130
8. How Different are Volatility and Uncertainty in Reality?................. 134
9. The Relationship of Uncertainty Across A reas................................. 136
10. The Relationship of Uncertainties AcrossPeriods............................. 138
V. THE RELATIONSHIP BETWEEN UNCERTAINTY AND
ECONOMIC STRUCTURE............................................................................. 140
1. The Relationship between Development and Uncertainty.................. 142
2. Insights from the Sign of the Correlation

Coefficients between Structure and Uncertainty............................... 144
3. The Relationship between Economic Structure and

Uncertainty tested by Group Averages............................................... 146
4. Regressing Uncertainty on Economic Structure................................ 147
5. Remaining Stylized F a cts...................................................................... 151
VI. INVESTMENT UNDER UNCERTAINTY.................................................. 154
1. Investment and Risk at the M icro-Level............................................ 155
2. Intuitive Explanations for a Positive Correlation.............................. 158
3. Empirical Analysis of Investment under Uncertainty....................... 160
4. How to Specify the Aggregate Domestic Investment Function 169
5. How Robust are Impacts of Macroeconomic Uncertainty? 179
VH. THE RELATIONSHIP BETWEEN INVESTMENT AND

DEVELOPMENT............................................................................................ 188
1. Investment and G row th.......................................................................... 189
2. Growth and Development..................................................................... 192
3. Investment as a Major Factor of Development................................. 195
vi
v m . CONCLUSION........................................................................................ 197
1. Connecting Major Results................................................................... 197
2. Lessons from an Optimistic Interpretation........................................ 201
3. Limitations and Suggestions for Further Research........................... 203
APPENDICES................................................................................................................206
1. Review of Export Supply Elasticities.................................................................206
2. Problems Related to Interest Rates D ata ......................................................... 208
3. Standard Definitions of Structural Variables.................................................. 212
4. Patterns of Development: Details of RelevantRegression Results................ 216
5. Uncertainty Measures of Different Time Periods........................................... 231
6. Results of Stationarity T ests............................................................................. 238
BIBLIOGRAPHY........................................................................................................ 239
LIST OF TABLES
Table Page
1. List of Countries.................................................................................................. 26
2. Economic and Social Structures Analyzed in Cheneryand Syrquin 41
3. Variables Characterizing Economic Structure................................................. 50
4. Group Averages and Panel Data Correlations................................................ 60
5. Empirical Results of the Structural Regressions

(t-statistics, significance, and range o f R2s ) .................................................... 62
6. A Comparison of Primary Export Shares........................................................ 87
7. Preliminary Uncertainty Measure..................................................................... 124
8. Correlation Matrices of Three Different Uncertainty

M easures............................................................................................................ 133
9. Correlation Matrices of Three Uncertainty and One Volatility Measure ... 135
10. Correlation Matrices of Uncertainties Across A re as..................................... 137
11. Summary of Average Uncertainties Across Periods...................................... 139
12. Macroeconomic Uncertainty and GDP per c a p ita .......................................... 143
13. Correlation Coefficients between Economic Structureand

Macroeconomic Uncertainty............................................................................. 145
14. Group Averages of Uncertainty by Structural V ariable................................. 148
15. Significant t-statistics based on Regression R esults...................................... 150
Table Page
16. Six Initial Specification for Aggregate Domestic Investment....................... 177
17. Results o f Six Initial Specifications...................................................... 178
18. Results for the Specification Excluding Interest R ates................................. 180
19. Results for the Specification Excluding Resource F lo w s............................ 181
20. Results for the Specification Including Debt to Reserve R atios................... 182
21. Testing the Robustness by Analyzing Sub-Periods....................................... 185
22. Testing the Robustness by Analyzing Fixed Private Investment.................. 187
23. Export Supply Elasticities of Developed C ountries....................................... 206
24. Export Supply Elasticities of Developing C ountries..................................... 207
25. Country Averages of Interest Rates for Industrialized Countries................ 209
26. Country Averages for Interest Rates for Developing Countries................. 210
27. Results for the Share of Agriculture................................................................ 216
28. Results for the Share of Industry.................................................................... 217
29. Results for the Share of Manufacturing.......................................................... 217
30. Results for the Share of Services..................................................................... 218
31. Results for the Share of Gross Domestic Investment to G D P .................... 218
32. Results for the Share of Gross DomesticFixed Investment to G D P 219
33. Results for the Share of Fixed Private Investment to G D P ........................ 219
ix
Table Page
34. Results for the Share of Total Consumption to G D P ................................... 220
35. Results for the Share of Private Consumption to G D P ................................ 220
36. Results for the Share of Savings to G D P ......................................................... 221
37. Results for the Share of Government Expenditure to G D P ........................ 221
38. Results for the Share of Government Revenues to G D P ............................. 222
39. Results for the Share of Government Tax Revenues to G D P ...................... 222
40. Results for the Share of Broad Money to G D P ........................................... 223
41. Results for Trade Intensity..................................................................... 223
42. Results for the Share of Current Account Balance to G D P ........................ 224
43. Results for the Share of Capital Account Balance to G D P ......................... 224
44. Results for the Share of Exports to G D P ...................................................... 225
45. Results for the Share of Merchandise Exports to G D P ............................... 225
46. Results for the Share of Machinery Exports to G D P ................................... 226
47. Results for the Share of Machinery Exports to Total Exports.................... 226
48. Results for the Share of Primary Exports to Total Exports........................ 227
49. Results for Export Product Concentration........................................... 228
50. Results for Export Market Power in World M arkets................................... 228
51. Results for Financial Market Development Indicator # 1 .............................. 229
Table Page
55. Nominal GDP Uncertainty Across Different Time Periods......................... 231
56. Real GDP Uncertainty Across Different Time Periods................................ 232
57. Inflation Uncertainty Across Different Time Periods................................... 233
58. Monetary Uncertainty Across Different Time Periods................................. 234
59. Nominal Exchange Rate Uncertainty Across Different Time Periods 235
60. Real Exchange Rate Uncertainty Across Different Time Periods................ 236
61. Terms of Trade Uncertainty Across Different Time Periods......................... 237
xi
LIST OF FIGURES
Figure Page
1. A Circular Relationship between Development,

Economic Structure, Uncertainty, and Investment......................................... 2
2. Scatter Diagrams of Sectoral Shares of G D P ................................................ 66
3. Scatter Diagrams of Investment Shares in G D P ............................................ 68
4. Scatter Diagram of the Shares of Savings in G D P ........................................ 70
5. Scatter Diagrams of the Shares of Total and Private

Consumption in G D P ....................................................................................... 71
6. Government Expenditures, Government Revenues, and Fiscal

Deficits as Shares of G D P .............................................................................. 74
7. Scatter Diagram of the Share of M2 in G D P ................................................ 77
8. Scatter Diagrams of Other Monetary Variables............................................ 77
9. Scatter Diagrams of Shares of Current and Capital

Account Balances in G D P ................................................................................ 79
10. Shares of Total and Merchandise Exports in G D P ...................................... 84
11. Scatter Diagrams of the Share of Machinery Exports in GDP,

the Share of Machinery Exports in Total Exports, and
the Share of Primary Exports in Total E xports............................................. 85
12. Scatter Diagrams of Export Product Concentration....................................... 88
13. Scatter Diagrams of Export Market P o w er.................................................... 90
14. Scatter Diagrams of Financial Market Development.................................... 91
xii
Figure Page
15. The Sectoral Shares E xam ple............................................................................ 198
16. The Nexus Between High Consumption and High GDP Uncertainty 199
17. Linking Government Revenues to Monetary Uncertainty............................ 199
18. The Export Composition Exam ple.................................................................... 200
19. The Importance of Financial Market Development........................................ 201
xiii
CHAPTER I
INTRODUCTION
I. Objectives
The general objective of this dissertation is to show that economic structure and
macroeconomic uncertainty have a significant impact on aggregate domestic investment.
It is suggested that the relationship between economic structure and investment under
uncertainty is part of a four-step interrelationship between economic structure,
macroeconomic uncertainty, investment, and development, as it is illustrated in Figure
1. As will be shown below in more detail, the interrelationship between underdeveloped
economic structure, macroeconomic uncertainty, subsequent low investment and
underdevelopment could be interpreted as a vicious circle of perpetuating
underdevelopment. It is in this sense that we refer to a circular relationship.
Relationship 1 shows the correlation between development and the corresponding
changes in economic structure, in other words, the patterns of development, as they
have been established in the development literature. As Chenery and Syrquin (1975, p.
4) have pointed out, "a development pattern may be defined as a systematic variation
2
in any significant aspect of the economic or social structure associated with a rising
level of income or other index of development." Some 47 relationships between
development and economic structure will be examined in Chapter III.
Figure 1: A Circular Relationship between Development, Economic Structure,

Uncertainty, and Investment
uncertainty
relationship 2 relationship 3
economic investment
structure
development
Relationship 2 is the correlation between economic structure and macro-
economic uncertainty. As will be shown in Chapter V, economic structures of less
developed countries increase macroeconomic uncertainty.
3
Relationship 3 shows the negative impact of macroeconomic uncertainty on
investment, as it is suggested by the real option theory of investment under uncertainty.
The impact of a variety of macroeconomic uncertainties will be reexamined in some
detail in Chapter VI.
Finally, relationship 4 is the generally positive relationship between investment
and development, which will be reviewed in Chapter VII.
Concerning these four relationships, the dissertation has four main objectives:
(a) to examine some old and new stylized facts of development patterns
(relationship 1 in Figure 1);
(b) to establish some new stylized facts about the relationship between economic
structure and macroeconomic uncertainty (relationship 2 in Figure 1);
(c) to examine if macroeconomic uncertainty has a significant negative impact on
investment (relationship 3 in Figure 1); and
(d) to review the recent literature on the relationship between investment and
development (relationship 4 in Figure l ) .1
In short, the objectives are to learn which economic structures are related to
which kind o f macroeconomic uncertainty, and which kind of macroeconomic
uncertainty is a crucial determinant of domestic investment. The three analytical parts
1 The scope of the fourth objective has been limited to a short review of the
literature, as there have been recent contributions to the relationship between investment
and growth, and also the relationship between growth and development. In addition, we
will be able to draw some tentative conclusions of the relationship between investment
and growth from our empirical analysis of the investment function in Chapter VI.
4
(relationships 1 to 3) are based on the recent experiences o f up to 93 countries over the
last 25 years (1970 to 1994).
2. Context
In order to evaluate the importance and novelty of this dissertation, it is
important to put the proposed research in the context of existing economic theories and
their latest developments. It is possible to relate the proposed research to three recent
developments in economic theory: (a) the wide-spread neglect of economic structures in
the new growth theory, (b) the real option theory of investment under uncertainty, and
(c) the recent work on macroeconomic volatility. The dissertation will combine various
aspects o f these developments. It will build a bridge between traditional structural
analysis and the new investment theory by analyzing the relationships between economic
structure, macroeconomic uncertainty and domestic investment.
2.a. Economic Structures and the New Growth Theory
The idea that economic structure matters for macroeconomics of developing
countries achieved great importance through the seminal work of Lewis (1954), the
many contributions of Simon Kuznets, Hollis B. Chenery and Moshe Syrquin, and
especially the contributions of Lance Taylor.2 The relevance of economic structure for
investment and growth has been touched on by Cooper (1971), Krugman and Taylor
(1978), Branson (1983), Katseli (1983) and Taylor (1983). Branson (1986) showed that
2 Please see the first section of Chapter II for a short review of the previous
literature analyzing patterns of development.
5
less developed countries' trade structures differ in terms of elasticities of demand and
supply for exports and imports from the assumptions of the monetary model and that
devaluations therefore have potentially stagflationary effects. The relevance of financial
structure has been stressed by Goldsmith (1969) and Stiglitz (1988) and various recent
contributions by Ross Levine (please see more details below).
The importance of economic structure is becoming clear in the recent work of
institutional economics. But beyond the new institutional literature even traditional areas
such as tax policy are geared to the understanding of differences in economic structure.
For example, Tanzi (1993, pp. 35-36) has stated that "differences in tax levels and in
tax structure between industrial and developing countries are probably explained more
by differences in economic structures than by differences in policies." In other words,
that economic structure matters for development is a well established fact.
However, structural analysis has not received enough attention since the
emergence of the so-called new or endogenous growth theory. Endogenous growth
theory relates economic growth to production functions with either increasing returns to
scale or non-diminishing returns to reproducible capital. It stresses the importance of
human capital, and more recently, also other factors influencing the effectiveness of
physical and human capital, such as income distribution and political instability.3 Even
though old and new growth theories have been criticized repeatedly for not taking into
3 The main contributions along these lines are Alesina and Perotti (1993), Barro
(1997), Borner, Brunetti, and Weder (1995), Campos and Root (1996), and Tavares and
Wacziarg (1996).
6
account changes in sectoral composition,4 even the new growth literature has limited the
analysis of economic structures to investment shares/ various trade shares,6 and
financial market development.7 Only two aspects of economic structure have been
analyzed systematically across countries within the investment literature: international
trade structures and financial market development. The two studies which center on the
topic of financial market development and investment are Greenwood and Jovanovic
(1990), and Bencivenga and Smith (1991). However, neither study has made the
connection to the real option theory of investment under uncertainty.
2 .b . The Emergence of the Real Option Theory
Traditional investment theories at the aggregate level are the accelerator theory,
the neo-classical theory of investment, and Tobin's q-theory. The early accelerator
4 See for example, Pack (1994), p. 68, who refers explicitly to Denison (1985).
Denison (1985) finds that intersectoral shifts in production explain part of aggregate
growth. Pasinetti (1994), p. 356, has criticized new growth theory for being "essentially
one-commodity models, with no structural change".
5 The literature relating investment to growth will be reviewed in more detail below
in Chapter VII.
6 There is some discussion on what impact trade intensity and outward orientation
really had on growth. For example, while Dollar (1992) provides strong evidence for a
positive relationship between outward-orientation and growth, Singh (1994) is more
critical on how open the East Asian miracle countries really were. More recently,
Vamvakidis (1996) has concluded that the benefits of an open economy depend on the
openness of the economy's trading partners.
7 The impact of financial market development on growth has been analyzed

especially by Ross Levine and others [Levine and Renelt (1992), King and Levine
(1993), Demirguc-Kunt and Levine (1996), Levine and Zervos (1996) and Levine
(1997)].
7
theory of investment captured the dependence of investment on expected returns.
According to the accelerator model, investment is related linearly to past changes in
output. Output growth rates are expected to persist through time. No account is taken of
the influence of the cost of capital on the investment decision. The neo-classical theory
o f investment, which followed, addressed the latter shortcoming but continued to
assume that output was determined outside the model. According to the neoclassical
model, investment spending depends on the user cost of capital and is geared to
maintaining the optimal capital stock and an associated level of output.8
Based on the poor empirical performance of traditional investment theories,
recent research o f the investment theory has led to a revised and extended account of the
determinants of investment. It has especially focused on the gain from postponing
investment in periods of heightened uncertainty. In contrast to earlier theories that
assume perfect markets for capital goods, the new approach emphasizes that investment
decisions are inherently irreversible, that agents typically have some discretion over the
timing of investments, and that the uncertainty of investment returns is positively related
to macroeconomic and political uncertainty. The idea of postponing irreversible
investment in periods of profound uncertainty has been formalized using option pricing
theory by Avinash Dixit and Robert Pindyck (1994). The ability to wait is an option to
invest later; the higher the uncertainty, the greater the value of the option. In periods of
heightened uncertainty, there is a high marginal return from waiting until the
8 For recent reviews and issues of measurements of Tobin's q, see Caballero (1996),
and Lewellen and Badrinath (1997).
8
uncertainty is resolved. As Federer (1993) has demonstrated, the real option theory of
investment under uncertainty is a more sophisticated version of what many Post-
Keynesians were emphasizing for a long time: (a) that investment is irreversible,9 and
(b) that the ability to wait is an option to invest later.10
The real option theory of investment under uncertainty is especially important
for developing countries for at least three reasons. First, as markets and institutions are
underdeveloped, information is more costly. Second, as will be verified empirically in
this dissertation, LDCs face higher macroeconomic uncertainties in terms of their
income, inflation, money supply, exchange rates, and terms of trade. Third, as LDCs
are generally more highly indebted than industrialized countries, they face a higher
uncertainty about their debt sustainability."
2.c. Recent Work on Macroeconomic Volatility
Related to the emergence of the real option theory of investment under
uncertainty, there has been some new interest in the analysis of macroeconomic
volatility. As will be shown below, volatility and uncertainty are two distinct concepts,
9 In the words of Davidson (1972, p. 16): "It may be costly if not even impossible to
restore the pre-existing circumstances."
10 In the words of Davidson (1972, p 16): "In a world of uncertainty, he who

hesitates is saved to make a decision another day."
" Note that the recent debt forgiveness proposal for heavily indebted poor countries
(HIPC) put forward by the World Bank and the International Monetary Fund (IMF) in
Fall 1996 has been based on the argument that the current debt levels of highly indebted
countries would deter investment and long-run growth.
9
even though there is a large literature which does not differentiate appropriately between
volatility and uncertainty. In any case, more recent work on macroeconomic volatility
has — in contrast to some of the older literature — shown that macroeconomic volatility
does undermine investment in less developed countries. This is more carefully analyzed
below in Chapter IV.
Finally, it is interesting to note that the source of macroeconomic volatility has
almost always been related to economic policies but not to economic structures. This
dissertation concentrates solely on economic structure and makes little reference to
economic policies. However, the neglect of economic policies should not be interpreted
as meaning that economic policies are unimportant. We neglect economic policies
simply because we are interested in the role of economic structures. The quantification
of policies would raise more questions then it would answer for our purpose.
In reality, policies and economic structure are both relevant for development.
The relative importance of policies versus structures depends on (a) the degree to which
good or bad policies have been applied, (b) the prevailing economic structure of the
country, and (c) the specific topic under consideration. Most of the policy analyses have
ignored economic structure, which may explain why policies often fail to achieve their
objectives or have unexpectedly high costs. A better understanding of structure can lead
to more effective policy. This is the heart of development economics: to understand
economic structure and macroeconomic uncertainty in order to achieve equitable,
participatory and sustainable development. While much seems to be jargon at this point,
we will have a clearer picture at the end of this dissertation.
CHAPTER n
HOW TO RELATE DEVELOPMENT TO ECONOMIC STRUCTURE
There are obviously many problems in analyzing the theoretically indefinite number
of relationships between the level of development and economic structure. An initial
problem is related to the definitions of development and economic structure. There is no
accepted definition o f either development or o f economic structure. The more accurate a
definition is from a theoretical point of view, the more difficult it becomes for practical
purposes to find comparable data across countries and time. But even if there would exist
perfect definitions and data, there remains a large number o f problems related to the
research methodology. Some of these methodological complications are due to the fact
that the causality runs in both directions, that most of the time-series data is not stationary
(and taking first differences would be inconsistent with the objectives of this analysis), that
development is determined by many highly correlated structural variables, and that
development is certainly influenced by economic policies, institutional characteristics and
cultural differences, most of which are basically impossible to quantify.
While this chapter will address some of these issues, its main purpose is to outline
the extent and limitations of the structural analysis and to provide some explanation for the
10
11
deliberately chosen simplicity in analyzing the relationship between development and
economic structure. Given the extreme complexity of relationships as pointed out in some
of the earlier literature (see details below), the goal of this first analytical part is to
establish simple stylized facts1 o f relationships between development and economic
structure based on the experience o f 93 countries over the last 25 years (1970-94). We are
interested to show that differences in economic structures between richer and poorer
countries remain relevant, but we are not interested in the determination of either the
causality or the source o f these differences.
First, we provide an overview of the major contributions to the pattern of
development literature. This overview will concentrate on what has been analyzed and
how it has been analyzed. Then, we will explain our tools and our methodology for
analyzing simple relationships between development and economic structure. Finally, this
chapter will address the issues related to the definition of economic structure including the
problems related to data constraints. The relevant results of the earlier literature will
briefly be compared to our own results in the next chapter.
1. Outline o f Some Major Contributions
Within the long history of economic analysis, the systematic analysis of patterns of
development across countries is relatively new. While single aspects of development
1 As Syrquin (1994) has pointed out, it was Nicholas Kaldor who coined the term
‘stylized facts’ in his summary of observations about the growth o f industrial economies at
the IEA Corfu conference in 1958. These ‘facts’ are empirical regularities observed "in a
sufficient number o f cases to call for an explanation that would account for them ...
12
patterns have been analyzed for more than 140 years,2 it was 58 years ago that a variety of
patterns were analyzed systematically by Clark (1940). Single aspects remained to be
analyzed in the following years [see Kuznets (1955) and Houthakker (1957) below], but
the systematic analysis took off with Kuznets' series of 10 articles on "Quantitative
Aspects of the Economic Growth of Nations.” The analysis of development patterns
matured to a uniform analysis o f the principal changes in economic and social structure in
the seminal work of Chenery and Syrquin (1975). While some of these patterns were re
examined over the following years, the interest in structural analysis slowed down during
the 1980s. Only recently has the structural approach re-emerged as a powerful and fruitful
analysis in explaining differences in the impact o f adjustment on income distribution,
growth and development. The following review outlines some of the major contributions
since Clark (1940).
l.a. Clark's Conditions of Economic Progress
The first systematic analysis o f patterns of development was undertaken by Colin
Clark in 1940, in his first edition o f The Conditions o f Economic Progress. The
completely rewritten second edition was published in 1951, and the third revised edition in
1957. Clark's goal was to base economics on the collection and examination o f actual
independently of whether they fit into the general framework of received theory or not,"
Kaldor (1985), pp. 8-9.
For example, Engel's law, stating that the proportion of income spent on food
declines as income rises, was introduced by Ernst Engel (1857).
13
facts, instead of "upon speculations and theoretical reasonings.”3 For this purpose, he
collected a great deal of the existing statistical data from around the world.
While Clark's research methodology has been limited to the illustration and
interpretation of empirical data using simple descriptive statistics and graphical
illustrations, his analysis is far too specific to be summarized here.4 However, Clark's work
is the first systematic analysis of patterns of development across countries in the areas of
(i) the productivity o f various industries,
(ii) the consumption o f principal classes o f goods and services as a function
of real income,
(iii) the sectoral composition of employment, and
(iv) the distribution o f income between factors of production and between
persons.
O f interest is also the fact that Clark paid considerable attention to the problems
regarding the international comparison of income, especially the purchasing power of
money and the real national product per man-hour worked. He provided a standard of
comparison of purchasing power by taking the quantity of commodities exchangeable for
$1 in the United States over the average period 1925-34, which he called the
"International Unit.”
3 Clark (1951), page v.
4 The third edition o f The Conditions o f Economic Progress contains more than 250
tables, more than 50 diagrams and more than 400 pages of text.
14
l.b. Inequality and Growth: Kuznets (1955)
One of the most famous hypothesis in the field of development economics is
Kuznets' (1955) inverted U curve, portraying the relationship between growth and
inequality. Inequality increases initially, but decreases in later stages of development. This
hypothesis has been disputed over the years, especially by Papanek and Kyn (1986 and
1987). Papanek and Kyn concluded that there is little or no evidence for the intertemporal
Kuznets curve. They even concluded that cross-country comparisons show little evidence
for the relationship, and that more interventionist governments do not achieve greater
equality. Instead of growth, they find that the socio-political dualism (an elite drawn from
an ethnic minority) is an important factor in explaining inequality. Nevertheless, consistent
with the earlier findings, they do find a positive relationship between the spread of
education and equality, as well as a negative relationship between the share o f primary
exports in gross domestic product (GDP) and equality.
More recently Fishlow (1996) reviewed the literature on inequality and
development, and concluded that a complete dismissal of the original Kuznets (1955)
parabolic relationship between inequality and growth is an error because there is some
evidence o f its validity when economic policy is taken into account. He concludes that
new research findings which demonstrate a positive relationship between income growth
and greater equality are questionable, and depend on the Latin American experience
during the lost decade, when negative growth rates went hand in hand with increasing
inequality. Fishlow also confirms a positive relationship between inequality and education
and a negative relationship between inequality and the share of primary export. However,
15
Fishlow concludes that reduced poverty and improved income equality are compatible
goals and have been achieved in the past. To a similar conclusion comes the Human
Development Report 1996. James Gustave Speth writes in the Foreword o f the report:
"Contrary to earlier theories, new theory and evidence suggest that growth and equity
need not be contradictory goals."5
Finally, the relationship between growth and inequality has been reexamined by
Deininger and Squire (1997). Using a new database [Deininger and Squire (1996)], they
found no evidence o f the Kuznets curve. Their analysis went even one step further to
make more direct inferences regarding the relationship between growth and poverty. We
will return to this point in more detail in Chapter VII, when reviewing the relationship
between growth and development.
I.e. Houthakker's (1957) Reexamination of Engel's Law
Commemorating the centenary of Engel's law,6 Hendrik S. Houthakker (1957)
compared the household expenditure patterns, based on about 40 surveys from about 30
countries between 1901 and 1955. The survery date varied from country to country, the
sample size varied between 35 and 21,964 observations. The international comparison of
elasticities for food, clothing, housing and miscellaneous items with respect to total
expenditure confirms Engel's law.
5 UNDP (1996), page iii.
6 Engel’s law [based on the study by Ernst Engel (1857)] states that the proportion of
income spent on food declines as income rises.
16
l.d. Kuznets' Quantitative Aspects
The "Quantitative Aspects o f the Economic Growth of Nations,” were presented
in 10 studies by Simon Kuznets, published in various issues of Economic Development
and Cultural Change between 1956 and 1967,7 adding up to 1,056 pages of detailed
analysis. As footnote 7 illustrates, Kuznets's studies covered the following areas: levels
and variability of growth rates, the distribution o f national product, income, and labor
force, the comparison and trends of capital formation properties, the share and structure of
consumption, and the comparison and trends of level and structure o f foreign trade.
The historical growth patterns and the structural transformation of output and
production as a whole were then presented by Kuznets (1966) and Kuznets (1971). While
Clark used some graphical illustrations, Kuznets1 methodology was limited to verbal
description and interpretation of tables. While this is richly detailed, it clearly lacks the
provision o f generalizations which would have been useful to summarize Kuznets's work.
Also, most o f Kuznets’ analysis is limited to the development patterns of the industrialized
countries.
7 These 10 studies are:

I. Levels and Variability of Rates of Growth [Kuznets (1956)],
n. Industrial Distribution of National Product and Labor Force [Kuznets (1957)],
m. Industrial Distribution of Income and Labor Force by States, 1919-55 [Kuznets (1958)],
IV. Distribution of National Income by Factor Shares [Kuznets (1959)],
V. Capital Formation Properties: International Comparisons for Recent Years [Kuznets (I960)],
VI. Long-Term Trends in Capital Formation Proportions [Kuznets (1961)],
VII. The Share and Structure of Consumption [Kuznets (1962)],
Vm. Distribution of Income by Size [Kuznets (1963)],
IX. Level and Structure of Foreign Trade: Comparisons for Recent Years [Kuznets (1964)], and
X. Level and Structure of Foreign Trade: Long-Term Trends [Kuznets (1967)].
17
I.e. Chenery and Syrquin's Patterns of Development
Influenced by Kuznets' worlc, there were a number of studies related to the analysis
of development patterns, like Chenery (1960), Chenery and Taylor (1968), and Taylor
(1969). However, the problems with all these earlier studies were their lack of either data
availability or a consistent methodology.
The major break-through came with the seminal work Patterns o f Development,
1950-1970 by Chenery and Syrquin (1975). This work analyzed the patterns of
development in six chapters. The first chapter provides the basis for their comparative
analysis. The second chapter analyzes the uniformity of development patterns of the
accumulation and allocation processes, while the third chapter analyzes the uniformity of
development patterns of the demographic and distributional processes. The fourth chapter
then investigates the systematic differences in the development patterns. The fifth chapter
compares time-series and cross-section results, and finally, the sixth chapter provides
conclusions.
The importance of this work rests upon the uniform analysis of the principal
changes in economic and social structure. It is the uniform analysis of structural
characteristics which provides a consistent description o f a number of interrelated types o f
structural changes and also identifies systematic differences in development patterns
among countries that are following different development strategies.8
8 Chenery and Syrquin (1975), p. 1.
18
1. f. Patterns of Development Re-examined
The transformation of the economic structure was further elaborated in Chenery,
Robinson and Syrquin (1986). Syrquin and Chenery (1989) re-examine a reduced set of
development patterns, focusing on economic structure related to the sectoral allocation of
resources for 108 economies during the period 1950-1983. As in Chenery and Syrquin
(1975), real GDP per capita serves as the measure o f the level o f development, and the
sectoral allocation of resources is usually expressed in percent of GDP.
l.g. McCarthy, Taylor, and Talati's Trade Patterns
Trade patterns were thoroughly analyzed in McCarthy, Taylor, and Talati (1987).
These authors pooled data from fifty-five developing countries for 1964-82 to run
multivariate regressions for import and export shares of 14 commodity and service
categories. Though their results are too detailed to be presented here, they provide strong
support for the hypothesis that trade shares are related to the level o f development. What
is remarkable is that there is no systematic analysis of the relationship between export
concentration and the level of development, even though there is a long history of studies
related to export concentration.9
9 The pioneering work of geographic concentration in the foreign trade is Albert O.

Hirschman's (1945) study on National Power and the Structure o f Foreign Trade. More
than 15 years after Hirschman's study, Michaely (1962, p. 1) pointed out that while
references to trade concentration were ample in discussion o f both trade theory and trade
policy, "none of these discussions contains a rigorous and systematic analysis of
concentration. Moreover, both exception, geographic concentration have not yet been
subjected to meaningful quantitative investigations." The various contributions by
Michaely (1958, 1960, 1962, 1984) have closed much o f the quantitative gap. Though
Michaely has shown that high export product concentration is an attribute of
19
l.h. Counterfactual Analysis using the Maquette
Bourguignon, Branson, and de Melo (1989) constructed a simulation model, called
the "maquette,” to derive orders of magnitude about the likely distributive implications of
alternative adjustment strategies. They called the model a maquette because it provides
both structure and solution procedures, but leaves the provision o f parameters and initial
distributive shares to the user of the model.
While the lack of data does not allow to parameterize and calibrate the maquette
for all countries, Bourguignon, de Melo and Suwa (1991) developed archetype versions of
the maquette for African and Latin American economies. They used the stylized facts
developed by Chenery, Robinson and Syrquin (1986) for low- and middle-income
economies and added assumptions applicable to Africa and Latin America. Following
Bourguignon, de Melo and Suwa (1991), Jayarajah, Branson, and Sen (1996) have
constructed an East Asian archetype using structural data from Indonesia, Malaysia, the
Philippines, and Thailand. Representing differences in economic structure, the different
archetypes demonstrate that the same adjustment policies can have completely different
impact on the distribution o f income.
A similar categorization o f developing countries has been used in the
undergraduate development textbook by Auty (1995). Auty classifies the developing
underdeveloped economies, there does not seem to exist a systematic relationship between
the level o f development and export concentration. Instead, what has been studied is (a)
the income levels of trade by good, see chapters 6 to 10 of Michaely (1984), and (b) the
impact of trade concentration on export earnings instability, with different results. For
example, in contrast to previous studies, Love (1986) has indicated the existence o f a
statistically significant, positive relationship between concentration and export instability.
20
countries according to country size and natural resource endowment. He describes five
types of developing countries: low-income Africa, low-income large Asia, newly
industrializing East Asia, mid-income Latin America, and the mineral economy. Auty also
analyzes the patterns of development, though the approach is different from the earlier
patterns of development literature as Auty analyzes the patterns of development along five
central themes: rural neglect, income inequality, hyper-urbanization, unequal terms of
trade, and the role o f government.
I.i. Structuralist Models of North-South Trade Liberalization
Building on the important contributions by Taylor (1983) and Marglin (1984),
there exist also a variety o f structuralist models which analyze the impact of trade
liberalizations. Similar to the structural assumptions of the maquette, Amitava K. Dutt
(1989 and 1990) has provided a convenient framework for making different structural
assumptions about the North and South using the method of ‘alternative closures’ of a
common model o f growth and distribution. The latest contribution along this line of
structuralist models is Blecker (1996). Robert A. Blecker adapts a structuralist North-
South trade model to incorporate the features of the new economic integration, that is,
“capital flows and technology transfers into the South are now creating up-to-date
manufacturing capacity that competes with similar capacity in the North.”10 Analyzing the
implications based on short-, medium-, and long-run models, Blecker (1996, pp. 343-344)
shows that
10 Blecker (1996), p. 323.
21
economic integration is likely to be a much more bumpy and contradictory process

than many of its promotors have hitherto admitted. At the same time, popular fears
o f a ‘giant sucking sound' of massive job losses in the North are probably
exaggerated, and there are many respects in which the North stands to gain from
present integration efforts, especially in the long run (although the interests of
Northern capital and labor may diverge).
l.j. Sundrum's Structural Transformation
Sundrum (1991) wrote three chapters on "Growth and Structural Transformation"
in his recent book, Economic Growth in Theory and Practice, where he specifies a multi
sector model which takes into account changes in the economic structure. In a nutshell, his
model consists o f three key parameters: the allocation of labor to various sectors, the
growth rates of labor productivity over time, and the changing income elasticities of
demand. Sundrum studies the behavior of the model through simulation exercises because
there are too many parameters to attempt an analytical solution. Sundrum's empirical data
is taken from earlier studies o f Kuznets and Chenery et al. Cornwall and Cornwall (1994)
use a structuralist model of Sundrum (1991) to show that while new growth theory
endogenizes growth, it neglects aggregate demand and distributional shifts in output and
employment. By analyzing the European integration, they show that economic structure
matters. Pauly (1994) addresses the relationship between the structures of national
financial markets and capital mobility. He concludes that "notwithstanding the increasing
mobility of capital, asymmetries in structures persist and have important consequences for
the rules of the international economic game as they are now evolving."11
11 Pauly (1994), page 342.
22
l.k. Agenor and Montiel's Development Macroeconomics
A good sununary o f the differences in economic structures can be found in the
textbook Development Macroeconomics by Agenor and Montiel (1996). Agenor and
Montiel (1996) justify their textbook by arguing that "the structural differences that
differentiate a ‘representative’ developing economy from the textbook industrialized
model cover a wide spectrum.”12 Agenor and Montiel go beyond the usual definition of
"economic structure,” by including differences in exchange rate systems, economic
policies, institutional features and behavioral relationships. Regarding the more traditional
structural features of an economy, Agenor and Montiel (1996) conclude that developing
countries tend to be capital importers and have larger government sectors than
industrialized countries. They depend more on imported intermediate goods, and have less
developed financial markets. Agenor and Montiel (1996) also provide convincing evidence
for the case that developing nations tend to be substantially more open than the seven
major industrialized countries, where openness is defined as the sum of import and export
shares to GDP. The question which will be addressed below is one whether or not this
holds true concerning smaller industrialized countries. In other words, is there a systematic
relationship between trade intensity and the level of development?
2. Tools of Analysis
Following Clark and Kuznets, first scatter diagrams and country-group averages
are used to familiarize the reader with general characteristics of the data. Then, following
12 Agenor and Montiel (1996), p. 16.
23
Chenery and Syrquin (1975), we apply a simple but consistent set of regression
specifications for all variables. Additional specifications are provided to check for
robustness. Time-series analysis is excluded for reasons which will be explained below.
But we use correlation coefficients based on panel data to check some dynamic aspects.
The following sub-section provides the rationale for the use o f these four tools o f analysis.
Then, each tool will be described in more detail.
2.a. Rational for Simplicity
The four tools o f analysis are relatively simple, reflecting the limited goal of this
part o f the dissertation and the desire to follow earlier specifications to allow for
comparisons. The first analytical part o f the dissertation will determine some of the main
differences in economic structure between richer and poorer countries, based on the recent
(1970-94) experience o f all countries for which there is available a critical mass of data.
After excluding the 58 economies with a population o f less than 1 million, such data is
mostly available for 93 countries.13
Some o f the main consequences o f this analysis are that we do not need to worry
about issues related to causality or the underlying sources of differences in economic
structure beyond GDP per capita. As Chenery and Syrquin (1975) have pointed out, the
income level incorporates other factors determining economic structure in a single income
13 See Table la on page 222 of the World Development Report 1996 for the list of
these 56 countries with a population of less than 1 million. Most of the other countries for
which there is generally too little data available are currently or formerly socialist
economies. The 93 countries included in the analysis are classified below when defining
the country groups.
24
effect. This is likely the case after controlling for economies of scale. Furthermore, the
precise relationship between development and economic structure is irrelevant as long as
one relationship can be established with sufficient evidence. For example, the relationship
may be stronger at a low level of income and less strong at high level o f income. Thus, the
estimators of our regressions do not have to be the best linear unbiased estimators
(BLUE), as they are not supposed to be used for any prediction or forecasting. Finally, we
will limit our analysis to the testing o f a linear, log-linear, or quadratic relationship and a
limited number of exogenous variables in each regression. If none of these three
relationships can be established, then no simple relationship exists. The non-existence of a
simple relationship should not be interpreted as that there is no relationship at all.
2.b. Scatter Diagrams
Two-dimensional scatter diagrams of cross country averages will be examined to
determine if a particular relationship is linear, log-linear, or quadratic. Together with the
analysis o f country-group averages, the results of the scatter diagrams will then determine
the linear, log-linear, or quadratic regression specification. The relevant scatter diagrams
are provided later in Chapter ID.
2.c. Country-Group Averages
Within the first analytical part o f the dissertation, there will be four consistently
assembled country-group averages for all structural variables under consideration. The
25
four country-groups are based on the World Bank's14 classification o f low-income, lower-
middle-income, upper-middle-income, and high-income countries. Table 1 provides the
names and classification of the 93 countries for which there is sufficient amount of data
available. We will explain the data source below when defining the various economic
structures analyzed in this dissertation. While some data would have been available for
some further countries, it was preferred to keep a consistent set o f countries throughout
this dissertation.
Among these 93 countries, the low-income group contains thirty-nine countries,
the lower-middle-income group contains twenty-two countries, the upper-middle-income
group contains eleven countries, and the high-income group contains twenty-one
countries. These differences in numbers across country groups reflect the relative size of
these country groups and should not be interpreted as a possible source of a country
selection bias.15
14 World Bank, World Development Report 1996. This implies that the countries are
classified by income levels as they existed at the end of the time period under
consideration. The reason for not using the classification of income levels as they existed
in the middle of the time period (1982) is that a few developing countries improved their
classification up to 1982, but fall back in the years after. Overall, there are actually only a
couple of countries which jumped income categories since 1970.
15 For example, excluding economies with a population of less than one million, there
exist 14 upper-middle-income countries, excluding the Czech Republic, Hungary, and
Slovenia. Data was available for 11 o f these 14 upper-middle-income countries.
26
Table 1: List of Countries
Country Classification
Algeria lower-middle-income LDC

Argentina upper-middle-income LDC
Australia high-income DC
Austria high-income DC
Bangladesh low-income LDC
Belgium high-income DC
Benin low-income LDC
Bolivia lower-middle-income LDC
Botswana lower-middle-income LDC
Brazil upper-middle-income LDC
Burkina Faso low-income LDC

Burundi low-income LDC
Cameroon low-income LDC
Canada high-income DC
Central African Republic low-income LDC
Chad low-income LDC

Chile upper-middle-income LDC
China low-income LDC
Colombia lower-middle-income LDC
Congo low-income LDC
Costa Rica lower-middle-income LDC

Cote d'Ivoire low-income LDC
Denmark high-income DC
Dominican Republic lower-middle-income LDC
Ecuador lower-middle-income LDC
Egypt low-income LDC

El Salvador lower-middle-income LDC
Finland high-income DC
France high-income DC
Gabon upper-middle-income LDC
Gambia, The low-income LDC

Germany high-income DC
Ghana low-income LDC
Greece upper-middle-income DC'
Guatemala lower-middle-income LDC
Table 1 continues on next page.
27
Table I(cont):
Guinea-Bissau low-income LDC

Haiti low-income LDC
Honduras low-income LDC
India low-income LDC
Indonesia lower-middle-income LDC
Ireland high-income DC
Italy high-income DC
Jamaica lower-middle-income LDC
Japan high-income DC
Kenya low-income LDC
Korea, Republic of upper-middle-income LDC

Lesotho low-income LDC
Madagascar low-income LDC
Malawi low-income LDC
Malaysia upper-middle-income LDC
Mali low-income LDC

Mauritania low-income LDC
Mauritius upper-middle-income LDC
Mexico upper-middle-income LDC
Morocco lower-middle-income LDC
Myanmar low-income LDC

Nepal low-income LDC
Netherlands high-income DC
New Zealand high-income DC
Nicaragua low-income LDC
Niger low-income LDC

Nigeria low-income LDC
Norway high-income DC
Pakistan low-income LDC
Panama lower-middle-income LDC
Papua New Guinea lower-middle-income LDC

Paraguay lower-middle-income LDC
Peru lower-middle-income LDC
Philippines lower-middle-income LDC
Portugal high-income DC
28
Table 1(cont.):
Rwanda low-income LDC

Saudi Arabia upper-middle-income LDC
Senegal low-income LDC
Sierra Leone low-income LDC
Singapore high-income LDC2
Somalia low-income LDC

Spain high-income DC
Sri Lanka low-income LDC
Sudan low-income LDC
Sweden high-income DC
Switzerland high-income DC
Syrian Arab Rep. lower-middle-income LDC
Tanzania low-income LDC
Thailand lower-middle-income LDC
Togo low-income LDC
Tunisia lower-middle-income LDC

Turkey lower-middle-income LDC
United Kingdom high-income DC
United States high-income DC
Uruguay upper-middle-income LDC
Venezuela lower-middle-income LDC

Zaire low-income LDC
Zambia low-income LDC
Notes:
1 Within this sample of 93 countries, Greece is the only industrialized country which is
not a high-income country but an upper-middle-income country.
2 Within this sample of 93 countries, Singapore is the only high-income country which is
classified by the United Nations or otherwise regarded by their authorities as developing.
Source: World Bank, World Development Report 1996, Table I, pp. 188-189.
29
For the group averages, a monotone relationship between group and group
averages is required for a possible linear or log-linear relationship.16 For example, if the
group o f low-income countries has an average fiscal deficit o f 4.0 percent, and the group
of the richest countries has an average fiscal deficit of 2.0 percent, we would require that
groups 2 and 3 have fiscal deficits of about 3.3 percent and 2.7 percent.
2.d. Regression Analysis: General Specification Issues
2.d.i. Specifications o f Earlier Studies
The systematic specifications for all structural variables under consideration
follows the principal specification of Chenery and Syrquin (1975) and o f Syrquin and
Chenery (1989), which builds on the regression specifications of earlier studies, especially
Houthakker (1957), Chenery (1960), and Chenery and Taylor (1968).
Houthakker's (1957) international comparison of household expenditure patterns
estimated the following functional form using ordinary least squares regressions:
lnX = a + p i n Y + yl n N
where X is the share of the expenditure group under consideration, Y is total expenditure,
and N is family size.
Chenery (1960) has used the same functional form as Houthakker (1957) though
the variables have changed. In Chenery (1960), X is the dependent structural variable,
16 We have also considered the possibility o f U-shaped or inverse U-shaped

relationships. However, since no such relationship could be established for any of the 47
structural variables, we have dropped the further explanation of that methodology.
30
usually taken as a ratio to GDP; Y is the income level measured as GNP per capita; and N
is the country's population. Chenery and Taylor (1968) added a nonlinear income term to
the specification. Chenery and Syrquin (1975) added a nonlinear country size term and a
variable controlling for the net resource flow (F):
In X = Yi + y2 In Y + y3 (In Y)2 + y4 In N + y5 (In N)2 + y6 F
where X, Y and N are defined as above in Chenery (1960), and F is the net resource
inflow measured as imports minus exports of goods and non-factor services as a share of
GDP.
Though the inclusion of a linear and a nonlinear income and country size term in
one equation raise serious questions about multicollinearity and parameter interpretation,
they are meant to capture an upper and lower asymptote. The upper asymptote reflects the
industrialized countries' experience of a period of fairly rapid change followed by
deceleration. The lower asymptote reflects the case of many now upper-middle-income
developing countries experiencing a period of accelerated structural change after an initial
stage of stagnation.
Finally, McCarthy, Taylor, and Talati (1987) dropped the exogenous variable
controlling for the net resource inflow. However, they added two new explanatory
variables: the log of capacity utilization and the log of the real effective exchange rate.
Furthermore, the import function adds the log o f the real GNP per capita of the OECD
countries as an additional exogenous variable. As there is no empirical data on capacity
levels, McCarthy, Taylor and Talati (1987) estimate the potential real GNP as a linear
31
envelope of actual levels over time, which is then used to calculate the capacity level as
the ratio of actual to the estimated potential output.
2.d.ii. Defining Economic Development
There is no perfect indicator for the level of development at the national level, or
among countries at the international level. The usual measure in the patterns of
development literature is either real GDP17 per capita or real GNPIg per capita, either
based on nominal exchange rates or purchasing power parity adjusted exchange rates.
There are well-known limitations of income per capita as a measure of development.
Income per capita excludes non-market transactions; says nothing about the distribution of
income; and does not take into account environmental pollution or degradation, the loss of
natural resources, changes in the quality of life, or other externalities. Finally, it is not
without statistical and measurement problems.
There are now many more comprehensive indicators o f development, for example,
the Human Development Index (HDI) of the United Nations Development Programme
(UNDP) and the Overseas Development Council's Physical Quality of Life Index (PQLI).
However, problems remain, particularly those related to the consistency in the definition
o f these composite indicators and to the availability of these variables over a long period,
17 Gross domestic product (GDP) measures the total output of goods and services for
final use occurring within the domestic territory o f a given country, regardless of the
allocation to domestic and foreign claims.
18 Gross national product (GNP) measures the total domestic and foreign value added
claimed by residents. It comprises GDP plus net factor income from abroad, which is the
income residents receive from abroad for factor services (labor and capital) less similar
payments made to nonresidents who contributed to the domestic economy.
32
as well as the more obvious correlation of any composite index with structural variables.
Therefore, it is still more appropriate to use real GDP per capita for the purpose of
analyzing patterns of development.
In order to control for differences in purchasing power, we generally use
purchasing power adjusted real GDP per capita (Y)19 as the level o f development.
However, we will also use non-adjusted real GDP per capita (Z)20 in one of the
specifications for each structural variable under consideration for reasons of comparison.
2.d.iii- Controlling for Economies of Scale
It is interesting to note that while N is supposed "to allow for effects of economies
of scale and transport costs on patterns of trade and production,"21 N is defined as the
country's population but not as the country's size in terms of land area or as a country's
population density. Later studies, like Syrquin and Chenery (1989), and even the analysis
o f trade patterns by McCarthy, Taylor, and Talati (1987) have continued to use population
as the variable controlling for country size or effects o f economies o f scale. Therefore, we
have used population as an exogenous variable in our regression analysis. We have also
substituted population with population density (D) in other specifications as population
19 A country's real GDP per capita, adjusted for differences in purchasing power, is
taken from the World Penn Data set, also called the Summers/Heston data set. For a
detailed description of this data set, see Summers and Heston (1988).
20 A country's non-adjusted real GDP per capita is converted into US dollars by using
the market exchange rate as it is published within the IMF's International Financial
Statistics.
21 Chenery and Syrquin (1975), page 17.
33
density is, from a theoretical point of view, a better proxy for economies of scale than
population by itself.
2.d.iv. Controlling for Country Size
In addition to population or population density, we also add a country's land area
as an exogenous variable in trade equations where the endogenous variable is defined as a
ratio of GDP. The same applies to the endogenous export variables expressed as shares of
world exports. In cases where the endogenous trade variables are defined as percentage of
total imports or total exports, it will not be necessary to add the country's land area as an
additional exogenous variable since a country's land area is introduced only to take care of
the fact that large countries tend to trade less than small countries.22 By defining a trade
variable as a share of another trade variable, whether a country's trade share is large or
small, for what ever reason, is already taken into account. The definition of a trade share
in total trade is therefore a more general definition than defining a trade share in GDP and
controlling then for country size.
2.d.v. Controlling for Net Resource Flows
Following McCarthy, Taylor, and Talati (1987), the net resource flow (F) defined
as imports minus exports, has been excluded as an exogenous variable from all equations
with trade shares as endogenous variables. This makes sense as F is basically the same as
the capital account balance, and thus more or less the negative of the current account
22 •
Another reason to control for country size would be to control for natural resource
abundance as is reflected in the Heckscher-Ohlin model, which is however not crucial to
our analysis.
34
balance. Similarly, in countries where exports are highly concentrated in one good, the net
resource flow is closely related to changes in exports of this good. In other words, in
countries with an export concentration in one good, F cannot be considered exogenous to
the share o f the major export good. Even though this may be the case for a few countries
only, it is better not to include F as an exogenous variable in the equations explaining the
major export concentration.
2.d.vi. Problems related to Multicollinearitv
While we continue to use Chenery and Syrquin's (1975) exogenous variables, we
use a consistent set of different specifications, allowing for different combinations o f the
exogenous variables but without running into problems related to multicollinearity. In
other words, the regression specifications will include the exogenous variables, Y and N,
as well as Y2 and N2, though not in one specification.23
23 While the introduction of linear and quadratic terms are often introduced to allow for
some non-linear relationship, there are better alternatives to test for non-linear
relationships. For example, a neat specification for the testing of a U-shaped relationship
which does not need any rearrangement in the data set would be to run the following two
regressions on the whole data set:
LH: In (X-X*) = a + p In (Y-Y*)
RH: In (X-X*) = a + P In (Y*-Y)
where Y* is the value representing the minimum of the U of the independent variable, and
X* is its corresponding value (not the minimum or maximum) of the independent variable.
Specification LH estimates the left-hand part o f the U, while specification RH estimates
the right-hand part of the U. Both specifications can be run with the data of the whole U
since talcing logs excludes automatically the data of the irrelevant part of the U.
35
2.e. Regression Analysis: Linearity vs Non-Linearity
As mentioned earlier, depending on the result of the scatter diagram and the
country-group averages, we have chosen between a linear, log-linear, or U-shaped
regression specification.
2.e.i. Linear Specifications
Though the later analysis may not use a linear specification, because economic
patterns may not be linear (different forms of scatter diagrams will serve as indicator for
which specification to be the more appropriate one), the 8 specifications for a linear
relationship would be LEN1 to LEN8:
LDSTl: X=a+pZ
LIN2: X = a + |3 Y
LIN3: X = a + P Y + yN
LIN4: X= a + P Y +y D
LINS: X = a + P Y +y N + 5 F
LIN6: X= a + p Y+yD +6F
LIN7: X = a + p Y2 + y N 2
LBM8: X = a + P Y2 + y D 2
where X, Y, Z, N, D and F are defined as above.
36
2.e.ii. Log-linear Specifications
In cases where scatter diagrams seem to indicate that the relationship between
structural variable and the level of development to be log-linear, the basic regression
specifications are loglinl to loglin8:
loglinl. lnX = a + P InZ
loglin2: lnX = a + p i n Y
loglin3: lnX = a + P In Y + y InN
loglin4: In X = a + P In Y + y In D
loglin5: InX = a + p l n Y + Y ln N +
loglin6: In X = a + P In Y + Y In D +
logIin7: In X = a + P (In Y) l + y (InN)2
loglin8: In X = a + P (In Y):1+ y (In D)2
where X, Y, Z, N, D and F are defined as above.
In the few cases where the endogenous structural variable (X) can take on positive
or negative values, obviously no log will be taken of the endogenous variable, even though
the right-hand side variables will be specified in log version.24
24 The few cases where the endogenous variable can take on positive as well as
negative values are the fiscal deficit, the current account balance, and the capital account
balance.
37
2.f. Regression Analysis: Cross Country vs Times Series
All the regressions will be based on cross country regressions, using averages of
annual data from 1970-1994, since most time series data are generally not stationary and
the use o f time series data has become a controversial and complicated issue. For example,
while it is a common practice to take first differences to remove most of the trend, this has
led to increased criticism that first-differencing implies an important loss of information.
Furthermore, the method of first differencing is certainly unacceptable or the estimation
o f patterns o f development, as first differencing would turn the
pattern of development analysis into a growth analysis.25
Given this situation and the limited goal of this part o f the dissertation, we can
therefore neglect the analysis of time series regressions. Nevertheless, we can check for
some dynamic aspects o f development patterns through correlation coefficients based on
panel data. Panel-data-based correlation coefficients are therefore a limited fourth tool of
analysis.26
25 While the averaging over 25 years is helpful to detect long-run relationships, it may
also obscure some structural changes, especially in samples of only a few countries. Given
that our sample consists of 93 countries, this seems to be less a problem in our case.
Furthermore, differencing over a long time period, like 1970-75 averages to 1989-94
averages, could be helpful for identifying relationships concerning structural change.
However, as will be seen below, the gains of this additional work may not be worth the
effort for our analysis as the results are quite strong.
26 Note that the correlation coefficients provide some additional information beyond
the coefficients of determination from the cross-country regressions, as the correlation
coefficients are based on panel data observations, while the bivariate regressions are based
on cross sectional averages over the whole time period of 1970-1994. The correlation
coefficient o f cross-country averages would not provide any further information, as it is by
38
2.g. Robustness of Patterns of Development
The various tools of analysis, especially the many different specifications of the
regression analysis will provide a considerable basis for evaluating the robustness of the
results related to the relationship between economic structure (X) and the level of
development (Y). Obviously, we are not concerned about the robustness o f any of the
other exogenous variables, like population, population density, or resourceflow.
Furthermore, we are not anxious about the value of the parameter (b), though we are
interested in the value and sign of the t-statistic o f 3 - In addition, o f some interest are the
coefficients of determination (the R2s) of all 8 specifications. For a regressor to be
considered significant we have required the usual significance level of 95 percent. Of some
interest are also the coefficients o f determination (the R2s). Based on the F-statistic, it is
possible to determine a threshold level of R2, which indicates if a coefficient of
determination is "high" or "low.” Depending on the specific regression specification, this
threshold R2 is between 0.07 and 0.08 for a 99% significance level.
As is well known, there is always the possibility for two kinds of errors in the
testing of a hypothesis. One is that a hypothesis can be true even though the empirical
results do not support the hypothesis, the so-called error of type 1 (or A). As the main
objective o f this analytical part is to show that economic structure matters, there is little to
worry about possible errors of type 1. The goal here is to establish some simple stylized
facts.
definition the square root of the coefficient of determination (R-squared) in a bivariate

regression.
39
The other type of error is that a hypothesis can be wrong even though the
empirical results support the hypothesis (the so-called error of type 2 or B). The
dissertation tries to minimize the possibility for errors o f type B not only by using data
from 93 countries over a time period o f 25 years, but also by requiring a consistency of the
results from the various tools and specifications. This requirement also implies that the
regression analysis may be dropped in cases where the scatter diagram and country-group
averages clearly indicate that the relationship is none of the following four: linear, log-
linear, U-shaped, or inverse U-shaped.
3. Defining Economic Structure
The term "economic structure" is generally used for a wide variety of
characteristics of an economy. The most traditional measures of economic structure are
sectoral shares of the labor force, consumption patterns, and variables measuring income
distribution. All three categories have been analyzed in Clark (1951). Kuznets'
"Quantitative Aspects of the Economic Growth of Nations" examined these three
categories in more detail and added the analysis of sectoral shares of GDP and trade
related variables. As Table 2 shows, Chenery and Syrquin (1975) have added 5 more
categories of variables: investment, government revenues, education, urbanization, and
demographic transition; which amount to a total of 27 economic and social variables.
3.a. Data Constraints of Social Variables
There are considerable problems related to the data quality of the social variables.
The problem related to the quality o f social data is also reflected in the limited data
40
availability. As Table 2 shows, Chenery and Syrquin (1975) had more than 1,000
observations for many of their macroeconomic variables, though they had only 213
observations for the birth and death rate and only 66 observations for their two income
distribution variables.
In addition to the social variables included in Chenery and Syrquin (1975), there is
now a wide variety of other social and institutional characteristics, which are sometimes
included in the term "economic structure.” Examples are fertility rates, central bank
independence, and institutional development. For many of these social and institutional
variables, the data quality and availability have improved considerably over the last 20
years. For example, recently new data sets on school enrollment [Barro and Lee (1996)]
and income inequality (Deininger and Squire (1996)] have been assembled.
However, compared to most o f the macroeconomic variables, there are still
considerable gaps in the quality and availability of social and institutional variables. Given
these data constraints, this dissertation has excluded all these and other social and
institutional characteristics. Instead of relying on debatable social and institutional
variables, we have chosen to concentrate on macroeconomic variables.
Like in the case of neglecting economic policies, this should not be interpreted as
social and institutional characteristics having no impact or are unrelated to development.
Indeed, as will be shown below, macroeconomic uncertainty has a negative impact on
investment, and though we have not analyzed political instability and uncertainty, it is
suspected that economic and political uncertainty are highly correlated to each other.
41
Table 2: Economic and Social Structures Analyzed

in Chenery and Syrquin (1975)
Structural Variable Observations
A. Accumulation Processes
1. Investment (as % of GDP)
a. Gross domestic saving 1,432
b. gross domestic investment 1.432
c. Capital inflow (net import of goods and services) 1.432
2. Government revenue (as % of GDP)

a. Government revenue I, IU
b. Tax revenues 1,111
3. Education
a. education expenditure by government as % of GDP 794
b. Primary and secondary school enrollment ratio 433
B. Resource Allocation Processes

4. Structure of domestic demand (as % of GDP)
a. Private consumption 1,508
b. Government consumption 1,508
c. Food consumption 642
5. Structure of production (as % of GDP)
a. Primary output 1,325
b. Industry output 1,325
c. Utilities output 1.325
d. Services output 1.325
6. Structure of trade (as % of GDP)
a. Exports 1,432
b. Primary exports 413
c. Manufactured exports 413
d. Services exports 413
e. Imports 1,432
C. Demographic and Distributional Processes

7. Labor allocation
a. Share of primary labor 165
b. Share of industry labor 165
c. Share of service labor 165
8. Urbanization
Urban % of total population 317
9. Demographic transition
a. Birthrate 213
b. Death rate 213
10. Income distribution
a. Share of highest 20% 66
b. Share of lowest 40% 66
Source: Chenery and Syrquin (1975), Table 1, page 9.
42
3.b. Data Constraints of Economic Variables
Even after excluding all social and institutional characteristics, in theory, there are
still unlimited possibilities of characterizing a country's economic structure. In practice,
however, the set of variables with which to describe the structure of an economy
consistently across a large sample of countries is quite limited. By "consistently" we mean
to have a nearly complete set of annual data available for the time period of 1970-1994.
This is generally the case for all the macroeconomic variables which are part o f the
national account, but unfortunately not for many other economic variables. The following
8 sub-sections describe the data constraints related to:
(i) differences in relative prices;
(ii) the sectoral distribution o f the labor force;
(iii) dualism, labor market segmentation, and factor market rigidities;
(iv) export supply elasticities;
(v) nominal and real interest rates;
(vi) private and public investment;
(vii) government sector variables; and
(viii) exports o f primary products and machinery.
3.b.i. Differences in Relative Prices
Most o f the structural variables are expressed in percentage shares of GDP, and
were calculated from current domestic currency values in order to avoid any bias related
to the conversion of currencies. There nevertheless remains a problem related to the
43
differences in relative prices. For example, it is well-known that prices for investment
goods are relatively higher in developing than in industrialized countries. This implies that
the real share of investment is actually lower in countries with relatively high prices of
investment goods than in countries with relatively low prices of investment.
Unfortunately, there seems to be little empirical data to correct for this distortion.
As seen in Schydlowsky and Syrquin (1972), one possibility for correcting for differences
in relative prices would be to deflate nominal values by 1 plus their effective rates of
protection which can be regarded as the ratio of the internal and world prices of a unit of
value-added.27 However, it seems likely that the bias caused by these differences in relative
prices actually works in favor o f our analysis, as it is usually the case that prices are low
where the share or sector is large and inversely, prices are high where the share or sector
is small: investment goods are most expensive in countries with low investment shares and
agricultural goods are cheapest in countries with large agricultural sectors. We therefore
do not have too much about distortion due to differences in relative prices.
3.b.ii. Sectoral Distribution of the Labor Force
There are still considerable gaps in the availability o f the sectoral distribution of the
labor force. For example, even the World Bank's "World Data 1995" does not have data
on the sectoral composition o f the labor force beyond 1980. Given our sample of 93
27
A similar approach has also been used by Koss (1983) to explain choice techniques
of production. However, as Schydlowsky and Syrquin point out for the case of
estimations of CES production functions, the error due to differences in relative prices is
common to virtually all studies.
44
countries, this provides 1031 observations and is a considerable improvement over the
only 165 observations in Chenery and Syrquin (1975). However, it would not be
appropriate to use data ranging from 1970-80 to construct an average of data from 1970-
1994. Furthermore, assuming that more developed countries have generally higher levels
o f productivity than less developed countries, there seems to be little need to analyze the
sectoral composition of labor in cases where there is also an analysis o f sectoral shares of
output, for which there are basically no data constraints.
3.b.iii. Dualism. Labor Market Segmentation, and Factor Market Rigidities
Most developing countries are characterized by a dual economy. The term "dual
economy" was originally coined by Boeke (1953) to represent an economy divided
between traditional and modem sectors. The term became popular when Lewis (1954)
used it in his classic model o f development. An economy is dualistic in the classical
definition when a significant part o f it operates under a paternalist or a quasi-feudalist
regime, while another significant part operates under a system of wage employment. In the
Lewis model, decisions in the manufacturing sector are made with the objective of
maximizing profits, while in the agricultural sector the distribution of products is based on
conventional norms rather than on marginal products.
Parallel to the existence o f a dual economy is the segmentation of labor markets.
There are many sources for labor market segmentation. For example, Mezzera (1981) has
shown that imbalances in non-labor markets can lead to labor market segmentation.
Manove and Papanek with Dey (1985) have shown how the concept of tied rents can
45
account for higher reservation wages, thus causing too little migration from the work and
income sharing sector to the commercial sector. While it is generally assumed that a more
developed economy is less dualistic, less segmented and less rigid, it would be interesting
to examine this relationship in a systematic way. Unfortunately, there is not sufficient
comparable data available for such an analysis.
3.b.iv. Export Supply Elasticities
Following Lance Taylor's (1983) textbook Structuralist Macroeconomics, there is
a wide variety o f elasticities which are considered to be important structural
characteristics. A serious attempt has been made to collect data on export supply
elasticities. The result o f this effort is presented in Appendix 2. There are further empirical
estimations of export supply elasticities for the developing countries using single equation
estimations. However, these single equation estimations imply an identification problem,
and subsequently a simultaneous system bias. Appendix 2 reports therefore only those
results which have been estimated using at least two-stage least square estimations.
While the data is not sufficient to include it in our systematic analysis of Chapter 3,
there is some possibility for systematic differences which are proposed as very tentative
questions:
• Are long-run supply elasticities higher the more developed a country is?
• Are long-run supply elasticities higher the larger a country is in terms o f land
area?
46
• Are long-run supply elasticities higher the higher a country’s share of exports
to GDP is?
• Are long-run supply elasticities higher the higher a country’s export product
diversity is?
Given the data constraints, answers to these questions remain to be the analysis of
some future research which would first need to estimate unbiased and consistently defined
export supply elasticities.
3.b.v. Nominal and Real Interest Rates
There is still a considerable gap in the availability of interest rate data, whether real
or nominal, deposit or lending rates. Looking at our 93 countries and time period from
1970-94, the World Bank's World Data 1995 had 1465 observations o f the nominal
deposit rate and 1299 observations for the nominal lending rate. The problem is worsened
by the fact that the gap in data availability is not due to problems related to data
collection, but due to problems related to economic situations, for example economic
crises. The data availability is further reduced for real interest rates, as there are some
additional lacks in the data availability of inflation rates. Based on the analysis of interest
rates in Appendix 2, it has been decided to exclude interest rates from the further analysis.
3.b.vi. Private and Public Investment
As Table 3 will demonstrate, there is a nearly complete set of data available for
gross domestic investment (2274 observations). However, there is considerable gap in the
number of observations for domestic fixed private investment (1011 observations) and
47
domestic fixed public investment (875 observations). A careful analysis of the data gaps
shows that there is no data for domestic fixed private investment for any of the 21
industrialized countries and neither for Argentina, Chad, Congo, Ghana, Guinea-Bissau,
Morocco, Nicaragua, Niger, Rwanda, and Somalia. While the non-availability of data for
these 31 countries implies a loss o f 775 observations, it also implies that the data for the
remaining 62 countries is more complete than what the number initially indicates. The
level o f data availability for the 62 countries with data is on average above 65 percent.
The same applies to data for domestic fixed public investment, where no data is
available for the 21 industrialized countries and neither for 18 developing countries. The
non-availability of data for these 39 countries implies a total loss of 975 observations.
Again, the data for the remaining 54 countries is far from complete, but still at a level of
more than 64 percent. Given these low levels of data availability, the variables would have
been excluded if the dissertation would not be on investment under uncertainty.
3.b. vii. Government Sector Variables
There are generally three issues related to government variables: data availability,
data quality, and the issue of decentralization. Given the high level of aggregation for the
government variables under consideration, the advancement of statistical offices even in
the poorest developing countries, as well as the continued surveillance by the World Bank
and the IMF, neither the data availability nor the data quality is an issue.
However, there may still be a problem of comparability as countries have different
levels o f decentralization and as most government variables are calculated using data only
48
on the central government, which is related to the inadequate statistical coverage of state,
provincial, and local governments. While many developing countries have started to
decentralize their governments over the last few years, most notably Argentina, it is
nevertheless the case that the more developed countries (like the United States, Japan and
Germany) are generally the more decentralized countries.28 Thus, the industrialized
countries' data is more understated than the developing countries' data. We will see below
that industrialized countries generally have larger government sectors than developing
countries. The bias caused by decentralization is therefore not the source of these
differences. Indeed the industrialized country government variables would be larger if local
governments were included.
3.b. viii. Exports of Primary Products and Machinery
The limited data availability for exports, both of primary product and machinery is
not based on economic reasons but on the collection bases of 1970, 1975, 1980, 1985,
1990, 1992, and 1993. Therefore, it is less likely to constitute a selection bias than if the
data would be missing because of economic turmoil. A second reason to include these
variables is that they can easily be expressed as ratios to total exports and avoid problems
related to biases caused by the country size.
3.c. Variables and Areas of Considerations
Given the constraints of data availability and the limited goal o f this part of the
dissertation, the analysis of economic structure is confined to 45 macroeconomic variables
28 Exceptions may be Chile and China.
49
as listed in Table 3. The 45 variables characterizing economic structure are grouped into
10 areas of considerations: the sectoral composition of output, shares of investment to
GDP, shares of savings and consumption to GDP, shares of government expenditures and
revenues to GDP, inflation and money supply, overall trade related variables, import
related variables, export related variables, export product concentration, market power in
world export markets, and financial market development.
The data source for the first seven areas o f considerations is the World Bank’s
World Data 1996, the data source for areas 8 and 9 (export product concentration and
export market power) is the United Nations’ Comtrade Database. The 4 indicators of
financial market development have been calculated from the IMF's International
Financial Statistics.
These 45 variables represent probably one of the most extensive analysis of
economic structure in a world-wide cross-country analysis. With the exception o f the few
variable definitions outlined below, the definitions follow the standard definition (see
Appendix 4) and therefore need no further explanation. The exceptions are the definitions
of:
0) trade openness,
(ii) export product concentration,
(in) export market power, and finally,
(iv) financial market development.
50
Table 3: Variables Characterizing Economic Structure
Area/Variable Acronym Observations
1. Sectoral Composition of Output

a. share of agriculture in GDP (in percent) AGRI 2044
b. share of industry in GDP (in percent) INDUS 2032
c. share of manufacturing in GDP (in percent) MANU 1907
d. share of services in GDP (in percent) SERVICE 2032
2. Investment Shares in GDP (as % of GDP)

a. gross domestic investment INVES 2274
b. gross domestic fixed investment INVFI 2072
c. domestic fixed private investment INVFIPR 10111
d. domestic fixed public investment INVFIPU 8751
3. Consumption and Savings (as % of GDP)

a. total consumption CONS 2273
b. private consumption CONSPR 2249
c. general government consumption CONSGO 2219
d. gross domestic savings SAVI 2273
4. Government Expenditures and Revenues2

a. total government expenditure (as % or GDP) GOVEXP 1566
b. total revenue (as % or GDP) GOVREV 1591
c. tax revenue (as % or GDP) GOVT AX 1595
d. fiscal deficit (as % or GDP) FISDEF 1523
5. Inflation and Money Supply

a. inflation rate (%) based on GDP deflator INFGDP 2321
b. inflation rate (%) based on CPI INFCPI 2070
c. money supply M l (as % of GDP) MON 2209
d. annual growth rate o f M l (percent) MONGR 2209
e. money supply M2 (as % of GDP) MONB 2168
f. annual growth rate of M2 (percent) MONBGR 2168
6. Overall Trade and Import Variables

a. current account balance (as % of GDP) CUAB 2244
b. capital account balance (as % of GDP) CAAB 2174
c. sum o f exports and imports (as % o f GDP) OPEN3 2265
d. total imports (as % o f GDP) IMPPER 2272
e. merchandise imports (as % of GDP) IMPMER 2267
51
Table 3 (cont.)
7. Export Variables
a. total exports (as % of GDP) EXPPER 2274
b. merchandise exports (as % o f GDP) EXPMER 2273
c. exports of machinery (as % o f GDP) EXPMACH 5854
d. exports of machinery (as % o f total exports) EXPMACHEX 5854
e. primary exports (as % of GDP) EXPPRIM 6304
f. primary exports (as % of total exports) EXPPRIMEX 6304
8. Export Product Concentration5

(as % of total exports, TE)
a. share of the major export good in TE EX1EX 1922
b. share of the second major export good in TE EX2EX 1922
c. share of the third major export good in TE EX3EX 1922
d. export product concentration index EXSEX 1922
9. Market Power in World Export Markets6

(as % of the world's export in that good, WE)
a. share of the major export good in WE EX1W 1922
b. share of the second major export good in WE EX2W 1922
c. share of the third major export good in WE EX3W 1922
d. index of market power in WEs EXSW 1922
10. Financial Market Development7

a. the ratio of liquid liabilities to GDP FIN1 2222
b. the ratio of deposit money bank domestic
assets to deposit money bank domestic assets
plus central bank domestic assets FIN2 2166
c. the proportion o f credit allocated to
private enterprises by the financial system FEND 2231
d. the ratio of claims on the nonfinancial
private sector to GDP FEN4 2222
Notes:
1 See chapter n, section 3.b.iv. above for an explanation of data constraints.
2See Chapter n , section 3.b.v. above for an explanation o f data constraints.
3 See Chapter n, section 3 .c.i. below for further definitions.
4 See Chapter n, section 3 .b. vi. above for an explanation of data constraints.
5 See Chapter n, section 3 .c.ii. below for further definitions.
6 See Chapter n, section 3 .c.iii. below for further definitions.
7 See Chapter n , section 3 .c.iv. below for further definitions.
52
3.c.i. Defining Trade Openness
Trade has long been viewed as the engine o f growth; therefore trade openness
becomes an important structural characteristic. Unfortunately, as Learner (1988) has
pointed out, there is no perfect measure o f trade openness. Yet, there are two extreme
definitions of trade openness with a range of definitions between. At the one extreme,
openness is defined as trade intensity, that is, the factual realization o f import and export
shares o f GDP without any corrections for tariffs and other trade restrictions.29 At the
other extreme, trade openness is defined as a measure of trade restrictions without taking
into account the actual size of import and export shares.30
The first extreme definition (trade intensity) has three advantages: availability,
conceptual simplicity, and theoretical purity. However, the major problem of this
definition is a bias caused by trade restrictions, country size, and the availability of
resources. Learner (1988) tried to overcome these problems by using data on supplies of
productive resources and distances to markets to derive an adjusted trade intensity ratio as
29
For example, the Summers and Heston data set has followed this simple definition of
openness.
30 For example, Sachs and Warner (1995) have defined an openness variable which
measures the proportion of years in which an economy is open to trade. An economy is
deemed to be open to trade if it satisfies four tests: (1) average tariff rates are below 40
percent, (2) average quota and licensing coverage o f imports are less than 40 percent, (3)
a possible black market premium is less than 20 percent, and (4) there are no extreme
controls (taxes, quotas, state monopolies) on exports. Besides the fact that it is usually
very difficult to calculate average tariff and licensing coverage rates, the definition also
neglects considerable differences in openness of countries with tariff rates of less than 40
percent. For example, it does not seem to be appropriate to put a country with no tariffs,
no quotas, no licenses, and no black market premium at the same level as a country with
an average of 35 percent tariff rates and 35 percent import quotas.
53
an alternative openness measure. However, even this sophisticated measure o f openness
has earned a variety of criticism.31
The problems o f the second definition are related to (a) the non-availability of data
(especially after the rise in the relative importance of non-tariff barriers), and (b) the
conceptual complexity of how to combine the different trade restrictions with each other.
Learner (1988) has also provided a measure of overall trade interventions, by taking the
difference between the degree o f trade intensity predicted in his model and the actual
degree o f trade intensity. Clearly, this definition depends on the accuracy of the predicted
trade intensity.
The most important contribution of Learner’s analysis is its strict differentiation
between the two definitions of openness: trade intensity and the inverse o f overall trade
intervention. Nevertheless, the most o f the later trade literature has continued to mix up
the two definitions by combining aspects of trade intensity with aspects of trade
interventions. The resulting measure o f such a mix is completely arbitrary. For example, a
large country (in terms of land area) can have low levels of exports and imports expressed
as percentages o f GDP, even though it may have neither tariffs nor other trade restrictions.
On the other hand, a tiny country with high tariff rates may nevertheless have high trade
shares to GDP, simply because of its small size. Depending on how trade intensity and
31 See the comments on Learner's article by Brown (1988). For example, Brown (1988,
page 201) writes that "this procedure will tell us how a country's trade pattern is deviating
from the average trade pattern for countries similarly endowed,” but doubts that this is
really what we mean by openness.
54
trade restrictions are combined with each other, the large or the small country may appear
to be more open than the other one.
In conclusion, it is suggested to confine the analysis to the unadjusted trade
intensity ratio, defined as the sum of export and import shares to GDP. There are three
reasons for this limitation: (a) it is inappropriate to mix trade intensity and trade
intervention measures, (b) there is no objective way to measure trade interventions, and
(c) there remains criticism o f Learner's suggestion to adjust trade interventions.
3.c.ii. Defining Export Product Concentration
An important indicator of economic structure is the degree of a country's export
product concentration, which is (analogous to the Herfindahl index of market
concentration) defined as the sum of the shares of the three major export goods in total
exports o f each country. The major export good o f a country is simply the export good
with the highest share in total export of each country, based on annual data. The second
and third major export goods of a country are the export good with the second and third
highest export shares, again based on annual data. The export good classification is based
on SITC 3 digits, revision 1. The original source, from which this data was necessary to
extract country by country and year by year, is the United Nations Trade Data base.
3.c.iii. Defining Export Market Power
Similar to the definition of a country's export product concentration, it is possible
to derive a country's export market power in world exports. The most simple definition of
such a characteristic would be to express a country's major export good as a percentage of
55
world exports o f that good and year under consideration. However, as for any
concentration measure, the quality of such a measure could be improved by expressing
also the second and third major export goods as a percentage of world exports of those
two goods and by adding up the three shares to achieve a more comprehensive
concentration measure. Again, this has been done country by country and year by year,
based on SITC 3 digits, revision 1.
3.c.iv. Defining Financial Market Development
The recent literature has also given importance to the degree of financial market
development. Following the influential work o f King and Levine (1993) and reiterated in
Levine (1997), there are four indicators of financial market development. The first
indicator (FTN1) measures the size of financial intermediaries. FIN I is therefore defined as
the ratio of liquid liabilities of the financial system to GDP, whereby the liquid liabilities
are defined as currency plus demand and interest-bearing liabilities o f banks and nonbank
financial intermediaries. The second indicator (FTN2) measures the degree to which the
central bank versus commercial banks are allocating credit. FIN2 is therefore defined as
the ratio of deposit money bank domestic assets to deposit money bank domestic assets
plus central bank domestic assets. The third financial market indicator (FIN3) measures
the degree to which credit is allocated to private enterprises. FIN3 is therefore defined as
the ratio o f claims on the nonfinancial private sector to total domestic credit, excluding
credit to money banks. Finally, the fourth indicator (FIN4) measures the size o f the private
56
credit market. It is defined as the ratio of claims on the nonfinancial private sector to
GDP.
One nice property of these indicators is that they can be easily calculated using
data readily available from the IMF's International Financial Statistics, which results in at
least 2166 observations per indicator. Another likeable property of FIN 1 and FIN4 is that
they measure financial market development as a ratio to GDP. This is consistent with the
definition o f most other structural variables.
Rather unusual is the definition of FIN2 and FIN3, as both measure the relative
size of the private sector. As FIN2 measures the relative size o f credit allocated by private
banks and the size o f private banks constitute financial development, FEN2 makes intuitive
sense. However, this is not the case for FIN3, as FIN3 measures the relative size of credit
allocated to the private sector, which is related to the size of the private sector compared
to the size o f the government sector. FIN3 could therefore be interpreted as an indicator
measuring the relative credit activity o f the private sector to the relative credit activity of
the government sector. Consequently, it is difficult to see that FIN3 is an appropriate
measure o f financial market development. We will see below, if FIN3 is nevertheless
correlated to the level of development or if it really should be discarded as a measure of
financial development.
4. Conclusion
As Gottschalk and Smeeding have argued in the case o f cross-national income
inequality, full comparability of economic structures will never be attainable as long as
57
prices, surveys and institutions differ across countries. While these limitations must be
kept in mind, strong patterns emerge out o f these admittedly noisy data. "The issue in not
the existence o f noise, which surely exists in all data sets, but the relative size of the signal
to the noise."32
32 Gottschalk and Smeeding (1997), p. 641.
CHAPTER m
EMPIRICAL RESULTS: PATTERNS OF DEVELOPMENT, 1970-1994
The organization of this chapter follows the ten areas of economic structure of
Table 3. After presenting the results of the patterns o f development from 1970-94 by
topic and variable in the first ten sections, a last section will summarize the results in
the form of stylized facts. However, in order to avoid the repetition of similar results in
each of the 10 sections, it is appropriate to outline some general results.
First, most of the results are consistent across various tools of analysis. For
example, significant parameters (ps) are associated both with high values of coefficients
of determination (R s) from the cross-section regressions and high correlation
coefficients from the panel data. Second, the values and significance of the income
parameters (Ps) are generally consistent across the six or eight different specifications.
Third, the results are consistent across structural variables. For example, the positive
relationship between the level o f development and the shares of manufacturing in GDP
is consistent with the result that LDCs' shares of machinery exports are lower than the
shares of industrialized countries. Fourth, the results are consistent with the results of
the earlier literature as far as they have been analyzed until now.
58
59
Group averages for each of the 45 structural variables are provided in Table 4.
It shows the parallel between clear trends of country group averages and relatively high
correlation coefficient of the panel data. In order to clarify the trends based on group
averages, variables with a consistently positive trend are underlined, variables with a
consistently negative trend are bold, and variables without a clear trend appear in
italics. Table 5 provides the t-statistics, significance levels, and the range o f R2s of the
336 regressions. Further details of the relevant regressions are provided in Appendix 5.
1. Sectoral Composition o f Output
Results of earlier studies demonstrated that the share o f agriculture is decreasing
and that the shares of industry, manufacturing, and services are generally increasing as
development proceeds. All tools of analysis confirm these relationships. The results of
the eight log-linear regression specifications are especially strong, as they provide
always significant ps (income parameters) at the 99% level. The coefficients of
determination vary between 0.81-0.83 for the regression related to the share of
agriculture, and between 0.33-0.43, 0.25-0.45, and 0.47-0.52 for the shares or
industry, manufacturing, and services, respectively. However, two qualifications seem
to be appropriate. First, while group averages and scatter diagrams would allow for an
inverse U-shaped relationship in the case of industry and manufacturing, the cross-
60
Table 4: Group Averages and Panel Data Correlations
lower- upper- correlation

structural low middle middle high coefficient
variable income income income income (panel)
AGRI 36.63 19.24 12.68 4.56 -0.595

INDUS 21.71 31.72 35.211 35.002 0.458
MANU 11.81 16.44 20.87 22.51 0.371
SERVICE 41.66 49.04 50.611 57.782 0.409
INVES 19.37 24.02 24.45 24.33 0.350

INVFI 18.10 22.32 23.60 23.59 0.407
INVFIPR 9.30 14.49 17.09 28.03 0.466
INVFIPU 9.01 7.61 6.36 9.99 0.013
CONS 92.06 79.65 77.841 76.03 -0.289

CONSPR 78.35 66.20 59.32 58.87 -0.387
CONSGO 14.18 13.45 14.00 17.15 0.282
SAVI 7.95 20.32 22.321 24.06 0.289
FISDEF -5.19 -2.73 -4.67 -3.95 0.067

GOVEXP 23.57 21.82 25.83 33.16 0.288
GOVREV 16.79 19.34 22.52 30.79 0.517
GOVTAX 13.96 15.41 18.89 27.59 0.584
INFGDP 53.94 59.29 105.97 7.28 -0.041

INFCPI 84.78 67.18 99.73 7.41 -0.049
MON 18.51 18.22 13.60 23.31 0.104
MONGR 10.24 10.51 12.84 12.00 0.017
MONB 29.55 36.35 41.68 72.04 0.535
MONBGR 11.50 13.75 17.45 12.46 0.020
CUAB -11.92 -5.15 -1.98 -1.11 0.335

CAAB 5.53 3.26 2.12 1.09 -0.173
OPEN 60.07 70.41 66.36 82.13 0.125
IMPPER 36.55 38.15 34.09 41.75 0.046
IMPMER 24.37 25.07 21.14 29.94 0.037
61
Table 4 (coot.)
lower- upper- correlation

structural low middle middle high coefficient
variable income income income income (panel)
EXPPER 23.74 31.88 32.26 40.37 0.192

EXPMER 17.13 22.11 25.38 27.81 0.178
EXPMACH 0.24 1.07 2.50 8.73 0.424
EXPMACHEX 1.35 3.89 8.03 21.57 0.737
EXPPRIM 13.35 14.08 18.35 9.34 -0.184
EXPREMEX 53.54 51.93 51.45 22.57 -0.504
EX1EX 46.15 37.52 31.593 14.73 -0.414

EX2EX 16.85 16.12 11.24 8.05 -0.382
EX3EX 9.15 9.10 6.71 5.93 -0.269
EXSEX 72.15 62.73 54.47 28.72 -0.518
EX1W 5.13 5.95 8.083 9.83 0.155

EX2W 2.78 4.31 7.22 13.82 0.368
EX3W 2.42 2.98 6.68 12.90 0.301
EXSW 10.33 13.24 23.71 36.55 0.386
FIN1 0.27 0.33 0.36 0.64 0.606

FIN2 0.59 0.64 0.69 0.89 0.493
FIN3 0.62 0.72 0.73 0.81 0.109
FIN4 0.18 0.23 0.29 0.59 0.693
Notes:
1 after excluding Gabon and Saudi Arabia, which because o f their large oil
industry, have far too high industry shares, too low service shares, too low
consumption shares and too high savings share, compared to any other upper-
middle-income country.
2 after excluding Ireland, which has an industry sector of 9.9% and a service
sector of 81.22% (please see the explanation below).
3 after excluding Saudi Arabia, which share of oil exports in total exports is over
85%, and constitutes 27.10% o f the world's oil exports.
62
Table 5: Empirical Results of the Structural Regressions

(t-statistics, significance, and range of R s>)
stru ctu ral t - s t a t l s t i c s and s i g n i f i c a n c e range o f
v a r i a b le a d ju sted Ra
l o g l i n l lo g l i n 2 l o g l l n 3 l o g l l n 4 lo g l in S l o g l l n 6 lo g l i n 7 l o g l i n 8 (Mm-MMC)
agri - 2 0 .1 0 - 1 9 .7 3 -1 9 .4 7 - 1 9 .4 3 - 1 4 .4 0 - 1 4 .8 1 - 2 0 .1 7 - 2 0 .1 9 0 .8 0 - 0 .8 2
*** *** *** *** • ** *** *** *•*
Indus 6 .7 0 7 .3 9 7 .1 8 7 .6 0 4 .4 9 4 .5 8 6 .9 0 7 .1 9 0 .3 2 - 0 .4 2
*** **• *** **• »** **» *** *»*
manu 5 .5 7 6 .3 2 6 .2 8 6 .1 3 5 .8 7 4 .64 6 .1 8 5 .8 9 0 .2 5 - 0 .4 3
*** *** *•# *** *** *** *«• **• *•*
serv ic e 9 .2 0 8 .9 9 9 .4 8 8 .7 7 8 .0 0 8 .4 1 9 .3 7 8 .6 1 0 .4 6 -0 .5 0
*** ... **# »** **• ***
in v e s 3 .6 8 3 .8 9 3 .9 6 3 .8 8 4 .79 4 .8 1 3 .7 7 3 .6 5 0 .1 2 -0 .1 9
*** ... ... *** *** *** *#* **• *•*
in v fi 4 .3 3 4 .5 9 4 .6 0 4 .5 1 4 .8 6 4 .8 8 4 .4 0 4 .2 5 0 .1 6 - 0 .2 1
... *** ... *** *** ***
ln v fip r 5 .6 4 5 .1 8 5 .1 3 5 .1 3 4 .3 3 4 .5 3 5 .0 2 4 .6 6 0 .2 8 -0 .3 4
*** • ** *** *** • ** *** «*•
in v f lp u - 1 .8 4 -1 .7 4 - 1 .7 3 -1 .7 0 -0 .6 6 - 0 .9 2 - 1 .6 2 - 1 .6 4 0 .0 1 -0 .0 5
cons - 7 .0 6 - 7 .1 0 - 6 .8 9 -7 .0 9 - 2 .1 0 - 2 .4 6 - 6 .6 9 -6 .7 6 0 .3 2 -0 .5 8
*** »• •* *** »*»
con sp r - 8 .5 7 - 8 .1 5 - 8 .0 8 -8 .6 8 -3 .7 2 - 4 .5 7 -7 .9 4 -8 .3 3 0 .4 0 -0 .5 6
*** *** *•* *** **• *** **• ***
consgo 3 .3 2 2 .4 6 3 .2 1 3 .1 2 4 .4 6 5 .3 5 3 .3 6 3 .3 5 0 .0 5 -0 .3 1
*** ** *** *•* *** *** #*
savl 6 .2 0 6 .9 7 6 .9 0 6 .91 3 .3 1 3 .3 3 6 .61 6 .6 2 0 .3 0 -0 .4 3
*** ... *** **# *** #** *** ttt
fisd e f 0 .9 5 0 .9 3 0 .8 6 1 .01 - 2 .2 3 - 1 .8 9 0 .9 3 0 .8 6 0 .0 0 -0 .1 7
** *
govexp 4 .1 5 3 .5 2 4 .4 3 3 .7 3 5 .9 9 6.41 4 .4 9 3 .8 3 0 .1 1 - 0 .3 6
*•* *•* *•* t«t • ** *** »** *** ***
govrev 7 .1 5 6 .4 4 7 .3 5 6 .5 3 6 .5 3 6 .94 7 .3 4 6 .5 0 0 .3 1 - 0 .4 1
*** *** *** *#* • ** • ** »•* ***
g o v ta x 8 .3 0 7 .0 7 8 .0 9 7 .1 3 6 .7 9 7 .19 8 .1 9 7 .2 3 0 .3 5 -0 .4 5
• ** **• *** *** *** *** **• *•** ***
63
Table 5 (cont.):
stru ctu ral t - s t a t 1 s t i e s and s 1 g n 1 f 1 c a n c e range o f
v a r i a b le a d ju s te d R3
l o g l i n l l o g l i n 2 l o g l l n 3 lo g l l n 4 lo g lin S lo g l in 6 l o g l i n 7 l o g l i n S (Hn-Mu)
in fg d p - 1 .8 4 - 1 .3 4 -1 .3 7 - 1 .0 5 - 0 .4 3 - 0 .5 3 -1 .5 4 -1 .1 1 0 .0 0 - 0 .0 8
in fcp i - 1 .9 3 -1 .4 5 -1 .5 1 - 1 .2 7 -0 .3 9 -0 .6 0 - 1 .6 7 - 1 .3 2 0 .0 0 - 0 .0 6
*
non 2 .1 5 1 .8 4 1.6 4 1 .6 6 3 .1 7 2 .58 1 .7 2 1 .7 3 0 .0 2 - 0 .1 1

** **
mongr 2 .4 3 2 .1 1 2 .0 5 2 .0 6 1.3 8 1 .3 6 2 .0 7 2 .0 4 0 .0 1 - 0 .0 6
** ** *• ** •* **
stonb 8 .6 6 9 .2 4 9.10 9 .3 1 9.7 1 9.00 9 .1 3 9.1 1 0 .4 4 - 0 .5 8
*** *** ••• *** • ** *** *•* *** ***
monbgr 1 .8 0 1 .8 4 1.67 1 .7 2 0 .6 5 0.42 1 .5 5 1 .5 3 0 .0 2 - 0 .0 5
cuab 6.7 9 7 .4 3 7.90 7 .2 3 7 .7 6 7.0 2 0 .3 2 - 0 .5 5

**• *** *** **• *** »**
caab - 4 .1 5 -4 .4 4 -4 .2 1 - 4 .2 2 - 4 .2 9 -4 .2 0 0 .1 4 - 0 .2 6
• ** • ** *** *** *** • **
open 2 .3 2 1 .8 6 3.24 3 .2 4 3 .2 4 3.2 4 0 .3 1 - 0 .5 3

** * *** *** • ** *** **#
impper 0 .6 5 0 .1 7 1.37 1 .3 7 1 .2 9 1 .3 0 0 .3 2 - 0 .5 7
**«
Inpmer 0 .5 2 0 .2 4 1.02 1 .0 2 1 .0 2 0 .9 7 0 .3 9 - 0 .5 4
• **
expper 4 .2 5 3 .8 1 5.07 5 .0 7 4 .8 6 4 .7 0 0 .3 2 - 0 .4 7
• ** *•* tit *•» *** *•* ***
expmer 3 .8 3 3 .2 5 4.42 4 .4 2 4 .2 3 3 .9 9 0 .2 1 - 0 .3 9
*** *** • ** *** *»* • **
expmach 1 0 .0 2 9 .4 9 9.32 9 .3 2 9 .54 9.39 0 .5 0 - 0 .5 5
• ** *•* **• »*» »** *** ***
expmachex 8 .1 3 7 .9 5 8.27 7 .7 7 8 .5 0 7.89 0 .4 1 - 0 .5 2
*** *** *** »** ***
expprlm - 1 .6 0 -1 .6 9 -1 .1 3 - 1 .1 3 - 1 .3 5 - 1 .0 8 0 .0 2 - 0 .3 8
exprlm ex - 5 .6 2 -5 .3 7 -5 .2 6 -5 .4 3 - 5 .4 1 -5 .3 7 0 .2 3 - 0 .4 3
*** •* * *•* *•* *** *** ***
64
stru ctu ral t - s t a t 1 s t i c s an d s i g n 1 f 1 c a n c e ran ge o f

v a r ia b le a d ju s te d RJ
l o g l i n l l o g li n 2 l o g l i n 3 l o g l i n 4 l o g l i n 5 l o g l i n 6 l o g l i n 7 l o g lin S (Mat—Hue)
ex lex - 7 .7 9 -7 .6 6 - 7 .9 5 -7 .5 6 -7 .3 0 - 5 .8 0 -8 .1 6 -7 .5 5 0 .3 9 - 0 .5 3
*** •** *** • ** **# *** *** *** ***
ex 2 ex - 5 .9 1 -6 .1 5 - 6 .2 8 -5 .9 4 -5 .0 8 - 3 .8 3 - 6 .4 9 -6 .0 4 0 .2 7 - 0 .4 4
*** *** *** *** *** *** *** *** +*#
ex 3 ex -3 .1 2 -3 .5 6 - 3 .4 1 -3 .4 1 - 2 .3 2 - 2 .1 7 -3 .5 2 -3 .4 7 0 .0 8 - 0 .1 3
*** *** • ** *** • ** *** ***
exsex - 8 .8 8 -8 .9 1 -9 .7 2 -8 .9 5 -8 .9 7 - 6 .8 3 -10.05 -8 .9 4 0 .4 6 - 0 .6 3
*** *** •** • ** *** *** *•* *** ***
ex lw 4 .2 6 4 .77 4 .5 2 4 .5 2 4 .4 9 4 .22 0 .2 7 - 0 .4 6
**• *** *** *** **• *** * +*
ex2w 8 .1 6 9 .16 1 0 .0 8 1 0 .0 8 5 .9 1 5 .9 1 9 .9 4 9.59 0 .4 6 - 0 .7 2
*** *** *** *** *** *** »** • *» ***
ex3w 8 .7 2 9.50 1 0 .6 7 1 0 .6 7 6 .70 6 .70 1 0 .5 5 10.23 0 .5 2 - 0 .7 4
*** *#* * •* *** *** *** *#+
exsw 7 .1 2 7.90 8 .4 6 8 .4 6 4 .5 5 4 .5 5 8 .3 7 8 .14 0 .4 3 - 0 .6 9
*** *** *** *** *** *** *** ***
ffin l 9 .2 2 9.54 9 .4 1 9 .7 5 8.84 8 .5 0 9 .5 3 9.58 0 .4 7 - 0 .5 9

**• *** *** **• *** *** *** ***
fin 2 5 .5 7 5 .26 5 .2 3 5 .2 5 3 .5 2 3 .7 3 5 .3 6 5.34 0 .2 1 - 0 .2 6
*** *** *** • ** »** *** *** * +*
fin 3 2 .4 0 2 .02 2 .1 6 2 .1 0 1 .5 5 1 .8 3 2 .1 5 2 .02 0 .0 1 - 0 .0 5
** •* ** ** * ** **
f in 4 9 .9 7 9.33 9 .1 2 9 .1 1 8 .1 1 7 .7 6 9 .1 1 9.00 0 .4 8 - 0 .5 2
*** *** *** *** *** *** *** *•* ***
Notes: * significant at the 90% level

** significant at the 95 % level
*** significant at the 99% level
65
country regressions did not confirm such a relationship. However, time-series data of
the most advanced countries seem to indicate that an inverse U-shaped relationship may
become relevant in the future.
Second, we have to consider the impact of outliers. The first scatter diagram of
Figure 2 demonstrates that there is very little deviation from the best fitted line,1
indicating a negative relationship between agriculture and development. On the other
hand, the other three scatter diagrams o f Figure 2 show that there is some deviation
from the general trend for the sectoral shares of industry, manufacture, and services.
There are three outliers: Gabon, Saudi Arabia, and Ireland. Gabon and Saudi Arabia’s
industrial shares (53% and 66%, respectively) are extremely high, compared to the
other upper-middle-income countries. Such high industry shares of Saudi Arabia and
Gabon are due to large petroleum industries. Given these high industry sectors, the
manufacturing and service sectors are subsequently too low compared to other middle-
income countries. Ireland’s share o f industry is inconsistent with any other industrial
country. Excluding Ireland, the industrialized countries sectoral shares of industry are
within the range of 28% (Netherlands) and 42% (Japan). Ireland's average share of
industry from 1970-94 is less than 10%. Accordingly, Ireland's manufacturing sector is
also very low (less than 3.5%), while its service sector is very high (higher than 81%).
1 The best fitted line of the scatter diagrams is always just based on a simple
regression of the log of a structual variable on the right-hand side and a constant and
the log of GDP per capita on the left-hand side of the equation. It does not control for
any of the other explanatory variables.
66
Figure 2: Scatter Diagrams of Sectoral Shares of GDP
4.5 — Saudi
. Arabia
4.0- Gabon
3.5-
3.0-
2.5-
* Ireland
2.0
LOG(GDPCAP)
4.0 4.5
* Ireland
3.5 -
3.0 - 4.0-
- Gabon
2 .0 -
Gabon 3.5- i
* Saudi Arabia ■ Saudi
I Arabia
. Ireland
1.0 3.0
LOG(GDPCAP)
2. Investment Shares in GDP
While there is some evidence of a positive relationship between the level of
development and the ratio of gross domestic investment (total, fixed, or private) to GDP,
there is no indication of any systematic relationship between the level of development and
the ratio of public investment to GDP. Chenery and Syrquin (1975) analyzed gross
domestic investment as a share to GDP, and Syrquin and Chenery (1989) analyzed the
67
share of private investment to GDP. Both studies came to the same conclusion o f a
positive relationship between investment shares and GDP per capita. Within our
analysis, the positive relationship is the strongest between private fixed investment and
development, though it must be taken into account that there is no comparable data
available for the industrialized countries. Let's analyze the results of each of the four
tools of analysis in more detail.
2.a. Scatter Diagrams
The first three scatter diagrams of Figure 3 reflect the positive relationships
between development and the ratio of gross domestic investment to GDP for total, total
fixed, or fixed private investment. The last scatter diagram of Figure 3 also shows that
there is no relationship between the level of development and public fixed investment.
2.b. Country Group Averages
The group averages of lower-middle, upper-middle, and high-income countries
are very close to each other for the ratios of gross domestic investment to GDP (24.02,
24.45, 24.33) as well as for the ratios of domestic fixed investment to GDP (22.32,
23.60, and 23.59). In both cases, it is the group of upper-middle-income countries,
which has the highest group average. This casts some doubt on a strictly positive
relationships and indicates inverse U-shaped relationships. For the shares of fixed
private investment to GDP, the country group averages are fully consistent with a
positive relationships between the ratio of fixed private investment to GDP and GDP
68
per capita. The group averages for the ratio of fixed public investment to GDP are
consistent with the scatter diagram of no systematic relationship.
Figure 3: Scatter Diagrams of Investment Shares in GDP
4.0 4.0
3.5 3.5
<n
3.0 3.0
25 25
20 20
4 6 S 10
LOG(GDPGAP) LOG(GOPCA P )
4 3.5
3.0 -
3
s
& 25
u_
2 i
I i? “ ■
t
1.5 -
0
LOG(GDPCAP) LOG(GDPCAP)
2.c. Cross Country Regressions
Clearly, the strongest results come from the 8 different specifications regressing
development on the four investment shares. While the Ps are never significant for
public fixed investment, they are significant at the 99% level for all 8 different
specifications of gross domestic investment, for all 8 specifications of domestic fixed
investment, and for all 8 specifications of fixed private investment.
69
2.d. Coefficients of Determination and Correlation
The trends and degrees of relationships are confirmed by the relative size of the
R2s, as they are always below 0.08 for fixed public investment, between 0.12 and 0.21
for gross domestic investment, between 0.17 and 0.23 for domestic fixed investment
and between 0.30 and 0.35 for fixed private investment. The levels of R2s are matched
by corresponding levels of correlation coefficients from the panel data (see Table 4).
Finally, as will be seen in Chapter VI below, regressions which include macroeconomic
uncertainty as explanatory variables push up the R2s considerably, without decreasing
the significance of the income term.
3. Savings and Consumption
3.a. Savings
The strong positive relationship between development and private investment is
paralleled by a robust positive relationship between development and savings. All tools
of analysis are consistent with each other. For example, the Ps o f the eight different
specifications are always significant at the 99% level and the R2s vary between 0.30
and 0.45. Finally, saving shares have also been analyzed by Kuznets (I960 and 1961)
and by Chenery and Syrquin (1975), who came to the same conclusion of a positive
relationship between saving shares and GDP per capita.
2 , 2
The highest R are obtained for specifications controlling for the net resource
inflows. The lowest R2 are obtained for the bivariate specifications.
70
Figure 4: Scatter Diagram of the Shares of Savings in GDP 3
oP
3 o
0 0 9
2 00
1
4 6 8 10
LOG(GDPCAP)
3.b. Total and Private Consumption Shares of GDP
Given the robust positive relationship of development and savings, and the
positive relationship of development and investment (shown in the previous section), it
is not necessarily surprising but consistent to find robust negative relationships between
3 Note that the scatter diagram reflects that the log o f a log-linear specification may
provide an even better fit; however, we have preferred to keep the methodology
consistent as it does not make a difference for our purpose of examining simple
relationships.
71
development and shares of total4 and private consumption to GDP. All tools of analysis
(scatter diagrams o f Figure 5, country group averages of Table 4, and regression results of
Table 5) confirm such negative relationships. Moreover, the result is consistent with the
results o f earlier studies, particularly those of Engel (1857), Houthakker (1957), and the
two studies by Chenery and Syrquin. Finally, the result is consistent with the Keynesian
consumption function and the limited ability to smooth consumption due to credit
constraints.
Figure 5: Shares of Total and Private Consumption in GDP
5.5 5.0
Lesotho
5.0. Lesotho 4.5
so
o 4.5. 4.0.
S'
o_I
4.0- 3.5-
3.5 3.0
4 6 8 10 4 6 8 10
LOG(GDPCAP) LOG(GDPCAP)
4 Analog to the footnote o f Figure 7, Gabon and Saudi Arabia's shares of total
consumption are artificially pushed down (Gabon: 51.9%, Saudi Arabia: 57.8%) and are
therefore excluded from the country group averages. The four countries with consumption
shares of more than 100% are included in the low-income group as their exclusion would
not change the result significantly. For example, excluding Lesotho's total consumption
share o f 159% would only push down the low-income country group average by 1.76%
(from 92.06 to 90.30%).
72
3.c. Share of Government Consumption to GDP
The country group averages of Table 4 do not allow us to establish any
systematic relationship between the level o f GDP and the share of government
consumption. Again, this is consistent to the two earlier studies (Chenery and Syrquin
(1975) and Syrquin and Chenery (1989)).
4. Government Expenditures and Revenues
4.a. Government Expenditure
The average shares of government expenditure in GDP are 23.6% for low-
income countries, 21.8% for lower-middle-income countries, 25.8% for upper-middle-
income countries, and 33.2% for high-income countries. While these averages suggest
that there is a considerable difference between developing and industrialized countries,
the value for the group of low-income countries (23.6%) is slightly higher than the
value for the group of lower-middle-income countries (21.8%) and does therefore not
allow to draw the conclusion of an overall positive relationship between the share of
government expenditures and the level of development. However, a careful analysis of
the group of low-income countries shows that three low-income countries bias the
average o f the low-income countries upward and that the averages indicate a positive
relationship after excluding these three countries.5 The regression results based on all
5 The three countries are Mauritania (41%), Egypt (45%), and Guinea-Bissau
(52%), which have higher shares of government expenditures than any other
developing country. Excluding these three countries, the group average for the low-
income countries would be 21.66%.
73
93 countries confirm the positive relationship, though the t-statistics are not always
significant at the 99% level as long as the three outliers are included. Overall, we
conclude that there is some weak support for the hypothesis that "Wagner1s law"6 holds
even across countries.
4.b. Government Revenue
While there was only a weak positive relationship between development and the
share of government expenditures, the data on the shares of government revenues
demonstrate a robust positive relationship between the degree of development and the
share of total government revenue in GDP. This positive relationship is even stronger
between GDP per capita and the share of tax revenues in GDP. The average of the
poorest countries is just about half of the average of the industrialized countries (14%
versus 28%). There is of course some variation. Still, the high correlation coefficients
and the regression results7 are sufficient to establish stylized facts that total government
and tax revenues are both positively related to increases in GDP per capita. Our results,
which are based on data from 1970-94, are also consistent with the only other major
6 A hypothesis, advanced in the late nineteenth century by the German economist

Adolph Wagner, which stated that the development of an industrialized economy would
be accompanied by a rising share o f public expenditures in GDP. Neither Kuznets nor
the two studies by Chenery and Syrquin have analyzed the share o f government
expenditures.
7 The Ps of all eight regression specifications are always significant at the 99% level
for the share of total government revenues as well as for the share o f tax revenues.
74
cross-country study (Chenery and Syrquin (1975)) analyzing the relationship between
government revenues and GDP per capita.
Figure 6: Government Expenditures, Government Revenues, and Fiscal Deficits
4.c. Fiscal Deficit
Considering the stylized fact that LDC governments have lower shares of
government revenues to GDP, it is tempting to assume that there is also a systematic
relationship between the level of GDP per capita and the share of the fiscal balance to
GDP. However, the data does not enable us to draw such a conclusion. The 25-year
averages (1970-94) are 5.19%, 2.73%, 4.667% and 3.95% for the low-income, lower-
middle-income, upper-middle-income, and high-income countries, respectively. The
scatter diagram of Figure 6 also reflects the fact that there is little or no systematic
relationship between the fiscal deficit and the level of development. What is indeed
75
surprising is that it is the group o f lower-middle-income countries which has the lowest
share of fiscal deficits to GDP and that there is a difference of more than 1.2% to the
average of the high-income countries. None of the earlier studies o f development
patterns has either analyzed or reported a relationship between GDP per capita and the
share of the fiscal balances to GDP.
5. Inflation and Money Supply
5.a. Inflation
While the country group averages of Table 4 indicate that the three low- and
middle-income groups have 10 times higher inflation rates than the high-income group,
the variation within and across the three low and middle-income groups indicates that
there is no systematic relationship. Any comparison of inflation rates across a large
number of countries is complicated by the fact that many developing countries
experienced very high inflation rates at some point of time which distorts the average
inflation rate. Argentina, Bolivia, Brazil, Nicaragua, Peru, and Zambia all experienced
at some point of time inflation rates of more 1,000% per year.8 However, even after
excluding these six hyperinflationary countries, average inflation rates remain distorted
8 This last fact is independent o f calculating inflation rates based on the GDP
deflator or the CPI. Considering the income level of these six countries, wee see that
two are upper-middle-income countries (Argentina and Brazil), two are lower-middle-
income countries (Bolivia and Peru) and two are low-income countries (Nicaragua and
Zambia). Interestingly, five o f the six countries are in Latin America. There have been
some explanations o f Latin America's affinity to high inflation based on institutional
characteristics and macroeconomic populism, see for example, Dombusch and Edwards
(1990, 1994), Fisher (1990), and Sachs (1989).
76
and the hypothesis that inflation rates fall as development proceeds can neither be
confirmed nor rejected. 9
5.b. Money Supply
5.b.i. Money supply broadly defined fM2)
There is strong evidence of a positive and robust relationship between money
supply broadly defined (measured as a ratio to GDP) and GDP per capita. This is
evident by the following tools: the scatter diagram (Figure 7), the country group
averages (29.55, 36.35, 41.68, 72.04), the Ps and R2s of the regression analysis,10 or
the correlation coefficient based on panel data (0.535).
5 b.ii. Other monetary variables
The data does not indicate any systematic relationship between GDP per capita
and the growth rate of M2, the share of money supply (M l), or the growth rate of M l.
While some of the country group averages allow for some kind of a U-shape
relationship, neither the scatter diagrams (Figure 8) nor the appropriate regression
analysis support such a relationship as there is too much variation within each of the
9 Using the limited data availability for the inflation rate based on the CPI, 33
developing countries have experienced an annual inflation rate of at least 40%. Using
inflation data based on the GDP deflator, 38 developing countries have experienced an
annual inflation rate o f at least 40% at some point during 1970-94.
10 All eight Ps are statistically significant at the 99% level and the coefficients of
determinations are within the range o f 0.45 to 0.59.
77
four country groups. Neither inflation rates nor monetary aggregates have been
analyzed in any o f the previous pattern o f development studies.
Figure 7: Scatter Diagram of the Share of M2 in GDP
LOG(GDPCAP)
Figure 8: Scatter Diagrams of Other Monetary Variables
4.5 4.0 4.0
4.0 3.5
3.0 3.0
IS
5 15
3.0
2.0
15
2.0 t.O
78
6. Overall Trade and Import Variables
6.a. Current and Capital Account Balances
While current and capital account balances have not been analyzed directly
within the literature related to patterns of development, the country group averages of
provide some evidence for the literature on stages in the balance o f payments." Table 4
shows that current account balances are positively related, while capital account
surpluses are negatively related, to GDP per capita. The group averages of current
account balances are -11.92%, -5.15%, -1.98%, and -1.12% for the low-, lower-
middle-, upper-middle-, and high-income-countries, respectively. The corresponding
shares o f capital account surpluses are 5.53%, 3.26%, 2.12%, and 1.09%.
An initially perplexing observation is that the averages of all four country
groups imply current account deficits and capital account surpluses. There are three
explanations for this observation. First, as our sample contains only 93 countries, it is
possible that the excluded countries would make up for the gaps. A second, more
plausible explanation is that the current account balances are expressed as shares of
each country's GDP without weighing these shares by the relative size of GDP. This
implies that one large exporter can compensate for many small importers.12 Finally, it
11 See Fischer and Frenkel (1974) for an early analysis.

12 For example, assume that Germany is a net exporter to Belgium, the Netherlands,
and Luxembourg (the BENELUX countries) and that Germany's trade surplus is 3%
expressed in Germany's GDP. Given the fact that Germany's GDP is about 3 times the
sum o f the BENELUX'S GDP, this would allow trade deficits of about 9% for each of
the three BENELUX countries. However, adding up three times current account
79
is well known that when official current account figures for all nations are added up,
the result is — because of statistical and measurement problems — a current account
deficit for the world. 13
The scatter diagrams of Figure 9 also show a positive relationship between GDP
per capita and current account balances, and a negative relationship between GDP per
capita and capital account balances.
Figure 9: Shares of Current and Capital Account Balances in GDP
20-
i -2 0 .
cn 10-
•40 -
-60 -10
4 6 8 10
LOG(GDPCAP)
Six semi-log regression specifications (similar to loglinl - loglin4 and loglin7 -
loglin8) provide consistently significant 3s at the 95% level.14 The R2s vary between
deficits of 9% with one current account surplus of 3% would give an average of current
account deficits of 6% [ 3x(-9) + 3 = -24; -24/4 = -6].
13 This seems to be the main reason. See the IMF's (1987) Report on the World
Current Account Discrepancy for more details. As Abel and Bemanke (1995, p. 153,
Box 5.1) illustrate: "IMF projections for 1993 were that industrial countries would
have a collective $35.6 billion current account deficit, developing countries would have
a $52.9 billion deficit, and former centrally planned economies would have a $24.6
80
0.33-0.56 for the current account, and between 0.15-0.27 for the capital account. The
difference between the goodness of fits between the current and capital account variable
is also visible in the scatter diagrams of section 11. Nevertheless, there is sufficient
evidence that current account deficits and capital account surpluses decrease at higher
income levels. The United States is one o f the exceptions to this general observation.
6.b. Openness Defined as Trade Intensity
As mentioned in chapter n, section 1 above, Agenor and Montiel (1996) show
that "developing nations tend to be substantially more open than the major
industrialized countries,” 15 whereby openness is defined as the sum of import and
export shares in G D P.16 Using the same definition of "openness,” our empirical results
of 21 industrialized and 72 developing countries suggest that developing countries are
less open than industrialized countries, especially after controlling for country size.
Even without controlling for country size, the country group averages (60.07%,
70.41%, 66.36%, and 82.13%) tend to indicate a rather more positive than negative
billion deficit, all of which adds up to a current account deficit for the world as a whole
of $113.1 billion."
14Note that the regression specifications are semi-log, as we take the log of the
exogenous variables, but not o f the (positive and negative) endogenous variables. Note
also that we have not run specifications similar to LOGLIN7 and LOGLIN8 as they
contain the net resource flow as an exogenous variable.
15 Agenor and Montiel (1996), p. 16.
16 The issues related to the definition o f openness were discussed above in chapter
2, section 3.c.i.
relationship. In order to control for country size, we added the country's land area in
square kilometer as an additional exogenous variable to all 6 specifications:
loglinl': In X = a + P In Z + s In CS
loglin2': In X = a + 3 In Y + s In CS
Ioglin3': In X = a + P In Y + y ln N + 8 In CS
loglin4': In X = a + p ln Y + y In D + e In CS
loglin7': In X = a + p (ln Y )2 + y (ln N )2 + e In CS
loglin8': In X = a + P (In Y )2 + y (In D )2 + 8 In CS
where X, Y, Z, N, D and F are defined as above, and
CS is country size in square kilometer (km2).
The first observation is that the parameter of CS is negative and significant at
95% (or higher) for all six specifications. Second, whenever the specifications have
included population or population density, their parameters are also always statistically
significant at the 99% level. Lastly, Z and Y are also significant at the 95% level. The
exception is specification loglin2', where Y is only significant at the 90% level. The
coefficients of determination vary from a minimum of 0.333 (of specification loglin2')
to a maximum of 0.540 (of specifications loglin3' and loglin4'). In conclusion, there is
ample evidence of a positive relationship between trade intensity and the level of
development. Excluding the limited comparison by Agenor and Montiel (1996), there
does not seem to be any previous literature analyzing this important and robust
relationship between trade intensity and development. As was mentioned in Chapter I,
82
there are however some controversial studies which have tried to assess the impact of
openness on growth. O f interest for this dissertation is a study by Dani Rodrik (1996),
who has shown a positive and robust partial correlation between openness (measured by
the share of trade in GDP) and the scope of government (as measured by the share of
government expenditure in GDP). Rodrik even provides some evidence that the
causality runs from openness to the scope of government, which he explained with
larger governments serving as absorbers of external shocks.
6.c. Total Import and Merchandise Import Shares in GDP
Neither total import shares nor merchandise import shares in GDP seem to
follow any systematic (linear, log-linear or U-shaped) relationship to GDP per capita.
Though not shown, the scatter diagrams look like a circles with observations more or
less equally distributed within the circle. Looking at the movement of country group
averages from low to high-income groups, they first increase by about 5%, then drop
by about 10%, but then rise sharply by more than 20%. Moreover, none of the six
usual regression specifications provide any significant p. Even after correcting for
country size, the Ps remain insignificant for all 6 specifications. All this is consistent
with low correlation coefficients of 0.05 for total import shares and 0.04 for
merchandise import shares in GDP. While our results are also consistent with
inconclusive results of Chenery and Syrquin (1975) and Syrquin and Chenery (1989),
they contradict the results of McCarthy, Taylor, and Talati (1987), henceforth MTT,
83
who have found support for a positive relationship between the share of imports to
GDP and GDP per capita.
7. Export Shares
The first two export shares we analyze are analogous to the two import shares:
total exports in GDP and merchandise exports in GDP. Furthermore, there is some data
for the analysis o f machinery exports and primary exports, expressed as either a share
in GDP or a share in total exports. As far as these patterns have been analyzed in the
past, our results are generally consistent to the earlier literature. 17
7.a. Total Export and Merchandise Export Shares in GDP
In contrast to the result derived from the import shares, all tools of analysis
show that there are positive relationships between the level of development on the one
hand and, and the shares of total exports in GDP and merchandise exports in GDP, on
the other hand. The relationships are evident even when not controlling for country
size, though the adjusted Rrs are generally higher in the six specifications including
country size. The Ps are always significant at the 99% level. Even the country group
averages, which do not control for country size show a positive relationships between
the share of exports to GDP and GDP per capita.
17 The exception is Syrquin and Chenery (1989), who could not establish a clear export
pattern, while Kuznets (1964 and 1967), Chenery and Syrquin (1975), and especially
MTT showed that there is a strong relationship.
84
Figure 10: Shares of Total and Merchandise Exports in GDP
6 5
5
4
4
3
3
2
2
LOG(GDPCAP) LOG(GOPCAP)
N ote te a t te e s a sc atter d i g r a m s do no t reflect the robust

reteflonsN ps betw een GO P p e r capita and te e two export
s h a r e s v ery w e le s te e y do n o t control for c o w try itte .
7.b. Exports of Machinery
We analyze two different shares of machinery exports: (i) the share of
machinery exports in GDP (EXPMACH), and (ii) the share of machinery exports in
total exports (EXPMACHEX). For each of these two variables, we run two times six
regressions: specifications loglinl-loglin4 and loglin7-loglin8 (without controlling for
country size) and specification loglinl'-loglin4' and loglinT-loglin8' (including country
size as an exogenous variable). All tools of analysis provide strong support for the
hypothesis that the share o f machinery exports increases as the level of GDP per capita
increases. Very convincing is the fact that all 24 regressions testing such a hypothesis
provide significant ps at the 99% level.
85
Figure 11: Scatter Diagrams of the Share of Machinery Exports in GDP,

the Share of Machinery Exports in Total Exports, and the
Share of Primary Exports in Total Exports
•2 . -2-
LOG(GOPCAP)
7.c. Exports of Primary Products
Analogous to the exports of machinery, we analyze two different shares of
primary exports: (i) the share of primary exports in GDP (EXPPRIM), and (ii) the
share of primary exports in total exports (EXPRIMEX). We use specifications loglinl'-
Ioglin4' and loglinT -loglin8' which are controlling for country size in the regressions
for the share of primary exports in GDP, and specifications loglinl-loglin4 and loglin7-
loglin8 which do not control for country size in the regressions for the share of primary
exports in total exports.
In contrast to MTT, but consistent to the other literature, we conclude that
there is no systematic relationship between the level of development and the share of
primary exports in GDP (EXPPRIM). The differences between our and MTT’s analysis
are that (a) we control for country size while MTT do not, and (b) our sample consists
86
of 72 developing countries and 21 industrialized countries, while MMT’s sample
consists of only 55 developing countries. On the other hand, we find strong evidence
for a negative relationship between the level of development and the share of primary
exports in total exports (EXPRIMEX).
Three remarks are appropriate. First, the result is intuitive and consistent with
the results derived for the sectoral distribution of GDP in section 1 of this chapter.
Second, it makes even more sense to analyze the share o f exports in total exports than
to analyze the share of exports in GDP, as the latter may be biased by country size. It
is therefore interesting that none of the earlier literature has analyzed the share of
primary exports in total exports. Third., please note that there is no contradiction of
results with regards to EXPPRIM and EXPRIMEX within our analysis. This is because
the industrialized countries' higher total export shares imply that the shares of primary
exports in total exports are considerably lower than for LDCs. Table 6 explains this
seemingly contradiction with the real world examples of Canada and Chad. Canada and
Chad have about the same share of primary export in GDP (9.53% vs 9.72%).
Canada's total export share in GDP is over 25%. Thus, Canada's share of primary
exports in total export is only 35%. However, as Chad's total export share in GDP is
less than 19%, Chad's share of primary exports in total exports is more than 53% .18
18 Note that the example is not biased by country size, as Canada (9,220,000 km2) is
about 7 times larger than Chad (1,259,000 km2).
86
87
Table 6 : A Comparison of Primary Export Shares
Share Canada Chad
Total Exports in GDP 27.19 18.28
Primary Exports in GDP 9.53 9.72
Primary Exports
in Total Exports 35.07 53.16
8. Export Product Concentration fEPC)
The panel data correlation coefficient between EPC and GDP per capita is
-0.517. The sum of the shares o f the three major export goods in total exports is
72.15%, 62.73%, 54.48%, and 28.72% for the tow-income, lower-middle-income,
upper-middle-income, and high-income countries, respectively. The regressions provide
significant Ps for all 8 specifications at the 99% level, and the R2s vary between 0.47
and 0.64. It is therefore concluded that there is a negative relationship between a
country's level of development and a country's export product concentration.
This stylized fact is very robust, as it holds even for each of the three main
export goods individually. For example, the shares of the first major export good in
88
total exports are 46.15%, 37.52%, 31.59%,19 and 14.73% for the low-, lower-middle-,
upper-middle-, and high-income countries respectively. All the Ps remain significant
for all 8 specifications: for the first and second individual export concentration shares
(EX1EX and EX2EX) at the 99% level, and at the 95 significance level for the third
individual export concentration measure (EX3EX). It is only natural, that the
significance level decreases as we move from the first to the second and to the third
major export good. None of the previous literature on patterns of development has
attempted or reported any analysis of export product concentration.
Figure 12: Scatter Diagrams of Export Product Concentration
19 After excluding Saudi Arabia, whose share of oil exports in total exports is over
85%. Nevertheless, the trend remains even with Saudi Arabia. Including Saudi Arabia,
the group average for the upper middle-income countries would be 36.52%.
89
9. Market Power in World Export Markets
While developing countries have higher shares of export product concentration,
their exports represent less market power in world exports compared to industrialized
countries. This may not be surprising since it is a well-known fact that the
industrialized countries dominate world trade,20 however, what is surprising is that
there is a systematic relationship between market power in world export markets and
GDP per capita. Considering the empirical results, there is a robust positive
relationship between a country's market export power and its level of development
defined as GDP per capita.
There is a robust positive relationship even within the group of the developing
countries, as can be concluded from the scatter diagrams, the group averages, the panel
data correlation coefficient, and especially from the multiple regression analysis. We
have tested such a relationship with a total of 30 different regressions, which provided
always significant Ps at the 99% level or higher. These 30 regressions control for a
variety of different specifications. Please see Appendix 4 for detailed results of the
overall index of export market power. Like in the case of export product concentration,
there does not seem to be any previous literature which has systematically analyzed the
relationship between export market power in world exports and the level of GDP per
capita across countries.
20 As is for example reported in Appleyard and Field (1995) the developing

countries accounting for about 70% of world trade in recent years.
90
Figure 13: Scatter Diagrams of Export Market Power
•*.
10. Financial Market Development
We discussed earlier in section 5 of this chapter, the share of money supply
broadly defined (M2) in GDP is positively related to the level of GDP per capita. As
M2 comprises money and quasi monetary liabilities of a country's financial institutions
to residents other than the central government, the share of M2 in GDP can easily be
defined as a measure of financial market development. However, the recent literature
provides four alternative indicators of financial development (FIN1 to FIN4).
Following the theoretical discussion of Chapter II, section 3.c.iv., the empirical results
support the hypothesis that indicator FIN3 is not necessarily a good measure of
financial market development.
91
All tools of analysis provide consistent results for indicators FIN1, FIN2 and
FIN4. However, they are less convincing for indicator FIN3. For example, all 8
different specifications provide 3s which are always significant at the 99% level for
indicators FIN1, FIN2 and FIN4. For indicator FIN3, the 3s are statistically significant
at the 95% level for specifications loglinl to loglin4, Ioglin7, and loglin8; but
insignificant for specifications loglin5 and loglin6. While this casts doubt on the
appropriateness o f FIN3 as a measure of financial market development, it can be
concluded that there is a positive relationship between financial market development
and the level of GDP per capita.
Figure 14: Scatter Diagrams of Financial Market Development
0.5
0.5
0 .0 .
0 .0 .
-0.S.
-1 0 .
-1.0 .
•1 0.
- 2. 5 .
J to n o ta
92
11. Summary o f Results: Stylized Facts
The previous 10 sections have shown that there are many strong relationships
between development (measured by the level of GDP per capita), and economic
structure (measured by 45 structural variables). We find that 34 structural variables are
systematically related to the level of GDP per capita. Given the robustness of these 34
relationships and the fact that the analysis is based on the actual experience of 93
countries from 1970-94, it is justified to define these relationships as stylized facts of
development patterns for 1970-94.
Stylized Fact 1: Sectoral Shares o f GDP
Consistent with the previous literature, we conclude that the share of agriculture
in GDP falls and the shares of industry, manufacturing, and services in GDP rise, as
GDP per capita rises.
Stylized Fact 2: Total and Private Investment
Also consistent with the previous literature is that the shares of gross domestic
investment in GDP, domestic fixed investment in GDP, and fixed private investment in
GDP are higher at higher levels of GDP per capita.
Stylized Fact 3: Savings. Total and Private Consumption
The savings ratio is higher the higher the level of GDP per capita. Consistent to
the patterns of the savings ratio, the shares of total and private consumption in GDP
93
falls as GDP per capita rises. Both stylized facts are consistent with previous studies.
Stylized Fact 4: Government Expenditures and Government Revenues
The shares of (a) government expenditures, (b) total government revenues, and
(c) tax revenues are generally higher for countries with high levels of GDP per capita
than for countries with low levels of GDP per capita. Part (a) is new, parts (b) and (c)
have been established before.
Stylized Fact 5: Current Account. Capital Account, and Trade Intensity
A new stylized facts is that current account deficits and capital account surpluses
decrease while trade intensity (openness) increases as development proceeds.
Stylized Fact 6: Total and Merchandise Exports
Consistent with most of the previous literature, we conclude that the shares of
total exports and merchandise exports in GDP increase as GDP per capita rises.
Stylized Fact 7: Composition of Exports
As GDP per capita rises, the share of machinery exports in GDP as well as in
total exports increases and the share of primary exports in total exports decreases. The
two relationships with respect to total exports are new contributions to the literature of
patterns of development.
94
Stylized Fact 8: Export Product Concentration
A new stylized fact is also that a country's export product concentration is
higher, the poorer the country is.
Stylized Fact 9: Export Market Power
The final new stylized fact with regards to international trade is that a country’s
export market power in world exports is higher the more developed a country is.
Stylized Fact 10: Financial Market Development
Though partly established before within the new growth literature, our last new
stylized fact is that there is a strong positive relationship between development defined
as GDP per capita and appropriately defined financial market development.
CHAPTER IV
HOW TO MEASURE MACROECONOMIC UNCERTAINTY
Measuring macroeconomic uncertainty is a rather complicated issue. The main
problem is that most o f the literature has not dealt with the differences between risk,
volatility, and uncertainty. The first section o f this chapter will therefore point out the
main difference between risk and uncertainty. The second section will explain the
difference between volatility and uncertainty. Section 2 will also provide a general
overview of the recent literature on volatility and uncertainty. However, that review will
not go into details o f the investment literature, as the relevant investment literature will be
analyzed in Chapter VT. Section 3 will provide a comprehensive overview o f the
development of volatility and uncertainty measures. Based on the definitions and reviews
of the first three sections of this chapter, section 4 will suggest a country-specific, time-
invariant macroeconomic uncertainty measure. The variables for which such an uncertainty
measure will be calculated are defined in section 5. Section 6 will examine the
circumstances under which it will be necessary to modify the earlier suggested uncertainty
measure in order to avoid biases related to arbitrary differences in the level o f a variable.
95
96
The consistency o f the modified uncertainty measure will be checked in section 7.
Section 8 will examine how different volatility and uncertainty are, by comparing the three
uncertainty measures with the corresponding volatility measure. Finally, the last two
sections will compare the uncertainty measures (a) across different macroeconomic
variables, and (b) across different time periods. The three time periods under consideration
are 1970-94 as well as two twelve-year sub-periods, 1970-81 and 1983-1994.
1. Uncertainty versus Risk
l.a. Knight's Classical Distinction
Based on the pioneering work of Frank H. Knight (1921), economists are
supposed to distinguish between uncertainty and risk. Investment which is risky does not
need to be uncertain and vice versa. Uncertainty is a situation in which the likelihood of an
event occurring is not known at all; no probability distribution can be attached to the
outcomes. Investment under uncertainty implies that the conceivable returns are known
while their probability o f occurrence is unknown. According to Knight, risky investment
means that the returns are subject to a known probability. In other words, the difference
between uncertainty and risk is not based on a difference in the knowledge about
conceivable returns, but in the knowledge about probability. The conceivable returns are
assumed to be known in both cases.
97
l.b. Knightian-Keynesian Uncertainty
Keynes (1936) followed Knight's classical definition of uncertainty. As Davidson
(1991, p. 131) points out, Keynes (1937) explained that by uncertainty, he did
not mean merely to distinguish what is known for certain from what is only
probable. The game of roulette is not subject, in this sense, to uncertainty. ... The
sense in which I am using the term is th a t... there is no scientific basis on which to
form any calculable probability whatever.
Furthermore, Keynes used the term "animal spirits" as an explanation for what
determines investors' expectations o f the marginal efficiency of capital. While it is not clear
if Keynes used the term to indicate that an investor's decision may be based on other
investors' decisions,1 the term clearly indicates that an investor's decision cannot be
captured by a probability distribution. Investment decisions are based on subjective
guesses.
I.e. Uncertainty Today
Though the world has seen considerable progress in knowledge and especially the
access of knowledge and information, the investment decision continues to be
characterized by uncertainty. Even with the best information and forecasting methods,
knowledge about an investment's underlying probability distribution is limited to cases of
controlled experiences. In the words of Pollin (1997): "Because future economic
1 Such an effect is now well-known as bandwagon effect, especially in the consumer

theory.
98
outcomes are necessarily uncertain (...) the degree of risk can never be known or even
calculated with the precision of a probability distribution."2
I.d. Uncertainty in Reality
Considering the Knightian-Keynesian definition o f uncertainty, it is even more
realistic that the exact value of the conceivable investment return is not known. Thus,
reality is not only characterized by a lack of knowledge about the probability distribution,
but also about the conceivable return.
Overall, it is possible to distinguish four different cases:
(i) the conceivable return is known but subject to a known probability
(risk in the classical sense);
(ii) the conceivable return is known but subject to a unknown probability
(uncertainty in the classical sense);
(iii) the exact conceivable return is not known and subject to a known probability
(risk in reality); and finally,
(iv) neither the exact conceivable return, nor the underlying probability distribution are
known (uncertainty in reality).
From the practical point o f view of the investor, cases (ii), (iii), and (iv) are more
or less the same, as they all do not allow calculating an expected return. In the case of
classical uncertainty [case (ii)], it is not possible to calculate an expected return because of
the lack in knowledge about the probability distribution. In the case o f realistic risk [case
2 Pollin (1997) p. 16.
99
(iii)] it is not possible to calculate an expected return because o f the lack of knowledge
about the conceivable return. Finally, the case o f realistic uncertainty [case (iv)] lacks
knowledge about the conceivable return and the underlying probability distribution.
I.e. Uncertainty in Economic Modeling
Though case (a) is the least realistic, it has received the most attention in economic
modeling, because it allows the comparison of risky situations to non-risky, certain
situations. In other words, since it is not possible to attach a probability distribution to
uncertainty, uncertainty seems to be a useless concept for those interested in economic
modeling. The only choice left for those interested in modeling uncertainty is therefore to
substitute uncertainty with risk. However, the question emerges: does substituting
uncertainty with risk enable realistic conclusions? Given the qualitative difference between
risk and uncertainty, Keynes warned that "the hypothesis of a calculable future leads to a
wrong interpretation of the principles of behavior."3 Following Keynes, most of the Post-
Keynesians strictly rejected the deduction of the investment decision to an application of
probability theory.4 Only recently, Dixit and Pindyck (1994) have shown that the real
option theory allows us to use probability theory in a way that it is consistent with the
concept of uncertainty.
3 Keynes (1937, p. 122); as quoted by Davidson (1991), p. 131.
4 See Davidson (1991) for further details. For a recent evaluation o f Minsky's original
dissertation Induced Investment and Business Cycles, see Delli Gatti and Gallegati (1997).
100
l.f. Conclusion
In conclusion, while the terms "risk" and "uncertainty" are often used as synonyms,
especially in circumstances related to daily life, they are different concepts and need to be
treated as such. Uncertainty implies that no probability distribution can be attached to the
outcomes. However, it is possible to use probability theory to analyze investment under
uncertainty without attaching a probability distribution to outcomes, as is the case with the
real option theory of investment under uncertainty.
2. The Different Concents of Volatility and Uncertainty
2.a. Uncertainty, Risk and Volatility
There are two ways to look at uncertainty. One way is to look at uncertainty as a
situation or a state of the world. The other way is to look at uncertainty as a motion or
movement of a variable. When comparing states of the world, the difference is made
between risk and uncertainty. When comparing movements o f a variable, the difference is
made between volatility and uncertainty. Risk and volatility are thus comparable in the
sense that risk and volatility are measurable. Uncertainty is a more complicated concept,
either when referred to as a situation or a variable's movement.
2.b. Uncertainty versus Volatility
Similar to the difference between risk and uncertainty, there is —at least at the
theoretical level -- a very clear distinction between the concept of volatility and the
concept of uncertainty. As Hausmann and Gavin (1995) have pointed out: "Volatility and
101
uncertainty are in principle distinct concepts: volatility refers to the tendency of a variable
to fluctuate, while uncertainty is present only when those fluctuations are unpredictable."5
Note that uncertainty refers now to the behavior o f a variable not to a situation.
I.e. An Overview of the Volatility Literature
Though Hausmann and Gavin (1995) have stressed the important difference
between volatility and uncertainty, they continue to say that volatile quantities also tend to
be unpredictable, so that in practice, the distinction is less relevant. We will show below
that this is not necessarily the case. The remainder o f this section provides a more general
overview of the recent volatility and uncertainty literature.
The most outstanding and comprehensive analysis of macroeconomic volatility is
Hausmann and Gavin (1995). Hausmann and Gavin analyze the causes and consequences
of a great variety of volatilities and instabilities. Their main analysis centers on five
volatilities: monetary policy, fiscal policy, real exchange rate, terms of trade, and GDP.
Here volatility is defined as the standard deviation o f the variable under consideration.
Hausmann and Gavin's main conclusion is that volatility has imposed enormous costs on
the Latin American economies. In more detail, volatility is bad for economic growth, bad
for productivity, bad for investment, weakens the financial system, undermines educational
progress, is bad for the distribution of income, and finally, even increases poverty.
Towards, the end of the study, they discuss specific institutional and policy reforms that
can improve the shock resistance of the Latin American economies. In contrast to the
5 Hausmann and Gavin (1995), p. 193 (Box I).
102
comprehensive work of Hausmann and Gavin (1995), most earlier studies limit their
analysis to specific areas. The three traditional and two new areas of studies of
macroeconomic volatility and uncertainty are:
(i) prices, inflation, and the money supply;
(ii) the real and nominal exchange rate;
(iii) the terms o f trade;
(iv) output volatility; and
(v) capital flow volatility.
2.c.i. Prices, inflation, and money
There is a long history o f analysis o f effects o f price or inflation volatility.6
Following Friedman's (1977) Nobel Lecture, it is now well-known -- though perhaps not
generally accepted —that increased volatility of inflation reduces the efficiency of market
prices as coordinators of economic activity. This hypothesis actually goes back to von
Hayek (1945). Similarly, Barro (1976, 1980) points out that monetary volatility adds noise
to the process of extracting the correct relative price signals needed for an efficient
resource allocation. These and other hypotheses have been confirmed based on the post
war data from 47 countries in the growth analysis o f Kormendi and Meguire (1985), as
well as by Grier and Tullock (1989). A more sophisticated analysis of the impact of
monetary policy uncertainty on growth and investment is Aizenman and Marion (1993).
6 For a review of the early contributions by Keynes (1924) and von Hayek (1945), see
Blejer and Leiderman (1980). For a review o f other important contributions, especially of
103
Aizenman and Marion's study and other follow-up studies will be reviewed in more detail
later when reviewing the empirical investment literature.
Most recently, Aspergis (1997) has extended a money demand equation to include
volatility of inflation. The empirical analysis shows that an increase in inflation volatility
has a negative impact on the demand for money. Finally, an interesting connection
between inflation volatility and exchange rate volatility has been made by Bahmani-
Oskooee (1991a). Bahmani-Oskooee concluded that with the current floating exchange
rates system, exchange rate variability contributes to inflation variability.
2.c.ii. Real and nominal exchange rates
There is a considerable theoretical and empirical literature analyzing the impacts of
exchange rate instability. The largest share of this literature deals with the effects of
exchange rate volatility on trade. Here the proponents and opponents of a flexible
exchange rate system have come to completely different results, ranging from negative
impacts, to no impacts, and even positive impacts.7
Following Friedman and Hayek, Krugman (1989) argued from a theoretical point
o f view that higher exchange rate volatility could adversely affect the efficiency of the
economic system by reducing the information content in relative prices. This reduction in
the information content is also the reason for why nominal exchange rate instability may be
more important than real exchange rate instability. Akhtar and Hilton (1984) have also
the 1970s and the early 1980s, see Fischer (1981). A recent review of the latest
developments can be found in Glezakos and Nugent (1996).
7 For a more detailed review of the earlier literature, see Arize (1996).
104
argued that since there is no unique or precise way to measure uncertainty, it is more
appropriate to use nominal exchange rate volatility rather than real exchange rate volatility
as a proxy of exchange rate uncertainty.*
O f the latest empirical contributions, Arize (1995) concluded that exchange rate
volatility has a negative impact on U.S. trade. Arize (1996) extended the same result for
eight European countries. Finally, Arize (1997) uses a measure o f conditional exchange
rate volatility to show its negative impact on the volume of foreign trade for seven
industrialized countries. However, Friberg and Vredin (1996) concluded that there is no
strong empirical evidence that exchange rate volatility hampers trade of the Swedish
economy.9 While the economic literature is far from having reached a consensus for the
industrialized countries, the conventional wisdom seems to have moved towards the
proposition that exchange rate volatility has a negative impact on international trade.
The picture is slightly more homogenous for the few multi-country analyses of
developing countries. Caballero and Corbo (1989) concluded that despite the ambiguity of
theory, the empirical relation between investment and exchange rate uncertainty is strongly
negative. However, Bahmani-Oskooee (1991b) has shown that while most estimates are
consistent with the arguments o f opponents of flexible exchange rates, some positive
8 In the words of Akhtar and Hilton (1984, p. 10): "We reject the use of the real
exchange rate variability as the relevant proxy for uncertainty in our empirical work. Since
'true' uncertainty is not measurable, that leaves us with the observed nominal exchange
rate variability."
9 For recent theoretical contributions along this line, see Broil and Wahl (1995), Broil
and Zilcha (1995), and Neumann (1995). For aspects o f portfolio substitution and
exchange rate volatility, see Sibert and Ha (1997).
105
effect on the exports of Brazil and Korea can be seen. Finally, Grobar (1993), using a
variety o f volatility and uncertainty measures of the real exchange rate, concluded that
some categories o f LDC manufactured exports are negatively affected. Overall, analyzing
the impact on world trade, de Grauwe (1988) concluded that about 20% of the 5.7
percentage point slowdown in the growth o f world trade since 1973 can be explained by
exchange rate variability.
2.c.iii. Terms of trade volatility
Within the recent literature analyzing terms of trade volatility, Mendoza (1994 and
1997) demonstrates that the variability of the terms of trade has ambiguous effects on the
savings and growth rates, depending on the degree of risk aversion. But Lutz (1994) has
shown that there is a statistically significant link between income terms of trade volatility
and lower growth rates. Chang (1995) analyzes the relationship between export
diversification and international debt under terms of trade volatility. Also, Mendoza
(1995) demonstrates that terms of trade shocks account for nearly 1/2 o f actual GDP
variability. Finally, van Wincoop (1992) shows that increased terms o f trade volatility
leads to a higher fraction of the labor force in the non-tradeables sector.
2.c.iv. Output volatility
Similar to Hausmann and Gavin (1995), there are now a few studies analyzing
output volatility. Basically all of these studies are part of the new growth literature.
However, the empirical results are inconclusive. For example, Ramey and Ramey (1995)
conclude that in a sample of 92 countries, countries with higher output volatility have
106
lower growth rates. However, Dawson and Stephenson (1997), examining the relationship
using state data o f the United States, find no evidence o f such a relationship once other
correlates o f growth are incorporated into the analysis.
2.c v. Capital flow volatility
Following the December 1994 Mexican peso crisis, there is now emerging a fast
growing literature on the economic impacts o f capital flow volatility. While most of the
theoretical issues have been analyzed in a systematic way, as for example in the most
recent World Bank (1997) research policy report on Private Capital Flows to Developing
Countries: The Road to Financial Integration, there is still a gap of empirical
examinations at the cross-country level. Most o f the literature analyzes either the cost and
benefits of capital inflows, or concentrates on how to prevent rapid capital outflows,
leading not only to currency crisis but also banking crisis.10 For an examination of capital
movements, investment and animal spirits, see Velasco (1996).
3. Measures of Volatility and Uncertainty
This section provides an overview o f the measurement o f volatility and
uncertainty. This overview discusses the major contributions to, and reviews of, volatility
and uncertainty measures. Though uncertainty and volatility are clearly distinct concepts,
most of the literature does not differentiate appropriately between them. For example,
10 A good overview of these issues can be obtained from the many contributions in
Calvo, Goldstein, Hochreiter (1996) Hausmann and Rojas-Suarez (1996). The lessons of
capital flow volatility for debt management can be found in Calvo and Gunter
(forthcoming).
107
Fischer (1993, p. 11) writes that "the variability of inflation might serve as a more direct
indicator of the uncertainty o f the macroeconomic environment." While there is little
doubt that economists like Stanley Fischer are well aware of the difference between
volatility and uncertainty, the widespread ignorance about the difference can be explained
as follows.
First, there is little agreement on how to measure volatility in empirical studies. As
will be shown later in more detail, the usual measures are deviations from the trend, the
standard deviation, and variance of a variable. Second, there is even less agreement about
how to measure uncertainty other than by volatility. Indeed, there simply does not exist a
measure of uncertainty which fully accounts for the conceptual difference between
volatility and uncertainty. Third, macroeconomic uncertainty is not always expressed best
by a sophisticated measure o f volatility or uncertainty. For example, high levels of inflation
may imply high macroeconomic uncertainty even if high inflation is quite predictable. A
second example is economic uncertainty caused by high levels of indebtedness. The level
o f debt to GDP ratios may serve as a better measure of uncertainty, than the volatility of
the debt to GDP ratio. A third example may be to measure exchange rate uncertainty by
the level o f the black market premium.11 Finally, it is generally assumed that there is little
difference between volatility and uncertainty in praxis.
11 The possibility and limitations of the black market premium have been recognized by
Fischer (1993), p. 488: "I use the black market premium on foreign exchange as an
indicator o f the sustainability and appropriateness of the exchange rate. The black market
premium is a good indicator o f a distorted or controlled market for foreign exchange, but
is less good as an indicator o f the unsustainability of the exchange rate, since an exchange
rate may be overvalued and unsustainable even when there is no black market premium."
108
However, given the long debate on the appropriate measure of uncertainty,
especially within the exchange rate literature, it is possible to distinguish between
(a) measures which are closer to the theoretical concept of volatility
(i.e., measures of dispersion around the mean), and
(b) measures which are closer to the theoretical concept of uncertainty
(i.e., measures which attempt to measure unpredictable parts o f the fluctuations).
3.a. Measures More Closely Related to Volatility
3a.i. Early Measures of Economic Instability
Most of the early literature (that is before the collapse o f the Bretton Woods
exchange rate system) used indices of economic instability rather than the variance or
standard deviation. As reviewed by Knudsen and Pames (1975), these indices were:
• the sum o f squared deviations from an exponential trend line, fitted by minimizing
sum of squared residuals;
• the sum o f squared deviations from a linear trend line, fitted by minimizing the sum
of squared residuals;
• the sum of absolute deviations from an n-year moving average, whereby the value
of n is determined a priori and controls the degree of smoothing;
• the index developed by Coppock (1962), which consists of deviations from a trend
line determined by first and last observations;
• the sum o f the average percentage deviations from a linear trend;
109
the United Nations' (1952) index, the sum of year-to-year absolute deviations as a
percentage of the larger of the two yearly observations; and
• the IMF's (1966) index, the sum of a three-year weighted average, whereby the
weights are 0.5 for the current year and 0.25 for each o f the two previous years.
Because of problems related to these indices, and the relative popularity of the
permanent income theory in the early 1970s, Knudsen and Pames (1975, pp. 93-94)
provided their own index o f economic instability: the transitory income index. This index
is defined as the normalized variance of transitory income, that is:
export instability is measured in terms of transitory income, or the part of income

that is unpredictable or temporary and does not enter consumption decisions. An
index of instability is then defined as the sum of square of transitory export
earnings, normalized by permanent export earnings.
Note that this normalization implies that countries with higher growth rates of
export earnings have a higher export instability than countries with low or no growth in
the share of export earnings. We will see in Chapter VI that this definition has important
implications for the effect o f export instability on investment and growth.
3.a.ii. The rise o f volatility
With the collapse o f the Bretton Woods exchange rate system, it became
fashionable to analyze the impacts of exchange rate volatility. At an initial level, there
exists a universally accepted measure of volatility, which is either the variance (s2) or its
110
square root, the standard deviation (s).12
Inevitably, economists debated about which exchange rate is the most appropriate:
is it the nominal exchange rate, the real exchange rate, or any of the effective exchange
rates? The next question then was whether or not it is more accurate to calculate the
volatility from the level of the exchange rate or from its annual or quarterly changes.
These and other questions finally led Lanyi and Suss (1982) to address the question of
which exchange rate indices should be used for measuring exchange rate variability. While
Lanyi and Suss refer to different interpretations o f nominal versus real exchange rate
variability, they do not address issues related to measuring the unpredictable part of
exchange rate fluctuations.
Medhora (1990) provides a brief discussion of the more traditional measures, such
as the standard deviation, deviations from the trend, Gini mean difference coefficient,
coefficient of variation, and the scale measure of variability. Recently, Glezakos and
Nugent (1996) have calculated relative price variability using a divisia-type index as the
weighted sum o f squared relative price changes o f individual commodities. As we will see
12 The variance is defined according to the formula:
Si (Xi - X )2
s = ------ (discrete case)
N
or _
s2 = o (Xi - X)2 f(X) dX (continuous case)
where X is the variable mean,

X; is the rth observation of the variable,
N is the number o f observations, and
f(X) is the probability density function.
I ll
in Chapter V, these relatively simple methods, particularly the conventional standard
deviation, are still commonly used. Indeed, they are the majority of measures used in the
empirical investment literature.
3.a.iii. Volatility defined as the standard error
Three years before Lanyi and Suss, Peter Kenen (1979) had already written a
similar research memorandum, which subsequently found its way into a more formal
working paper by Kenen and Rodrik (1984). A summarized version of the 1984 working
paper was formally published as Kenen and Rodrik (1986), and turned out to be an
important contribution to the measurement of macroeconomic uncertainty. Kenen and
Rodrik (1986) suggested three different volatility measures, the last two being initial
proxies o f uncertainty:
The first measure is the standard deviation of the monthly (percentage) change in
the real exchange rate. ... The second measure is the standard deviation (error) of
the real exchange rate obtained from a log-linear trend equation. ... The third
measure is the standard deviation (error) of the real exchange rate from a first-
order autoregressive equation.13
13 Kenen and Rodrik (1986), p. 311. Note that the description for the second and third
measured is not very precise. Instead of defining measured 2 and 3 as "the standard
deviation (error) of the real exchange rate obtained" from either a log-linear trend
equation or a first-order autoregressive equation, it would have been more precise to state
that these measures are the standard errors o f the residuals obtained from either a log-
linear trend equation or a first-order autoregressive equation. Is seems unlikely that Kenen
and Rodrik were fully aware o f their contribution of moving from the standard deviation
o f the variable to the standard error of the residuals. The standard error of residuals is a
more appropriate measure o f uncertainty than the usual standard deviation. As a result of
Kenen and Rodrik's ambiguous definitions, most of the later reviews dealing specifically
with the question of how to measure exchange rate uncertainty, have completely neglected
Kenen and Rodrik's contribution.
112
Kenen and Rodrik's approach has been applied to investment under uncertainty by
Aizenman and Marion (1993), who use the method to analyze the impact of policy
uncertainty on growth and investment. The variable characterizing policy uncertainty is a
structural variable. For example, monetary policy uncertainty is measured by the growth
rate o f the money supply. To estimate policy uncertainty with a structural variable is
problematic. As Fischer (1993, p. 488, n.10) points out: "Aizenman and Marion (1991)
attempt to quantify policy uncertainty by estimating autoregressive processes for policy
variables and using the standard deviations of policy surprises as a measure of uncertainty.
This is a promising approach, which however does not distinguish contemporaneous
variability caused by responses to exogenous shocks from purely random variability."
Anyway, Aizenman and Marian show that "policy" uncertainty is negatively correlated to
both investment and growth.
Avenell, Leeson and Wood's (1989) review o f uncertainty measures explicitly
recognizes that exchange rate uncertainty is a far more complicated concept than exchange
rate volatility. However, the review makes no attempt to suggest a measure of uncertainty.
Neither do they mention Kenen and Rodrik's work. Instead, Avenell, Leeson and Wood
(1989, p. 34) argue that since exchange rate uncertainty "cannot be directly observed ... it
is common to choose and estimate measures of exchange rate volatility.”
Pritchett (1991) analyzes the distribution of real exchange rates of 56 developing
countries and explains why it is not justified to use the standard deviation as a measure to
compare real exchange rate uncertainty across countries. Pritchett points out that the
higher order moments (skewness and kurtosis) are at least as important as the standard
113
deviation. He stresses that the use o f the standard deviation fails to demonstrate the
negative impact of real exchange rate variability on export performance. However, higher
order moments are highly significant. He emphasizes that it would be wrong to conclude
that exchange rate volatility has little impact on export performance as long as the
conclusion is drawn solely from the insignificance of the exchange rate's standard
deviation.14 The problem with Pritchett's paper is that it neither refers to the classical
contributions like Knight and Keynes, nor does it refer to the major recent contributions
dealing specifically with the measurement of economic instability, for example, those of
Engle (1982), Kenen and Rodrik (1986), and Pagan and Ullah (1988).
3.b. Measures Related More Closely to Uncertainty
3.b.i. ARCH models
A major break-through in the search for a measure of uncertainty is the seminal
work o f Engle (1982). The important novelty of the autoregressive conditional
heteroscedasticity (ARCH) model is its use of the systematic information contained in the
error terms. As Bomberger (1996, p. 381) points out: "The estimated conditional variance
at each date is assumed to reflect the uncertainty of a representative forecaster." The
remaining errors are a better measure o f the unexplained fluctuations of a variable than the
unconditional standard errors. In the words of Huizinga (1993, p. 528): "This measure
seems to best account for the idea for series whose deviations from the unconditional
14 In the words of Pritchett (1991, p. 21): "The conclusion from a regression with no
significant coefficient on s that the pattern of instability of the real exchange rate behavior
had little impact on export performance would be quite wrong."
114
mean can be reliably predicted, it is not fluctuations around an average value that are of
concern (that is, the unconditional variance) but rather fluctuations about a predicted
future path." Hence, the conditional standard error can be considered to be the first
uncertainty measure. This is also confirmed by Ghosal and Loungani (1996, p. 220), who
write that "the basic idea of measuring uncertainty as a conditional standard deviation is
consistent with suggestions from the theoretical work (see Dixit and Pindyck)."
While ARCH models have been widely applied to financial markets in the business
literature,15 it is not clear why Engle's work has been completely ignored in most of the
macroeconomic literature of exchange rate uncertainty. A possible explanation could be
that it was initially applied only to the analysis o f inflation volatility. However, the basic
ARCH model specification has been applied recently to investment theory by Huizinga
(1993), Ghosal and Loungani (1996), and by Ogawa and Suzuki (1997).16
In Ghosal and Loungani (1996), price uncertainty is measured by the conditional
standard error of four different price "forecasting" equations. The initial forecasting
equation is determined by the log o f equally weighted industry product prices o f previous
periods and a linear time trend. The first alternate measure is based on adding lagged
values o f unit variable cost to the forecast equation. The second alternate measure is
derived by expressing the forecasting equation in growth rates (instead of in logs). Finally,
15 See Bollerslev, Chu, and Kroner (1992) for a review o f this extensive literature in
the financial literature. For a response to the critics o f the ARCH models, see Anderson
and Bollerslev (1997).
16 Ogawa and Suzuki (1997) also use more simple approaches of the unconditional
variance and the conventional way o f computing the standard deviation.
115
the third uncertainty measure is based on the initial forecasting equation, though more
weight is given to the more recent periods.
3.b.ii. Conditional variance via IV estimator
One o f the major contributions to the econometric analysis of models with risk
terms is Pagan and Ullah (1988). Pagan and Ullah recommend applying an instrumental
variable (IV) estimator, with instruments constructed from the information set, to exploit
the existing orthogonality conditions between variables in the information set and higher-
order moments of the unanticipated variable density.
Bini-Smaghi (1991) provides a useful review of the major drawbacks of earlier
uncertainty measures. He blames the use o f wrong measures for difficulties in finding
empirical relationships between exchange rate variability and trade. Bini-Smaghi (1991)
follows the suggestion of Pagan and Ullah (1988) and uses an instrumental variable
technique to construct proxies for what he calls risk variables. "This approach has been
used here to try to decompose exchange rate variability into an expected and unexpected
component, by regressing the contemporaneous values on lagged values of that variable
and on the level of the exchange rate."17
Finally, Arize (1997) offers a thorough discussion of alternative measures of
exchange rate volatility. Building on the critique of Pagan and Ullah (1988) that even "an
ARCH proxy for risk does not escape the classic errors-in-variables problem as the ‘true’
17 Bini-Smaghi (1991), p. 934.
116
variable will still be measured with error,” Arize uses a similar IV procedure to obtain a
consistent estimator.
3.b.iii. Further extensions of the ARCH model
There were two major reviews of uncertainty measures in 1995: Seabra (1995) and
Caporala and Doroodian (1995). Seabra (1995) derives an ex ante measure of exchange
rate uncertainty for 11 Latin American countries by estimating the expected exchange rate
uncertainty according to an extended version of the ARCH model, building on Engle's
(1982) ARCH model and its extension since then.1® The contribution by Caporala and
Doroodian (1995) goes one step further. The study suggests estimating a generalized
autoregressive conditional heteroscedasticity (GARCH) measure of exchange rate
uncertainty since it enables the conditional variance in the exchange rate to be consistently
parameterized. Caporala and Doroodian follow the arguments put forward by Akhtar and
Hilton (1984), Medhora (1990), and Kumar and Dhawan (1991), and use the nominal not
the real exchange rate as the variable o f consideration. An extension of Pagan and Ullah's
(1988) ARCH model is applied to investment under uncertainty by Price (1995),
suggesting a GARCH-M model. The GARCH-M model (generalized autoregressive
conditional heteroskedasticity model in mean) adds the conditional variance of the
regression as an additional regressor to the equation. This an attractive form in financial
applications since it is natural to suppose that the expected return on an asset is
18 Engle and Bollerslev (1986), Poterba and Summers (1986), and Lastrapes (1989).
117
proportional to the expected risk of the asset. The expected risk of the asset is proxied by
the conditional variance o f the regression.
3.b.iv. Other suggestions on uncertainty measures
There are five further interesting suggestions for how to measure uncertainty:
Bomberger (1996), Conway (1991), Federer (1993a and 1993b), and Leahy and Whited
(1996). Bomberger (1996) suggests disagreement as a measure o f uncertainty. Conway
(1991) derives real exchange rate uncertainty from optimal forecasting rules, whereby he
uses a Kalman filter to derive an optimal forecast as the expected mean of the real
exchange rate conditional on information available at that time.19
Federeris (1993a, p. 32) uncertainty measure is rather complicated and relies on a
variety of assumptions. The initial idea is to employ the term premium embedded in the
term structure of interest rates "to measure uncertainty about interest rates and other
macroeconomic variables." Based on mean-variance portfolio analysis, the term premium
can be viewed as a risk premium.20 In order to estimate the risk premium, Federer follows
Shiller and McCulloch (1989), who provided an estimation method for the risk premium
based on the linearized expected excess holding period return associated with buying and
selling bonds. However, this equation requires a value of the market's interest rate
19 Conway (1991) refers to Harvey (1989) for an excellent reference on Kalman

filtering applications in statistical inference.
20 The theoretical foundation o f this uncertainty measure builds on earlier work by

Engle, Lilien, and Robbins (1987), who have shown that there is a positive relationship
between backward-looking ARCH measures o f interest rate uncertainty and the risk
premium embedded in the term structure for U.S. Treasury bills.
118
expectation, which Federer takes from the median interest rate forecasts collected by the
Goldsmith-Nagan Bond and Money Market News Letter (now called the Washington
Bond & Money Market Report). Federer's uncertainty measure constitutes a considerable
contribution at the theoretical level. However, from a practical point of view, it is
questionable how much better such an uncertainty measure really is compared with other
more easily obtainable uncertainty measures.
Federer (1993b) measures macroeconomic uncertainty as the cross-forecaster
standard deviations o f five macroeconomic variable forecasts from the Blue Chip survey.
He builds on earlier empirical work by Zamowitz and Lambros (1987), which suggests
that forecaster discord provides a good proxy for group uncertainty. However, as Federer
admits, the main disadvantage of this method is that it relies on the availability o f good
survey data, which is a problem for most developing countries.
Finally, an interesting exposition on how to measure uncertainty is found in Leahy
and Whited (1996, pp. 68-69).
Since uncertainty relates to expectations and not to actual outcomes, it would be

incorrect to use the ex post volatility asset returns as a measure of variability o f the
firm's environment. We therefore need an ex ante measure. On solution to this
problem is to extract a forecast o f return variance from option prices.
Given the considerable data constraints of such a measure, Leahy and Whited
(1996) propose an alternative solution by constructing forecasts of volatility using a vector
autoregression technique.
119
4. A Preliminary Measure of Uncertainty
The review of the last section has demonstrated the different ways of measuring
uncertainty. Some progress is evident in the measurement o f uncertainty, starting from the
standard errors of ARCH models, to GARCH models, and using more and more
appropriate estimation methods, for example, the IV estimation. However, the economics
profession is far from having reached an agreement about how to measure uncertainty
best.21 The most important selection criterion for an uncertainty measure is its ability to
reflect uncertainty accurately. The second most important criterion is the ability of the
measure to be expressed numerically. A theoretical construct may be an excellent
theoretical contribution; however, it will be useless if it cannot be used for empirical
analysis. The measure we are interested in is must be applicable to our cross-country
analysis comparing 93 countries from 1970-94. In addition, the measure should not be
dependent on arbitrary differences related to the scale of the variable under consideration.
For example, the measure should be unbiased with respect to country size. Finally, the
uncertainty measure must be robust in the sense that small changes in the specification
should not lead to large differences in the size of the uncertainty measure. The last two
issues o f unbiasedness and robustness will be dealt with separately in sections 5 and 6. In
21 Like the GARCH-M model, there are now many other extensions o f the original
ARCH model, like the EGARCH (exponential GARCH) model, see Nelson (1991), or the
TARCH (threshold ARCH) model, see Glosten, Jananathan and Runkle (1993). The
TARCH and EGARCH models are ways which take into account that downward
movements in the market o f equities are followed by higher volatilities than upward
movements o f the same magnitude.
120
this section we concentrate on the first two issues and suggest a preliminary measure of
macroeconomic uncertainty.
4.a. Criteria 1: Reflecting Reality
There are two major differences in how to measure uncertainty. One is based on
past experiences, and the other is based on future expectations. All the measures related to
ARCH and/or GARCH models are measures which define uncertainty as the conditional
standard error calculated from past experiences. In contrast, the two measures suggested
by Federer (1993a and 1993b) are built on differences of expected values from realized
values. In reality, an accurate uncertainty measure should be a combination o f both,
experiences of the past and expectations of the future. These two requirements make it
very difficult to provide a good uncertainty measure that varies over time. Even if there is
a solution to optimally combining aspects o f past experiences with aspects o f future
expectations, the question which would need to be answered for a time-variant measure is
the following: how many years should be looked back and how many years forward?
Depending on the number of years looked backward and forward, the time-variant
uncertainty measure may take on any value.
Clearly, there is no easy solution. However, a reasonable solution may be to move
from a time-variant uncertainty measure to a uncertainty measure calculated for a specific
time period, for example, an uncertainty measure built on information of 1970-1994.
Obviously, we would still lose the information from years before the time period begins,
e.g. the impact of 1969 experiences on investment of 1970. Still, the loss of information
121
due to years before the time period becomes more and more irrelevant the longer the time
period under consideration. The same applies to the loss o f information about expectations
following the time period, e.g. the impact o f 1995 expectations on investment of 1994.
The rationale of such a country-specific but time-invariant measure is that
countries with many shocks and crises during the time period of consideration have higher
macroeconomic uncertainty than countries which have been relatively stable. Furthermore,
it is assumed that this time-invariant long-term uncertainty measure is a good proxy for the
long-term (or overall) effects of uncertainty. Hence, such a long-term measure can cope
with situations where the economy is still booming shortly before the collapse. Finally, a
time-invariant uncertainty measure avoids problems related to the determination of the
asymmetric time-lag after which uncertainty affects investment.22
4.b. Criteria 2: Quantification
As the examples o f Federer (1993a and 1993b) show, most sophisticated
uncertainty measures cannot be applied to most countries, simply because there is no way
to satisfy the data requirement. Even the more sophisticated ARCH and GARCH models
are at a level of sophistication that make sense only in cases where there is data which is
both complete and of high quality. For example, assume data is missing for every third
year, i.e., data is missing for t-2, t-5, t-8, etc. While this would not pose a serious problem
for ARCH (1) estimations, it would not be applicable at all for any higher order ARCH
22 It is often the case that bad news has an immediate impact, while good news takes
much more time. There are many more asymmetries which the time-invariant uncertainty
122
estimation. It is therefore suggested to start measuring uncertainty by the standard error of
ARCH (1) estimations:
St = aSu + et
where St is the structural variable under consideration, and et is the error term.
The ARCH specification assumes that S is normally distributed with mean (s) and
variance (h), where h is conditional on the past error terms, i.e.:
St ~ N(s,ht) and ht = bi + b2 (et-i)2
Note that it is necessary to estimate the standard error separately for each country,
that is, it will be necessary to run 93 of these ARCH(1) equations with time series data
from 1970-94 for each macroeconomic uncertainty measure under consideration.
S. Areas of Macroeconomic Uncertainty
There are unlimited areas of economic uncertainty. We intend to capture the most
important areas, which are -- analogous to the three traditional areas of volatility --
uncertainty o f inflation, money supply, nominal and real exchange rates,23! Madagascar
measure can easily cope with, while any of the time-variant measures can deal with only
very limited.
23 Since exchange rates are expressed as local currency per dollar, it was impossible
to calculate exchange rate uncertainties of the United States and Panama. Excluding
Panama and the United States, all other countries have been included in the analysis, even
though many countries had limited the flexibility of their exchange rate in one or the other
way for some period. The most important constraints on exchange rate flexibility are:
The Dominican Republic, El Salvador, Guatemala, Haiti, Honduras, and Paraguay

had a fixed exchange rate to the US dollar during the whole time period o f 1970-81, and
have thus a nominal exchange rate uncertainty o f zero for that period.
123
left the Franc zone in 1982.24 and the terms o f trade. Furthermore, it would be of interest
to have some index o f overall macroeconomic uncertainty. Hausmann and Gavin (1995)
suggested that real GDP uncertainty is a proxy for overall macroeconomic uncertainty.
Following the discussion about the relevance o f nominal versus real variables, we will use
nominal GDP uncertainty as a second proxy of overall macroeconomic uncertainty. In
conclusion, we define macroeconomic uncertainty for the following 7 variables:25 (a)
nominal GDP, (b) real GDP, (c) inflation (in %), (d) money supply (Ml as a% of GDP),
(e) nominal exchange rate, (f) real exchange rate, and (g) terms of trade. The results of
this preliminary uncertainty measure for each of these seven areas are listed in Table 7.
6. How to Obtain an Unbiased Uncertainty Measure?
6.a. Bias Related to Different Country Levels
The problem with some of the preliminary uncertainty measures in section 5 is that
they may be biased by the level of the variable. For example, the United States’ GDP is
about $3.4 trillion, while the Gambia's GDP is about $0.2 billion. Therefore, the U.S. GDP
is about 15,000 times the size of Gambia's GDP. Without controlling for this difference in
the level o f GDP, the standard errors are obviously much larger for the United States than
• Within our sample of 93 countries, 12 African countries (Benin, Burkina,

Cameroon, Central African Republic, Chad, Congo, Cote d'Ivoire, Gabon, Madagascar
Niger, Senegal, and Togo) fixed their exchange rate to each other and have therefore the
same value o f nominal exchange rate uncertainty in respect to the US dollar.
25 Given that we are going to analyze investment under uncertainty in Chapter VI, it
may also have been desired to analyze interest rate uncertainty, however, the problems are
here the same as in Chapter II, and therefore no attempt is made to measure interest
rate uncertainty.
124
Table 7: Preliminary Uncertainty Measures
Countiy GDP HESGDP INF MON NEX REX TOT
Australia 1.75E+10 4.77E+09 2.769547 0.993413 0.102338 9.07639 16.32314

Austria 1.02E+10 1.44E+09 1.567153 0.922579 1.884077 15.55275 2.85615
Belgium 1.40E+10 2.41E+09 2.161067 1.035175 5.562862 15.67294 2.500856
Canada 2.50E+10 1.04E+10 2.181801 1.105808 0.054055 5320707 4.446326
Denmark 8.69E+09 13IE+09 2.445407 1.690866 0.944068 15.09007 4.933207
Finland 1.16E+10 2.47E+09 2.774314 3.696691 0.532398 13.42036 4.001985
France 8.09E+10 1.10E+10 2.258529 1.235881 0.815028 14.43051 4.911145
Germany 1.10E+11 1.55E+10 1.2052 1.143813 0.27351 15.4473 4.869466
Greece 4.04E+09 1.50E+09 5519442 0.663664 11.43093 9.723722 6.283059
Ireland 3.73E+09 6.22E+08 3.698998 0.912699 0.081883 1238009 7.594944
Italy 8.63E+10 130E+10 3.555396 2.645976 177.0373 14.72817 6.555959
Japan 1.71E+11 236E+10 435206 1.755044 24.52468 16.15336 10.49949
Netherlands 1.95E+10 2.97E+09 1.466114 1.098452 0.300979 14.47562 2.155402
New Zealand 2.87E+09 1.08E+09 4.207389 3.04593 0.146584 12.60466 13.41105
Norway 5.88E+09 1.08E+09 2390195 2.69661 0.609743 9.191771 10.28629
Portugal 6.04E+09 2.00E+09 6.402178 3.10429 153176 13.4253 5.070693
Spain 4.01E+10 7.05E+09 2.93698 1.544283 15.37375 15.03174 7.206018
Sweden 1.99E+10 1.97E+09 3.005179 NA 0.785099 12.97227 3.830244
Switzerland 1.56E+10 2.69E+09 2.029943 2.896326 0.258332 17.46757 7.01357
UK 6.49E+10 1.67E+10 4.993049 3.873996 0.061778 11.15079 4.920448
USA 8.11E+10 9.04E+10 2.436568 0.83354 NA NA 5.352401
Algeria 5.20E+09 2.07E+09 5.707059 5.944965 3.655391 18.03919 29.2281
Argentina 2.32E+10 8.64E+09 7483223 4.807036 0.154361 2836568 13.58843
Bangladesh 2.25E+09 7.64E+09 12.96707 2.0034 1.606995 21.46473 14.54921
Benin 1.99E+08 1.97E+08 NA 2.52908 72.50592 17.04654 21.56578
Bolivia 2.87E+08 3.21E+08 2774355 6.637423 0341132 12.59994 22.62891
Botswana 1.89E+08 2.09E+08 2.849603 1.817933 0.176799 13.67183 18.20998
Brazil 4.53E+10 2.27E+10 826.9938 1.643141 0.000318 10.27539 11.3386
Burldna Faso 1.71E+08 1.22E+08 12.6235 1.147359 72.50592 18.3517 7.037958
Burundi 80354358 1.36E+08 10.16326 1.595576 10.57479 8.913935 47.57157
Cameroon 1.18E+09 7.37E+08 6315966 1.204834 72.50592 22.46612 20.70359
CAR 1.04E+08 71520322 7.4449 1.940245 72.50592 15.40945 9.779169
Chad 1.14E+08 2.40E+08 11.39698 3.234622 72.50592 13.28802 8.001623
Chile 3.81E+09 3.12E+09 91.92008 3.043555 16.50072 11.68897 21.9989
China 2.20E+10 4.88E+10 6.106432 5.174551 0.603288 9.708121 6.364969
Colombia 2.79E+09 1.48E+09 6.468104 0.697173 54.45518 7.27469 22.8705
Congo 2.99E+08 2.88E+08 6.751975 1.659314 72.50592 17.82392 24.23492
Costa Rica 6.10E+08 3.27E+08 21.76893 1.736742 6.770672 14.29367 14.60859
Cote d'Ivoire 1.15E+09 1.01E+09 8.219496 1.281742 72.50592 19.287 14.53689
Dominican Rep. 7.80E+08 7.26E+08 16.14359 4.006204 1.18065 14.73882 58.41601
Ecuador I.25E+09 1.02E+09 14.24886 1.414095 1153924 12.20089 20.12042
Egypt 2.21E+09 2.29E+09 5.531359 12.39119 0379506 32.63896 17.60611
El Salvador 231E+08 3.88E+08 7318616 2.389798 0.663442 11.30423 12.12958
Gallon 5.88E+08 3.79E+08 7.887071 1.964811 72.50592 18.97631 25.66478
Gambia 25811286 56368428 15.76192 3.489287 0.758294 13.19867 2439624
Ghana 5.87E+08 7.94E+08 45.55336 16.02676 33.66923 1438447 29.20464
Guatemala 8.41E+08 5.78E+08 1232578 1329269 0.437479 9.445755 17.138
125
Table 7: Preliminary Uncertainty Measures (cont)

C ountty GDP HESGDP INF MON NEX REX TOT
Guinea-Bissau 20328692 45497037 45.75197 4.779108 577.1721 15.16241 13.06769

Haiti 4.00E+08 1.92E+08 11.61773 2.918241 1.227294 12.66767 12.35451
Honduras 2.73E+08 1.57E+08 8308234 2.018435 0.729939 16.86838 11.2482
India I.73E+10 2.04E+10 8.44273 0303088 1.582397 7.872551 9.682867
Indonesia 8.61E+09 735E+09 8.48543 1.132323 95.5971 10.02116 21.62313
Jam aica 4.27E+08 2.62E+08 17.9773 1.963247 2.591615 15.08106 26.48735
Kenya 7.97E+08 7.27E+08 8.173291 1.588126 6.140987 18.60015 12.52101
Korea I.02E+10 4.95E+09 7.280092 0.958736 54.27493 7.81867 7.284446
Lesotho 45399342 75083756 433785 1.84605 0.214223 13.06697 NA
Madagascar 3.95E+08 3.69E+08 11.0711 1.827761 2383797 11.41887 13.90154
Malawi 2.08E+08 I.76E+08 10.74187 1.413836 0.8397 9.939994 13.11197
Malaysia 2.57E+09 3.21E+09 4.102603 I.S82U8 0.143625 8.03876 15.98853
Mail 2.94E+08 1.66E+08 NA 2.645714 92.71978 32.03753 9.353979
M auritania 78218359 94098745 8353855 1.806305 8.268807 11.18981 10.69424
Mauritius 1.44E+08 2.87E+08 11.00607 1.989594 1.076721 11.43604 23.67333
Mexico 3.18E+10 1.88E+10 29.40639 1308894 0.270071 10.79993 18.40407
Morocco 1.94E+09 I.58E+09 4.756079 2.711481 0.738742 10.13584 20.06604
M yanmar 2.I9E+09 1.65E+09 1038315 4.62053 0.576207 16.69903 14.69079
Nepal 1.84E+08 I.02E+09 7364314 1.625572 2.264499 7.564786 17.47507
Nicaragua 5.61E+08 5.75E+08 2888.81 70.81483 0.920498 524.2396 17.17254
Niger 2.84E+08 3J4E+08 11.54923 1.158145 72.50592 21.41234 12.63284
Nigeria 1.14E+10 6.95E+09 18.43875 1306347 1.902903 15.43647 28.96034
Pakistan 2.I5E+09 2.65E+09 6.043896 4.021171 1.111068 10.2974 13.94314
Panama 2.85E+08 4.47E+08 4.004717 1.175508 NA NA 10.07254
Papua New Guinea 3.10E+08 1.98E+08 5.103209 1300736 0.062803 9305085 19.01154
Paraguay 6.44E+08 5.03E+08 9.820315 1.703113 110.7458 14.68437 12.62905
Peru 4.87E+09 3.69E+09 1689.248 3.678439 0.195414 23.35624 13.3336
Philippines 2.92E+09 3.67E+09 14.26666 0.968463 1.666566 6.97174 23.7528
Rwanda 1.78E+08 233E+08 9.118478 1.717309 2037624 21.77136 38.51504
Saudi A rabia 1.73E+10 935E+09 7.705808 3.264068 0.259673 31.05644 32.40051
Senegal 6.07E+08 2.11E+08 11.63898 2.066803 72.50592 19.72649 18.05721
Sierra Leone 2.47E+08 3.45E+08 45.45693 3.026788 6036501 20.4737 7.982648
Singapore 1.37E+09 6.15E+08 7.069791 1.954714 0.099459 5.948708 3.443567
Somalia I.I6E+08 6.05E+08 29.17579 11.79527 39.78581 19.14656 7.938464
Sri Lanka 4.79E+08 7.87E+08 7351816 1.177323 1.751558 10.02352 20.07062
Sudan I.78E+09 U 1E + 09 20.56163 17.61113 19.48833 61.57197 13.80197
Syrian A rab Rep. 1.83E+09 2.83E+09 12.56609 21.87305 1.623016 35.08164 19.62057
Tanzania 6.96E+08 5.83 E+08 730217 2.703559 12.79238 17.74629 18.65826
Thailand 2.62E+09 4.18E+09 6.220652 0.701017 0.957035 5.877485 14.82302
Togo 1.73E+08 1.05E+08 7.538535 4.206919 72.50592 19.3636 62.40941
Tunisia 8.21E+08 4.61 E+08 1.86752 1.624042 0.061351 7.21683 15.62345
Turkey 1.66E+10 5.60E+09 25.94703 1.505372 1907.256 14.69976 10.00655
Uruguay I.51E+09 7.42E+08 233803 1.407281 0.213464 1634935 26.84293
Venezuela 8.53E+09 5.87E+09 22.65146 3.483354 11.5073 11.65415 28.92576
Zaire I.53E+09 7.13E+08 4414.417 15.43494 22.20993 2575.876 19.053
Zambia 5.09E+08 2.94E+08 30.8216 2.410902 57.97078 14.92959 34.74287
126
for Gambia. Such a bias would obviously not allow for a comparison across countries.
However, there is a simple solution. The residuals can be expressed as percentages of the
levels. This will successfully remove the bias.
6.b. Bias Related to Different Levels within a Country
Using percentage residuals even removes differences in the relative size of the
residuals within a country. This is crucial though not limited to any of the nominal
variables, especially the nominal exchange rate, which level may differ considerable within
a country. Furthermore, to express the residuals as percentages of levels is consistent with
the concept o f uncertainty as higher levels of a variable will have an impact on the
expected changes.
6.c. New Problems Related to Percentage Residuals
There are two problems with the method of expressing the current residuals as
percentages of current levels. The first minor problem is that the calculation of the
uncertainty measures becomes considerably more cumbersome as most regression
packages calculate the standard error of the regressions automatically. However, if the
residuals need to be divided by the variable level, these calculations will need to be done
before calculating the standard error.
The second problem is that the percentage residuals dampen the size of the true
uncertainty in cases where a large jump was completely unexpected. There is a variety of
possibilities for expressing the residuals as percentages which will reduce this downward
bias. In the method described above, the residuals are expressed as percentages of current
127
levels, that is et is expressed as a percentage o f St. Other options would be to express et as
percentage of St-i, or of the mean of the variable (s). These other options would reduce
the problem of downward biases but are then subject to problems o f overestimating
uncertainty in cases where a jump was expected. There is no perfect solution. The
advantage of an uncertainty measure which underestimates the true uncertainty is that a
possible impact o f uncertainty is not the result o f an exaggerated uncertainty measure. The
disadvantage of an uncertainty proxy which tends to underestimate the true uncertainty is
that it may not fully reflect the degree of impact uncertainty has. The parameter for
uncertainty may turn out to be insignificant, even though true uncertainty has a significant
impact. On the other hand, if it can be shown that there is a negative impact on investment
with conservatively measured uncertainty proxies, in can be concluded that the true effect
of uncertainty is generally even larger than with the conservatively measured uncertainty
proxies.
6.d. Where is the Level-Bias of Relevance?
In principle, the bias caused by having different levels of variables applies to each
o f the seven areas o f uncertainty. However, there may be economic reasons for differences
in levels which may be relevant for an assessment of uncertainty. Expressing the residuals
as a percentage o f the levels may actually cause more harm than good. This possibility is
analyzed within the following three paragraphs for the real exchange rate uncertainty,
terms o f trade uncertainty, and inflation uncertainty.
128
6.d.i Level-bias of the real exchange rate
Given the fact that the real exchange rate is an index which is normalized at
1987=100 for all countries, it does not seem appropriate to express the residuals of the
ARCH(1) estimations of the real exchange rate as percentages. However, there exist
considerable differences in levels of the real exchange rate even though they are all
normalized at 1987=100. For example, Nigeria had a real exchange rate index of less than
30 from 1978-84; Japan, Saudi Arabia, and Switzerland had real exchange rate indices of
higher than 200 for 1970. Such differences in levels have an impact on the levels of the
corresponding residuals, which may not necessarily reflect real exchange rate uncertainty.
The conclusion is therefore that it is appropriate to express the residuals of the real
exchange rate as percentages of their contemporaneous levels.
6.d.ii Level-bias of the terms of trade
Regarding the terms of trade, which are also normalized at 1987=100, the situation
is similar to the real exchange rate. Countries with overall falling terms of trade have
relatively high levels of terms of trade before 1987, while countries with improving terms
o f trade have relatively low terms of trade before 1987. These differences in levels of the
terms o f trade would bias the residuals if not expressed as percentages. Given the fact that
there have been a few studies demonstrating that decreasing terms of trade have a negative
impact on growth and investment, such a level bias could pick up the impact of decreasing
terms o f trade but not of terms of trade uncertainty. Again, it is concluded that it is
129
appropriate to express the residuals o f the ARCH(1) estimations o f the terms o f trade as
percentages o f the contemporaneous levels.
6.d.iii Level-bias of inflation
There are several reasons for not expressing the residuals of ARCH(1) estimates of
annual inflation rates as percentages of contemporaneous levels. First, inflation rates are
annual increases in the price level which do not carry over values from previous periods.
In other words, inflation rates are already differenced. The picture would be different if the
uncertainty would be calculated based on ARCH(1) estimations of the normalized
consumer price index. The case of the consumer price index would be comparable to the
cases o f the real exchange rate index and the terms o f trade index. Second, to express the
residuals of ARCH(1) estimations o f inflation rates as percentages would neglect the fact
that high inflation itself implies increased uncertainty. This is best explained by an example.
Assume that country A's inflation rates are 40% in year 1, 42% in year 2, and 45% in year
3. Country B's inflation rates are 2%, 3% and 4%, respectively. Common sense would
indicate that the high-inflation country faces more inflation uncertainty than the low level
inflation country. However, expressing the residuals as percentages of levels would
indicate that the low inflation country exhibits more inflation uncertainty. The remaining
problem of extreme high levels of residuals for hyperinflationary countries is better solved
by expressing the standard errors as logs, then by expressing the residuals as percentages
o f the levels.
130
6.e. Conclusion
As there is no perfect solution, it is suggested to use uncertainty proxies based on
absolute residuals for variables where the differences in levels are based on economic
reasons, and to use uncertainty proxies based on percentage residuals for variables where
differences in levels are either arbitrary or based on indexation. Therefore, measures of
inflation uncertainty will be based on absolute residuals of ARCH(1) estimations, measures
o f monetary uncertainty, real and nominal exchange rate uncertainty, real and nominal
GDP uncertainty, and terms of trade uncertainty will be based on percentage residuals.
7. Checking for Consistency
This section checks the uncertainty measure for consistency. It is based on
comparing the above uncertainty measures based on three different estimation methods.
ARCH(l), AR(1), and AR(2) estimations. While there are obviously some changes in the
resulting uncertainty measures, the relevant question is whether or not there are systematic
differences across estimation methods. For example, assume that GDP uncertainties of
four countries are first based on ARCH(1) estimations and provide uncertainty proxies of
3.2 for country I, 6.7 for country 2, 5.8 for country 3 and 7.9 for country 4. Then, assume
that the four uncertainty measures based on AR(l) estimations are 6.4, 13.4, 11.6, and
15.8, for the four countries, respectively. While this implies that all the proxies have
doubled, they reflect exactly the same relative uncertainty, and are thus not systematically
different. Hence, the absolute values of the uncertainty proxies are irrelevant because the
standard errors are only proxies of true but unmeasurable uncertainty. What matters is if
131
there would be huge relative differences in uncertainty across estimation method.
Therefore, an efficient and objective way o f comparison of the two sets o f uncertainty
proxies is to use correlation coefficients of uncertainty proxies received from the three
different estimation methods.
7.a. Comparing ARCH(l) with AR(1) Estimates
The first consistency check is to compare the standard errors of ARCH(1) with the
standard errors o f AR(1) estimations. That is, we compare the uncertainty measures based
on simple, unconditional AR(1) estimations to conditional ARCH(1) estimations. As can
be seen from Table 8 below, the correlation coefficients for ARCH (1) and AR(l)
standard errors are:
0.814 for inflation uncertainty
0.993 for money supply uncertainty,
0.999 for nominal exchange rate uncertainty,
0.991 for real exchange rate uncertainty,
0.997 for terms of trade uncertainty,
0.996 for nominal GDP uncertainty, and
0.999 for real GDP uncertainty.
Given these high correlation coefficients, it can be concluded that it is not
necessary to differentiate between uncertainty measures based on ARCH(l) or AR(1)
residuals. Obviously, this result would not hold true for a time-variant uncertainty
132
measure. Recall that the overall stability of the time-invariant proxy was the reason to
calculate it instead of a time-variant arbitrary uncertainty measure.
7.b. AR(1) versus AR(2) Uncertainties
The next question is whether or not there is a systematic difference between
uncertainty measures based on AR(1) and AR(2) residuals. We are also interested in the
relationship between uncertainty measures between ARCH(1) and AR(2) residuals. The
various correlation coefficients are shown in Table 8. It turns out that the correlation
coefficients of AR(1) and AR(2) uncertainty proxies are generally slightly higher than
between ARCH(1) and AR(1):
0.997 for inflation uncertainty
0.863 for money supply uncertainty,
0.959 for nominal exchange rate uncertainty,
0.994 for real exchange rate uncertainty,
0.972 for terms of trade uncertainty,
0.921 for nominal GDP uncertainty, and
0.941 for real GDP uncertainty.
As the correlation coefficients are still within the range of 0.863 to 0.997, it is
appropriate to conclude that non-time variant uncertainty measures either based on
ARCH(1) or AR(1) or AR(2) are consistent uncertainty measures. The later analysis of
investment under uncertainty can use uncertainty measures based on ARCH(1) residuals,
without implying that the result is specific to the ARCH(1) estimation method. The high
133
correlation coefficients between the three uncertainty measures also imply that the
practical relevance o f more and more sophisticated improvements o f ARCH measures is
quite limited for estimating uncertainty.
Table 8: Correlation Matrices of Three Different Uncertainty
UNCGDP UNCGDPAR1 UNCGDPAR2

UNCGDP 1.000 0.996 0.938
UNCGDPAR1 0.996 1.000 0.921
UNCGDPAR2 0.938 0.921 1.000
UNCHES UNCHESAR1 UNCHES AR2

UNCHES 1.000 1.000 0.943
UNCHESAR1 1.000 1.000 0.942
UNCHESAR2 0.943 0.942 1.000
UNCINF UNCINFAR1 UNCINFAR2

UNCINF 1.000 0.815 0.845
UNCENFAR1 0.815 1.000 0.997
UNCINFAR2 0.845 0.997 1.000
UNCMON UNCMONAR1 UNCMONAR2

UNCMON 1.000 0.993 0.870
UNCMONAR1 0.993 1.000 0.863
UNCMONAR2 0.870 0.863 1.000
UNCNEX UNCNEXAR1 UNCNEXAR2

UNCNEX 1.000 1.000 0.956
UNCNEXAR1 1.000 1.000 0.956
UNCNEXAR2 0.956 0.956 1.000
UNCREX UNCREXAR1 UNCREXAR2

UNCREX 1.000 0.991 0.974
UNCREXAR1 0.991 1.000 0.994
UNCREXAR2 0.974 0.994 1.000
UNCTOT UNCTOTAR1 UNCTOTAR2

UNCTOT 1.000 0.997 0.975
UNCTOTAR1 0.997 1.000 0.973
UNCTOTAR2 0.975 0.973 1.000
134
Though more sophisticated econometric methods may be more relevant for
forecasting purposes, they certainly feed into the view that econometricians are people
finding non-existent black cats in dark rooms. It is for this reason and our satisfactory
results just above that we are not discussing additional alternatives of specifying
uncertainty, like for example, the use o f vector autoregessions (VARs).
8. How Different are Volatility and Uncertainty in Reality?
This section will compare volatility measures with uncertainty measures. We have
calculated the simple standard deviations measuring volatility and compared them (via
correlation coefficients) to the three uncertainty measures obtained in section 7 of this
chapter. The results are presented in extended correlation matrices shown in Table 9
below. As can be seen from Table 9, the correlation between nominal GDP volatility
(VOLGDP) and any o f the three nominal GDP uncertainty proxies is negative (within the
range of -0.264 to -0.256). Interestingly, the correlation between real GDP volatility and
real GDP uncertainty is basically the same as for the correlation between nominal GDP
volatility and nominal GDP uncertainty. This indicates that the differences between
estimation methods are the same for real and nominal GDP. However, it does not imply
that real and nominal uncertainty measures are the same. The correlations between the
volatility and uncertainty o f the money supply are also negative (within the range of -0.104
to -0.098). The correlation between the volatility and uncertainty of the nominal exchange
rate is nearly zero (ranging between -0.001 to 0.01).
135
Table 9: Correlation Matrices of Three Uncertainty and One Volatility Measure
UNCGDP UNCGDPARI UNCGDPAR2 VOLGDP

UNCGDP 1.000 0.996 0.938 -0.264
UNCGDPARI 0.996 1.000 0.921 -0.261
UNCGDPAR2 0.938 0.921 1.000 -0.256
VOLGDP -0.264 -0.261 -0.256 1.000
UNCHES UNCHESAR1 UNCHESAR2 VOLHES

UNCHES 1.000 1.000 0.943 -0.270
UNCHESAR1 1.000 1.000 0.942 -0.268
UNCHESAR2 0.943 0.942 1.000 -0.258
VOLHES -0.270 -0.268 -0.258 1.000
UNCINF UNCINFAR1 UNCINFAR2 VOLINF

UNCINF 1.000 0.815 0.845 0.835
UNCINFAR1 0.815 1.000 0.997 0.998
UNCINFAR2 0.845 0.997 1.000 0.999
VOLINF 0.835 0.998 0.999 1.000
UNCMON UNCMONAR1 UNCMONAR2 VOLMON

UNCMON 1.000 0.993 0.870 -0.103
UNCMONAR1 0.993 1.000 0.863 -0.104
UNCMONAR2 0.870 0.863 1.000 -0.098
VOLMON -0.103 -0.104 -0.098 1.000
UNCNEX UNCNEXAR1 UNCNEXAR2 VOLNEX

UNCNEX 1.000 1.000 0.956 0.010
UNCNEXAR1 1.000 1.000 0.956 0.010
UNCNEXAR2 0.956 0.956 1.000 -0.001
VOLNEX 0.010 0.010 -0.001 1.000
UNCREX UNCREXAR1 UNCREXAR2 VOLREX

UNCREX 1.000 0.991 0.974 0.927
UNCREXAR1 0.991 1.000 0.994 0.965
UNCREXAR2 0.974 0.994 1.000 0.984
VOLREX 0.927 0.965 0.984 1.000
UNCTOT UNCTOTAR1 UNCTOTAR2 VOLTOT

UNCTOT 1.000 0.997 0.975 0.839
UNCTOTAR1 0.997 1.000 0.973 0.842
UNCTOTAR2 0.975 0.973 1.000 0.832
VOLTOT 0.839 0.842 0.832 1.000
136
However, there are strong correlations between volatility and uncertainty for the inflation
rate, the real exchange rate, and the terms of trade. Based on these empirical results, the
difference between volatility and uncertainty is of relevant for real and nominal GDP, for
the nominal exchange rate, and money supply, though less so for the inflation rate, the real
exchange rate, and the terms of trade.
9. The Relationship of Uncertainties Across Areas
This section will compare the modified uncertainty measures derived from the
residuals o f ARCH(l) processes across the seven areas o f consideration: nominal and real
GDP uncertainty, inflation uncertainty, monetary uncertainty, nominal and real exchange
rate uncertainty, and terms of trade uncertainty. The correlation matrix of Table 10 shows
all 21 relationships between these seven uncertainties. The most important observation is
that there are no negative correlations between any of the seven macroeconomic
uncertainty measures. This is indeed an important result as a negative correlation between
any two macroeconomic uncertainties would be inconsistent with the overall hypothesis of
a circular relationship between development, economic structure, macroeconomic
uncertainty and investment, as illustrated in Figure 1.
9.a. Positive Correlations
Nominal and real exchange rate uncertainty are (with a correlation coefficient of
0.96) highly correlated to each other. Nominal and real exchange rate uncertainty are also
closely related to inflation uncertainty (0.80 and 0.78) and to monetary uncertainty (0.53
and 0.65). Inflation uncertainty and monetary uncertainty are with a correlation coefficient
137
o f 0.62 also related to each other. Finally, there is a close relationship between nominal
and real GDP uncertainty.
9.b. Low or Zero Correlations
Terms of trade uncertainty does not show any strong relationship to any o f the
other 6 uncertainties. The correlations are basically zero (lower than 0.02 in absolute
value) between (a) real GDP and inflation uncertainty, (b) real GDP and nominal exchange
rate uncertainty, and (c) nominal exchange rate and terms of trade uncertainty.
Table 10: Correlation Matrix of Uncertainties Across Areas
uncgdp unches uncinf uncmon uncnex uncrex unctot
uncgdp 1.00 0.50 0.06 0.27 0.03 0.21 0.25
unches 1.00 0.02 0.35 0.01 0.16 0.33
uncinf 1.00 0.62 0.79 0.78 0.03
uncmon 1.00 0.53 0.65 0.20
uncnex 1.00 0.96 0.02
uncrex 1.00 0.09
unctot 1.00
138
10. The Relationship of Uncertainties Across Periods
The last section o f this chapter compares the suggested uncertainty measures based
on ARCH(1) residuals across three time periods: (a) the overall time period from 1970-94,
(b) the first 12-year sub-period from 1970-1981, and (c) the last 12-year sub-period from
1983-94. The break at and exclusion of 1982 not only splits the overall 25 year period into
two 12-year sub-periods, but also allows for a comparison o f uncertainty measures before
and after the debt crisis o f 1982. While the nearly 2,000 uncertainty measures are
presented in seven tables in Appendix 5, here we will only discuss the few major
differences o f uncertainty measures across time periods.
As Table 11 demonstrates, there are some differences in the average uncertainty
across time periods. The uncertainty of nominal GDP, money supply, nominal and real
exchange rate were on average more than 20% higher after the debt crisis than before the
debt crisis. On the other hand, real GDP uncertainty and terms of trade uncertainty were
about 20% lower after the debt crisis than before the debt crisis. Table 11 also displays the
average uncertainty per time period separately for the group o f industrialized and
developing countries. Though there are considerable differences within each group of
countries, the changes in the two groups of countries' macroeconomic uncertainty have
generally moved in the same direction. The exception is inflation uncertainty. The
industrialized countries' inflation uncertainty was about 50% lower between 1983-94 than
between 1970-81. The opposite is the case for the group o f LDCs.
139
Table 11: Summary of Average Uncertainties Across Periods

A v c n ra o f nom inal GDP uncertain tv
Periods 21 DCs 72 LDCs All 93
1970-81 936 11.99 11.40

1983-94 11.18 1530 1437
1970-94 10.80 15.41 1437
Averaees of real GDP uncertainty

1970-81 2.77 6.10 5.35

1983-94 2.16 5.09 4.43
1970-94 2.60 6.09 530
Averaees of inflation uncertainty

1970-81 4.41 19.99 16.14

1983-94 2.13 345.69 266.41
1970-94 3.08 198.10 154.07
Averaees o f money j u d d I v uncertainty

1970-81 6.66 13.02 11.61

1983-94 7.11 2132 18.23
1970-94 735 18.68 16.19
Averaees of nominal exchanec rate uncertainty

1970-81 9.07 15.37 13.99

1983-94 11.86 22.85 20.43
1970-94 10.81 3836 3233
Averaees o f real exchanee rate uncertainty

1970-81 9.44 13.74 12.80

1983-94 12.01 19.05 17.50
1970-94 10.67 1534 1434
Averaees of TO T uncertainty
1970-81 6.61 15.42 13.41

1983-94 4.31 11.05 9.51
1970-94 5.81 13.92 12.06
CHAPTER V
THE RELATIONSHIP BETWEEN UNCERTAINTY AND
ECONOMIC STRUCTURE
The goal o f this chapter is to show that some economic structures are
systematically related to macroeconomic uncertainty as far as this is relevant to the
relationship between development, economic structure, macroeconomic uncertainty and
investment. Such a circular relationship was proposed and illustrated in Figure 1. Since
there are 47 structural variables and 7 macroeconomic uncertainties, we could examine up
to 329 relationships. Obviously, it would go beyond the scope of this dissertation to
analyze 329 relationships in the same depth as it was done for the 47 relationships between
development and economic structure of Chapter HI. A strategy is therefore needed to
reduce and simplify this examination.
One way to reduce the number of relationships to analyze is to drop the
examination of structural variables for which no simple relationship could be established
between economic structure and development. The rationale here is that we are interested
in establishing the circular relationship of Figure 1. A second possibility for reducing the
number o f relationships to analyze is to drop the analysis of the four investment variables,
140
141
as the relationship between uncertainty and investment will be dealt with separately in
Chapter VI. Third, it is suggested to limit the examination o f export concentration and
export market power to the overall export concentration index (EXSEX) and the overall
index of export market power (EXSW). This leaves us with the examination of
relationships between 24 structural variables1 and 7 macroeconomic uncertainties, or in
other words, 168 relationships.
Given the large number of 168 relationships, it is appropriate to simplify and
reduce the tools o f analysis. First, since the uncertainty measures are time-invariant the
analysis of correlation coefficients is limited to correlation coefficients of cross-country
data. Second, preliminary examinations of scatter diagrams indicated that log-linear
specifications are generally the more appropriate regression specifications. It has therefore
been decided to analyze only log-linear relationships. The only two exceptions are for
relationships analyzing the current and capital accounts, which have been entered without
logs as these variables can take on negative values. Third, since there exist systematic
relationships between development and economic structure, we may have problems related
to multicollinearity if GDP per capita would be added as an exogenous variable.
Furthermore, besides controlling for a country's land area in equations where the
exogenous trade variable is expressed as percent of GDP, it is not clear what further
control variables are appropriate. The regressions of this chapter are therefore relatively
1agri, indus, manu, service, cons, conspr, savi, govexp, govrev, govtax, cuab, caab,
openness, expper, expmer, expmach, expmachex, expprimex, exsex, exsw, finl, fin2, fin4,
and monb.
142
simple. To compensate somehow for the simplicity of the regression, it will be required
that the coefficients are statistically significant at the 99% level.2 Fourth, it is suggested to
limit the analysis to simple monotone relationships. The examination of U-shaped or
inverse U-shaped relationships are postponed to a later, more detailed analysis.
The remainder of this chapter is organized as follows. The first section provides an
overview of the relationships between development (measured as real GDP per capita) and
macroeconomic uncertainty. Section 2 looks at cross-country correlation coefficients
between economic structure and macroeconomic uncertainty. Section 3 analyzes variable-
specific group averages o f uncertainty analog to the group average analysis of chapter III.
Section 4 examines the results o f the regressions and section 5 summarizes the
relationships between economic structure and macroeconomic uncertainty in 10 stylized
facts.
1. The Relationships between Development and Uncertainty
The first part o f Table 12 demonstrates that the correlations between seven
macroeconomic uncertainties and real GDP per capita [either non-adjusted (gdpcap) or
adjusted (hescap) for differences in purchasing power] are always negative. The strongest
relationships are between (a) real GDP uncertainty and GDP per capita, (b) terms of trade
uncertainty and GDP per capita, (c) monetary uncertainty and GDP per capita, and (d)
2 Obviously, it is generally not appropriate to consider a high level of significance as a

compensation for a poorly specified regression. Indeed, the requirement of a significance
level of 99% may imply that true relationships may be dropped from further
considerations. However, since the goal is to establish the most likely stylized facts this
143
nominal GDP uncertainty and GDP per capita. As the second and third parts of Table 12
demonstrate, these last three relationships are also consistent to group averages of the 31
countries with the lowest real GDP per capita (low-gdpcap and low-hescap), the 31
countries with a middle level of real GDP per capita (mid-gdpcap and mid-hescap), and
the 3 1 countries with the highest real GDP per capita (high-gdpcap and high-hescap).
Table 12: Macroeconomic Uncertainty and GDP per capita
Part 1: Correlation coefficients of uncertainty and real GDP per capita

gdpcap -0.22 -0.40 -0.09 -0.24 -0.06 -0.10 -0.32

hescap -0.23 -0.47 -0.11 -0.28 -0.09 -0.15 -0.33
Part 2: Group averages of uncertainties by income group (based on gdpcap)

GROUP gdpcap uncgdp unches uncinf uncmon uncnex uncrex unctot
low-gpdcap 273.50 5.73 6.80 156.68 19.48 565.94 20.22 13.29

mid-gdpcap 1063.57 14.76 5.47 249.83 18.19 32.18 15.11 13.82
high-gdpcap 10097.40 12.71 3.73 55.69 10.80 36.77 12.19 9.13
Part 3:Group averages of uncertainties by income group (based on hescap)

GROUP hescap uncgdp unches uncinf uncmon uncnex uncrex unctot
low-hescap 789.18 15.44 6.98 156.73 22.16 573.49 20.44 12.39

mid-hescap 2256.17 14.49 5.14 248.62 14.76 23.41 14.70 14.34
high-hescap 9132.32 13.29 3.88 56.85 11.46 37.58 12.48 9.48
Note: gdpcap = real GDP per capita not adjusted for differences in purchasing power
hescap = real GDP per capita adjusted for differences in purchasing power
strategy seems to be appropriate. It will be left to a more detailed analysis to establish

further stylized facts.
144
2. Insights from the Sign of the Correlation Coefficients between Economic

Structure and Macroeconomic Uncertainty
The signs of the correlation coefficients o f Table 13 provide an interesting initial
insight into the possible relationships between economic structures and macroeconomic
uncertainty. Looking at Table 13, we can make three interesting observations. First, we
notice some empty spaces in Table 13. These gaps are due to the fact that correlation
coefficients smaller than 0.10 in absolute value have been dropped from the analysis in
order to neglect correlation coefficients which are close to zero. Second, comparing the
correlations coefficients across columns we notice that there is a striking consistency in a
sense that if a correlation coefficient is positive for GDP uncertainty, it is also positive for
any o f the other six macroeconomic uncertainties. Third, we notice that all structural
variables which are positively correlated to development (indus, manu, service, savi,
govexp, govrev, govtax, cuab, openness, expper, expmer, expmach, expmachex,
exsw, flnl, fin2, fin4, and monb) are negatively correlated to macroeconomic
uncertainty. On the other hand, all structural variables which are negatively correlated to
development Cagri. cons, conspr. caab. expprimex. and exsex) are positively correlated to
macroeconomic uncertainty. These differences in the signs of the correlation coefficients
are consistent with Figure 1, where it was suggested that underdeveloped economic
structures increase macroeconomic uncertainty.
145
Table 13: Correlation Coefficients between Economic Structure

and Macroeconomic Uncertainty
agri 0.13 0.56 0.24 0.10 0.11

indus -0.28 -0.17 -0.17
manu -0.26 -0.59 -0.29 -0.17 -0.42
service -0.16 -0.60 -0.21 -0.13 -0.36
cons 0.12 0.31 0.26 0.13

conspr 0.15 0.26 0.10 0.28 0.15
savi -0.12 -0.31 -0.26 -0.13
govexp -0.14 -0.17 -0.15 -0.19 -0.15 -0.11

govrev -0.18 -0.34 -0.14 -0.26 -0.23 -0.19
govtax -0.18 -0.44 -0.16 -0.34 -0.25 -0.27
cuab -0.19 -0.44 -0.15

caab 0.10 0.25 0.21 0.12 0.15
openness -0.16 -0.12 -0.19 -0.11
expper -0.16 -0.10 -0.11 -0.21 -0.11
expmer -0.12 -0.17
expmach -0.29 -0.30 -0.17 -0.12 -0.36

expmachex -0.37 -0.49 -0.20 -0.16 -0.43
expprimex 0.27 0.47 0.15 0.34 0.14 0.55
exsex 0.33 0.68 0.33 0.18 0.66
exsw -0.19 -0.35 -0.19 -0.23
finl -0.33 -0.50 -0.17 -0.31 -0.10 -0.17 -0.34

fin2 -0.25 -0.37 -0.29 -0.49 -0.24 -0.31 -0.26
fin4 -0.32 -0.50 -0.17 -0.36 -0.12 -0.20 -0.39
monb 0.38 -0.46 -0.18 -0.28 -0.13 -0.20 -0.31
146
3. The Relationship between Economic Structure and Uncertainty

Tested bv Group Averages
The purpose of this section is to check the relationships between economic
structure and macroeconomic uncertainty by comparing the average uncertainties of three
variable-specific groups. While this approach is analog to the analysis of group averages of
Chapter HI, it is best explained by an example.
When analyzing the relationship between the share of agriculture and nominal GDP
uncertainty, the total o f 93 countries are divided into three groups: (a) the 31 countries
with the lowest shares o f agriculture and their corresponding average GDP uncertainty,
(b) the 31 countries with the mid-level shares o f agriculture and their corresponding
average GDP uncertainty, and (c) the 31 countries with the highest shares o f agriculture
and their corresponding average GDP uncertainty.
The consistency check of a positive relationship between the share o f agriculture
and GDP uncertainty requires that the group of 31 countries with the lowest shares of
agriculture has the lowest average GDP uncertainty. Similarly, the group of 31 countries
with the highest shares o f agriculture needs to have the highest average GDP uncertainty.
In other words, the three country group values will need to be monotone increasing for a
positive relationship between economic structure and uncertainty, and monotone
decreasing for a negative relationship. This consistency check o f variable-specific group
averages is applied to all structural variables.
While the group averages for all 168 relationships between economic structure and
147
macroeconomic uncertainty are provided in Table 14, we have used a large font for those
group averages where (a) the absolute value of correlation coefficients is at least 0.10, and
(b) where there are either monotone increasing or monotone decreasing group averages.
It turns out that this is the case for 95 relationships between economic structure and
macroeconomic uncertainty. It is striking to note that all these 95 structure-uncertainty-
relationships are consistent to the results of Chapter III and the overall relationship
illustrated in Figure 1. We now filter out the most robust structure-uncertainty-
relationships by testing the relationships with regression analysis.
4. Regressing Macroeconomic Uncertainty on Economic Structure
We have run 168 mostly bivariate regressions between the 24 structural variables
which established a stylized fact in Chapter III and the seven macroeconomic
uncertainties. All statistically significant t-statistics of these 168 regressions are reported in
Table 15. It turns out that 142 t-statistics are statistically significant at the 95% level. As
before, these t-statistics are marked with a double-star (**). After requiring a 99% level of
significance, the number reduces to 92, which are marked with a triple-star (***). As
before, the regressions should not be understood as a determination of causality,3 but as an
additional tool of analysis.
3 This does not contradict our above hypothesis that uncertainty is the partial result of
economic structures, as causality may run both directions. Economic structures have
implications for macroeconomic uncertainty and macroeconomic uncertainty has
implications for economic structures through the negative impact of uncertainty on
investment and growth, hence the circular relationship of Figure 1.
148
Table 14: Group Averages of Uncertainty by Structural Variables
group and group value uncgdp unches uncinf uncmon uncnex uncrex untot
low-agri 5.77 13.12 3.87 58.98 11.72 38.78 12.41 9.65
mid-agri 20.84 14.43 5.10 388.88 16.99 56621 19.84 13.74
high-agri 40.71 15.66 7.04 14.33 19.74 30.03 1529 12.42
low-indus 17.36 15.38 6.92 14.21 20.56 31.57 15.51 12.24

mid-indus 29.56 14.02 4.43 333.17 14.78 581.30 19.78 12.06
high-indus 39.85 13.89 4.65 114.82 12.75 39.40 12.58 11.89
low-manu 8.24 16.08 7.45 15529 22.25 590.91 21.54 14.91
mid-manu 16.50 12.77 4.24 98.16 13.77 19.31 11.65 12.08
high-manu 25.00 14.49 4.07 22229 11.72 43.67 14.88 8.89
low-service 36.88 15.07 7.08 10326 17.91 22.55 15.48 14.35
mid-service 48.80 15.93 5.14 349.31 20.03 594.71 20.67 13.30
high-service 6 0.21 12.19 3.79 9.64 10.58 16.30 11.32 8.58
low-cons 72.38 12.90 4.48 56.70 11.68 36.72 11.96 11.31
mid-cons 81.24 14.37 4.47 156.16 13.38 22.86 12.36 12.69
high-cons 96.36 15.99 7.05 249.33 22.60 574.85 23.21 12.19
low-conspr 55.83 12.88 4.46 29.81 11.73 15.17 12.30 10.81

mid-conspr 67.27 15.23 4.82 186.61 12.20 45.01 14.77 12.81
high-conspr 83.37 15.17 6 .7 1 245.79 24.37 57427 20.55 12.59
low-savings 3.65 15.99 7.05 249.33 23.66 574.85 23.21 12.19

mid-savings 18.93 14.50 4.43 156.91 13.48 23.66 12.31 12.78
high-savings 27.53 12.77 4.53 55.96 11.59 35.92 12.00 1122
low-govexp 9.78 15.00 5.20 259.88 18.46 58623 18.49 12.96

mid-govexp 14.41 14.68 4.98 100.90 15.49 28.11 13.04 12.17
high-govexp 19.98 13.51 5.70 104.43 14.06 19.80 16.16 11.06
low-govrev 11.93 14.81 6.20 182.07 19.71 25.42 15.07 12.71
mid-govrev 19.78 15.23 5.14 130.12 13.98 51.11 1528 12.19
high-govrev 30.10 11.46 3.99 7.18 10.51 13.76 10.40 8.84
low-govtax 10.16 15.06 6.47 181.83 20.20 25.53 15.30 13.09
mid-govtax 16.65 13.96 5.04 129.13 14.08 49.97 14.51 11.41
high-govtax 25.52 12.52 3.83 8.41 9.91 15.92 11.05 9.25
low-cuab -14.80 16.24 6.70 195.79 18.82 34.06 1726 1326
mid-cuab -4.95 13.63 5.39 209.30 18.62 580.65 17.93 12.27
high-cuab -0.36 13.45 3.91 57.11 11.06 37.75 12.53 10.70
low-caab 0.06 13.50 4.24 57.58 12.68 38.91 13.09 10.50

mid-caab 3.29 13.67 5.58 64.74 17.90 26.05 12.90; 12.44
high-caab 7.41 16.06 6.10 339.88 17.94 587.76 21 .88! 13.20
Table 14 continues on the next page.
149
Table 14: Group Averages of Uncertainty by Structural Variables (cont.)
g r o u p a n d g ro u p v a lu e uncgdp unches uncinfnp uncmon uncnex uncrex untot

tow-open 0.35 14.96 5.36 258.54 20.38 612.00 19.81 11.84
mid-open 0.57 13.62 4.99 102.26 16.17 22.03 12.36 12.11
high-open 1.13 14.68 5.60 101.39 11.88 18.83 15.71 12.26
low-expper 13.57 14.62 5.74 258.57 20.35 612.69 19.93 10.82

mid-expper 24.61 15.29 5.32 196.34 17.06 24.39 15.60 13.56
high-expper 53.12 13.31 4.90 7.29 11.03 15.70 12.24 11.80
low-expmer 9.88 14.51 5.77 116.12 18.94 52.46 14.75 10.78

mid-expmer 18.48 14.92 5.06 337.91 17.55 565.17 20.54 11.99
high-expmer 36.72 13.81 5.13 8.17 12.00 17.12 12.32 13.03
low-expmach 0.10 16.82 7.04 303.27 23.43 596.27 23.83 13.95

mid-expmach 0.47 15.22 5.30 63.43 14.36 42.26 13.43 13.95
high-expmach 7.24 9.77 2.81 93.79 8.95 11.14 9.14 6.93
low-expmachex 0.41 16.57 6.86 248.02 23.17 593.27 23.52 15.90
mid-expmachex 2.26 14.93 5.40 91.85 14.04 23.43 13.64 12.27
high-expmachex 21.31 11.48 3.20 133.55 10.60 37.33 10.26 7.37
lo w -e x p p rim e x 19.61 12.12 3.61 6.23 10.29 13.80 11.44 7.96
m id -e x p p rim e x 48.48 14.52 5.13 203.39 16.57 585.29 17.69 11.94
h ig h - e x p p rim e 77.13 16.88 7.10 277.75 22.00 35.30 18.75 16.53
lo w -e x s e x 30.31 11.74 3.12 56.02 9.47 35.86 10.94 6.64
m id -e x s e x 59.76 15.43 5.44 248.60 16.52 22.53 15.55 13.15
h ig h - e x s e x 85.15 16.17 7.39 168.20 23.06 614.82 21.68 16.71
low-exsw 3.51 15.93 6.79 106.42 16.01 24.72 16.62 13.18
mid-exsw 14.24 14.36 4.70 297.66 20.93 574.69 18.40 13.62
high-exsw 39.61 12.92 4.23 61.87 11.72 36.59 12.66 9.21
low-tin1 0.20 15.78 6.52 327.45 20.19 592.37 20.11 14.65

mid-fin 1 0.31 16.30 5.95 130.56 19.15 29.22 16.92 13.19
high-fin1 0.62 11.07 3.51 4.19 9.11 12.25 10.59 8.24
low-fin2 0.47 16.25 6.36 369.42 24.68 599.32 22.77 13.25
mid-fin2 0.69 15.22 5.75 87.95 14.44 22.81 13.44 13.43
high-fin2 0.90 11.43 3.47 4.72 8.75 11.95 10.80 8.87
low-fin4 0.11 16.46 7.43 300.55 25.26 576.45 21.13 14.67

mid-fin4 0.22 14.91 4.93 127.52 13.58 22.61 15.62 13.36
high-fin4 0.55 11.91 3.66 34.13 9.43 35.76 10.67 8.24
low-money 21.18 15.92 6.47 298.35 19.86 566.87 19.31 14.53
mid-money 35.26 16.15 6.10 159.15 19.10 54.99 17.62 13.25
high-money 70.14 11.09 3.42 4.70 9.50 12.84 10.72 8.45
150
Table 15: Significant t-statistics based on Regression Results
uncgdp unches uncinfnp uncmon uncnex uncrex unctot

AGRI **2.23 *** 6.75 ***3.01 ***4.15 **2.36 **2.11 — 4.74
INDUS ***-3.25 **-2.29
MANU **-2.41 ***-6.20 *** -3.51 **-2.45 *** -3.75
SERVICE ***-6.77 **-2.29 ***-3.49 —-2.05 ***-4.26
CONS ***2.75 **2.41 **2.15 **2.11
CONSPR ***2.59 *** 2.63 *** 2.60 **2.58
SAVI ***-2.51 *** -2.68 ** -2.27
GOVEXP **-1.94 ***-2.74 *** -2.78 ** -2.07

GOVREV *** -3.66 *** -3.54 *** -3.68 — -2.83
GOVTAX *** -4.73 *** -3.89 *" -5.00 *** -3.31
CUAB **-1.96 *** -5.10 **-2.19 **-2,28 **-2.23

CAAB **2.03 ***3.59 *** 3.59 **2.08 **2.52 -2.29 **2.46
OPENNES;s *** -2.55 ** -2.04 **-2.05
EXPPER **-2.06 *** -3.05 *** -2.86 *** -2.56 **-2.16

EXPMER **-1.99 **-1.96 **-2.26
EXPMAC *** -3.95 *** -8.14 ***-3.61 *** -5.75 ***-4.10 *** -4.54 *** -6.15
EXPMAC *** -4.17 *** -8.46 ***-2.70 *** -5.06 ***-3.28 *** -4.41 *** -6.48
EXPPRIM *** 3.33 ***5.43 *** 4.29 *** 4.68 — 2.87 *** 2.58 *** 5.72
EXSEX *** 3.66 *** 9.74 ***3.17 *** 6.30 — 2.61 *** 3.30 *** 9.67
EXSW *** -4.90 **-2.02 ***-3.09 **-2.29 — -2.78
FIN1 ***-3.51 *** -6.83 *** -5.07 *** -5.44 *** -3.91 *** -3.17 ***-5.11
FIN2 **-1.98 *** -3.58 *** -5.40 *** -5.84 *** -4.45 *** -3.27 — -3.24
FIN4 *** -3.90 *** -7.31 *** -4.61 *** -6.86 ***-4.13 *** -4.35 *** -5.33
MONB | *** -3.73 *** -6.06 *** -4.41 *** -4.81 — -3.91 *** -3.20 *** -4.52
151
S. Remaining Stylized Facts
Finally, we require a consistency between the group averages of section 3 and the
highly significant correlation coefficients of section 4. This reduces the number of robust
relationships to 73. Table 15 has marked the t-statistics of these 73 relationships in bold
and a slightly larger font. Finally, looking at the consistency of these 73 relationships
across similar variables, the overall 10 strongest stylized facts of structure-uncertainty-
relationships are as follows.
Stylized Fact #1:
The share of agriculture is positively related to real GDP uncertainty and money
supply uncertainty.
Stylized Fact #2:
The share of manufacturing is negatively related to real GDP uncertainty and
money supply uncertainty.
Stylized Fact #3:
The share of private consumption to GDP is positively related to real GDP
uncertainty, inflation uncertainty, money supply uncertainty, and nominal exchange rate
uncertainty.
152
Stylized Fact #4:
The shares of government revenues (either total or tax revenues) are negatively
related to real GDP uncertainty, inflation uncertainty, money supply uncertainty, and also
to terms of trade uncertainty.
Stylized Fact #5:
Current account deficits and capital account surpluses are both positively related to
real GDP uncertainty. In other words, countries with current account surpluses face lower
real GDP uncertainty than countries with current account deficits.
Stylized Fact #6:
The share of machinery exports is negatively related to nominal and real GDP
uncertainty, monetary uncertainty, real exchange rate uncertainty and terms o f trade
uncertainty. This stylized fact holds both for machinery exports expressed as a share of
GDP and machinery exports expressed as a share o f total exports.
Stylized Fact #7:
Consistent with the last stylized fact, the higher the country's share o f primary
exports in total exports, the higher the country’s uncertainty in terms of nominal and real
GDP, money supply, real exchange rate, and terms o f trade.
153
Stylized Fact #8:
Similar to the last stylized fact, it is also the case that a high export product
concentration is positively correlated with uncertainty of nominal and real GDP, money
supply, real exchange rate, and terms of trade.
Stylized Fact #9:
The higher a country's export market power, the lower a country's real GDP
uncertainty.
Stylized Fact #10:
Financial market development, appropriately defined by FTN1, FIN2, FIN4, and
the share o f broad money to GDP, is negatively related to real GDP uncertainty, inflation
uncertainty, money supply uncertainty, and real exchange rate uncertainty.
CHAPTER VI
INVESTMENT UNDER UNCERTAINTY
How does uncertainty affect the investment decision? There are over 100 studies
which claim to analyze investment under uncertainty. However, most o f this literature
does not differentiate appropriately between risk, volatility and uncertainty. Indeed, most
o f the investment literature is at the microeconomic level, analyzing the impact of risk on a
firm's investment decision from a theoretical point of view. This chapter deals with the
issue of investment under uncertainty at the macro level, but not with issues of classical
risk. The latter will therefore only be reviewed very briefly in the first section.
The intuitive yet wrong explanations for a positive correlation between uncertainty
and investment are examined in section 2. The empirical literature on investment under
uncertainty and volatility will then be reviewed in section 3. This will, together with
Chapter IV, provide some inputs for the initial specification o f the aggregate domestic
investment function (section 4), which will be estimated using panel data o f 93 countries
from 1970-94. The results of the initial specification will then be reported in section 5.
Finally, section 6 will test the robustness of the results of the initial investment
function by running regression for (a) a variety of slightly modified specifications, (b) two
154
155
12-year sub-periods instead of the whole 25-year period, and (c) substituting gross
domestic investment with fixed private investment.
1. Investment and Risk and the Micro Level
l.a. Risk Aversion
After the defining contribution of Knight (1921), the first major microeconomic
analysis of risk and investment is Hart (1942). Hart considered flexible production
decisions under risk in a two-period model. Assuming that investors are risk averse, risk
has a negative impact on investment. However, even risk averse investors are not scared
off if possible returns are high enough to compensate for risk. Much of the early literature
is along these lines.
l.b. Risk as an Incentive to Invest
The next major contribution to the risk and investment literature is Hartman
(1972), who argued that industries must be imperfectly competitive in order for risk to
reduce investment. According to Hartman, uncertainty1 increases investment of
competitive risk-neutral firms. As Federer (1993a) points out, this last result is due to
Jensen's inequality: if marginal revenue product of capital is convex in price, then a mean-
preserving increase in price risk raises the expected payoff to marginal units of capital and
stimulates investment. Most of the theoretical literature of the 1980s is along this line,
especially Abel (1983, 1984, and 1985), Bemanke (1983), and to some degree also
Cukierman (1980). As all of this literature has substituted risk with uncertainty, the then
1What Hartman meant is risk, not uncertainty.
156
dominant view was one where uncertainty has a positive impact on investment. This view
got challenged towards the late 1980s, when it became popular to analyze issues o f
investment irreversibilities.
I.e. Irreversible Investment and the Real Option Theory
As Solimano (1992) has pointed out, irreversibility is after risk aversion, the
second justification for risk to affect investment adversely. Irreversibility is the main
explanation for the emergence o f the real option theory of investment under uncertainty.
While most o f the mainstream literature of the 1970s and 1980s has neglected the issue of
irreversibility, a few studies applied the option theory to environmental issues of
irreversible investment in previously undisturbed land or sea. Arrow and Fisher (1974), as
well as Henry (1974) dealt with the irreversible tradeoff between development and
preservation. Other major contributions along this line are Fisher and Hanemann (1986,
1990) and Hanemann (1989). However, it was not until the early 1990s, that the
implications of irreversibility became part of the mainstream literature on investment under
uncertainty. Due to the contributions of Bertola and Caballero (1994), Dixit (1989), Dixit
and Rob (1994) Ingersoll and Ross (1992), Pindyck (1990, 1991, and 1993), Pindyck and
Solimano (1993), and Rodrik (1991)2 the real option theory of investment under
uncertainty became the major challenge to the earlier view of a positive impact of
uncertainty on investment.
2 Developing a simple model to link policy uncertainty to the private investment

response, Rodrik demonstrates that even moderate amounts of policy uncertainty can act
157
Recently, the microeconomic analysis of investment under uncertainty has swapped
over to macroeconomic applications. For example, Fischer (1993) distinguishes between
policy-induced macroeconomic uncertainty and temporary uncertainty about the
macroeconomy. As Fischer (1993, p. 488) points out, both have a negative impact on
investment and growth:
First, policy-induced macroeconomic uncertainty reduces the efficiency of the price

mechanism, as in the classic Lucas (1973) contribution. (...) Second, temporary
uncertainty about the macroeconomy tends to reduce the rate o f investment, as
potential investors wait for the resolution of the uncertainty before committing
themselves [Pindyck and Solimano (1993)].
A financial channel through which uncertainty affects investment has been
suggested recently by Mishkin (1997, p. 40):
A dramatic increase in uncertainty in financial markets, due perhaps to the failure

of a prominent financial or nonfinancial institution, a recession, political instability,
or a stock market crash, makes it harder for lender to separate good from bad
credit risks, the increase in uncertainty therefore makes information in the financial
markets even more asymmetric and worsens the adverse selection problem. The
inability o f lenders to solve the adverse selection problem renders them less willing
to lend, leading to a decline in lending, investment, and aggregate economic
activity.
What becomes clear is that economic theory took a large U turn in how it viewed
the impact o f uncertainty on investment. In difference to the original view of a negative
impact,3 the 1970s and 1980s were dominated by the view of a positive impact. Only since
as a hefty tax on investment. A similar though more general approach has been applied by
Faini and de Melo (1990).
3 As Federer (1993a) has shown, the negative impact o f uncertainty on investment can
be traced back at least to Alfred Marshall.
158
Dixit and Pindyck's (1994) Investment under Uncertainty, it is now again the dominant
view that uncertainty has a negative impact on investment.
2. Intuitive Explanations for a Positive Correlation?
There are generally two intuitive explanations for a positive correlation between
uncertainty and investment. The first intuitive explanation is related to the impact of
uncertainty on precautionary savings, the second intuitive explanation is related to the
positive correlation between high growth and high volatility.
2.a. The Impact of Precautionary Savings
The positive impact of uncertainty on precautionary savings was originally
suggested by Sandmo (1970). For the most recent confirmation o f the positive impact of
macroeconomic uncertainty on precautionary savings, see Gosh and Ostry (1997). The
increase in precautionary savings has then been used as an explanation for a positive
relationship between uncertainty and investment: "Instability reduces the propensity to
consume by requiring larger reserves for unexpected or temporary declines in income. The
higher rate of savings, which results from the lower propensity to consume, makes
possible higher investment and growth."4
There are major problems with this first explanation o f a positive relationship
between uncertainty and investment. First of all, the argument that increased savings
automatically leads to increased investment is false. The main reason for increased savings
under uncertainty is to have some reserves for unexpected or temporary declines in
4 Knudsen and Pames (1975), pp. 131-132.
159
income. These reserves would not be available immediately if invested. Indeed, the
precautionary reason which leads to increased savings is the very same reason which
makes investors more conservative about investing. This role of money as a hedge against
an uncertain future enabled Keynes (1936) to postulate a liquidity preference theory of the
rate of interest emphasizing the distinction between decisions to save and decisions to
invest. Furthermore, a lower propensity of consumption does not necessarily imply higher
absolute savings and investment, as lower levels o f consumption may imply lower levels of
income.
2.b. The Relationship between Volatility and Growth
The second intuitive explanation is based on the relationship between volatility and
growth. As Ramey and Ramey (1995, p. 1138-1139) point out "Black (1987) has argued
that countries may have a choice between high-variance, high-expected-retums
technologies and low-variance, low-expected-retums technologies. In such a world,
countries with high average growth would also have high variability." This second
explanation demonstrates how important it is to separate between volatility and
uncertainty. While higher growth rates may imply higher GDP volatility, they may also
imply less uncertainty.5 As will be shown in the next section, it is the wrong measure of
uncertainty, which misled many authors to conclude that export instability has a positive
impact on investment and growth.
5 On the other hand, there may be a positive correlation between higher volatility and
higher uncertainty in cases of large negative growth rates. It is this asymmetry which
makes it so difficult to relate volatility to uncertainty.
160
3. Empirical Analysis of Investment under Uncertainty
Based on the difference between uncertainty and volatility, it is possible to classify
the empirical investment under uncertainty literature into two categories. The first
category includes studies claiming to analyze investment under uncertainty, though they
actually analyze the relationship between investment and volatility or risk or some other
kind o f economic instability. The second category includes contributions following the
Knightian-Keynesian definition of uncertainty. The usual methodology of empirical studies
is to add one variable, which captures either institutional, political, or economic instability,
to the more traditional investment determinants.
3.a. Investment and Economic Instability
Excluding the now quite considerable literature on the negative impact of debt
overhang on investment6 and the few studies analyzing investment under Knightian-
Keynesian uncertainty, there are currently less than 30 empirical studies analyzing the
impact o f economic instability on domestic investment.7 The measures indicating economic
instability in these studies are related to relative prices and inflation, exchange rates, terms
o f trade and exports, and aggregate volatility in terms o f GDP or stock markets.
6 The major contributions here are Borensztein (1990a and 1990b), Cardoso (1993),
Cohen (1985 and 1993), Faini and de Melo (1990), FitzGerald, Jansen, and Vos (1994),
Greene and Villanueva (1990), Oshikoya (1994), Sawides (1992), Schmidt-Hebbel and
Muller (1992), Serven and Solimano (1992), (1993a) and (1993b), and Warner (1992).
The variable under consideration is usually the ratio of foreign debt to GDP. The debt
overhang can be interpreted as uncertainly related to the possibility of a country to repay
its debt.
7 There are also a few studies analyzing the impact o f exchange rate volatility on
foreign investment, see for example, Goldberg and Kolstad (1995).
161
3.a.i. Price and inflation volatility
There seem to be a total of 5 empirical studies dealing with the impact of price and
inflation volatility on domestic investment: Behrman (1972), Glezakos and Nugent (1996),
Hadjimichael and Ghura (1995), Serven (1997) and Serven and Solimano (1992, 1993a
and 1993b).8 Behrman (1972) estimated sectoral real physical capital investment functions
from time-series data for postwar Chile. In addition, he reviews and contributes to the
controversy o f investment determinants. Behrman (1972) seems to be the first to show
empirically that relative output price volatility (measured by standard deviation) has a
negative impact on domestic investment. Similarly, Serven and Solimano (1992, 1993 a
and 1993b) analyze the impact of inflation volatility of 15 developing countries from 1976-
88. Also, Glezakos and Nugent (1996) found negative impacts of both inflation volatility
and relative price volatility in an empirical study for the United States using quarterly data
from 1960-90. The analysis of 32 African economies by Hadjimichael and Ghura (1995)9
showed that the variability of inflation has a strong adverse impact on private investment
performance between 1986-1992. Finally, Serven (1997) showed that inflation volatility
has a negative impact on private investment for 40 economies in Sub-Saharan Africa and
about 40 developing countries in other regions.
8 The 1992 publication of Serven and Solimano's article in a World Bank symposium
on Adjustment Lending Revisited: Policies to Restore Growth is identical to the article in
the 1993 World Bank publication on Striving fo r Growth after Adjustment: The Role o f
Capital Formation, edited by Serven and Solimano (1993b). Serven and Solimano
(1993a) is basically a summary published in World Development, in January 1993.
9 See also the more comprehensive review of the African experience by Hadjimichael,
Nowak, Sharer Tahari et al. (1996).
162
3.a.ii. Exchange rate volatility
Three of the five above studies have also included the volatility o f real exchange
rate volatility as a determinant of private investment, though the results of these three
studies are mixed: Hadjimichael and Ghura (1995) find significantly negative impacts; the
results in Serven and Solimano (1992, 1993a and 1993b) depend on the estimation
model;10 while Serven (1997) does not find real exchange rate volatility to be significant.
However, there are seven further empirical studies analyzing the impact o f real exchange
rate volatility on investment, which all but one find a significant negative impact:
• Cardoso (1993) for private investment of six Latin American countries from
1970-85";
• Cottani, Cavallo and Khan (1990) for net investment of 24 developing
countries from 1960-83;
• Faini and de Melo (1990) for private investment in 20 developing countries
between 1970-86;
• Ghura and Grennes (1993) for total domestic investment of 33 Sub-Saharan
African economies during 1972-87;
• Goldberg (1993) finds inconclusive results for a study o f the United States;
10 In one model real exchange rate volatility is significant in another model it is not.
11 In Cardoso (1993) the real exchange rate volatility is part of an overall

macroeconomic instability index comprising the level of inflation rates and the debt ratio.
In the case o f including the terms of trade as a determinant, the macroeconomic instability
index is insignificant.
163
• Hausmann and Gavin (1995) for domestic investment of about 90 countries
from as early as 1960-85; and
• Larrain and Vergara (1993) for Korea, Singapore, Thailand, and Malaysia from
1971-88.
Furthermore, Serven (1997) has included the black market premium variability as a
determinant of private investment, which is however not significant.
3.a.iii. Terms of trade and export instability
Besides the variability of inflation, the variability of the real exchange rate, and the
variability o f the black market premium, Serven's (1997) fourth and final macroeconomic
instability measure is the variability in the terms of trade. His study finds that the terms of
trade variability has a significant negative impact on private investment.
As seen in Chapter IV, there are a number of studies which have analyzed the
impact of export instability on investment. The latest review of the more recent literature
is Love (1989).12 As in earlier studies, Love's analysis of 12 countries between 1960 and
the early 1980s is completely inconclusive. The main problem of most o f this literature is
that it defines export instability as deviations from the trend without correcting for
differences in the level of exports. This implies that countries with high export shares
exhibit high export instability. The subsequent mostly positive relationship between such
defined export instability and investment seems to be a reflection o f a positive relationship
12There are a number of more recent studies [e.g. Fernandez (1992) and El-Shamhouri
(1994)], who have however not analyzed the impact of export instability on investment.
164
between the level of export shares and the level of investment shares but should not be
interpreted as a positive relationship between export instability and investment.
Similarly, Knudsen and Pames (1975) used a measure o f export instability which is
biased not by the level but by the growth rate of the export sector. Countries with high
growth rates o f the export sector have higher export instability than countries with low
export sector growth rates. Using such a wrong measure o f uncertainty resulted in the
empirically significant positive correlation between export instability and investment.
3 a.iv. Aggregate domestic volatility
Only two empirical studies which have explicitly analyzed the impact of GDP
instability on investment. First, Solimano (1992) showed that output variance has a
statistically significant negative impact on Chile's private investment even though it did not
affect private capital formation.
Second, Hausmann and Gavin (1995) suggested that macroeconomic volatility,
including the standard deviation of real GDP growth, undermines investment in less
developed countries.13
13 There are two studies closely related to the analysis o f output volatility and
investment. Fist, Blejer and Khan (1984) used the cyclical deviation of GDP as a
determinant of private investment to control for the possibility that private investors
respond more rapidly to changes in desired investment during the expansionary phase of
the business cycle than in the contractionary phase. Second, Ramey and Ramey (1995)
include the investment share as a control variable in analyzing the relationship between
GDP growth and the standard deviation of real GDP growth. They conclude prematurely
that there is no relationship between GDP volatility and investment, since including
investment as a control variable did not alter the negative relationship between growth and
growth volatility.
165
3.a.v. Other volatility/instability measures
There are a few other instability measures which have been regressed on some kind
o f domestic investment. For example, Dailami (1987) concluded that stock market
volatility has a negative impact on private investment in Brazil. Sundararajan (1987)
showed that two dummy variables, representing economic policy changes and general
uncertainty, had a negative impact on investment in Korea.14 In addition to output
volatility, Solimano (1992) also analyzed the impact of the variance of Tobin's q. While
the variance o f Tobin's q bore a negative sign in the investment function, it was not
statistically significant. Finally, Aizenman and Marion (1993a and 1993b) used the
residuals of AR(1) processes of a variety of macroeconomic policy instruments15 to show
that volatility has a negative impact on private investment.
14 Sundararajan (1987) analyzed empirically the impact of interest rates on overall cost
o f capital, saving, investment and growth in the Korean economy during 1963-81. The
first dummy variable assumes a value o f unity in 1980 and 1981 and zero in other periods,
and represents the political and economic uncertainties that dominated those years
following the second oil shock and a change in political leadership. The second dummy
assumes the value o f unity from 1973 onward and zero in all previous years, and reflects
the shift in investment strategy to favor more capital intensive sectors, and the 1972
Presidential Decree that initiated substantive changes in financial sector policies.
15 In Aizenman and Marion (1993a), the policy instruments are:

a) the share of government consumption expenditure in GDP,
b) share of public investment in GDP,
c) the unexpected parts of domestic credit expansion, and
d) money growth. In Aizenman and Marion (1993b) they add:
e) the growth in the share of government consumption expenditures,
f) the average tax rate (as measured by the ratio of government revenue to GDP),
and
g) the government budget deficit scaled by GDP.
Reproduced with permission of the copyright owner. Further reproduction prohibited w ithout permission.
166
3.b. Investment under Knightian-Keynesian Uncertainty
As indicated in Chapter IV, there seem to be only seven empirical studies analyzing
investment under Knightian-Keynesian uncertainty, in the sense that they at least attempt
to measure the unpredictable part of fluctuations, instead of relying on some kind of
volatility measure.
The basic ARCH model uncertainty measure was applied to investment theory in
Huizinga (1993), Ghosal and Loungani (1996), and in Ogawa and Suzuki (1997). All
three studies estimate the impact of price and inflation uncertainty on investment in the
manufacturing sector —Huizinga, Ghosal and Loungani for the United States, Ogawa and
Suzuki for Japan. Huizinga (1993) and Ogawa and Suzuki
(1997) confirm that uncertainty has a negative impact on investment. The result of Ghosal
and Loungani (1996) is slightly more complicated. The overall result is that uncertainty
does not have an appreciable impact on investment when data for all industries are pooled.
However, for the set of industries characterized by low seller competition and vigorous
product market competition an increase in price uncertainty lowers investment.
An extension of Pagan and Ullah's (1988) ARCH model has been applied in the
investment under uncertainty analysis of Price (1995), suggesting a GARCH-M model.
Price (1995, p. 147) concludes that the level of aggregate uncertainty has a negative effect
on manufacturing investment. "In particular periods there were very large effects - on
average, uncertainty reduces investment by about 5%, but in 1974 the effect peaked at
48%." Similarly, Leahy and Whited (1996) demonstrated that stock market uncertainty
has a negative effect on investment of U.S. manufacturing firms.
167
Finally, Federer (1993a, 1993b) examines the impact of macroeconomic
uncertainty (measured by either a risk premium or the cross-forecaster standard deviation)
on domestic investment and provides strong support for a negative impact. Federer
(1993 a) shows that the impact of uncertainty on investment spending is larger than the
cost of capital ratio (Tobin's average q). Federer (1993b) reveals that the negative impact
o f macroeconomic uncertainty remains significant even after controlling for movements in
stock returns, accelerator effects, and the cost of capital.
3.c. Overall Results and Gaps
The review of the last two sections disclosed that there are many more empirical
studies analyzing investment under volatility than there are empirical studies analyzing
investment under uncertainty. It also revealed that while the results are mixed in the case
of volatility, they are more uniform in the case of uncertainty. Furthermore, the review has
shown that there is a gap of analyzing the impact of nominal GDP and nominal exchange
rate uncertainty on investment. The arguments here are similar to the arguments used in
the literature analyzing the impacts of nominal exchange rate uncertainty on trade. There it
was argued that real exchange rate uncertainty may seriously underestimate the degree of
uncertainty.
Concerning investment under nominal exchange rate uncertainty is important for at
least three cases. First, investment may depend on foreign capital inflows, either in form of
complementary foreign direct investment or portfolio investment providing necessary
foreign exchange. However, nominal exchange rate uncertainty may scare of foreign
168
investors to other markets. Second, the investment project may depend on imported inputs
of production, which may be too expensive in the downswings o f the nominal exchange
rate. Third, outputs generated by the investment may be exported, at least partly, but the
prospects to export the produced good decreases as exchange rate uncertainty has a
negative impact on trade.16
Concerning investment under nominal GDP uncertainty, the argument is similar to
the cornerstone o f the equilibrium or monetary business cycle theory of Lucas (1972,
1973) and Barro (1976, 1980). The main argument here is that even though agents have
rational expectations in these models, the lack of timely information on monetary shocks
implies that agents erroneously perceive price level movements as representing changes in
relative prices. Applied to the theory of investment under uncertainty, the higher nominal
GDP uncertainty, the less investors will be able to differentiate between changes in
nominal or real GDP. Therefore, nominal GDP uncertainty has a negative impact on
investment. It is surprising that this basic feature of imperfect information has yet not been
applied to the analysis of investment under uncertainty. Alternatively, it is also possible to
explain the importance of nominal GDP uncertainty with a New Keynesian argument. New
Keynesians17 assert that fluctuations in output arise largely from fluctuations in nominal
aggregate demand. These fluctuations have real effects because nominal wages and prices
are rigid.
16 For the most recent evidence and further references, see Adam-Muller (1997).
17 See Mankiw and Romer (1991), especially the Introduction in Volume 1.
169
4. How to Specify the Aggregate Domestic Investment Function?
The specification o f an aggregate domestic investment function is a difficult and
controversial issue. For example, Bleaney and Greenaway (1993) stated that "attempts to
ascertain the determinants o f investment in general, and private investment in fixed plant
and equipment in particular, have long been a preoccupation of macroeconomists with, it
must be said, no real 'general' theory having emerged."18 However, the complication is not
only due to the fact that there are many different kinds of investment, but also because
there are many different kind of investors with different motivations. As Blecker (1997, p.
191) has put it succinctly, decisions to invest are heterogeneous:
Government investment is determined by public policy decisions rather than by

private cost-benefit calculations. Private investment includes inventory
accumulation by business firms (both planned and unplanned), purchases of new
residences by households, and new plant and equipment expenditures of business
firms (business fixed investment). Business firms in turn are a diverse group of
entities, distinguished by such characteristics as whether they are incorporated or
unincorporated, large or small, rapidly of slowly growing, national or
multinational, and whether they have high or low dividend payouts.
Furthermore, as Dailami and Walton (1989) have shown for Zimbabwe, there are a
wide range o f socio-political and economic factors which have a considerable impact on
the determination of investment. Such factors are difficult to disentangle and even more
difficult to quantify.19 As outlined in Chapter I, the traditional investment literature has
18 Bleaney and Greenaway (1993) p. 579.
19 Dailami and Walton divided the socio-political and economic factors into three
categories: (i) supply-side factors, related primarily to shortage o f foreign exchange
necessary for imports of essential capital goods and industrial inputs; (ii) excessive
administrative intervention in the areas of investment decision making, labor relations, and
170
gone through considerable changes from the accelerator theory, to the neo-classical theory
o f investment and finally, Tobin's q-theory. However, based on the poor empirical
performance of these traditional investment theories, recent research of the investment
theory has led to a revised and extended account of the determinants o f investment. This
holds particularly true for the determinants of investment in developing countries.20
The specification of the investment function is even more challenging in cases of
studies using panel data o f a very heterogeneous group o f countries. While there is some
justified doubt over the homogeneity o f determinants of investment across a large variety
of industrialized and developing countries, there are generally some common determinants
o f investment across all countries. Advice can also be obtained from Martin Feldstein's
Fisher-Schultz Lecture given at the World Congress of the Econometric Society:
The investment process is far too complex for any single econometric model to be
convincing. (...) In practice all econometric specifications are necessarily ‘false’
models. They are false models not only in the innocuous sense that the residuals
reflect omitted variables but also in the more serious sense that the omissions and
other misspecifications make it impossible to obtain unbiased or consistent
estimates of the parameters even by sophisticated transformations of the data. The
applied econometrician, like the theorist, soon discovers from experience that a
useful model is not one that is ‘true’ or ‘realistic’ but one that is parsimonious,
plausible and informative.21
price controls; and (iii) socio-political factors reflecting the country's history, strategic
location, and political evolution.
20 A good review of empirical investment function specifications for developing

countries is provided by Rama (1993). Rama mentions that there were only 31 empirical
studies of the investment function for developing countries, o f which 7 studies included a
term controlling for economic instability, see Rama's table 5A.1.
21 Feldstein (1982), p. 829.
171
Following Feldstein’s advice, we will begin with a plausible and informative panel
data specification which includes nevertheless non-time varying macroeconomic
uncertainty. Then, a variety of modified specifications will be applied in later sections in
order to test the robustness o f results. As will be explained in the following five sub
sections (4.a to 4.e), a plausible and informative specification of aggregate domestic
investment includes some measure of (a) the income accelerator, (b) the inflation rate, (c)
the cost of capital, (d) the availability o f foreign exchange, and (e) macroeconomic
uncertainty. Section 4.f will present the results of the initial specifications and deal with
issues related to the possible lack of stationary times series data and the possible serial
correlation of error terms.
4.a. Income Accelerator
There is little doubt that the income accelerator is a considerable determinant of
investment. On simple and good measure o f the income accelerator is the current growth
rate of GDP. However, given the likely bivariate causality between the current growth rate
and investment, the growth rate needs to be lagged by one period. In difference to the
argumentation of the relevance of nominal GDP uncertainty, it is real not nominal growth,
that is the more appropriate variable in the determination o f investment. The lagged real
growth rate serves also as broad approximation of the availability o f investment funds,
which constitute another important determinant of investment.
172
4.b. Inflation
There are a number of studies demonstrating that high inflation rates have a
significant negative impact on investment and growth. For example, Greene and
Villanueva (1991, p. 41) argued that: "High rates o f inflation adversely affect private
investment by increasing the riskiness of longer-term investment projects, reducing the
average maturity o f commercial lending, and distorting the information content of relative
prices." Furthermore, Greene and Villanueva (1991, p. 41) declared that "high inflation
rates are often considered an indicator o f macroeconomic instability and a country's
inability to control macroeconomic policy, both of which contribute to an adverse
investment climate."
This is consistent with the well-known fact that the level o f inflation and inflation
volatility are highly correlated. Fischer (1993, p. 488) expressed that the inflation rate and
the variance o f the inflation rate are highly correlated in the cross-section, which makes it
difficult to disentangle the effects on growth of the level o f inflation from the effects of
uncertainty about inflation. In addition, section 8 of Chapter IV demonstrated that there is
also a high correlation (0.85-0.99) between inflation uncertainty and inflation volatility.
In any case, it seems highly appropriate to include either the inflation rate, inflation
volatility, or inflation uncertainty as a determinant of investment. We have included the
inflation rate as an explanatory variable in specification 5 as alternative specifications have
shown that it does not matter to include or exclude the inflation rate in specification 5.
173
4.c. Cost of Capital
The usual proxy for the cost of capital is the real interest rate. However, as seen
Chapter II, data on real interest rates are not only highly distorted but also highly
correlated with inflation rates.22 In addition, many earlier investment studies have shown
that real interest rates are hardly a significant determinant of investment.23 This is
consistent with the view that the cost of capital is determined by other factors besides
interest rates, they are not easy to get a hold of. Also, instead of using inflation biased real
interest rates, nominal interest rates may serve as a better proxy of both the cost of capital
and the availability of credit. It is therefore suggested that the nominal lending rate be
included in the initial specification of the investment function. However, nominal interest
rates will be excluded from some of the specifications of section 5.
4.d. Availability of Foreign Exchange
While the availability of foreign exchange may not be a constraint for most
industrialized countries, there is little doubt that the investment function of developing
countries is seriously misspecified when not controlling for the availability of foreign
This is due to the fact that many countries had high inflation rates and non-market
determined nominal interest rates.
23 For example, Blecker (1997, p. 193) points out that according to the structuralist
Keynesian view, "the decision to invest (at least in new plant and equipment in the
corporate sector) is determined primarily by factors other than interest rates or the cost of
capital, particularly the growth of demand (accelerator effect) and the cash flow available
for internal finance of investment."
174
exchange.24 However, only 8 of the 31 empirical studies of investment in developing
countries reviewed in Rama (1993) use some proxy to control for foreign exchange
availability. This is mainly due to problems of data availability. Following the definition of
net resource flow (F) in Chapter H, F will be used as a proxy for the availability of foreign
exchange in the initial specification of this section. Later specifications will test the
robustness of the results by excluding F.
4.e. Macroeconomic Uncertainty
According to the new investment theory we add macroeconomic uncertainty as a
determinant of aggregate domestic investment. However, given the fact that many o f the
seven macroeconomic uncertainties defined in Chapter V are correlated with each other,
we will examine each of the six non-inflation related macroeconomic uncertainties
separately.25 The inclusion o f only one uncertainty measure at a time follows the approach
taken by Cardoso (1993). On the other hand, to the degree that the level of inflation rate
serves also as a measure of general macroeconomic uncertainty, the initial specification
includes already two measures of macroeconomic uncertainty.
24 The importance o f financing constraints in general has been stressed by Fazzari,

Hubbard and Petersen (1988) and Fazzari and Athey (1987). More specifically, and quite
some while ago, the impact of foreign aid on investment has been analyzed by Papanek
(1972 and 1973), a more recent analysis has been provided by Movratas (1996).
25 (i) nominal GDP uncertainty, (ii) real GDP uncertainty, (iii) money supply
uncertainty, (iv) nominal exchange rate uncertainty, (v) real exchange rate uncertainty, and
(vi) terms o f trade uncertainty.
175
4.f. Initial Results and Further Considerations
Based on the above discussion, there are six basic regressions, each includes five
exogenous variables: the lagged growth rate, the level of inflation rate, the nominal
lending rate, the net resource flow, and one additional measure of macroeconomic
uncertainty. The exact specifications of these six initial regressions are listed in Table 16.
The results o f these six initial specifications are presented in part A of Table 17: all six
measures of macroeconomic uncertainties are highly significant at the 99 percentage level.
Also significant at the 99 percentage level are the lagged real growth rate (HESLAG), and
the lagged net resource flow (FLAG). While these initial results are quite promising, it is
necessary to address two further issues within this section, before testing the robustness of
these results.
The first issue is the possible non-stationarity of time-series data. Given the fact
that we use panel data o f 93 countries, we will test the time series data of the basic
investment function specification for unit roots using two standard tests: (a) the
Augmented Dickey-Fuller (ADF) test and the Phillips-Perron (PP) test. As can be seen
from Appendix 6, the results o f these two tests indicate that the relevant data is stationary
at the 1% significance level.
The second issue is the possible autocorrelation of error terms. The Durbin-
Watson statistics o f the initial six regressions range from 0.34 to 0.39 and indicate
autocorrelated errors. Based on a standard procedure correcting for autocorrelation, the
AR(1) specification is applied to the regression to take care of first order autocorrelation.
After invoking the AR(1) error correction for all six initial specifications, the Durbin
176
Watson statistics range from 2.36 to 2.49, thus indicate no further autocorrelation.
Actually, these numbers may indicate an over-correction or a different problem than non-
stationarity to cause the original low Durbin Watson statistics. It seems therefore
appropriate to use specifications with and without AR(1) corrections.26
The results for the six initial specifications after including the AR(1) error
correction are presented in part B of Table 17. It turns out that money supply uncertainty,
as well as nominal and real exchange rate uncertainty remain significant at the 99%
significance level. Nominal and real GDP uncertainty remain significant at the 95 and 90
percentage levels. However, terms of trade uncertainty is no longer significant, although
the coefficient remains negative. O f interest is also that the lagged growth rate and the
lagged resource flow remain highly significant (at the 99 percentage level) for all 6
specifications correcting for possible autocorrelation. On the other hand, correcting for
possible autocorrelation does not bear any any major change in the relevance of inflation
and interest rates as explanatory variables. While the t-statistics increase slightly for the
log of the inflation rate (however, whithout becoming statistically significant), the t-
statistics decrease further for the log of the lending rate.
26 Instead of assuming that the initial autocorrelation was due to inertia or prolonged
influences o f shocks, it is also possible that the low levels of the Durbin-Watson statistic
are due to the omission of a relevant independent variable. This possibility cannot be ruled
out, especially after taking into account Feldstein's statement o f the unavoidable
misspecification o f the investment function, as well as the high level of aggregation of
investment across 93 countries. However, as will be shown in more details in section 5,
even the inclusion o f further potential determinants, like the debt to reserve ratio, does not
improve the Durbin-Watson statistic. The best solution would thus be to stick with the
initial specification without the AR(1) correction.
177
Table 16: Six Initial Specifications for Aggregate Domestic Investment
(1) log(lNV) = a + B HESLAG + y log(INF) + 5 Iog(INT) + e FLAG + p. log(UNCGDP)
(2) log(INV) = a + B HESLAG + y log(INF) + 8 log(INT) + e FLAG + (x log(UNCHES)
(3) log(INV) = a + B HESLAG + y log(INF) + 8 log(INT) + e FLAG + jx Iog(UNCMON)
(4) Iog(INV) = a + B HESLAG + y log(INF) + 8 log(INT) + e FLAG + p. log(UNCNEX)
(5) log(INV) = a + B HESLAG + y log(INF) + 8 log(INT) + e FLAG + |x log(UNCREX)
(6) Iog(INV) = a + B HESLAG + y log(INF) + 8 log(INT) + e FLAG + |x log(UNCTOT)
where
INV = the ratio of gross domestic investment to GDP,
HESLAG = the lagged real growth rate (based on PPP adjusted GDP),
INF = the inflation rate,
INT = the nominal lending interest rate,
FLAG = the lagged net resource flow as percent of GDP,
UNCGDP = nominal GDP uncertainty
UNCHES = real GDP uncertainty,
UNCMON = money supply uncertainty,
UNCNEX = nominal exchange rate uncertainty,
UNCREX = real exchange rate uncertainty, and
UNCTOT = terms o f trade uncertainty.
A. Specification without Correcting for Possible Autocorrelation
heslag log(inf) flag log(intlen) log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.019 -0.007 0.006 -0.034 -0.104
t-stat. 10.74 -0.68 6.03 -2.29 -3.88
EQ.2 0.021 -0.007 0.009 -0.039 -0.161
Table 17: Results of Six Initial Specification
t-stat. 11.51 -0.63 8.25 -2.71 -7.03

EQ.3 0.020 0.009 0.009 -0.020 -0.240
t-stat. 11.29 0.67 8.77 -1.34 -12.4
EQ. 4 0.017 0.019 0.009 -0.018 -0.223
t-stat. 9.54 1.82 8.63 -1.29 -12
EQ. 5 0.018 0.003 0.009 -0.021 -0.348
t-stat. 9.93 0.32 8.30 -1.43 -11.52
EQ. 6 0.020 -0.007 0.003 -0.039 -0.051
t-stat. 11.49 -0.71 2.87 -2.74 -2.81
B. Specification with Correcting for Possible Autocorrelation

heslag log(inf) flag log(intlen) log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.007 -0.010 0.011 0.001 -0.244
t-stat. 9.43 -1.38 9.46 0.05 -2.21
EQ.2 0.007 -0.009 0.011 0.000 -0.161
t-stat. 9.41 -1.25 9.33 0.00 -1.73
EQ.3 0.007 -0.009 0.011 -0.002 -0.239
f-sfaf. 9.31 -1.23 9.28 -0.15 -3.41
EQ. 4 0.007 -0.01 0.011 0.004 -0.28
t-stat. 8.58 -1.49 10.03 0.34 -4.27
EQ.5 0.007 -0.012 0.011 0.000 -0.486
t-stat. 8.65 -1.68 10.03 -0.04 -4.28
EQ.6 0.008 -0.010 0.011 0.000 -0.073
t-stat. 9.53 -1.42 9.05 -0.04 -0.94
179
5. How Robust are the Impacts of Macroeconomic Uncertainty?
This section tests the robustness of the impacts of macroeconomic uncertainty by
(a) using a variety o f slightly different specifications, (b) analyzing two twelve-year sub
periods from 1970-81 and 1983-94, and (c) substituting gross domestic investment with
fixed private investment.
S.a. Testing Alternative Investment Specifications
Within this sub-section, we test the robustness of the above results when looking
at three variations o f each of the two initial investment specifications. The first variation
excludes nominal interest rates, the second variation excludes lagged net resource flows,
and the third variation includes the debt-to-reserves ratio as an additional regressor. The
debt-to-reserves ratio can either be interpreted as an additional measure of uncertainty
(related to a country's indebtedness27 and the subsequent proximity to a balance of
payments crisis), or as an alternative proxy of foreign exchange availability. The results of
all three variations are presented in Tables 18 to 20. Each table has two parts, reflecting
the results with and without the AR(1) error correction specification. The three tables
demonstrate that money supply uncertainty, nominal exchange rate uncertainty, and real
exchange rate uncertainty are always significant at the 99 percentage level. Consistent
with our earlier result, terms
27 The major reason against the reserves to debt ratio variable is that there is no data
available for 22 countries.
Table 18: Results for the Specification Excluding Interest Rates heslag log(inf) flag log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.017 -0.015 0.006 -0.115
t-stat. 12.74 -2.2 6.63 -5.17
EQ.2 0.019 -0.015 0.008 -0.165
t-stat. 13.86 -2.25 9.21 -9.35
EQ.3 0.017 0.006 0.007 -0.196
t-stat. 13.34 0.81 8.69 -14.75
EQ.4 0.016 -0.001 0.006 -0.083
t-stat. 12.04 -0.21 7.13 -9.61
EQ. 5 0.016 -0.005 0.006 -0.22
t-stat. 11.97 -0.73 7.61 -12.37
EQ. 6 0.019 -0.019 0.004 -0.066
t-stat. 13.72 -2.83 4.35 -4.69

heslag log(inf) flag log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.005 -0.009 0.011 -0.183
t-stat. 8.05 -1.81 12.74 -2.27
EQ.2 0.005 -0.009 0.011 -0.087
t-stat. 8.05 -1.84 12.76 -1.34
EQ.3 0.006 -0.011 0.011 -0.187
t-stat. 8.03 -2.09 12.29 -4.18
EQ.4 0.004 -0.009 0.011 -0.116
t-stat. 7.3 -1.81 13.2 -3.93
EQ.5 0.005 -0.010 0.011 -0.303
t-stat. 7.41 -1.89 13.22 -4.86
EQ.6 0.005 -0.010 0.011 -0.007
t-stat. 8.25 -1.91 12.24 -0.14
Table 19: Results for the Specification Excluding Resource Flows heslag log(inf) log(intlen) log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.020 0.002 -0.038 -0.081
t-stat. 10.62 0.17 -2.49 -3.00
EQ.2 0.021 0.003 -0.043 -0.093
t-stat. 11.06 0.23 -2.82 -4.27
EQ.3 0.020 0.017 -0.029 -0.193
t-stat. 11.01 1.52 -1.90 -10.07
EQ.4 0.018 0.025 -0.026 -0.176
t-stat. 9.48 2.26 -1.74 -9.56
EQ.S 0.018 0.012 -0.027 -0.280
t-stat. 9.78 1.42 -1.83 -9.31
EQ.6 0.021 -0.003 -0.041 -0.044
t-stat. 11.46 -0.31 -2.80 -2.45

heslag log(inf) log(intlen) log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.008 -0.011 -0.004 -0.015
t-stat. 9.36 -1.41 -0.35 -2.12
EQ.2 0.008 -0.011 -0.004 -0.017
t-stat. 9.34 -1.40 -0.36 -0.96
EQ.3 0.008 -0.010 -0.004 -0.183
t-stat. 9.36 -1.31 -0.29 -2.61
EQ.4 0.007 -0.012 -0.001 -0.199
t-stat. 8.58 -1.61 -0.10 -3.00
EQ.5 0.007 -0.013 -0.004 -0.344
t-stat. 8.63 -1.74 -0.04 -3.02
EQ.6 0.008 -0.011 -0.006 -0.042
t-stat. 9.42 -1.52 -0.47 -0.62
Table 20: Results for the Specification Including Debt to Reserve Ratios
heslag log(inf) flag log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot) log(deres)
EQ. 1 0.017 0 0.01 -0.056 0.035
t-stat. 10.51 0.03 8.44 -1.85 4.23
EQ.2 0.017 -0.005 0.011 -0.117 0.042
t-stat. 10.47 -0.58 9.39 -4.44 5.29
EQ.3 0.016 0.020 0.010 -0.197 0.029
t-stat. 10.48 2.43 8.9 -11.03 3.7
EQ.4 0.016 0.009 0.009 -0.048 0.029
t-stat. 10.08 1.03 7.98 -4.64 3.51
EQ.5 0.015 0.013 0.01 -0.194 0.025
t-stat. 9.93 1.62 9.15 -9.16 3.18
EQ.6 0.017 -0.006 0.007 0.068 0.027
t-stat. 10.95 -0.07 5.89 2.68 3.36

heslag log(inf) flag log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot) log(deres)
EQ. 1 0.005 -0.007 0.013 -0.153 0.032
t-stat. 6.88 -1.08 11.84 -1.55 2.65
EQ.2 0.005 -0.007 0.013 -0.086 0.034
t-sfat. 6.84 -1.13 11.75 -0.94 2.85
EQ.3 0.005 -0.006 0.013 -0.223 0.029
t-stat. 6.88 -1.37 11.25 -4.01 2.51
EQ.4 0.004 -0.007 0.013 -0.087 0.025
t-stat. 6.13 -1.16 11.93 -2.60 2.09
EQ.5 0.004 -0.007 0.013 -0.274 0.025
t-stat. 6.24 -1.18 12.09 -4.00 2.14
EQ.6 0.005 -0.007 0.013 0.088 0.034
t-stat. 6.94 -1.16 11.57 1.00 2.86
183
o f trade uncertainty is usually not significant in the case of correcting for the possible
autocorrelation with the AR(1) specification. The results for nominal and real GDP
uncertainty are slightly more complex. Excluding the debt-to-reserves ratio, which may
like the level o f inflation be considered an overall uncertainty measure similar to GDP
uncertainty, nominal GDP uncertainty is always significant at 95 percentage level. Real
GDP uncertainty is always significant at the 99 percentage level in the cases without
correcting for serial correlation, but is insignificant for all three variations with the AR( 1)
specification.
5.b. Testing for Robustness using Sub-Periods
The analysis of two twelve-year sub-periods used specific time-period relevant
uncertainty measures, as defined in Chapter V. The two investment specifications are as
defined in section 4 above, one with and one without the correction for the likely
autocorrelation. As Table 21 demonstrates, the results derived from analysis o f the sub
periods are overall quite similar to the results derived from the whole time period. Money
supply uncertainty is always significant at the 99 percentage level for the whole time
period as well as the two sub-periods, with exception of one case, where the significance
level is at the 95 percentage level. Real exchange rate uncertainty is always significant at
the 99 percentage level. Nominal exchange rate uncertainty and nominal GDP uncertainty
are, with one exception, always significant at least at the 95 percentage level. Real
exchange rate uncertainty and terms of trade uncertainty are both insignificant for both
sub-periods in cases of correcting for serial correlation.
184
5.c. Testing for Robustness using Fixed Private Investment
Finally, the last test of robustness is to substitute gross domestic investment by
fixed private investment. There are two major consequences of such a substitution: first, it
seems appropriate to place fixed public investment on the right hand side of the regression,
and second, due to data constraints, the number of countries and observations included in
this analysis is considerably lower than for the case of gross domestic investment. The
main data constraints are the none-availability of comparable data for any of the 21
industrialized countries and 18 developing countries. Nevertheless, there remains some
data for 54 countries. The results for the two specifications with and without correction
for serial correlation are displayed in Table 22. Nominal GDP uncertainty, money supply
uncertainty, and real and nominal exchange rate uncertainty are always significant at the
99 percentage level. Consistent to our earlier results is also that real GDP uncertainty and
terms o f trade uncertainty are not always significant.
5.d. Conclusion
There is sufficient evidence for a very robust negative impact of nominal GDP
uncertainty, monetary uncertainty, and real and nominal exchange rate uncertainty on
domestic investment. The robustness is considerably weaker in the case of real GDP
uncertainty, and non-existing for the terms of trade uncertainty.
A. Specification without Correcting for Possible Autocorrelation (1970-81)
EQ. 1 0.011 -0.006 0.006 -0.106
Table 21: Testing the Robustness by Analyzing Sub-Periods
t-stat. 6.26 -0.47 4.62 -4.12

EQ. 2 0.012 -0.008 0.008 -0.19
t-sfaf. 6.97 -0.66 6.7 -8.63
EQ. 3 0.012 0.000 0.007 -0.167
t-stat. 6.8 0.03 5.2 -7.83
EQ.4 0.011 -0.014 0.006 0.001
t-stat. 6.2 -1.13 4.35 0.8
EQ. 5 0.011 0.003 0.005 -0.156
t-stat. 6.06 0.29 4.07 -6.21
EQ. 6 0.012 -0.001 0.006 -0.119
t-stat. 7.01 -0.09 4.77 -6.06
B. Specification with Correcting for Possible Autocorrelation (1970-81)

EQ. 1 0.003 -0.008 0.011 -0.073
t-stat. 3.67 -1.09 9.59 -0.77
EQ. 2 0.003 -0.008 0.011 -0.028
t-stat. 3.67 -1.11 9.56 -0.33
EQ. 3 0.003 -0.008 0.011 -0.009
t-stat. 3.77 -1.04 9.07 -1.99
EQ.4 0.003 -0.022 0.011 0.000
t-stat. 3.55 -1.04 9.48 -1.97
EQ. 5 0.003 -0.007 0.011 -0.190
t-stat. 3.54 -1.01 9.46 -2.13
EQ.6 0.003 -0.007 0.010 -0.011
t-stat. 3.79 -1.01 9.17 -0.16
Table 21: Testing the Robustness by Analyzing Sub-Periods (cont.) A. Specification without Correcting for Possible Autocorrelation (1983-94)
EQ. 1 0.024 -0.015 0.006 -0.073
t-stat. 9.98 -1.77 5.08 -2.84
EQ. 2 0.025 -0.02 0.006 -0.048
t-stat 10.5 -2.41 5.41 -2.39
EQ. 3 0.024 0.016 0.008 -0.166
t-stat. 10.26 1.79 7.14 -9.27
EQ.4 0.022 -0.007 0.006 -0.088
t-stat. 9.19 -0.76 5.01 -6.08
EQ. 5 0.021 -0.012 0.007 -0.130
t-stat. 8.87 -1.3 8 5.58 -5.47
EQ. 6 0.025 -0.017 0.002 -0.019
t-stat. 11.14 -2.18 1.51 -1.06
B. Specification with Correcting for Possible Autocorrelation (1983-94)

EQ. 1 0.010 -0.003 0.010 -0.157
t-stat. 8.33 -0.40 6.40 -1.86
EQ. 2 0.010 -0.004 0.010 -0.028
t-stat. 8.45 -0.49 6.36 -1.70
EQ. 3 0.010 0.000 0.010 -0.179
t-stat. 8.30 -0.05 6.72 -3.62
EQ.4 0.008 -0.005 0.009 -0.140
t-stat. 7.30 -0.68 6.55 -2.22
EQ.5 0.008 -0.004 0.010 -0.258
t-stat. 7.29 -0.54 6.73 -2.91
EQ.6 0.010 -0.006 0.009 -0.040
t-sfaf. 8.50 -0.73 5.60 -0.66
Table 22: Testing the Robustness by Analyzing Fixed Private Investment
heslag log(inf) flag log(invfipu) log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.023 -0.020 -0.008 -0.239 -0.180
t-stat 7.27 -1.31 -3.32 -6.27 -3.27
EQ. 2 0.024 -0.034 -0.008 -0.235 -0.165
t-stat. 7.57 -2.29 -3.34 -6.17 -3.41
EQ.3 0.022 0.000 -0.006 -0.259 -0.285
t-stat. 7.26 0.03 -2.35 -7.16 -9.79
EQ.4 0.022 0.001 -0.008 -0.299 -0.115
t-stat. 6.88 0.08 -3.21 -7.69 -6.50
EQ.5 0.021 -0.013 -0.007 -0.273 -0.357
t-stat. 6.84 -0.84 -3.08 -7.24 -8.55
EQ. 6 0.023 -0.028 -0.009 -0.223 -0.121
t-stat. 7.36 -1.88 -3.62 -12.80 -2.90

heslag log(lnf) flag log(invfipu) log(uncgdp) log(unches) log(uncmon) log(uncnex) log(uncrex) log(unctot)
EQ. 1 0.004 -0.025 0.016 -0.194 -0.783
t-stat. 3.46 -3.08 8.60 -5.58 -2.24
EQ.2 0.001 0.008 0.002 -0.191 -0.498
t-stat. 3.44 -3.16 8.57 -5.51 -1.51
EQ.3 0.004 -0.025 0.016 -0.194 -0.488
t-stat. 3.45 -3.03 8.58 -5.60 -3.10
EQ.4 0.003 -0.024 0.016 -0.208 -0.184
t-stat. 2.95 -2.93 8.81 -6.02 -1.83
EQ.5 0.003 -0.025 0.016 -0.209 -0.505
t-stat. 3.03 -3.03 8.73 -6.04 -2.01
EQ. 6 0.004 -0.025 0.016 -0.187 -0.382
t-stat. 3.69 -3.14 8.53 -5.51 -1.38
CHAPTER VH
THE RELATIONSHIP BETWEEN INVESTMENT AND DEVELOPMENT
While there is broad agreement about an overall positive relationship between
investment and development, at least in the mainstream literature, there has been some
criticism about making too broad generalizations. Investment can be wasted in inefficient
projects. At a broader level, investment can be counterproductive to long-term
development if it is in sectors which do not reflect a country's comparative advantage, like
for example, in import substitution sectors.1 Furthermore, it has been argued that while
growth rates may have increased average GDP per capita, it did not promote development
as it did not reach the poor segments of society. In many of the least developed countries,
these segments constitute the vast majority of society.
A historical interpretation on this question has been suggested by Easterly (1997).
He writes that "early development economists were optimistic about how growth would
improve a wide range o f health and education indicators," while "the second generation of
development economists and political scientists fiercely challenged these conclusion."
188
1The literature along these lines is large, for a recent discussion see Schydlowsky
(1995).
189
The question which remains is where we stand today. Fortunately there have been
some major reviews of these topics in the recent literature. The contents of this chapter
has therefore been limited to a brief review of the recent literature. We divide this review
into two sections: (1) the relationships between investment and growth, and (2) the
relationship between growth and development. This will close the last link in the
relationship between development, economic structure, macroeconomic uncertainty, and
investment, as it was suggested in Figure 1 of the Chapter I.
1. Investment and Growth
The notion of a positive relationship between investment and growth is quite
intuitive and has a long history. For example, in the response by Keynes (1937, p. 221) to
critics o f his General Theory o f Employment, Interest and Money, he concluded that "the
theory can be summed up by saying that (...) the level of output and employment as a
whole depends on the amount of investment." Since then, there have been many studies
analyzing specifically the relationship between investment and growth. Indeed, there are
far too many country-specific studies to mention here. However, there are a limited
number of large cross-country studies examining the relationship between growth and
some kind of investment, i.e. equipment investment, private investment, and gross
domestic investment.
190
l.a. Equipment Investment
De Long and Summers (1991 and 1993) analyzed the relationship between
equipment investment and economic growth. Based on the empirical data for some sixty-
one non-oil-exporting economies from 1960-85, De Long and Summers (1991) found a
strong association between the GDP growth rate per worker and their estimates of the
share of GDP devoted to machinery investment. De Long and Summers (1993) sharpened
their estimates of equipment investment and extended their analysis to 88 non-oil-
exporting economies. Their (1993, p. 395) conclusion is even stronger:
There is a very strong growth-investment association even when rich industrialized

countries are not considered. Rapid growth is found where equipment investment
is high, and slow growth where equipment investment is low. If there is a region
where the post-WWH growth-equipment nexus is weak, it is the well-integrated
and very rich regions of western Europe -- not the developing world.
Lb. Private Investment
Greene and Villanueva (1991) analyzed the relationship between private
investment and growth for 23 developing countries from 1975-87, and concluded (p. 33)
that "econometric evidence indicates that the rate of private investment is positively
related to GDP growth." Our own empirical results of Chapter VI have provided the same
result for 54 developing countries with data from 1970-94.
191
I.e. Gross Domestic Investment
Kormendi and Meguire (1985, p. 157) analyzed the investment-growth nexus for
47 countries from 1950-77 and found that "the investment-to-income ratio has major
effects on economic growth". Levine and Renelt (1992, p. 959) examined how robust or
fragile the conclusions o f earlier studies are to small changes in the conditioning
information set. Based on the experience of 119 countries from 1960-1989, they found
that almost all results are fragile, however, there is "a positive and robust correlation
between average growth rates and the average share o f investment in GDP."
The experience o f 12 African countries from 1970-1991, has been examined by
Ojo and Oshikoya (1993). They conclude (p. 20) that "investment is positively related to
growth", but that the relationship is not very strong, as "the investment rate would have to
increase by about 10 percentage points in order to raise the per capita GDP growth rate by
2 percentage points."
On the other hand, though there may be factors which are stronger related to
growth than investment, our own empirical results o f Chapter VI have indicated that the
relationship is very robust. While our specification o f Chapter VI lagged the GDP growth
rate by one period, further preliminary regressions indicated that the robust relationship
remains independent of using lagged, current, or future growth rates.
192
1.d. Comprehensive Reviews
The most comprehensive review of the recent theoretical and empirical work on
the determinants o f saving and investment, as well as their links to growth, has been
provided by Schmidt-Hebbel, Serven, and Solimano (1996). They concluded that:
First, saving and growth reinforce each other — causality runs in both directions.
Second, saving and investment are highly correlated —due to low capital mobility,
domestic policies that restrict large current-account imbalances, or common
factors that push both variables in the same direction. Third, physical investment is
a necessary, but not sufficient, condition for growth. And fourth, human capital
investment, technological innovation, and appropriate policies are also necessary
for sustained high growth.2
Influenced by the work of Schmidt-Hebbel, Serven and Solimano (1996), the
World Bank's World Development Report 1995 concluded on page 21, that "investment in
physical and human capital is necessary but does not guarantee productivity growth".
2. Growth and Development
There is no doubt that economic growth causes development, as long as
development is defined as sustained increases in GDP per capita. This is simply a
definitional tautology. However, we have already outlined the problems of such a narrow
definition of development in Chapter n . Defining development as human development,
there is justified suspicion as to whether the positive relationship between growth and
development remains true. One relatively old concept is what has come to be known as
2 Schmidt-Hebbel, Serven and Solimano (1996), p. 108; originally presented as internal

research paper in September 1995.
193
"immiserizing growth". Other more recent concerns are related to the distributional and
environmental impacts of growth.
2.a. Immiserizing Growth
The concept of immiserizing growth has spooked for some decades through the
trade literature. It refers to the case where economic growth leads via repercussions
through trade to a situation of diminishing economic welfare. This is theoretically possible
in cases where the marginal propensity to import is high, and growth is biased towards the
export commodity, for which foreign demand is price inelastic. In more detail, the first
necessary condition is that growth in real output induces a demand for more imports. The
second necessary condition is that the country would need to export more in order to be
able to pay for the increased imports. The third and critical condition is then, that the price
elasticity o f export demand is so low, that exports will need to be sold at such a low price,
that the terms of trade deteriorate to a degree that this wipes out the gain from increased
output. Based on the contribution by Jagdish Bhagwati (1958), today, most economists
tend to conclude that cases of immiserizing growth are practically improbable.
2.b. Growth and Human Development
There have just been three extensive reviews o f the relationship between economic
growth and human development. The first two recent studies have already been mentioned
earlier in Chapter n, when we reviewed the literature on the relationship between growth
and inequality, the third and most recent examination o f the relationship between growth
and human development is Easterly (1997).
194
UNDP's Human Development Report 1996 provides a qualitative review on the
relationship between growth and human development. As James Gustave Speth writes in
the foreword o f the report:
The central message of Human Development Report 1996 is clear: there is no

automatic link between economic growth and human development, but when these
links are forged with policy and determination, they can be mutually reinforcing
and economic growth will effectively and rapidly improve human development.3
The report gives special attention to two topics: the translation of growth into
employment opportunities and the contribution of women to development. The first issue
is related to what has become known as "jobless growth", that is, even though there is
economic growth, employment opportunities have decreased. However, as Okun's law
indicates, the problem here is not growth itself, but too little growth. In order for growth
to reduce unemployment, growth rates must more than offset productivity growth.
Somehow different are aspects of job quality. As the report points out, more and more
jobs are less and less satisfactory, job security is being eroded, and employment is
increasingly part-time and in piecework in industrial countries and in the informal sector in
developing countries. Obviously there is no easy solution to this problem. However, there
are policy options which can foster an employment friendly growth strategy. As for the
contribution of women to development, the report concludes that investing in women's
capabilities and empowering them is the surest way to contribute to economic growth and
overall development.
3 UNDP (1996), p. in.
195
Deininger and Squire (1996 and 1997) analyze the relationship between growth
and human development by examining the impact o f growth on poverty. While they could
not find a systematic relationship between growth and inequality, they found a strong
positive relationship between growth and poverty reduction, which is indicated by an
increase in income o f the poorest quintile. Deininger and Squire (1997, p. 40) add
explicitly, that “even when inequality has worsened, its negative effect on the poor has
been more than outweighted by the positive effect o f growth.”
Finally, Easterly (1997) provides quantitative results o f 95 quality of life indicators,
using fixed effects and first differences instrumental variable estimations. While he finds
that surprisingly few quality of life indicators are statistically significantly related to
economic growth, it would certainly be inappropriate to conclude that growth harms
human development. Overall, we can conclude from the three studies that growth and
human development are generally positively correlated.
3. Investment as a Maior Factor of Development
Jere Behrman (1972, p. 837) wrote more than 25 years ago: "Although questions
have been raised about the widely assumed dominance o f real physical capital stock
investment in the economic development process, the importance o f real physical capital
stock as one o f the major factors in development still is widely accepted." Our review of
the more recent literature has shown that the widely assumed positive relationship between
physical investment and human development is generally justified. However, there is much
196
room for making physical investment even more relevant for human development than is
currently the case in most developing and industrialized countries.
CHAPTER Vffl
CONCLUSION
This dissertation has analyzed the relationship between development, economic
structure, macroeconomic uncertainty, and aggregate domestic investment. We established
10 stylized facts of relationships between development and economic structures in Chapter
in , and 10 stylized facts of relationships between economic structures and macroeconomic
uncertainties in Chapter V. Chapter VI provided evidence for the negative impact of
macroeconomic uncertainty on investment, and Chapter VH concluded that physical
investment has generally a positive impact on economic development.
I. Connecting Maior Results
While it would not make sense to repeat all the specific results at this point, we can
connect some of the results from different chapters to provide a set of more specific
variations of Figure 1. Although there are many alternative ways to specify the general
relationship of Figure 1. We have selected five intuitive examples. Our first example is
related to the characteristics of sectoral shares in GDP and is illustrated in Figure 15.
197
198
LDCs have high shares of agriculture and low shares of manufacturing to GDP Both
structural features are positively related to nominal GDP uncertainty, which seems to
have negative impacts on investment and development.
Figure IS: The Sectoral Shares Example
nominal GDP
uncertainty
high shares of agriculture low

low shares of manufacture investment
little
development
The second example (Figure 16 below) illustrates the circular relationship between
(a) high shares of private consumption to GDP, (b) high nominal GDP uncertainty, (c) low
domestic investment, and (d) low levels o f development.
199
Figure 16: The Nexus Between High Consumption and High GDP Uncertainty
lominal GDP
uncertainty
high shares of low

private consumption investment
little
development
Figure 17 links low levels of government revenues to high monetary uncertainty,
low levels of domestic investment, and perpetuating underdevelopment.
Figure 17: Linking Government Revenues to Monetary Uncertainty
monetary
uncertainty
low levels of
government low
rcveneues investment
relationship 1 r elationship 4
little
development
200
Figure 18 illustrates that LDCs have unfavorable export compositions, i.e., they
have low shares o f machinery exports, high shares o f primary exports, and a high degree
of export product concentration. We have shown in Chapter V that all three structural
features are positively correlated to real exchange rate uncertainty, which is correlated to
low investment and low development.
Figure 18: The Export Composition Example
real exchange
rate uncertainty
low shares of machinery exports

high shares of primary exports low
high export product concentration investment
relationship I relationship 4
little
development
Finally, our fifth example (Figure 19) illustrates the importance of financial market
development. Underdeveloped financial structures are strongly related to real GDP
uncertainty, monetary uncertainty, and real exchange rate uncertainty. Chapter VI has
shown that all three macroeconomic uncertainties have a negative impact on domestic
investment and thus on development.
201
Figure 19: The Importance of Financial Market Development
V real GDP uncertainty

monetary uncertainty
/ real exchange rate uncertainty
low financial low

market investment
development
little
development
2. Lessons from an Optimistic Interpretation
A pessimistic interpretation of the above circular relationships would be that the
least developed countries are trapped in a vicious circle o f underdeveloped economic
structures, high uncertainty, subsequent low investment and thus perpetuating low growth
and little development. However, a more optimistic interpretation would be to use the
above relationships as a tool to achieve development. While it seems hard to escape the
low-level viscious circle traps, it is not impossible to break out of them. To do so may
require big pushes on several fronts at once. This can be attempted by using appropriate
economic policies to (a) establish more beneficial economic structures, and (b) reduce
macroeconomic uncertainty. The optimistic interpretation implies at least three major
lessons.
202
Lesson number one is to take economic structures seriously. Though economic
structures cannot be changed in the short run, they are endogenous in the long run.
Appropriate macroeconomic policies can support the change in economic structure and
thus decrease macroeconomic uncertainty, increase investment, and thus achieve
development faster than by neglecting the relevance of economic structures. The three
most beneficial structural changes would imply increased industrialization, increased
international trade (openness), and financial market development.
Lesson number two is to take macroeconomic uncertainty seriously. Propositions
such as "policy makers do not need to worry about uncertainty" or "the effect of
uncertainty on investment is small" are inappropriate and misleading.1 More specifically,
our results clearly indicate that monetary and exchange rate uncertainty have a negative
impact on investment. Even though the market provides some mechanisms to reduce
exchange rate uncertainty, these mechanisms are costly and cannot substitute for exchange
rate stability. This should not be interpreted as an advice to fix the exchange rate
nconsistent to macroeconomic fundamentals, but some sand in the wheels o f foreign
exchange markets (like the Tobin tax) might need to be considered.2
1This conclusion is not obvious. For example, a recent IMF occasional paper, Mussa,
Goldstein, Clark, Mathieson, and Bayoumi (1994) concluded on the basis of just two
studies -- one purely theoretical [Aizenman (1992)] and another one with the U.S.
experience [Goldberg (1993)] —that the effect of exchange rate uncertainty on investment
is small.
2 See Felix (1993) and Eichengreen, Tobin, and Wyplosz (1995) for additional
supportive arguments; see Garber and Taylor (1995) for a more skeptical view on the
Tobin tax proposal. For a comprehensive review o f issues related to the reform of the
international monetary system, see Gunter (1996).
203
Third, we can draw lessons for the design of structural adjustment programs. The
neglect of the negative impacts of macroeconomic uncertainty may explain why the
conventional policy prescriptions did not work as well as they were supposed to. It is now
recognized that periods of structural adjustment were many times followed by investment
pauses. A few authors3 have suggested that the investment pause can be explained by
uncertainty caused by structural adjustment programs. Appropriate adjustment programs
would therefore try to minimize the uncertainties, not to increase them. Increased
communication and participation, with a subsequent clear signal of commitment for the
proposed strategy by all players involved seem to be a good recipe to more successful
structural adjustment.
3. Limitations and Suggestions for Further Research
Finally, a word of caution seems to be appropriate. Caution number one is that the
establishment of a stylized fact should not be interpreted as an iron law. The world is
changing continuously and current or past patterns of development do not need to hold
forever. However, it is also unlikely that all these relationships and stylized facts of the last
25 years will suddenly break down and become irrelevant. Constant checks and balances
will be needed to adjust policies to reflect new patterns which have not been analyzed yet
but which can determine if a country achieves equitable, sustainable and participatory
development or not.
3 For one of the earliest studies along this line, see Rodrik (1991); for more recent
contributions see Branson and Jayarajah (1995) and Dewatripont and Roland (1995).
204
Caution number two refers to sometimes considerable deviations from average
patterns of development as it is visible in the scatter diagrams. These deviations may
reflect differences in economic policies, as well as social, cultural and institutional settings,
all o f which have been neglected in this analysis. We have applied either the traditional or
relatively simple definitions of economic structure to avoid any possible criticism that the
results are based on selective definitions of economic structure. We have been more
innovative in defining macroeconomic uncertainty, as the traditional definition was
inappropriate to analyze the impact o f uncertainty (as distinct from volatility).
Furthermore, we have used simple tools of analysis within the first two analytical
parts and slightly more complicated panel data regressions in the third analytical part. The
reasons for this simplicity in methodology has been (a) to use an overall consitent
methodology for many different structures, and (b) to show that the results are not
dependent on sophisticated econometric methods and selected instrumental variables.
However, there is some room for a more sophisticated regression analysis in cases where
we are interested in some more specifics. A major step forward would be to construct a
time-variant measure of uncertainty, however, without being as arbitrary as time-variant
uncertainty measures are currently. The use of long-period differenced data could shed
further light on structural changes.
There are also many more interesting structures, such as differences in export
supply elasticities or the degree of labor market fragmentation, which have not been
analyzed in this dissertation, but could bear interesting results on their relationships to
205
uncertainty, investment and growth. Much remains to be done to include these less
objective explanatory variables into the analysis.
Our final warning is to be careful with generalizations. Like others, Helleiner
(1986) warned that:
’Norms' and averages' for the world, however fascinating to statisticians and
development economists, are dubious guides for policy-makers in individual
countries. (.... ) Among the most controversial of these prescriptions relates to the
desirable degree of'outward orientation' or the 'openness' of domestic production.4
Though not completely conclusive, Helleiner’s (1986) analysis tends to indicate
that the poorest countries did not benefit from increased outward orientation, which he
explained with negative impacts on investment from increased trade instability Helleiner's
conclusion for the poorest countries does not necessarily contradict our general pattern o f
development. First, we have shown that many of the poor countries' economic structures
increase macroeconomic uncertainty, which then has a negative impact on investment.
Second, given that the least developed countries have a high degree o f export product
concentration and low export market power in world exports, increased outward
orientation may indeed increase uncertainty in the short run.5 What would be desired is an
analysis over time, especially at the country- or country-group-level. Further research
along these lines would therefore be desirable.
4 Helleiner (1986), p. 139.
5 As mentioned in Helleiner (1986, p. 141), this has been rationalized by Michael

Michaely (1977) and William Tyler (1981), who suggested that some basic or minimum
level o f development is necessary for a country to benefit from export oriented growth.
APPENDIX 1: EXPORT SUPPLY ELASTICITIES
Table 23: Export Supply Elasticities of Developed Countries
Country Elasticity Level o f Aggregation A uthor
B e lg iu m 1 .2 ( I r ) to ta l e x p o r ts G o ld s te in a n d K h a n ( 1 9 7 8 )
C anada 4 .9 ( I r ) m in e ra ls a n d m e ta l s B asev i (1 9 7 3 )
( to n o n -U S A )
4 .9 (I r ) c h e m ic a ls & f e r tiliz e r s B asev i (1 9 7 3 )
(to U S A )
1 .2 7 (I r ) to ta l e x p o r ts ( 1 9 6 2 ) L a w ren c e (1 9 8 9 )
1 .3 4 < lr) to ta l e x p o r ts ( 1 9 6 4 ) L a w r e n c e (1 9 8 9 )
1 3 4 (Ir) to ta l e x p o r ts ( 1 9 6 6 ) L a w ren c e (1 9 8 9 )
1 .4 6 (I r ) to ta l e x p o r ts ( 1 9 6 8 ) L a w ren c e (1 9 8 9 )
1 .6 7 ( I r ) to ta l e x p o t ts ( 1 9 7 0 ) L a w re n c e (1 9 8 9 )
1 .4 4 ( I r ) to ta l e x p o r ts ( 1 9 7 2 ) L a w ren c e (1 9 8 9 )
1 .5 7 ( l r ) to ta l e x p o r ts ( 1 9 7 4 ) L a w ren c e (1 9 8 9 )
1 .9 6 ( I r ) to ta l e x p o r ts ( 1 9 7 6 ) L a w ren c e (1 9 8 9 )
2 .0 1 ( I r ) to ta l e x p o r ts ( 1 9 7 8 ) L a w re n c e (1 9 8 9 )
2 .3 0 ( l r ) to ta l e x p o r ts ( 1 9 8 0 ) L a w re n c e (1 9 8 9 )
F ran ce IS ( lr) to ta l e x p o rts G o ld s te in & K h a n ( 1 9 7 8 )
G erm an y 4 .6 ( l r ) to ta l e x p o r ts G o ld s te in & K h a n ( 1 9 7 8 )
4 .6 ( l r ) to ta l e x p o tts G e r a c i a n d P re w o ( 1 9 8 0 )
4 .6 ( l r ) n o n e le c tr ic a l m a c h in e r y A r tu s & S o s a ( 1 9 7 8 )
0 .8 ( I r ) to ta l e x p o rts G y lf a s o n ( 1 9 7 8 )
G reece 2 .1 ( I r ) to ta l e x p o r ts ( 6 0 s - 7 0 s ) B a la ssa e t al. (1 9 8 6 )
I ta ly I I O r) to ta l e x p o r ts G o ld s te i n & K h a n ( 1 9 7 8 )
0 .5 ( l r ) to ta l e x p o r ts G y lf a s o n ( 1 9 7 8 )
Japan 6 .7 ( l r ) to ta l e x p o r ts G e r a c i a n d P re w o ( 1 9 8 0 )
2 .9 ( I r ) to ta l e x p o rts G o ld s te i n & K h a n ( 1 9 7 8 )
1-7 ( l r ) to ta l e x p o rts G y lf a s o n ( 1 9 7 8 )
N e th e rla n d s 2 .5 ( l r ) to ta l e x p o rts G o ld s te i n & K h a n ( 1 9 7 8 )

1 .4 ( l r ) to ta l e x p o r ts G y lf a s o n ( 1 9 7 8 )
Sw eden 3 .5 4 ( lr ) m a n u fa c tu r e s E ttlin ( 1 9 7 7 )
1 .4 5 ( l r ) to ta l e x p o r ts L u n d b o rg (l9 8 l)
UK 4 .2 ( l r ) n o n e le c tric a l m a c h in e r y A r tu s & S o s a ( 1 9 7 8 )
1 .4 ( l r ) to ta l e x p o tts G o ld s te i n & K h a n ( 1 9 7 8 )
1 .4 ( l r ) to ta l e x p o r ts G e r a c i a n d P re w o ( 1 9 8 0 )
0 .8 ( l r ) to ta l e x p o rts G y lf a s o n ( 1 9 7 8 )
0 .7 ( lr ) to ta l e x p o rts D u n le v y ( 1 9 7 9 )
U SA 1 2 .2 ( l r ) to ta l e x p o rts G e r a c i a n d P re w o ( 1 9 8 0 )
H -5 (lr ) to ta l e x p o rts M a g e e (1 9 7 0 )
6 .6 ( l r ) to ta l e x p o tts G o ld s te i n & K h a n ( 1 9 7 8 )
3 .1 ( l r ) n o n e le c tr ic a l m a c h in e r y A r tu s & S o s a ( 1 9 7 8 )
2 .4 ( lr) to ta l e x p o rts G y lf a s o n ( 1 9 7 8 )
2 .1 ( I r ) m a n u fa c tu r e s D u n le v y ( 1 9 7 9 )
206
207
Table 24: Export Supply Elasticities of Developing Countries

Country Elasticity Level o f Aggregation Author
C o te d t v o i r e 1 .0 8 ( lr ) to t a l e x p o rts
(d is e q u ilib r iu m m o d e l) T e g en e(1 9 9 0 )
0 .8 6 (Ir) to t a l e x p o rts
( e q u ilib r iu m m o d e l) T e g en e(1 9 9 0 )
0 .5 2 ( lr ) to t a l e x p o rts A riz e ( 1 9 8 7 )
E th io p ia 1 .5 5 ( lr ) t o t a l e x p o rts (d is e q .) T e g e n e ( l9 9 0 )
1 .2 0 (Ir) to t a l e x p o rts (e q .) Tegene (1 9 9 0 )
H u n g a ry 0 .5 7 (Ir) to t a l e x p o rts H a lp e m & S z e k e ly ( 1 9 9 2 )
I n d ia 2 5 .5 ( lr ) to t a l e x p o rts K o sh a l e t a l . ; ( a s re p .
1.9 1 ( s r ) t o t a l e x p o rts in V in o d & M c C u llo u g h 1 9 9 4 )
- 1 4 5 .4 t o 1 6 9 .7 (Ir) t o t a l e x p o rts V in o d & M c C u ll o u g h ( 1 9 9 4 )
- 0 .1 5 2 t o 5 .0 2 1 (s r) to t a l e x p o rts V in o d & M c C u ll o u g h (1 9 9 4 )
0 .9 2 ( l r ) to t a l e x p o rts A r iz e ( 1 9 9 0 )
In d o n e s ia 2 .1 5 (Ir) to t a l e x p o rts A riz e ( 1 9 9 0 )
K enya 2 .3 3 ( lr ) t o t a l e x p o tts (d is e q .) T egene (1 9 9 0 )

2 .3 0 (lr ) to t a l e x p o rts A r iz e ( 1 9 8 7 )
2 .1 2 (Ir) to t a l e x p o rts (e q .) T egene (1 9 9 0 )
K o re a 2 .7 5 (Ir) to t a l e x p o rts (1 9 6 7 - 7 1 ) K ang & K w on (1 9 8 8 )

1 .0 5 ( lr ) to t a l e x p o rts ( 1 9 7 2 - 7 6 ) K a n g & K w o n (1 9 8 8 )
0 .7 9 (Ir) to t a l e x p o rts (1 9 7 7 - 8 1 ) K ang & K w o n (1 9 8 8 )
1.9 - 2 .4 ( lr ) to t a l e x p o rts (6 0 s - 7 0 s ) B a la s s a e t a l . ( 1 9 8 6 )
M a la w i 1 .1 3 (Ir) to t a l e x p o rts (d is e q .) T egene (1 9 9 0 )

0 . 9 7 (Ir) to t a l e x p o rts (e q .) T egene (1 9 9 0 )
0 .8 0 (Ir) to t a l e x p o rts A riz e ( 1 9 8 7 )
M a la y s ia 3 .3 0 (Ir) to t a l e x p o rts A riz e ( 1 9 9 0 )
M a u r itiu s 0 .6 8 ( lr ) to t a l e x p o rts (d ise q .) T egene (1 9 9 0 )

0 .5 0 (Ir) to t a l e x p o rts A riz e ( 1 9 8 7 )
0 . 4 7 (lr ) to t a l e x p o rts (e q .) T egene (1 9 9 0 )
M o ro c c o 0 .5 2 (Ir) to t a l e x p o rts A r iz e ( 1 9 8 7 )
P a k is ta n 0 .7 6 ( lr ) to t a l e x p o rts A riz e ( 1 9 9 0 )
P h ilip p in e s 0 .3 1 (lr ) to t a l e x p o rts A riz e (1 9 9 0 )
T h a ila n d 0 .2 7 ( lr ) to t a l e x p o rts A riz e (1 9 9 0 )
T r in id a d 0 .1 2 - 0 .1 7 (s r) su g arcan e G afar(1 9 8 7 )
& Tobago 0 .1 3 - 0 .7 8 ( lr ) su g arcan e G a fa r(1 9 8 7 )
T u n isia 0 .9 2 (Ir) to t a l e x p o r ts (e q .) T egene (1 9 9 0 )

0 .6 3 (Ir) to t a l e x p o r ts (d is e q .) T egene (1 9 9 0 )
U p p e r V o lta 0 .0 0 (Ir) to t a l e x p o rts A riz e ( 1 9 8 7 )
Z a m b ia 0 .9 9 ( lr ) to t a l e x p o rts A riz e ( 1 9 8 7 )
0 .8 8 (Ir) to t a l e x p o r ts (d is e q .) Tegene (1 9 9 0 )
0 .8 5 ( lr ) to t a l e x p o r ts (e q .) T egene (1 9 9 0 )
APPENDIX 2: PROBLEMS RELATED TO INTEREST RATE DATA
The main problem with interest rate data is not the limited data availability but the
questionable data quality. As is well-known, most developing country interest rates are
rarely the result of the market but of heavy government or central bank intervention,
which hampers a serious comparison of nominal as well as real interest rates. The real
interest rate seems to be the more appropriate variable for any kind of economic analysis.
However, the real interest rate is further distorted by the low data quality of inflation rates,
whether calculated on the basis o f the GDP deflator or the consumer price index (CPI).
The first two columns of Tables 25 and 26 provide the available country averages
from 1970-1994 for the nominal deposit rate in perecent and the nominal lending rate in
percent. Column 3 provides the real deposit rate based on the inflation rate o f the GDP
deflator, while column provides the real lending rate based on the inflation rate of the CPI.
While the data for most of the industrialized countries seem to be more or less plausible,
the developing country data make little sense. For example, the average nominal deposit
rate over 25 years is higher than 1,196% for Argentina, higher than 1,909% for Chile,
higher than 15,574% for Nicaragua, and higher than 576% for Peru. Even after deflating
the nominal deposit rates by the GDP deflator, the real deposit rates remain to be higher
than 736%, 1,137%, 11,449%, and lower than -857% for Argentina, Chile, Nicaragua and
208
209
Peru, respectively. Even though the data availability is further limited for the nominal or
real lending rate, Table 26 shows an even worse picture for the nominal or real lending
rates than for the nominal or real deposit rate. The picture gets worse if one looks at time-
series data instead of country averages. In order to get meaningful data of real interest
rates, it would be necessary to exclude every country which experienced an inflation rate
o f more than 40 percent in at least one year. However, this would leave us with data of
less than 39 developing countries. In short, there is insufficient data for a systematic
analysis of interest rates and development.
Table 25:
Country Averages of Interest Rates for Industrialized Countries (1970-94)
Country deposit deposit lending lending

rate rate rate rate
(nominal) (real) (nominal) (real)
Australia 9.71 0.94 14.68 6.15

Austria 3.86 -0.22 n.a. n.a.
Belgium 5.80 0.45 12.40 8.36
Canada 9.12 3.66 10.45 4.32
Denmark 8.58 3.53 13.74 7.92
Finland 7.33 1.76 9.77 3.26
France 6.08 -1.88 16.27 8.20
Germany 5.67 2.29 10.76 7.64
Greece 13.38 -1.89 18.61 2.55
Ireland 6.70 -2.26 9.57 7.28
Italy 9.20 -1.16 15.74 5.91
Japan 3.52 -0.75 6.84 1.87
Netherlands 4.17 1.46 10.95 7.94
New Zealand 10.55 3.01 13.09 4.16
Norway 7.86 2.47 13.86 7.17
Portugal 16.69 -0.25 19.91 3.41
Spain 10.35 1.12 14.47 4.72
Sweden 8.41 0.67 13.13 5.49
Switzerland 5.09 1.30 6.04 2.60
United Kingdom 8.42 -0.66 10.21 1.27
United States 7.02 1.65 9.50 3.78
210
Table 26:
Country Averages of Interest Rates for Developing Countries (1970-94)

rate rate rate rate
Algeria n.a. n.a. n.a. n.a.

Argentina 1196.09 736.52 739144.50 738599.10
Bangladesh 10.06 2.62 13.35 4.78
Benin 5.97 0.20 14.31 n.a.
Bolivia 38.26 -803.05 72.57 -853.67
Botswana 9.06 -2.64 11.88 0.39
Brazil 1909.89 1137.90 n.a. n.a.
Burkina Faso 5.97 0.23 14.31 7.85
Burundi 3.72 -5.29 12.00 1.04
Cameroon 7.44 1.37 14.81 6.95
CAR 7.08 -2.09 13.41 12.00
Chad 5.88 0.33 12.92 12.94
Chile 34.43 6.64 47.70 21.78
China 6.48 1.75 7.47 -0.66
Colombia 31.66 6.84 41.51 16.05
Congo 7.30 1.95 13.37 6.67
Costa Rica 17.63 -6.12 27.20 3.46
Cote d'Ivoire 5.97 -0.65 14.31 6.43
Dominican Rep. n.a. n.a. n.a. n.a.
Ecuador 31.14 - 11.00 26.63 -10.96
Egypt 9.75 -3.33 14.20 -1.42
El Salvador 14.43 -1.93 17.35 -0.99
Gabon 7.67 -2.37 13.61 8.21
Gambia 10.74 -1.39 22.04 8.61
Ghana 15.68 -26.94 20.67 -32.51
Guatemala 11.63 -2.84 16.37 1.18
Guinea-Bissau 38.12 -11.37 38.59 -7.60
Haiti 10.04 1.62 n.a. n.a.
Honduras 9.94 -0.80 18.16 6.66
India n.a. n.a. 16.41 7.60
Indonesia 13.63 -1.15 21.69 13.96
Jamaica 17.05 -5.71 24.57 0.97
Kenya 7.84 -1.27 12.36 0.36
Korea 12.92 0.62 11.04 3.09
Lesotho 10.29 -3.28 16.09 2.37
211
Table 26 (cont.)

rate rate rate rate
Madagascar 9.41 -7.77 13.84 -3.42

Malawi 13.51 -3.89 20.94 3.72
Malaysia 6.75 2.57 8.62 4.91
Mali 5.97 -1.07 14.31 n.a.
Mauritania 5.65 -2.71 11.38 3.40
Mauritius 10.47 1.62 15.40 7.37
Mexico 37.00 -7.92 38.05 -7.12
Morocco 7.02 -0.28 8.03 0.44
Myanmar 1.43 -10.10 7.76 -1.53
Nepal 5.23 -5.53 14.79 5.92
Nicaragua 15574.48 11449.82 17508.87 19460.03
Niger 5.97 -0.35 14.31 10.36
Nigeria 8.58 -12.29 12.32 -10.49
Pakistan n.a. n.a. n.a. n.a.
Panama 6.99 5.51 11.66 10.88
Papua New Guinea 8.42 3.58 12.52 6.56
Paraguay 22.16 -0.83 31.43 7.21
Peru 576.23 -857.44 846.35 -516.59
Philippines 12.91 0.07 17.68 4.37
Rwanda 5.98 1.79 13.86 6.25
Saudi Arabia n.a. n.a. n.a. n.a.
Senegal 5.97 -0.39 14.31 8.56
Sierra Leone 18.31 -28.35 21.46 -22.09
Singapore 5.04 1.43 7.95 4.94
Somalia 7.83 -23.10 13.13 -23.73
Sri Lanka 15.63 3.27 14.57 1.84
Sudan 9.16 -17.04 n.a. n.a.
Syrian Arab Rep. 6.12 -4.83 n.a. n.a.
Tanzania 6.70 -14.05 19.18 -8.79
Thailand 10.80 4.91 16.93 10.58
Togo 5.97 -0.58 14.47 9.24
Tunisia 4.52 -4.32 8.65 2.10
Turkey 35.30 -15.67 37.52 -10.46
Uruguay 59.12 -1.76 94.34 32.35
Venezuela 24.17 -8.87 23.12 -11.59
Zaire n.a. n.a. n.a. n.a.
Zambia 12.14 -21.83 21.91 -25.65
APPENDIX 3: STANDARD DEFINITIONS OF STRUCTURAL VARIABLES
A. Sectoral Composition of Output

1. Agriculture
Agriculture covers forestry, hunting, and fishing, as well as agriculture. In developing
countries with high level of subsistence farming, much of agricultural production is either not
exchanged, or not exchanged for money. This increases the difficulty of measuring the contribution
of agriculture to GDP and reduces the reliability and comparability off such numbers.
2. Industry
The industry sector comprises mining and quarrying; manufacturing; construction; and
electricity, gas and water.
3. Manufacturing
The manufacturing sector comprises all industrial commodities excluding mining and
quarrying; construction; and electricity, gas and water.
4. Services
The service sector includes all service activities, that is, transport, storage and
communications; wholesale and retail trade; banking, insurance, and real estate; ownership of
dwelling; public administration and defense; and other services.
B. Investment
1. Gross domestic investment
Gross domestic investment is the sum of gross domestic fixed investment (see definition
below) and change in stocks.
2. Gross domestic fixed investment

Gross domestic fixed investment comprises all outlays (purchases and own-account
production) on additions of new and imported durable goods to the stocks of fixed assets, less the
proceeds of net sales of similar second hand and scrapped goods. Outlays by general government
on durable goods primarily for military purposes are excluded. According to the System of
National Accounts (SNA), those outlays are treated as current consumption and classified under
government consumption.
3. Domestic fixed private/public investment

Gross domestic fixed private/public investment comprises all outlays (purchases and own-
account production) by private/public sector enterprises on additions of new and imported durable
goods to the stocks of fixed assets, less the proceeds of net sales of similar second hand and
scrapped goods.
212
213
C. Consumption and Savings

1. Total consumption
Total consumption is the sum of private consumption (see definition below) and general
government consumption (see definition below).
2. Private consumption
Private consumption expenditure comprises the market value of all goods and services
purchases or received as income in kind by individuals and non-profit institutions, including the
imputed rent of owner-occupied dwelling.
3. General government consumption

General government consumption covers all current expenditures for goods and services by
government bodies. Excluded are outlays of public nonfinancial and public financial enterprises.
According to the SNA, outlays of the general government on durable goods primarily for military
purposes are included under this item.
4. Gross domestic savings

Gross domestic savings is equal to gross domestic product minus total consumption.
P . Government Expenditures and Revenues

1. Total government expenditure
Total government expenditure includes all nonrepayable and nonrepaying payment by
government, whether for current or capital purposes. Expenditure excludes government
amortization payments (which are classified in financing) and government lending (which is
classified in lending minus repayments).
2. Total revenue
Total revenue include all receipts, whether requited or unrequited, other than grants.
Revenue is shown net of refunds and other adjustment transactions. Revenue is otherwise shown
gross except for the proceeds of departmental enterprise sales to the public, which are netted
against the corresponding operating expenditures.
3. Tax revenue
Total revenue is defined as all government revenues from compulsory, unrequited,
nonrepayable receipts for public purposes, including interest collected on tax arrears and penalties
collected on nonpayment or late payments of taxes.
4. Fiscal deficit
The sum of total expenditure and government lending minus repayments, less the sum of
revenue and all grants received. The primary source is the IMF's Government Finance Statistics
Yearbook (GFSY). GFSY data are reported by countries using the system of common definitions
and classifications found in the IMF Manual on Government Finance Statistics (1986).
214
E. Inflation and Money S upp Iv

1. GDP deflator
The GDP deflator is derived by dividing current price estimates of GDP at market prices
by constant price estimates, also called the implicit GDP deflator.
2. Consumer Price Index (CPI)

The consumer price index (CPI) is the most frequently used indicator of inflation and
reflects changes in the cost of acquiring a fixed basket of goods and services by the average
consumer. This item is equal to line "64...zf' in the IMF's International Financial Statistics
publication.
3. Money supply (M l)
Money is the sum of currency outside banks and demand deposits other than those of the
central government. This series, frequently referred to as Ml is a narrower definition of money
than M2 (i.e., money plus quasi money). Data are from the IMF's monetary survey. This item is
equal to line "34...zf' in the IMF's International Financial Statistics publication.
4. Money supply broadly defined (M2)

The money supply broadly defined, also called M2, comprise money and quasi monetary
liabilities of a country's financial institutions to residents other than the central government.
F. Trade Related Variables

1. Current account balance
The current account balance before official transfers is the sum of net exports of goods
and nonfactor services, net factor services, net factor service income, and net private transfers.
This item is equal to line "77a.d" in the IMF's International Financial Statistics publication.
2. Capital account balance

The capital account balance is calculated from the IMF's International Financial
Statistics as the sum of net direct investment (line item 77bad), net portfolio investment (line item
77bbd), the net of other capital investments (line item 77g.d) and net errors and omissions (line
item 77e.d).
3. Total imports
Total imports of goods and services is the sum of merchandise imports f.o.b., imports of
nonfactor services and factor payments.
4. Merchandise imports
Merchandise imports refer to all movable goods (including non-monetary gold) involved in a
change of ownership from nonresidents to residents. Merchandise imports are valued free on board
(f.o.b.) at the customs frontier of the exporting country. An f.o.b. price at the customs frontier
includes the value of the goods, and the value of outside packaging, and related distributive
services used up to, and including, loading the goods onto the carrier at the customs frontier of the
exporting country. The few types of goods not included in merchandise include travelers' purchases
abroad, which are included in travel, and purchases of goods by diplomatic and military personnel,
which are classified under other official goods, services and income.
215
5. Total exports
Total exports of goods and services is the sum of merchandise exports f.o.b., exports of
nonfactor services and factor receipts.
6. Merchandise exports
Merchandise exports refer to all movable goods (including non-monetary gold) involved in a
change of ownership from residents to nonresidents. Merchandise exports are valued free on board
(f.o.b.) at the customs frontier of the exporting country. An f.o.b. price at the customs frontier
includes the value of the goods, and the value of outside packaging, and related distributive
services used up to, and including, loading the goods onto the carrier at the customs frontier of the
exporting country. The few types of goods not included in merchandise include travelers' purchases
abroad, which are included in travel, and purchases of goods by diplomatic and military personnel,
which are classified under other official goods, services and income.
7. Exports o f machinery
Exports of machinery comprise commodities in the Standard International Trade
Classification (SITC) Rev. I, Section 7 (Machinery and Transport Equipment).
8. Primary exports (as % o f GDP)

Exports of primary products comprise commodities in SITC Rev. 1, Section 0 to 4 and 68
(Food and Live Animal, Beverages and Tobacco, Animal and Vegetable Oil and Fat, Crude
Materials, including Fuels, and Non-Ferrous Metals.
APPENDIX 4: PATTERNS OF DEVELOPMENT: DETAILS OF
RELEVANT REGRESSION RESULTS
Table 27:
Results for the Share of Agriculture
log of log of log of log of square of square of square of adjusted
gdpcap hescap pop density (flow togjhescap) logfpop) log(density) R-squared
logiini ■0.569
t-ta tM c -20.10 081
loglin2 -0.814
t-sta tto lc -19.73 0.81
loglin3 •0.815 0.003

r-ia rttrfc -19.47 0.09 081
logllnA -0.808 •0.032

MMWc -19.43 -1.09 0.81
logllnS •0.789 0.014 0006

r-<nestfc -T4.40 0.41 0.73 0.81
ioglln6 41.792 •0.030 0.004

r-tM stlc -14.81 -1.02 0.50 0.81
toglin7 •0.051 0.000

r-ta titttc -20.17 0.12 0.82
logllnS -0.052 -0.008

t- tt Mfetfc -20.19 -2.14 0.83
216
217
Table 28: Results for the Share of Industry

log of log of log of log of squaroof squaroof square of adjusted
gdpcap hescap pop density rflow logjhescap) log(pop) log(denslty) R-squared
logllnl 0.149
r-stadstfc 8.70 0.32
loglln2 0.228
t-natttO c 730 037
loglln3 0.221 0.037

r-statittfc 7.18 184 0.38
logllnS 0.234 -0.035

r-ta r/* 0c 7.80 -ISO 0.38
logllnS 0.179 0.018 •0.009

t-tuO tO c 4.49 0.74 -1.84 0.39
logllnS 0.176 •0.040 -0.012

e -ta tito c 4.58 -1.88 -2.44 0.41
loglln7 0.014 0.001

t-tta tM c 8.00 173 036
logllnS 0.014 -0.003

r-nadtdc 7.10 -1.30 0.35
Table 29: Results for the Share of Manufaturing

log of log of log of log of squaroof square of square of adjusted
gdpcap hescap pop density rflow log(hescap| log(pop) log(density) R-squared
logllnl 0.171
r-statftpc 5.57 0.25
loglin2 0.269
r-sradsOc 8.32 0 31
loglln3 0.246 0.122

r-sfadsffc 8.28 4.27 0.42
loglln4 0.252 0.088

t-ia u ttlc 8.13 3.03 0.38
logllnS 0.299 0.144 0.012

5.87 4.56 1.50 043
loglln6 0.246 0.088 •0.001

r-staowfc 4.84 2.07 ■0.18 0.35
loglin7 0015 0.004

t-sradstfc 8.18 4.23 0.41
logllnS 0.016 0.010

r-soufttfe 5.89 2.80 0.34
218
Table 30:
Results for the Share of Services
log of log of log of log of square of square of squaroof adjusted
gdpcap hascap POP density rflow log(hwcap) log(pop) log(density) R-squared
logllnl 0.106
t- ta O s O c 9.20 0.48
loglin2 0.151
t- ta o s O c 8.99 0.46
log!ln3 0.156 -0.029

r-scadsOc 948 -2.48 0.49
loglln4 0.146 0.012

t-stadslfc 8.77 1.02 0.48
lofllln5 0.172 -0.023 0.004

r-sadsdc 8.00 - 1 .8 8 1 .1 4 0.49
loglinfi 0.178 0.015 0.006

r-stadsde 8 .4 1 1 .2 8 2 .2 5 049
logiln7 0.010 •0.001

t- M O tittiC 937 -253 0.49
logllnS 0.009 0.001

t-stattstic B.61 0.70 0.45
Table 31:
Results for the Share of Gross Domestic Investment to GDP
log of log of log of log of squaroof square of square of adjusted
gdpcap hescap pop density rflow log(hescap) log(pop) log(denslty) R-squared
logllnl 0.070
t- t t a O t t ie 3.68 0.12
loglln2 0.105
r-stadsdc 3.89 0.13
loglln3 0.108 -0.017

r s a tM c 3.96 ■ 0 .8 7 0.13
logline 0.106 -0.007

ts a ta a c 3.88 -0.34 012
logllnS 0.165 0.008 0.013

e - ta a t tte 4.79 0.36 258 018
logllnS 0.162 -0.001 0.012

t- tn tM c 4 .8 1 -0.06 268 0.18
loglln7 0.007 0.000

t-S O O tU C 3.77 ■ 0 .8 5 0.12
logllnS 0.006 0.000

r-saOstfe 3.65 0.09 0.11
219
Table 32:
Results for the Share of Gross Domestic Fixed Investment to GDP
log of log of log of log of squaroof square of squaroof adjusted
gdpcap hoscap pop density rflow logjhescap) logjpop) log(denstty) R-squared
loglim 0.084
t-s a e to c 4.33 0.18
loglJn2 0.127
t-ta a x o c 4.50 019
logllnS 0.129 -0.012

t-ta ttta c 4.60 -057 017
loglln4 0.126 0.002

r-stadtde 4.51 0.11 017
logllnS 0.178 0.008 0.012

r -ta tttttc 4.86 0.34 2.03 0.20
loglln6 0.175 0.007 0.011

r-tttO tO c 4.88 0.36 2.11 0 20
loglin7 0.008 0.000

t-s M tttc 4.40 -0.30 0.16
logllnS 0.008 0.001

tstattsU e 4.25 0.46 0.16
Table 33:
Results for the Share of Fixed Private Investment to GDP
gdpcap hescap pop density rflow log(hescap) log(pop) log(denslty) R-squared
logllnl 0.318
r-ttufscfc 5.64 0.34
loglin2 0.382
t-a a ttsttc 5.18 0.30
loglln3 0.381 ■0.005

r-sratfsdc 5.13 -0.14 0.29
logllna 0.382 0.001

r-starfstfc 5.13 0.03 0.29
logllnS 0.439 0.013 0.010

r-stattstfc 4.33 0.30 0.83 0.28
loglln6 0.428 0.001 0.009

c -to ttttte 4.53 0.02 0.70 0.28
loglln7 0.025 0.000

r-sraOstfc 5.02 ■0.06 028
logllnS 0.025 0.002

r-xfadsdc 4.66 0.32 0.28
220
Table 34:
Results for the Share of Total Consumption to GDP
log of log of log of log of square of square of squaroof adjustsd
gdpcap hsscap pop density rflow togPnscap) log(pop) log(dsnslty) R-squared
logllnl -0.057
t- ta tw c -7.06 0.35
loglln2 -0.082
r-ta u M c -7.10 0.35
loglin3 •0.079 •0.614

t-MOUtee -8.89 -1 71 0.36
loglin4 -0.083 0.005

t-MUttttiC -7.09 0.61 0.34
logllnS -0.026 0.009 0.012

t-*Utl*Oe -2.10 1.20 6.79 0.58
loglln6 -0.029 0.010 0.011

t-tu ttttie -2.46 1.52 716 0.58
loglinT -0.005 0.000

r -ta tittfc -6.69 -1.69 0.35
logllnS -0.005 0.000

ts M ttlc -6.76 0.04 0.33
Table 35:
Results for the Share of Private Consumption to GDP
log of log of log of log of square of square of square of adjustsd
gdpcap hascap pop dtnslty rflow log(hascap) log(pop) log(dtnslty) R-squared
logllnl •0.088
r-stsdstfc -6.57 0.44
loglin2 -0123
r-Msdstfc -6.15 0.42
loglln3 -0.123 0.003

t-ta tttU e -6.08 0.29 0 41
loglin4 •0.128 0.027

c-tM tO e •6.68 261 0.45
logllnS •0.066 0.028 0.013

r-atadtoc -3.72 248 5.07 0.54
loglin6 -0.077 0.032 0.011

r s a o tt/c -4.57 3.41 4.82 0.56
loglln7 -0.008 0.000

r-stadstfc -7.04 0.19 040
logllnS -0.008 0003

r-natfstfc ■8.33 1.90 0.42
221
Table 36:
Results for the Share of Savings to GDP
gdpcap hascap pop density rflow log(hascap) log(pop) log(dansity) R-squared
logllnl 0.295
r-taB sB c 8.20 0.30
loglin2 0.457
r-gnOgOc 8.97 0.35
loglin3 0.44a 0.081

t-MOtlMtk: 8.90 1.75 0.37
logllnS 0.458 -0.006

r-ta O tttc 8.81 ■0.13 0.35
logllnS 0.278 0.018 -0.046

t-a a ttttte 3.31 0.33 ■3.00 042
logllnS 0.272 -0.023 0049

r-tattsO c 3.33 •0 53 ■3.54 0.42
logllnr 0.027 0.003

t-ttaU gt/e 8.81 1.70 0.35
logllnS 0.028 0.001

t-stufstfe 6.62 0.22 0.33
Table 37:
Results for the Share of Government Expenditure to GDP
gdpcap hsscap pop dtnslty rflow log(hascap) log(pop) log(denslty) R-squared
logllnl 0.109
r-sradstfc 4.15 0.15
loglln2 0.138
r-gnttsttc 3.52 0.11
loglln3 0.160 -0.110

c-gaagOc 4.43 -4.28 0 26
logllnS 0.146 •0.041

t-gtaogtle 3.73 •1.48 0.13
logllnS 0.263 •0.064 0.024

ts M tttc 5.00 ,2.37 3.87 0.35
logllnS 0.284 -0.020 0.029

t-gtaUgttc 8.41 ■0.80 5.00 0.31
loglin7 0.010 -0.003

r-stadsde 4.40 -4.27 0.26
logllnS 0.010 -0.005

r-gtattgOc 3.83 -1.50 0.26
222
Table 38:
Results for the Share of Government Revenues to GDP
gdpcap hncap pop danslty rtlew log(haacap) log(pep) log(denslty) R-squared
logllnl 0.182
r-ttarfaOc 7.15 036
loglin2 0.245
r s a a ttte 8.44 0 31
logllnS 0.264 -0.605

e-tm a to c 7.35 -3.68 0.40
loglln4 0.252 •0.029

r-ta tttO c 553 -too 0 31
logllnS 0.305 -0.077 0.000

ts w tttic 553 ■ zee 1.35 0.40
loglln6 0.330 -0.017 0.017

rstaB sB c 804 -0.63 265 0.36
logllnT 0.017 -0.003

t-tU B tO c 734 ■3.71 040
logllnS 0.016 -0.003

t-s M ttic 6.50 -0.03 0.31
Table 39:
Results for the Share of Government Taxes to GDP
gdpcap hsscap pop danslty rflow log(hsscap) log(pop) log(denslty) R-squared
logllnl 0.204
CStM UtBC 8.30 0.43
loglln2 0.267
i-s tttls B c 707 0.36
loglin3 0.287 -0.098

t-tta tistK 8.00 ■3.87 0.44
logllnS 0.273 •0.026

t-ttaO tO c 7.13 -0.07 0 35
logllnS 0.315 -0.086 0.006

t- ta tM c 5.70 ■3.00 0.03 0.44
logllnS 0.343 •0.015 0.015

c-tatiMOc 7.19 -058 235 030
loglln7 0.018 -0.003

r s M tt/e 8.19 -4.00 0.45
logllnS 0.017 -0.004

r-saoMb 7.23 -1.12 0.36
223
Table 40:
Results for the Share of Broad Money to GDP
log of log of log of log of square of aquaraof aquara of adjusted
gdpcap haacap pop danslty rflow logjhescap) log)pop) log(danstty) R-squared
logllnl 0.238
t-tta ttfttc 8.88 0.45
loglin2 0.356
t-ta m a c 8.24 0.48
logllnS 0.338 0.068

t-ta m a c g.to 25< 0.51
loglirv* 0.332 0.097

r-stadftfc 0.31 3.86 0 55
logllnS 0.443 0.114 0.023

r-g a a ttfc 0.71 3.00 3.57 0.56
loglln6 0.399 0.104 0.014

t-ta tltO c 0.00 4.20 2.44 057
loglin7 0.021 0.002

t-statfstfc 0.13 253 0 51
logllnS 0.021 0 012

t-tu tttttc 0.11 3.S7 0.54
Table 41:
Results for Trade Intensity
log of log of log of log of log of aquaraof aquaraof square of adjusted
countrystze gdpcap haacap pop density (flow log(haacap) log(pop) fogfdensrty) R-squared
logllnl -0.173 0.072

r-siaasoe -8.48 232 0 34
loglln2 -0.174 0.084

r-ctaoauc -8.43 1.86 0.32
logllnS -0.068 0.124 •0.218

r-sratfstfe -244 3.24 -8.32 0.52
logllnd •0.286 0.124 ■0.218

t-ta O ttic -0.06 3.24 -8.32 0.52
logllnS •0.068 0.124 -0.217 0.000

r-nadstfe -243 247 -5.83 0.01 052
loglln6 -0.286 0.124 -0.217 0.000

t-tM ttte •8.88 247 ■5.83 O.Of 0.52
loglin7 -0.072 0.007 -0.006

r-stattelfc -255 -255 -6.17 0.52
logllnS -0.270 0.008 -0.024

r-MtatiMtlc ■8.61 -298 4.82 0.45
224
Table 42:
Results for the Share of the Current Account Balance to GDP
log of log of log of log of aquaraof aquaraof aquaraof adjusted
gdpcap noscap pop dtnslty rflow logjhascap) logjpop) log(danalty) R-squared
logllnl 2.947
ts a d e d c 0.70 0.33
ioglln2 4.484
i-s a tM c 743 0.37
loglin3 4.078 2.270

1-ta d xO c 7.00 8.05 055
loglirv* 4.285 1.048

c-ta d x d c 723 2.50 0.41
logllnS excluded
t-e a d e ttc excluded
toglin6 excluded
t-sa tisO c excluded
loglin7 0.257 0.066

ts a m tlc 7.78 5.62 053
loglln8 0.288 0.114

t-ta tttO c 7.02 2 07 0.39
Table 43:
Results for the Share of the Capital Account Balance to GDP
gdpcap hescap pop dtnslty (flow logjhescap) log(pop) log(denslty) R-squared
logllnl -1.052
t-ia ta ttc ■4.15 0.15
loglln2 -1.599
r-ttatfstfc -4.44 0.17
loglln3 -1.453 •0.820

r-stadatfc ■4.21 -3.27 025
loglint -1.521 -0.412

t-e te d tttc ■4.22 -1.01 0.18
logllnS excluded
r-etettsac excluded
loglln6 excluded
t-atatfatfe excluded
loglln7 -0.093 -0.024

r -ta d td c ■4.20 ■3.20 0.25
logllnS -0.095 -0.081

t-eted sd c -4.20 -1.85 0.19
225
Table 44:
Results for the Share of Exports to GDP
log of log of log of log of log of aquaraof aquaraof square o l adjusted
countrystzs gdpcap haacap pop danslty rflow log(hascap| log(pop) toqfdensJtyl R-squared
logllnl -0.165 0.146

r-sa tM c -5.58 4.25 0.34
loglln2 -0.168 0.192

r-iafliOc -5 5 2 3.81 032
loglln3 -0.067 0.229 •0.204

r-saasO c -2 0 2 5.07 •5.03 046
loglln4 -0.271 0.229 •0.205

MtUiHfc -8.01 5.07 -5.03 0.46
logllnS excluded
M U ttttfC excluded
logllnS excluded
HMMb excluded
loglin7 -0.071 0.014 -0.006

t-sta a ttlc -2 1 3 4.88 -4.88 045
logllnS -0.255 0.014 -0.022

r-nadtoc -7.01 4.70 -3.83 0.41
Table 45:
Results for the Share of Merchandise Exports to GDP
log of log of log of log of log of square of square of square o f adjusted
countrystza gdpcap Haacap pop density rflow logjhescap) logjpop) logfdenstty) R-squared
logllnl -0.127 0.135

t-staUsOc -4.10 3.83 0.25
loglin2 ■0.126 0.166

tsta tistic -4.14 325 0.22
loglln3 •0.027 0.207 -0.027

t-staustlc -0.80 4.42 -4.05 038
logllnS ■0.235 0.207 •0.201

t-staO sdc ■8.72 4.42 -4.95 0.38
logllnS excluded
tsta a sa c excluded
logllnS excluded
tsn a sO c excluded
loglln7 •0.031 0.013 •0.006

t-sM sd e -0.00 4.23 ■4.83 037
logllnS •0.212 0.013 •0.020

r-*ra#s0c -5.58 3.99 -3.45 0.30
226
Table 46:
Results for the Share of Machinery Exports to GDP
log of log of log of log of log of squaroof tq u areo f aquaraof adjusted
countryslzo gdpcap hascap pop density rflow log(hsscap) logfpop) toafdenxdyl R-squared
logllnl -0.057 0.688

t-s M ttic ■0.73 10.02 053
loglln2 -0.066 1.253

r-*od*Sc -0.84 0.48 050
loglin3 -0.201 1.210 0.266

t-etaO tdc - 2 10 0.32 238 0.53
loglin4 0.085 1.210 0.286

r-sntfadc 0.85 0.32 238 053
logllnS excluded
r-statfadc excluded
logllnS excluded
c-saOxOc excluded
loglin7 •0.205 0.077 0.009

c-tautsdc -218 0.54 242 054
logllnS 0117 0.076 0.044

r-*ndsdc 1.17 0.30 280 0.55
Table 47:
Results for the Share of Machinery Exports to Total Exports
gdpcap haacap POP denalty rflow log(haacap) log(pop) log(denalty) R-squared
logllnl 0.743
r-saflstfc 8.13 0.42
loglln2 1.052
t-statlstic 7.05 0.41
loglin3 1.006 0.388

t-ta O ttic 8.27 4.24 0.51
logllnS 1.014 0.208

t-tn tlsd c 7.77 2 23 0.44
logllnS excluded
c-taO tdc excluded
loglln6 excluded
r-stadstfe excluded
loglln7 0.065 0.011

r-sodxdc 8.50 4.25 0.S2
logllnS 0.064 0.028

r-sadsDc 7.80 234 0.45
227
Table 48:
Results for the Share of Primary Exports to Total Exports
log of log of log of log of squaroof square of aquaraof adjusted
gdpcap hsscap pop danslty rflow log(hescap) log(pop) log(denslty) R-squared
logllnl -0.252
r-sfaffsdc 5.02 026
loglin2 •0.349
m i aOsdc -5.37 0.24
loglln3 ■0.337 ■0.102

t-tatiM tic -5.20 -2.12 0.27
loglln4 -0.309 -0.224

t-to tM e -543 -5.43 0.43
logllnS excluded
t-tta lM c excluded
logllnS excluded
c-iorttO c excluded
loglln7 -0.022 -0.003

t-ta o to c 5.41 -2.10 028
logllnS •0.020 -0.027

t-tu tM c 537 -505 0.41
228
Table 49:
Results for Export Product Concentration
log of log of log of log of aquaraof aquaraof aquaraof adjuatad
gdpcap haacap pop danaity (flow logfhaacap) log(pop) log(danslty) R-squarad
logllnl •0.217
-8.88 0.46
loglln2 -0.312
r-toutxdc -8.91 0.47
logllnS -0.292 •0.131

t-ta tis tlc -0.72 -5.88 061
loglln4 •0.295 -0.088

r-sartsO c -8.95 •3.68 053
logllnS -0.348 -0.155 -0.013

r-soesBc -8.97 -8.35 -2.21 0.63
logl(n6 -0.292 •0.087 0.001

i-tu a tttc -8.83 ■3.61 0 14 0 53
loglln7 -0.019 -0.004

t-tttO tO e -10.05 -5.95 082
logllnS -0.019 -0.009

ts n o to c ■8.94 ■3.02 0.52
Table 50:
Results for Export Market Power in World Markets
log of log of log of log of log of aquaraof aquaraof tq u a n o f adjuatad
countiysiza gdpcap haacap pop danatty mow log(haacap) log(pop) log(6tns8y) R-squared
logllnl 0.270 0.485

t-su ttstfc 484 712 0.44
loglln2 0267 0.710
r-sn o stic 4.96 7.90 0.48
loglin3 0.048 0.864 0.453
t-sn ttstic 0.88 8.46 6.53 065
logllnd 0.501 0.634 0.453
r-sadadc 8.80 8.46 6.54 0.65
logllnS 0.054 0.421 0.357 -0.048
r-aatfrdc 1.03 4.55 5.04 -3 5 4 0.69
loglln6 0.410 0.421 0.357 •0.049
t-tM ttle 8.90 4.55 5.05 ■3.54 0.69
loglln7 0.055 0.040 0.013
t-tUO tU e 0.98 8.37 6.24 0.63
logllnS 0.519 0.038 0.062
t-ttMtjMtfC 9.13 8.14 6.96 0.66
229
Table 51:
Results for Financial Market Development Indicator #1
gdpcap haacap pop danaity (flow logjhaacap) logjpop) log(danalty) R-aquarad
logllnl 0.226
r-ia B trfc 9.22 048
loglin2 0.331
isnO M ttc 9.54 0.50
loglln3 0.319 0.064

t-S M M c 3.41 2.59 052
loglln4 0.312 0.098

t- to tM c 9.75 4.34 0.58
logllnS 0.383 0.092 0.014

r-ctufeoc 8.64 340 2 29 0 55
ioglln6 0.346 0.102 0.007

c-statltoc 8.50 4.48 1.3S 0 58
loglln7 0.020 0.002

r s ta tM c 3.53 zee 0.53
logllnS 0.020 0.012

t-tU O ttle 9.58 3.95 0.57
Table 52:
gdpcap haacap pop danaity rflow log(haacap) log(pop) logjdanaity) R-aquared
logliM 0.121
t-tta tixtlc 5.57 0.25
loglln2 0.166
r-sfaoadc 5.26 0.23
loglln3 0.168 -0.008

tstaO sO c 5.23 -0.35 0.22
loglind 0.168 -0.010

r-*tatf»ttc 525 -0.41 0.22
logllnS 0.147 -0.019 -0.005

t-atatfatfe 3.52 -0.69 -0.88 0.22
loglin6 0.152 -0.013

r-acatfatfc 3.73 ■0.53 ■0.63 0.22
loglin7 0.011 0.000

t-tta ttttic 5.36 -0.29 0.23
logllnS 0.011 -0.001

r-MtMOsrtc 5.34 ■0.13 0.23
230
Table 53:
log of log of log of log of aquaraof aquara of aquara o f adjuatad
gdpcap haacap pop danaity rflow logjheacap) log(pop) log(dartaity) R-aquared
logllnl 0.103
r-ta a tO c 240 0.05
togllnZ 0.12S
ts a O tttc 2 02 0.03
loglln3 0.134 -0.052

r-m u sttc 210 -1.15 0.04
logllrU 0.131 •0.033

i-ta o sO e 2 10 ■0.75 0.03
logllnS 0.127 -0.055 -0.002

r-cattftfc 1.55 -1.00 ■0.14 0.03
loglln6 0.147 -0.032 0.003

t-o a tttu c 1.83 ■0.71 0.31 0.02
loglln7 0.008 -0.001

r-stadatfc 215 -103 003
logllnS 0.008 -0.001

ts tu tttic 2 02 ■0.14 0.02
Table 54:
log of log of log of log of aquaraof aquara of aquara of adjusted
gdpcap haacap pop danaity rflow log(haacap) log(pop) logjdensity) R-aquared
logllnl 0.352
r-aradaflc 0.07 052
loglin2 0.488
r-stadsdc 9.33 0.48
loglln3 0.478 0.054

r-aadadc 0.12 1.41 0.49
loglln4 0.477 0.063

c-MaatOo 0.11 1.71 0.48
logllnS 0.551 0.086 0.016

r-atattrtfc 8.11 2 02 1.85 0.50
loglln6 0.518 0.067 0.008

7.76 1.81 1.02 0.49
loglln? 0.030 0.002

r-staswfe 0.11 155 0.49
logllnS 0.030 0.008

rsaO tO c 9.00 1.84 0.50
APPENDIX 5: UNCERTAINTY MEASURES OF DIFFERENT TIME PERIODS
Table 55: Nominal GDP Uncertainty Across Different Time Periods

U ncertainty M easure based on: U ncertainty M easure based on:
C ountry________ 1970-84 1970-81 1983-84_______ Country____________ 1970-94 1970-81 1883-94
A ustralia 9.89 8.30 9.72 G uinea-B issau 13.06 11.69 14.86

A ustria 11.40 11.40 11.49 Haiti 2233 7.48 29.12
Belgium 12.75 1Z71 11.76 H onduras 9.68 4.97 9.49
C anada 6.52 5.06 6.22 India 8.22 6.19 9.12
Denmark 11.21 11.33 11.24 Indonesia 1298 1242 10.09
Finland 13.92 8.18 15.79 Jam aica 15.44 13.04 18.40
France 11.59 11.14 11.52 Kenya 14.13 7.59 16.36
G erm any 12.14 12.33 12.17 K orea 9.12 10.84 7.16
G reece 9.49 8.80 9.90 Lesotho 15.18 13.86 15.54
Ireland 10.07 9.56 10.44 M adagascar 13.87 10.30 14.86
Italy 11.90 9.76 13.86 Malawi 15.76 5.84 19.46
Jap an 10.52 9.75 9.75 M alaysia 10.83 11.06 9.82
N etherlands 11.85 12.06 11.45 Mali 18.43 14.45 19.19
New Z ealand 10.01 8.47 10.69 M auritania 10.24 6.70 11.36
Norway 8.84 6.55 8.38 M auritius 1222 13.96 9.77
Portugal 12.54 8.40 14.92 Mexico 20.05 11.64 19.25
Spain 13.70 12.10 15.07 M orocco 11.77 1286 11.34
Sw eden 13.14 8.30 15.28 M yanmar 14.58 14.94 1284
Sw itzerland 11.83 11.02 12.64 Nepal 8.73 11.25 5.84
UK 10.96 9.98 10.66 N icaragua 39.97 15.72 50.34
USA 2.59 1.41 1.79 Niger 16.30 1217 17.74
Algeria 14.57 8.88 1286 Nigeria 26.55 1260 31.70
A rgentina 24.21 20.45 31.87 Pakistan 15.42 2218 6.01
B angladesh 26.92 39.77 5.53 Panam a 7.16 4.69 7.22
Benin 14.74 10.12 17.17 Papua New G uinea 11.24 11.31 10.56
Bolivia 9.12 5.85 8.38 Paraguay 16.98 6.22 18.21
Botsw ana 12.08 8.34 10.88 Peru 23.64 15.10 30.81
Brazil 15.37 6.89 19.77 Philippines 9.57 6.42 9.24
Burkina F aso 11.17 9.38 1250 Rwanda 13.96 11.43 13.98
Burundi 9.99 7.79 8.17 Saudi Arabia 21.48 17.68 13.04
Cam eroon 15.98 6.67 17.43 Senegal 16.63 1274 19.37
CAR 12.57 9.56 13.87 Sierra Leone 33.61 8.90 49.79
Chad 13.37 12.22 13.93 Singapore 8.89 8.26 8.88
Chile 22.13 26.37 14.18 Som alia 19.78 25.21 9.75
China 9.75 11.78 8.38 Sri Lanka 14.34 20.95 4.68
Colom bia 8.30 5.35 7.71 Sudan 18.57 10.75 24.45
C ongo 18.77 10.47 19.83 Syrian Arab R ep. 16.03 10.68 15.69
C osta Rica 21.69 32.28 6.13 Tanzania 16.98 4.37 17.30
C ote d'Ivoire 16.45 13.04 15.48 Thailand 7.41 6.98 7.64
Dominican Rep. 14.87 5.15 18.20 Togo 17.05 1255 18.39
E cuador 13.80 9.59 10.84 Tunisia 9.76 8.68 7.56
Egypt 8.88 9.21 8.38 Turkey 16.84 15.20 18.39
El S alvador 6.60 8.21 3.88 Uruguay 26.65 19.15 29.47
G abon 19.82 22.36 14.79 Venezuela 17.30 7.22 1921
G am bia 13.69 13.86 13.65 Zaire 15.34 12.80 15.55
G hana 11.08 10.13 1246 Zam bia 17.87 1275 2205
G uatem ala 11.93 5.68 14.43
231
232
Table 56: Real GDP Uncertainty Across Different Time Periods

U ncertainty M easure b a se d on: U ncertainty M easu re b ased on:
C ountry________ 1970-84 1970-81 1983-84________ C ountry__________ 1970-84 1970-81 1983-84

A ustria 1.98 2.49 1.10 H aiti 3.97 4.04 1.50
Belgium 2.38 2.90 1.73 H onduras 3.13 3.33 2.13
C anada 2.90 1.76 2.77 India 3.20 3.92 Z18
Denm ark 2.26 2.76 1.70 Indonesia 3.22 3.00 2.15
Finland 4.25 3.41 4.99 Jam aica 4.66 5.58 3.73
F rance 1.85 2.17 1.41 K enya 7.33 9.99 3.22
G erm any 2.18 2.57 1.37 K orea 4.43 5.94 1.44
G reece 3.16 3.81 1.73 L eso th o 7.42 8.33 5.08
Italy 2.59 3.34 1.11 M alawi 5.16 5.68 4.87
Jap an 2.26 2.84 1.44 M alaysia 5.76 5.07 6.26
New Zealand 3.25 3.98 2.40 M auritania 6.45 7.92 4.88
Norway 2.06 1.63 2.18 M auritius 7.26 8.81 5.49
P ortugal 3.93 3.93 4.39 M exico 4.82 2.58 4.80
Spain 2.54 3.03 2.08 M orocco 3.83 3.73 4.06
Sw eden 1.83 1.92 1.89 M yanm ar 7.54 2.27 12.86
Sw itzerland 3.12 4.08 1.58 N epal 8.04 9.47 0.82
UK 2.69 2.90 2.66 N icaragua 11.23 13.95 7.09
USA 2.57 2.60 2.08 N iger 10.87 13.25 5.94
A lgeria 5.29 6.55 2.70 N igeria 8.53 7.26 7.81
A rgentina 5.26 4.29 5.50 P ak istan 3.53 4.64 1.66
B angladesh 10.24 13.26 5.08 P anam a 6.91 3.85 9.33
Benin 4.96 4.06 4.79 P ap u a New G uinea 3.48 2.59 4.52
Bolivia 3.28 2.75 1.82 P araguay 6.77 5.68 6.00
B otsw ana 9.60 6.25 13.70 P eru 7.46 4.29 9.53
Brazil 5.17 4.95 4.19 P hilippines 4.17 1.55 5.38
B urkina Faso 3.95 4.20 2.98 R w anda 4.04 3.66 3.31
B urundi 8.36 11.14 4.18 S au d i Arabia 9.84 6.17 8.55
C am eroon 6.24 4.87 6.48 S enegal 3.41 3.83 2.52
CAR 4.61 3.89 5.55 S ierra Leone 10.11 11.22 9.59
C had 11.44 9.77 11.95 S ingapore 3.61 3.07 3.70
Chile 8.03 8.09 4.85 S om alia 13.85 16.58 11.44
China 4.78 5.22 4.55 S ri L anka 3.19 3.36 2.55
Colom bia 2.04 1.88 1.49 S udan 7.75 8.77 6.41
Congo 7.14 6.30 5.30 S yrian Arab Rep. 9.35 9.81 8.03
C osta R ica 4.20 4.11 2.10 T anzania 6.29 6.14 7.80
C ote d'Ivoire 7.04 8.64 3.47 T hailand 3.65 3.67 3.44
D om inican Rep. 5.59 5.04 5.82 T ogo 6.00 5.66 6.52
E cuador 5.19 5.25 3.42 T unisia 3.21 2.68 2.28
E gypt 3.29 3.39 2.77 T urkey 3.67 4.04 3.46
El S alvador 4.41 5.57 1.06 U ruguay 5.59 2.64 7.08
G abon 9.79 11.00 8.67 V enezuela 5.08 3.99 6.52
G am bia 10.05 11.06 9.57 Z aire 5.00 5.95 3.86
G hana 8.14 9.60 3.08 Zam bia 5.43 5.84 4.33
G uatem ala 3.49 2.31 3.22
233
Table 57: Inflation Uncertainty Across Different Time Periods

U ncertainty M easure based o n : U ncertainty M easure b ased on:
C ountry_________ 1970-84 1970-81 1983-94_______ C ountry____________1970-94 1970-81 1983-94
A ustralia 2.77 3.43 2.99 G uinea-B issau 45.75 na 45.75

A ustria 1.57 2.04 1.49 Haiti 11.62 11.91 13.05
C anada 2.18 2.47 1.77 India 8.44 13.50 4.08
Denmark 2.45 3.86 0.98 Indonesia 8.49 14.96 2.92
Finland 2.77 4.18 1.86 Jam aica 17.98 13.81 27.29
F rance 2.26 3.16 1.13 K enya 8.17 6.91 10.55
G erm any 1.21 1.41 1.26 K orea 7.28 11.69 2.47
G reece 5.92 9.29 4.31 L esotho 4.34 4.99 5.25
Italy 3.56 5.82 1.46 Malawi 10.74 na 9.90
Jap an 4.35 7.37 1.07 M alaysia 4.10 6.81 1.85
N etherlands 1.47 1.92 1.30 Mali 10.70 na na
New Z ealand 4.21 3.77 5.11 M auritania 8.95 na 8.95
Portugal 6.40 9.74 4.91 M exico 29.41 10.34 40.52
Spain 2.94 3.62 1.70 M orocco 4.76 7.33 4.04
Sw eden 3.01 3.54 3.38 M yanm ar 10.98 13.45 12.60
UK 4.99 8.09 2.46 N icaragua 2888.81 23.30 4985.95
USA 2.44 3.70 1.31 Niger 11.55 10.73 14.72
Algeria 5.71 4.61 8.07 N igeria 18.44 12.53 29.40
A rgentina 748.32 159.56 1233.82 P akistan 6.04 9.84 2.61
Benin 9.02 na na P apua New G uinea 5.10 9.37 2.35
B otsw ana 2.85 4.75 2.98 Peru 1689.25 11.35 2992.72
Burkina F aso 12.62 17.62 10.36 Rw anda 9.12 13.57 8.00
Burundi 10.16 16.33 7.06 S audi A rabia 7.71 12.07 2.34
CAR 7.44 na 5.90 S ierra Leone 45.46 9.44 70.27
Chad 11.40 na 11.40 S ingapore 7.07 12.39 1.64
Chile 91.92 168.95 7.89 Som alia 29.18 17.81 95.62
China 6.11 2.96 9.68 Sri Lanka 7.35 9.58 8.48
Colom bia 6.47 9.76 5.28 Sudan 20.56 14.01 32.52
Congo 6.75 9.05 6.99 S yrian A rab Rep. 12.57 10.34 19.14
C osta R ica 21.77 13.77 9.73 T anzania 7.90 12.54 5.55
Dom inican Rep. 16.14 7.99 25.22 Togo 7.54 13.14 2.17
E cuador 14.25 6.87 19.26 T unisia 1.87 na 1.87
Egypt 5.53 7.02 6.37 Turkey 25.95 37.70 25.21
G abon 7.89 8.71 7.74 V enezuela 22.65 5.36 28.83
Gambia 15.76 8.77 19.67 Z aire 4414.42 38.27 7965.83
G hana 45.55 48.43 21.26 Zam bia 30.82 151.24 55.57
G uatem ala 12.33 6.24 19.78
234
Table 58: Monetary Uncertainty Across Different Time Periods

U ncertainty M easure b ased on: U ncertainty M easure based on:
C ountry________ 1970-84 1970-S1 1983-84________ C ountry____________ 1970-94 1970-81 1983-94
A ustralia 7.70 7.49 6.20 G uinea-B issau 31.85 NA 25.79

A ustria 6.21 8.25 3.47 Haiti 18.84 11.64 24.64
C anada 7.89 7.46 6.16 India 5.56 8.51 5.22
Finland 15.73 9.12 20.73 Jam aica 12.81 8.64 16.39
F rance 4.35 5.46 3.14 K enya 9.86 9.37 10.80
G erm any 6.01 6.05 6.31 K orea 9.07 9.77 6.23
G reece 3.73 3.42 4.21 L esotho 9.56 NA 8.94
Italy 5.71 8.33 2.31 Malawi 11.93 12.13 12.18
Japan 5.39 7.25 3.57 M alaysia 7.60 7.09 8.52
N etherlands 5.26 6.22 3.36 Mali 15.01 18.74 1Z25
Portugal 8.02 9.01 6.96 Mexico 12.89 3.14 18.50
Spain 5.51 5,13 6.11 M orocco 7.32 6.06 7.50
Sw eden NA NA NA M yanmar 27.63 11.15 41.73
UK 12.18 5.77 15.50 N icaragua NA 5.78 127.32
USA 4.65 1.56 5.57 N iger 10.81 6.43 11.56
Algeria 12.11 10.61 12.54 Nigeria 14.39 13.94 14.12
A rgentina 40.87 30.05 28.58 Pakistan 14.35 18.97 9.15
Benin 15.28 16.34 14.72 P apua New G uinea 10.95 15.05 8.51
Botsw ana 15.55 15.38 12.56 Peru 26.16 9.80 36.04
Burkina F aso 8.85 11.25 6.05 Rwanda 17.24 22.47 10.57
Burundi 12.92 14.46 9.23 S audi Arabia 25.80 34.72 9.89
CAR 11.33 9.02 8.87 S ierra Leone 24.92 11.83 30.35
Chad 17.18 13.43 19.53 S ingapore 7.81 9.44 6.45
Chile 28.11 37.37 13.88 Som alia 28.70 18.64 27.22
China 14.02 10.94 12.81 Sri Lanka 9.17 11.51 6.87
Colom bia 5.40 4.65 6.39 Sudan 97.56 6.31 130.69
C ongo 11.20 7.62 14.67 Syrian Arab Rep. 29.15 8.98 46.49
C osta Rica 10.93 7.55 13.17 T anzania 10.59 8.60 10.03
C ote d lv o ire 6.19 5.67 6.61 Thailand 6.54 6.79 6.54
Dominican Rep. 29.40 8.57 43.08 Togo 21.49 19.39 22.08
E cuador 10.19 3.89 12.71 T unisia 6.08 4.98 6.26
Egypt 29.35 27.98 28.06 Turkey 10.16 7.43 12.02
G abon 16.01 20.13 11.73 V enezuela 23.66 10.60 21.42
Gambia 25.27 30.70 20.13 Z aire 89.72 14.46 139.02
G hana 41.81 23.25 53.20 Zam bia 13.91 14.58 11.43
G uatem ala 11.36 7.53 14.87
235
Table 59: Nominal Exchange Rate Uncertainty Across Different Time Periods
U ncertainty M easure b ased on: U ncertainty M easure based on:
C ountry________1970-84 1970-81 1983-94_________C ountry____________ 1970-94 1970-81 1883-94

A ustria 10.51 9.16 12.95 Haifa' 14.53 0.00 14.17
C anada 4.23 3.87 4.80 India 7.91 5.02 6.75
Finland 10.33 6.04 12.96 Jam aica 30.81 16.22 24.52
F rance 12.23 10.85 12.91 K enya 17.52 8.15 14.13
G erm any 10.78 9.79 12.97 K orea 8.20 7.74 6.68
G reece 11.08 8.54 10.22 L esotho 13.86 8.57 12.86
Italy 13.13 10.41 14.16 Malawi 20.85 5.98 16.98
Jap an 9.79 9.81 12.12 M alaysia 4.80 6.04 3.53
New Zealand 9.97 9.50 9.88 M auritania 10.25 5.98 9.78
Portugal 14.46 10.57 14.13 Mexico 46.91 11.45 24.08
Spain 13.36 11.23 13.83 M orocco 11.22 10.68 9.56
Sw eden 11.98 7.31 12.73 M yanmar 8.99 9.45 603
UK 10.21 9.56 10.11 N icaragua 53.50 9.27 50.49
USA NA NA NA Niger 22.95 11.25 20.28
Algeria 24.48 8.70 20.19 Nigeria 35.42 7.03 22.48
B angladesh 11.57 12.34 3.39 Panam a NA NA NA
Benin 22.95 11.25 20.28 P apua New Guinea 7.16 7.48 6.17
B otsw ana 12.42 7.77 12.02 Peru 47.22 22.28 39.86
Burkina Faso 22.95 11.25 20.28 Rw anda 15.71 5.26 15.98
Burundi 8.16 4.57 7.78 S audi Arabia 5.79 4.04 1.01
C am eroon 22.95 11.25 20.28 Senegal 22.95 11.25 20.28
CAR 22.95 11.25 20.28 S ierra Leone 58.76 8.68 27.54
Chad 22.95 11.25 20.28 S ingapore 3.73 4.83 3.04
Chile 33.04 37.06 12.34 Som alia 50.60 7.92 34.07
China 14.11 8.08 11.25 S ri Lanka 14.93 13.44 3.52
Colom bia 10.67 3.45 9.10 Sudan 262.47 10.35 47.57
C ongo 22.95 11.25 20.28 S yrian A rab Rep. 37.95 1.63 19.30
Dominican Rep. 45.00 0.00 24.49 Togo 22.95 11.25 20.28
E cuador 26.67 6.18 13.83 T unisia 9.12 8.73 7.42
Egypt 29.39 15.45 19.95 Turkey 41.37 42.40 26.51
El Salvador 20.31 0.00 15.98 U ruguay 33.11 19.80 14.02
G abon 22.95 11.25 20.28 V enezuela 34.86 1.02 26.65
G am bia 18.80 11.56 14.12 Z aire 168.44 98.18 135.08
G hana 35.33 22.37 27.27 Zam bia 57.78 9.69 37.09
G uatem ala 21.48 0.00 17.22
236
Table 60: Real Exchange Rate Uncertainty Across Different Time Periods
U ncertainty M easure b ased o n : U ncertainty M easure b ased on:
C ountry_________ 1970-94 1970-81 1883-94_______ C ountry____________ 1970-94 1970-91 1983-94

A ustria 11.45 10.12 13.38 Haiti 9.76 7.21 12.44
C anada 5.00 4.93 5.61 India 6.92 7.37 6.14
Finland 10.47 7.76 12.91 Jam aica 13.93 11.97 16.82
France 11.92 10.81 12.57 K enya 11.52 9.00 12.60
G erm any 11.82 10.94 13.55 K orea 7.33 8.44 6.46
G reece 9.46 8.79 10.47 L esotho 12.52 11.07 13.89
Italy 11.62 8.94 13.93 Malawi 9.13 4.41 11.80
Jap an 11.63 10.78 12.28 M alaysia 9.10 11.41 6.55
P ortugal 11.43 7.77 14.11 M exico 12.75 9.14 12.60
Spain 12.26 11.31 13.86 M orocco 10.11 10.80 9.81
Sw eden 10.89 7.49 12.95 M yanm ar 19.00 17.80 15.26
UK 10.06 9.86 10.01 N icaragua 96.53 7.95 133.80
USA NA NA NA Niger 17.41 19.53 14.94
Algeria 11.66 10.06 10.03 N igeria 18.45 12.85 21.59
B angladesh 29.23 25.18 4.32 Panam a NA NA NA
Benin 13.79 11.70 15.78 P apua New G uinea 10.10 12.64 6.85
B otsw ana 12.87 12.47 11.73 P eru 17.01 11.39 21.80
Burkina F aso 13.58 10.41 16.55 Rw anda 17.61 22.37 11.26
B urundi 8.70 10.31 5.93 Saudi A rabia 28.94 33.06 7.77
CAR 12.43 8.52 14.65 S ierra L eone 19.54 10.14 26.41
Chad 12.50 11.94 12.71 S ingapore 6.49 8.00 5.18
Chile 13.54 16.30 10.50 Som alia 28.47 24.90 24.27
China 9.40 7.48 10.99 S ri Lanka 10.03 14.42 4.16
Colom bia 7.69 5.89 8.55 S udan 24.33 6.68 34.94
C ongo 13.55 9.43 14.23 S yrian A rab Rep. 15.05 8.32 20.43
Dom inican R ep. 14.80 4.80 20.83 Togo 14.39 11.44 17.56
E cuador 12.48 11.32 12.73 T unisia 7.90 9.02 6.76
E gypt 14.68 13.46 18.63 Turkey 15.26 15.22 15.24
G abon 15.67 18.75 10.18 V enezuela 13.34 10.72 13.64
Gam bia 14.75 12.31 17.02 Z aire 17.05 103.50 289.69
G hana 30.34 28.00 21.35 Zam bia 18.33 12.89 22.47
G uatem ala 9.74 5.66 13.01
237
Table 61: Terms of Trade Uncertainty Across Different Time Periods

U ncertainty M easure based o n : U ncertainty M easure based on:
C ountry________ 1970-84 1970-81 1983-94_______ C ountry__________ 1970-94 1970-81 1983-94

A ustria 2.77 2.13 3.08 Haiti 13.97 10.23 17.60
C anada 3.48 4.80 1.92 India 7.80 9.33 5.36
Finland 3.80 3.43 4.36 Jam aica 17.23 23.37 10.54
F rance 4.94 6.26 3.17 K enya 8.70 9.90 7.70
G erm any 4.84 5.07 4.32 Korea 6.46 8.61 3.27
G reece 5.50 6.60 4.26 L esotho NA NA NA
Italy 6.43 7.16 4.45 Malawi 10.34 11.87 10.11
Jap an 11.70 13.67 7.95 M alaysia 13.38 16.69 9.71
Portugal 5.59 6.71 4.36 Mexico 15.90 1229 17.81
Spain 6.88 7.46 4.50 M orocco 14.14 20.54 4.73
Sw eden 3.63 4.35 2.77 M yanmar 10.21 10.32 7.60
UK 4.87 6.98 2.02 N icaragua 11.28 14.70 6.95
USA 4.77 5.65 2.78 Niger 9.04 10.54 5.76
A lgeria 25.92 20.25 27.11 Nigeria 25.58 19.64 27.25
A rgentina 8.03 8.16 7.79 Pakistan 8.30 11.61 3.74
Benin 16.40 12.29 18.08 Papua New G uinea 17.49 24.30 10.79
B otsw ana 11.98 11.47 11.14 Peru 9.02 9.12 9.51
Burkina F aso 5.91 6.03 5.69 Rwanda 19.18 25.61 12.32
Burundi 25.32 29.32 30.63 Saudi A rabia 33.61 23.17 37.32
C am eroon 13.17 15.98 10.86 Senegal 13.12 17.99 5.49
CAR 8.23 11.19 4.51 Sierra Leone 6.26 7.36 5.33
Chad 7.78 10.14 4.99 Singapore 2.65 222 1.87
Chile 12.29 13.98 7.62 Som alia 7.00 8.77 4.80
China 5.37 5.76 5.22 Sri Lanka 12.31 17.14 6.61
Colom bia 17.07 16.64 18.19 Sudan 9.42 9.70 8.80
C ongo 22.28 20.71 21.64 Syrian A rab Rep. 18.50 18.18 16.06
Dom inican Rep. 27.37 35.17 14.93 Togo 27.70 40.58 8.45
E cuador 18.22 18.91 14.71 Tunisia 13.91 15.83 9.90
Egypt 14.62 9.75 18.17 Turkey 9.18 10.42 5.91
El S alvador 9.21 10.42 8.74 Uruguay 17.34 2248 8.58
G abon 24.06 21.28 23.35 V enezuela 25.78 19.08 27.98
Gam bia 16.35 14.85 18.29 Zaire 13.37 18.09 5.65
G hana 17.32 23.16 10.11 Zam bia 21.68 27.11 10.69
G uatem ala 11.86 15.39 8.00
APPENDIX 6: RESULTS OF STATIONARY TESTS
1. Augmented Dickey-Fuller (ADF) Test:
Variable ADF test statistic 1% critical value result

INV -9.2877 -3.4367 stationary
INVFIPR -5.6340 -3.4413 stationary
INVFIPU -7.4902 -3.4424 stationary
HESLAG -26.2258 -3.4368 stationary
INF -19.5785 -3.4367 stationary
INT -31.8765 -3.4388 stationary
F (resource flow) -5.5261 -3.4371 stationary
DERES -9.5874 -3.4386 stationary
2. Phillips-Perron (PP) Test:
Variable PP test statistic 1% critical value result

INV -10.9724 -3.4366 stationary
INVFIPR -5.1557 -3.4407 stationary
INVFIPU -7.1842 -3.4417 stationary
HESLAG -41.5161 -3.4366 stationary
INF -30.0457 -3.4365 stationary
INT -132.3614 -3.4388 stationary
F -11.3901 -3.4371 stationary
DERES -19.0489 -3.4377 stationary
Note that all the uncertainty measures can be excluded from these tests as we do
not have a time series for the uncertainty measures.
238
BIBLIOGRPHY
Abel, Andrew B.,"Optimal Investment under Uncertainty,” American Economic

Review, Vol. 73, Number I (March 1983), pp. 228-233.
Abel, Andrew B.,"The Effects of Uncertainty on Investment and the Expected Long-
Run Capital Stock,” Journal o f Economic Dynamics and Control, Vol. 7, Number I
(February 1984), pp. 39-54.
Abel, Andrew B.,"A Stochastic Model of Investment, Marginal Q and the Market
Value o f the Firm,” International Economic Review, Vol. 26, Number 2 (June 1985),
pp. 302-322.
Abel, Andrew B. and Ben S. Bemanke, Macroeconomics, 2nd edition, (Reading, MA:
Addison-Wesley Publishing, 1995).
Adam-Muller, Axel F.A ., "Export and Hedging Decisions under Revenue and
Exchange Rate Risk: A Note,” European Economic Review, Vol. 41, Number 7 (July
1997), pp. 1421-1426.
Agenor, Pierre-Richard and Peter J. Montiel, Development Macroeconomics,

(Princeton, NJ: Princeton University Press, 1996).
Aizenman, Joshua, "Exchange Rate Flexibility, Volatility, and Domestic and Foreign
Direct Investment,” IMF Staff Papers, Vol. 39, Number 4 (December 1992), pp. 890-
922.
Aizenman, Joshua and Nancy P. Marion, "Macroeconomic Uncertainty and Private

Investment,” Economics Letters, Vol. 41, Number 2 (1993), pp. 145-163.
Aizenman, Joshua and Nancy P. Marion, "Policy Uncertainty, Persistence and

Growth,” Review o f International Economics, Vol. I, Number 2 (1993), pp. 145-163.
Akhtar, M. A. and R. Spence Hilton, "Effects of Exchange Rate Uncertainty on

German and U.S. Trade,” Quarterly Review (of the Federal Reserve Bank of New
York), Vol. 9, Number 1 (Spring 1984), pp. 7-16.
239
240
Alesina, Alberto and Roberto Perotti, "Income Distribution, Political Instability, and
Investment,” NBER Working Paper #4486, October 1993.
Anderson, Torben G. and Tim Bollerslev, "Answering the Critics: Yes, ARCH Models
Do Provide Good Volatility Forecasts,” National Bureau of Economic Research,
Working Paper Series, No. 6035 (April 1997).
Apergis, Nicholas, "Inflation Uncertainty, Money Demand, and Monetary

Deregulation: Evidence from a Univariate ARCH Model and Cointegration Tests,”
Journal o f Policy Modeling, Vol. 19, Number 3 (June 1997), pp. 279-293.
Appleyard, Dennis R. and Alfred J. Field, International Economics, 2nd ed. (Chicago
e ta l.: Irwin, 1995).
Arize, Augustine C., “The Supply and Demand for Imports and Exports in a Simultaneous
Model,” Applied Economics, Vol. 19, Number 9 (September 1987), pp. 1233-1247.
Arize, Augustine C., “An Ecnometric Investigation o f Export Behavior in Seven Asian
Developing Countries,” Applied Economics, Vol. 22, Number 7 (July 1990), pp. 891-904.
Arize, Augustine C ., "The Effects of Exchange-Rate Volatility on U.S. Exports: An

Empirical Investigation,” Southern Economic Journal, Vol. 62, Number 1 (July 1995),
pp. 34-43.
Arize, Augustine C ., "Real Exchange-Rate Volatility and Trade Flows: The Experience
o f Eight European Economies,” International Review of Economics & Finance, Vol. 5,
Number 2 (1996), pp. 187-205.
Arize, Augustine C ., "Conditional Exchange-Rate Volatility and the Volume of

Foreign Trade: Evidence from Seven Industrialized Countries,” Southern Economic
Journal, Vol. 64, Number 1 (July 1997), pp. 235-254.
Arrow, Kenneth J. and Antony C. Fisher, "Environmental Preservation, Uncertainty,

and Irreversibility,” Quarterly Journal of Economics, Vol. 88, Number 2 (May 1974),
pp. 312-319.
Artus, J.R. and S.C. Sosa, “Relative Price Effects on Export Performance: The Case
of Nonelectrical Machinery,” IMF Staff Papers, Vol. 25, Number 1 (March 1978), pp.
25-47.
Auty, Richard M., Patterns of Development: Resources, Policy, and Economic Growth,
(London: Edward Arnold, 1995).
241
Avenell, Simon, Robert Leeson and Gavin Wood, "Measuring Exchange Rate
Uncertainty: The Case of the Australian Dollar 1969-1987,” The Australian Economic
Review, No 86 (Winter 1989), pp. 34-42.
Bahmani-Oskooee, Mohsen, "Effects of Exchange Rate Variability on Inflation

Variability,” World Development, Vol. 19, Number 6 (June 1991), pp. 729-733.
Bahmani-Oskooee, Mohsen, "Exchange Rate Uncertainty and Trade Flows of

Developing Countries,” Journal o f Developing Areas, Vol. 25, Number 4 (July 1991),
pp. 497-508.
Balassa, Bela, Voloudahis, E.P. Fylathos, and Sub Suhtai, Export Incentives and Export
Growth in Developing Countries: An Econometric Investigation, DRD Discussion Paper,
DRD159, (Washington, DC: World Bank, 1989).
Bardhan, Pranab, "The Contributions of Endogenous Growth Theory to the Analysis of

Development Problems: An Assessment, in Jere R. Behrman, and T.N. Srinivasan
(eds.) Handbook o f Development Economics, Vol. Ill, Chapter 46, (Amsterdam,
Elsevier Science B.V., 1995), pp. 2983-2998.
Barro, Robert J., "Rational Expectations and the Role o f Monetary Policy" Journal of
Monetary Economics, Vol. 2, Number 1 (January 1976), pp. 1-32.
Barro, Robert J., "A Capital Market in an Equilibrium Business Cycle Model"
Econometrica, Vol. 46, Number 5 (September 1980), pp. 1393-1417.
Barro, Robert J., Determinants o f Economic Growth: A Cross-Country Empirical

Study, Cambridge, MA: MIT Press, 1997).
Barro, Robert J. and Jong Wha Lee, "International Measures of Schooling Years and
Schooling Quality,” American Economic Review, Vol. 86, Number 2 (May 1996), pp.
218-223.
Basevi, Giorgio, “Commodity Trade Equations in Project LINK,” in Robert Ball (ed.),
The International Linkage of National Economic Models, (Amsterdam: North-Holland,
1973), pp. 227-281.
Behrman, Jere R., "Sectoral Investment Determination in a Developing Economy,”

American Economic Review, Vol. 62, Number 5 (December 1972), pp. 825-841.
Bencivenga, Valerie R. and Bruce D. Smith, "Financial Intermediation and

Endogenous Growth,” Review o f Economic Studies, Vol. 58(2), Number 194 (April
1991), pp. 195-209.
242
Bemanke, Ben S., "Irreversibility, Uncertainty, and Cyclical Investment,” Quarterly

Journal of Economics, Vol. 98, Number 1 (February 1983), pp. 85-106.
Bertola, Guiseppe and Ricardo J. Caballero, "Irreversibility and Aggregate

Investment,” Review o f Economic Studies, Vol. 61, Number 2 (April 1994), pp. 223-
246.
Bhagwati, Jagdish, "Immiserizing Growth: A Geometrical Note,” Review o f Economic

Studies, Vol. 25(3), Number 68 (June 1958), pp. 201-205.
Bini-Smaghi, Lorenzo, "Exchange Rate Variability and Trade: Why is it so Difficult to

Find any empirical Relationship?,” Applied Economics, Vol. 23, Number 5 (May
1991), pp. 927-936.
Bleany, Michael and David Greenway, "Adjustment to External Imbalance and

Investment Slumps in Developing Countries,” European Economic Review, Vol. 37,
Numbers 2-3, (April 1993), pp. 577-585.
Blecker, Robert A., "Policy Implications of the International Saving-Investment

Correlation,” in Robert Pollin (ed.) The Macroeconomics of Saving, Finance, and
Investment, (Ann Arbor: University of Michigan Press, 1997), pp. 173-229.
Blecker, Robert A., "The New Economic Integration: Structuralist Models of North-
South Trade and Investment Liberalization,” Structural Change and Economic
Dynamics, Vol. 7, Number 3 (September 1996), pp. 321-345.
Blejer, Mario I. and Leonardo Leiderman, "On the Real Effects of Inflation and
Relative-Price Variability: Some Empirical Evidence,” Review of Economics and
Statistics, Vol. 62, Number 4 (November 1980), pp. 539-544.
Blejer, Mario I. and Mohsin S. Khan, "Government Policy and Private Investment in
Developing Countries,” IMF Staff Papers, Vol. 31, Number 2 (June 1984), pp. 379-
403.
Boeke, Julius H., Economics and Economic Policy of Dual Societies as Exemplified by
Indonesia, (New York: Institute of Pacific Relations, 1953).
Bollerslev, Tim, Ray Chu, and Kenneth Kroner, "ARCH Modelling in Finance: A
Review o f the Theory and Empirical Evidence,” Journal of Econometrics, Vol. 52,
Numbers 1/2 (April/May 1992), pp. 5-59.
243
Bomberger, William A., "Disagreement as a Measure of Uncertainty,” Journal of

Money, Credit, and Banking, Vol. 28, Number 3, Part I (August 1996), pp. 381-392.
Borensztein, Eduardo, "Debt Overhang, Debt Reduction and Investment: The Case of
the Philippines,” IMF Working Paper, No. 90/77, (Washington, D.C.: IMF,
September 1990).
Borensztein, Eduardo, "Debt Overhang, Credit Rationing, and Investment,” Journal of

Development Economics, Vol. 32, Number 2 (April 1990), pp. 315-335.
Bomer, Silvio, Aymo Brunetti, and Beatrice Weder, Political Credibility and Economic
Development, (New York: St. Martin's Press, 1995).
Bourguignon, Francois, Jaime de Melo, and Akiko Suwa, "Distributional Effects of

Adjustment Policies: Simulations for Archetype Economies in Africa and Latin
America,” World Bank Economic Review, Vol. 5, Number 2 (May 1991), pp. 339-366.
Bourguignon, Francois, William H. Branson, and Jaime de Melo, "Adjustment and

Income Distribution: A Counterfactual Analysis, ” World Bank, Policy, Planning, and
Research Working Papers, WPS 215 (May 1989); also published as "Adjustment and
Income Distribution,” Journal o f Development Economics, Vol. 38, Number 1
(January 1992), pp. 17-39.
Branson, William H ., "Economic Structure and Policy for External Balance,”

International Monetary Fund Staff Papers, Vol. 30, Number 1 (March 1983), pp. 39-
66 .
Branson, William H., "The Limits of Monetary Coordination as Exchange Rate

Policy,” Brookings Papers on Economic Activity, Vol. 1986, Number I, pp. 175-194.
Branson, William H ., "Stabilization, Stagflation, and Investment Incentives: The Case

of Kenya, 1979-1980,” in Sebastian Edwards and Liaquat Ahamed (eds.), Economic
Adjustment and Exchange Rates in Developing Countries, (Chicago: University of
Chicago Press, 1986), pp. 267-288.
Broil, Udo and Jack Wahl, "Exports Under Exchange Rate Uncertainty and Hedging
Markets,” Journal o f International Theoretical Economics (Zeitschriftfur die gesamte
Staatswissenschaft), Vol. 148, Number 4 (December 1992), pp. 577-587.
Broil, Udo and Itzak Zilcha, "International Production, Investment and Borrowing with
Exchange Rate Risk and Futures Markets,” Metroeconomica, Vol. 46, Number 1
(February 1995), pp. 90-106.
244
Brown, Drusiila K., "Comment on Measures of Openness,” in Robert E. Baldwin

(ed.), Trade Policy Issues and Empirical Analysis, (Chicago and London: University of
Chicago Press, 1988), pp. 200-204.
Caballero, Ricardo J., "Fixed Costs: The Demise of Marginal Q ,” National Bureau of
Economic Research, Working Paper Series, No. 5508 (March 1996); also published as
Harvard Institute of Economic Research, Discussion Paper Series, No. 1765 (June
1996).
Caballero Ricardo J. and Vittorio Corbo, "The Effect of Real Exchange Rate
Uncertainty on Exports: Empirical Evidence,” World Bank Economic Review, Vol. 3,
Number 2 (May 1989), pp. 263-278.
Calvo, Guillermo A., Morris Goldstein, and Eduard Hochreiter (eds.), Private Capital
Flows to Emerging Markets after the Mexican Crisis, (Washington, DC: Institute for
International Economics, 1996).
Calvo, Sara and Bernhard G. Gunter (eds.), Debt Management and Capital Flow
Volatility, (Washington, DC: World Bank, forthcoming 1998).
Campos, Jose E. and Hilton Root, The Key to the Asian Miracle: Making Shared
Growth Credible, (Washington, DC: Brookings Institution, 1996).
Caporale, Tony and Khosrow Doroodian, "Exchange Rate Regimes and Uncertainty,”
Weltwirtschaftliches Archiv, Vol. 131, Number 3 (September 1995), pp. 569-575.
Cardoso, Eliana, "Private Investment in Latin America,” Economic Development and

Cultural Change, Vol. 41, Number 4 (July 1993), pp. 833-848.
Chang, P.H. Kevin, "Export Diversification and International Debt under Terms-of-
trade Uncertainty: An Intertemporal Approach,” Journal o f Development Economics,
Vol. 36, Number 2 (October 1991), pp. 259-279.
Chenery, Hollis B., "Patterns of Industrial Growth,” American Economic Review, Vol.
50, Number 3 (September 1960), pp. 624-654.
Chenery, Hollis B., Sherman Robinson, and Moises Syrquin, Industrialization and
Growth: A Comparative Study, (New York: Oxford University Press, 1986).
Chenery, Hollis B. and Moises Syrquin, Patterns of Development: 1950-1970, (New

York: Oxford University Press for the World Bank, 1975).
245
Chenery, Hollis B. and Lance Taylor, "Development Patterns: Among Countries and
Over Time,” Review o f Economics and Statistics, Vol. 50, Number 4 (November
1968), pp. 391-416.
Chhibber, Ajay, Mansoor Dailami and NematShafik (eds.), Reviving Private

Investment in Developing Countries: Empirical Studies and Policy Lessons,
(Amsterdam:, North Holland, 1992).
Clark, Colin, The Conditions o f Economic Progress, (London: Macmillan, 1940 (2nd
ed. 1951, 3rd ed. 1957).
Cohen, Daniel, "How to Evaluate the Solvency of an Indebted Nation,” Economic

Policy: A European Forum, Vol.l, Number 1 (November 1985), pp. 140-167.
Cohen, Daniel, "Low Investment and Large LDC Debt in the 1980s,” American
Economic Review, Vol. 83, Number 2 (June 1993).
Conway, Patrick, "Implications of Relative Price Uncertainty for Private Investment

Expenditure in Turkey,” mimeo, University of North Carolina at Chapel Hill,
November 1991.
Cooper, Richard M., "Currency Devaluation in Developing Countries,” Princeton

Essays in International Finance, No. 80 (June 1971).
Coppock, Joseph D., International Economic Instability, (New York: McGraw-Hill,

1962).
Cornwall, John and Wendy Cornwall, "Growth Theory and Economic Structure,”
Economica, Vol. 61, Number 242 (May 1994), pp. 237-251.
Cottani, Joaquin A., Domingo F. Cavallo and M. Shahbaz Khan, "Real Exchange Rate
Behavior and Economic Performance in LDCs,” Economic Development and Cultural
Change, Vol. 39, Number 1 (October 1990), pp. 61-76.
Cukierman, Alex, "The Effects of Uncertainty on Investment under Risk Neutrality

with Endogenous Information,” Journal o f Political Economy, Vol. 88, Number 3
(June 1980), pp. 462-475.
Dailami, Mansoor, "Expectations, Stock Market Volatility, and Private Investment

Behavior: Theory and Empirical Evidence for Brazil,” (mimeo, Washington, DC:
World Bank, Country Economics Department, 1987).
246
Dailami, Mansoor and Michael Walton, "Private Investment, Government Policy and
Foreign Capital in Zimbabwe,” World Bank Policy Research and External Affairs
Working Papers, No. 248, (Washington, D.C.: World Bank, Country Economics
Department, 1989).
Davidson, Paul, Money and Real World, (London: Macmillan, 1972).
Davidson, Paul, "Is Probability Theory Relevant for Uncertainty,” Journal of

Economic Perspectives, Vol. 5, Number 1 (Winter 1991), pp. 129-143.
Dawson, John W. and E. Frank Stephenson, "The Link between Volatility and
Growth: Evidence from the States,” Economic Letters, Vol. 55, Number 3 (September
1997), pp. 365-369.
De Grauwe, Paul, "Exchange Rate Variability and the Slowdown in the Growth of
International Trade,” International Monetary Fund Staff Papers, Vol. 35, Number 1
(March 1988), pp. 63-84.
De Gregorio, Jose and Pablo E. Guidotti, "Financial Development and Economic

Growth,” World Development, Vol. 23, Number 3 (March 1995), pp. 433-448.
De Long, J. Bradford and Lawrence Summers, "Equipment Investment and Economic

Growth,” Quarterly Journal o f Economics, Vol. 106, Number 2 (May 1991), pp. 445-
502.
De Long, J. Bradford and Lawrence Summers, "How Strongly Do Developing

Economies Benefit from Equipment Investment?,” Journal o f Monetary Economics,
Vol. 32, Number 3 (December 1993), pp. 395-416.
Deininger, Klaus and Lyn Squire, "A New Data Set Measuring Income Inequality,”
World Bank Economic Review, Vol. 10, Number 3 (September 1996), pp. 565-591.
Deininger, Klaus and Lyn Squire, "Economic Growth and Income Inequality:
Reexamining the Links,” Finance & Development, Vol. 34 Number l(March 1997, pp.
38-41.
Delli Gatti, Domenico and Mauro Gallegati, "At the Root of the Financial Instability
Hypothesis: 'Induced Investment and Business Cycle',” Journal o f Economic Issues,
Vol. 31, Number 2 (June 1997), pp. 527-534.
Demirguc-Kunt, Asli and Ross Levine, "Stock Market Development and Financial
Intermediaries: Stylized Facts,” World Bank Research Observer, Vol. 10, Number 2
(May 1996), pp. 291-321.
247
Denison, Edward, Trends in American Economic Growth, 1929-82, (Washington. DC:

The Brookings Institution, 1985).
Dewatripont, Mathias and Gerard Roland, "The Design of Reform Packages under
Uncertainty,” American Economic Review, Vol. 85, Number 5 (December 1995), pp.
1207-1223.
Dixit, Avinash and Robert Pindyck, Investment under Uncertainty, (Princeton, NJ:
Princeton University Press, 1994).
Dixit, Avinash and Robert Pindyck, "Options Approach to Capital Investment,”

Harvard Business Review, Vol. 73 (May-June 1995), pp. 105-115.
Dixit, Avinash and Rafael Rob, Switching Costs and Sectoral Adjustments in General
Equilibrium with Uninsured Risk,” Journal of Economic Theory, Vol. 62, Number 1
(February 1994), pp. 48-69.
Dollar, David, "Outward-Oriented Developing Economies Really Do Grow More

Rapidly: Evidence from 95 LDCs, 1976-1985,” Economic Development and Cultural
Change, Vol. 40, Number 3 (April 1992) pp. 523-544.
Dornbusch, Rudiger and Sebastian Edwards, "Macroeconomic Populism, Journal of

Development Economics, Vol. 32, Number 2 (April 1990), pp. 247-277; also published
in Jorge I. Dominguez (ed.) Economic Strategies and Policies in Latin America, (New
York: Garland Publishing, 1994), pp. 277-307.
Dunlevy, James A., “Export Demand, Export Supply, and Capacity Pressure: A
Simultaneous Equations Study of American and British Export Performance, ” (Auburn,
Alabama: Auburn University, 1979).
Dutt, Amitava K., “Uneven Development in Alternative Models of North-South

Trade,” Eastern Economic Journal, Vol. 15, Number 1 (March 1989), pp. 91-106.
Dutt, Amitava K., Growth, Distribution, and Uneven Development, (New York:
Cambridge University Press, 1990
Easterly, William, "Life During Growth,” mimeo, World Bank, October 1997.
Eichengreen, Barry, James Tobin and Charles Wyplosz, "Two Cases for Sand in the
Wheels of International Finance,” Economic Journal, Vol. 105 (January 1995), pp.
162-172.
248
El-Samhouri, Mohammed, "Flexible Exchange Rates and Export Instability: The

Impact o f the Post-1973 International Monetary System on the Developing Countries,”
Journal o f Developing Areas, Vol. 28, Number 4 (July 1994), pp. 503-520.
Engel, Ernst, "Die Produktions- und Konsumptionsverhaltnisse des Konigreichs

Sachsen, ” Zeitschrift des Statistischen Bureaus des Koniglich Sachsischen Ministerium
des Inneren, Nos. 8 and 9 (November 22, 1857).
Engle, Robert F., "Autoregressive Conditional Heteroskedasticity with Estimates of the

Variance of United Kingdom Inflation,” Econometrica, Vol. 50, Number 4 (July
1982), pp. 987-1007.
Engle, Robert F., "Estimates of the Variance of U.S. Inflation Based upon the ARCH
Model,” Journal o f Money, Credit, and Banking, Vol. 15, Number 3 (August 1983),
pp. 286-301.
Engle, Robert F. and Tim Bollerslev, "Modelling the Persistence of Conditional

Variances,” Econometric Reviews, Vol. 5, Number 1 (1986), pp. 1-50.
Engle, Robert F., David Lilien, and Russell P. Robbins, "Estimating Time Varying
Risk Premia in the Term Structure: The ARCH-M Model,” Econometrica, Vol. 55,
Number 1 (March 1987), 391-407.
Ettlin, Franz, “Arbetskostnader, Priseroch Exportmarknadsandelar,” SEB’s

Kvartalsskrift, Number 1-2, 1977.
Faini, Ricardo and Jaime de Melo, "Adjustment, Investment, and the Real Exchange
Rate in Developing Countries," Economic Policy, Vol. 5, Number 2 (October 1990,
011), pp. 492-512.
Fazzari, Steven M., R. Glenn Hubbard and Bruce C. Petersen, "Financing Constraints
and Corporate Investment,” Brookings Papers o f Economic Activity, (1988, Part I), pp.
141-206.
Fazzari, Steven M. and Michael J. Athey, "Asymmetric Information, Financing

Constraints and Investment,” Review of Economics and Statistics, Vol. 69, Number 3
(August 1987), pp. 481-487.
Federer, J. Peter, "The Impact of Uncertainty on Aggregate Investment Spending: An

Empirical Analysis,” Journal of Money, Credit, and Banking, Vol. 25, Number 1
(February 1993), pp. 30-48.
249
Federer, J. Peter, "Does Uncertainty Affect Investment Spending?,” Journal of Post

Keynesian Economics, Vol. 16, Number 1 (Fall 1993), pp. 19-35.
Feldstein, Martin, "Inflation, Tax Rules and Investment: Some Econometric

Evidence,” Econometrica, Vol. 50, Number 4 (July 1982), pp. 825-862.
Felix, David, "Suggestions for International Collaboration to Reduce Destabilizing

Effects of International Capital Mobility on the Developing Countries" in United
Nations Conference on Trade and Development (UNCTAD), International Monetary
and Financial Issues for the 1990s, Research Papers for the Group o f Twenty-Four,
Volume III, (New York: United Nations, 1993), pp. 47-70.
Fernandez, Raquel, "Terms-of-trade Uncertainty, Incomplete Markets and

Unemployment,” International Economics, Vol. 33, Number 4 (November 1992), pp.
881-894.
Fischer, Stanley, "Relative Shocks, Relative Price Variability, and Inflation,”

Brookings Papers on Economic Activity, (1981, Part 2), pp. 381-431.
Fischer, Stanley, "The Role of Macroeconomic Factors in Growth,” Journal of

Monetary Economics, Vol. 32, Number 3 (December 1993), pp. 485-512.
Fisher, Antony C. and W. Michael Hanemann, "Option Value and the Extinction of
Species,” \n Advances in Applied Microeconomics, Vol. 4, (Greenwich, CT: JAl Press,
1986), pp. 169-190.
Fisher, Antony C. and W. Michael Hanemann, "Optimal Investment in Product-

Flexible Manufacturing Capacity,” Management Science, Vol. 36, Number 4 (April
1990), pp. 449-466.
Fisher, Bernhard, "Causes and Consequences of Inflationary Policies in Latin

American countries in the 1980s,” Konjunkturpolitik: Zeitschriftfo r Angewandte
Wirtschaftsforschung, Vol. 36, Number 1 (1990), pp. 43-62.
Fishlow, Albert, "Inequality, Poverty and Growth: Where Do We Stand?,” Annual

World Bank Conference on Development Economics 1995, (Washington, DC: World
Bank, 1996), pp. 25-39.
FitzGerald, E.V.K., Karel Jansen, and Rob Vos, "External Constraints on Private
Investment in Developing Countries,” in Jan Willem Gunning, Henk Kox, Wouter
Tims, and Ynto de Wit (eds.), Trade, Aid and Development: Essays in Honour of Hans
Linnemann, (New York: St. Martin's Press, 1994), pp. 185-220.
250
Friberg, Richard and Anders Vredin, "Exchange Rate Uncertainty and the
Microeconomic Benefits of EMU,” Stockholm School o f Economics: Working Paper
Series in Economics and Finance, Working Paper No. 127 (September 1996).
Friedman, Milton, "Nobel Lecture: Inflation and Unemployment,” Journal of Political

Economy, Vol. 85, Number 3 (June 1977), pp. 451-472.
Gafar, John, “The Supply Response for Sugar Cane in Trinidad and Tobago: Some
Preliminary Results,” Applied Economics, Vol. 19, Number 9 (September 1987), pp.
1221-1231.
Garber, Peter and Mark P. Taylor, "Sand in the Wheels of Foreign Exchange Markets:
A Sceptical Note,” Economic Journal, Vol. 105, Number 428 (January 1995), pp.
173-180.
Geraci, Vincent J. and Wilffied Prewo, “An Empirical Demand and Supply Model of
Multilateral Trade,” (Austin: University of Texas, March 1980), also published in
Review o f Economics and Statistics, Vol. 64, Number 3 (August 1982), pp. 432-441.
Gertler, Mark, "Financial Structure and Aggregate Economic Activity: An Overview,”

National Bureau o f Economic Research (NBER) Working Paper #2559, (Cambridge,
MA: NBER, April 1988)
Ghosal, Vivek and Prakash Loungani, "Product Market Competition and the Impact of
Price Uncertainty on Investment: Some Evidence from US Manufacturing Industries,
Journal of Industrial Economics, Vol. 44, Number 2 (June 1996), pp. 217-228.
Ghura, Dhaneshwar and Thomas J. Grennes, "The Real Exchange Rate and
Macroeconomic Performance in Sub-Saharan Africa, ” Journal of Development
Economics, Vol. 42, Number 1 (October 1993), pp. 155-174.
Glezakos, Constantine and Jeffrey B. Nugent, "Relative Price Variability, Inflation

Rate Uncertainty, and Postwar Investment of the United States,” Journal of Post
Keynesian Economics, Vol. 19, Number 2 (Winter 1996/97), pp. 181-194.
Gloston, Lawrence R., Ravi Jaganathan, and David E. Runkle, "Relationship between
the Expected Value and the Volatility of the Nominal Excess Return on Stocks,”
Journal o f Finance, Vol. 48, Number 5 (December 1993), pp. 1779-1801.
Goldberg, Linda S., "Exchange Rates and Investment in United States Industry,”
Review o f Economics and Statistics, Vol. 75, Number 4 (November 1993), pp. 575-
588.
251
Goldberg, Linda S. and Charles D. Kolstad, "Foreign Direct Investment, Exchange

Rate Variability and Demand Uncertainty,” International Economics, Vol. 36, Number
4 (November 1995), pp. 855-873.
Goldsmith, Raymond W., Financial Structure and Development, (New Haven and
London: Yale University Press, 1969).
Goldstein, Morris and Mohsin S. Khan, “Income and Price Effects in Foreign Trade,” in:
Jones, R.W. and P.B. Kenen (eds.) Handbook o f International Economics, Vol. II,
(Netherlands, Elsevier, 1985), pp. 1041-1105.
Gosh, Atish R. and Jonathan D. Ostry, "Macroeconomic Uncertainty, Precautionary

Saving, and the Current Account,” Journal o f Monetary Economics, Vol. 40, Number
I (September 1997), pp. 121-139.
Gottschalk, Peter and Timothy M. Smeeding, "Cross-National Comparisons of

Earnings and Income Inequality,” Journal o f Economic Literature, Vol. 35, Number 2
(June 1997), pp. 633-687.
Greene, Joshua and Delano Villanueva, "Private Investment in Developing Countries,”

IMF Staff Papers, Vol. 38, Number 1 (March 1991), pp. 33-58.
Greenwood, Jeremy and Boyan Jovanovic, "Financial Development, Growth and

Distribution of Income, Journal of Political Economy, Vol. 98, Number 5, Part 1
(October 1990), pp. 1076-1107.
Grier, Kevin B. and Gordon Tullock, "An Empirical Analysis of Cross-National

Economic Growth 1951-80,” Journal of Monetary Economics, Vol. 24, Number 3
(September 1989), pp. 259-276.
Grobar, Lisa Morris, "The Effect of Real Exchange Rate Uncertainty on LDC
Manufactured Exports,” Journal of Development Economics, Vol. 41, Number 2
(August 1993), pp. 367-376.
Gunter, Bernhard G., "Reforming the International Monetary System Towards a World
Central Bank: A Summary of Proposals and Fallacies,” in: Jo Marie Griesgraber and
Bernhard G. Gunter (eds.) The World's Monetary System: Toward Stability and
Sustainability in the Twenty-first Century, (London: Pluto Press, 1996), pp. 115-135.
Gylfason, Thorvaldur, The Effect of Exchange Rate Changes on the Balance o f Trade
in Ten Industrial Countries, (Washington, DC: International Monetary Fund, 1978).
252
Hadjimichael, Michael T. and Dhaneshwar Ghura, "Public Policies and Private Savings
and Investment in Sub-Saharan Africa: An Empirical Investigation,” IMF Working
Paper, W P/95/I9 (1995).
Hadjimichael, Michael T., Dhaneshwar Ghura, Martin Muhleisen, Roger Nord, and E.
Murat Ucer, "Sub-Saharan Africa: Growth, Savings and Investment, 1986-93, IMF
Occasional Paper, No. 118, (Washington, D.C.: IMF, January 1995).
Hadjimichael, Michael T., Michael Nowak, Robert Sharer, AmorTahari, etal.,

"Adjustment for Growth: The African Experience,” IMF Occasional Paper, No. 143
(Washington, D.C.: IMF, October 1996).
Hagiwara, May, "Volatility in the Terms of Trade with Non-identical Preferences,”

Journal o f International Money and Finance, Vol. 13, Number 3 (June 1994), pp. 319-
341.
Halpem, Laszlo and Istvan P. Szekely, “Export Supply and Import Demand in Hungary:
An Econometric Analysis for 1968-1989,” Discussion Paper No. 620 (London: Centre for
Economic Policy Research, 1992).
Hanemann, W. Michael, "Information and the Concept of Option Value,” Journal of

Environmental Economics and Management, Vol. 16, Number 1 (January 1989), pp.
23-37.
Hart, Albert G., "Risk Uncertainty, and the Unprofitability of Compounding

Probabilities,” in Oscar Lange, Francis McIntyre, and Theodore O. Yntema (eds.)
Studies in Mathematical Economics, (Chicago: University of Chicago Press, 1942).
Hartman, Richard, "The Effects of Price and Cost Uncertainty on Investment,” Journal
of Economic Theory, Vol. 5, Number 3 (October 1972), pp. 358-366.
Harvey, Andrew C., Forecasting, Structural Time Series Models and the Kalman
Filter, (Cambridge and New York: Cambridge University Press, 1989).
Hausmann, Ricardo, Michael Gavin et al., "Overcoming Volatility in Latin America"

in Inter-American Development Bank, Report on Economic and Social Progress in
Latin America: 1995, (Washington, D.C.: Inter-American Development Bank, 1995),
pp. 189-256.
Hausmann, Ricardo and Liliana Rojas-Suarez (eds.), Volatile Capital Flows: Taming
their Impact on Latin America, (Washington, DC: Inter-American Development Bank,
1996).
253
Helleiner, Gerald K., "Outward Orientation, Import Instability and African Economic
Growth: An Empirical Investigation,” in Sanjaya Lall and Frances Stewart (eds.)
Theory and Reality in Development: Essays in Honour of Paul Streeten, (New York:
St. Martin's Press, 1986), pp. 139-153.
Henry, Claude, "Investment Decisions under Uncertainty,” American Economic

Review, Vol. 64, Number 5 (December 1974), pp. 1006-1012.
Hirschman, Albert O ., National Power and the Structure of Foreign Trade, (Berkeley
and Los Angeles: University of California Press, 1945).
Houthakker, Hendrik S., "An International Comparison of Household Expenditure

Patterns, Commemorating the Centenary of Engel's Law,” Econometrica, Vol. 25,
Number 4 (October 1957), pp. 532-551.
Huizinga John, "Inflation Uncertainty, Relative Price Uncertainty, and Investment in

U.S. Manufacturing,” Journal of Money, Credit, and Banking, Vol. 25, Number 3,
Part 2 (August 1993), pp. 521-557.
Ingersoll, Jonathan E. and Stephen A. Ross, "Waiting to Invest: Investment and

Uncertainty" Journal o f Business, Vol. 65, Number 1 (January 1992), pp. 1-29.
International Monetary Fund, Compensatory Financing and Export Fluctuations,

(Washington, DC: International Monetary Fund, 1966).
International Monetary Fund, Report on the World Current Account Discrepancy,

(Washington, DC: International Monetary Fund, 1987).
International Monetary Fund, International Financial Statistics, (Washington, DC:

International Monetary Fund, various issues).
Jayarajah, Carl, William H. Branson, and Binayak Sen, Social Dimensions of

Adjustment: World Bank Experience, 1980-93, (Washington, DC: World Bank 1996).
Kaldor, Nicholas, Economics without Equilibrium, (New York: M.E. Sharpe, 1985).
Kang, Jung M. and Jene K. Kwon, “An Estimation o f Import Demand, Export Supply and
Technical Change for Korea,” Applied Economics, Vol. 20, Number 12 (December 1988),
pp. 1661-1674.
Katseli, Louka T., "Devaluation: A Critical Appraisal of the IMF's Policy

Prescriptions,” American Economic Review, Vol. 73, Number 2 (May 1983), pp. 359-
363.
254
Kenen, Peter B. and Dani Rodrik, "Measuring and Analyzing the Effects of Short-term
Volatility in Real Exchange Rates,” Review o f Economics and Statistics, Vol. 63,
Number 2 (May 1986), pp. 311-315.
Keynes, John Maynard, The General Theory o f Employment, Interest, and Money,
(London: MacMillan and Co. 1936).
Keynes, John Maynard, "The General Theory of Employment,” Quarterly Journal of

Economics, Vol. 51, Number 2 (February 1937), pp. 209-223.
King, Robert G. and Ross Levine, "Finance and Growth: Schumpeter Might Be
Right,” Quarterly Journal o f Economics, Vol. 108, Number 2 (August 1993), pp. 717-
738.
Knight, Frank H., Risk, Uncertainty, and Profit, (Boston: Houghton Mifflin, 1921).
Knudsen, Odin and Andrew Parnes, Trade Instability and Economic Development: An
Empirical Study, (Lexington: Lexington Books, 1975).
Kormendi, Roger C. and Philip G. Meguire, "Macroeconomic Determinants of

Growth: Cross-country Evidence,” Journal of Monetary Economics, Vol. 16, Number
2 (September 1985), pp. 141-163.
Koss, Esther Lee, "Variability of Input-Output Coefficients Across Firms and

Countries: An Application of a Hedonic Methodology to Explain Choice Techniques of
Production,” Unpublished Dissertation, Boston University, Graduate School, 1983.
Krugman, Paul, Exchange Rate Instability, (Cambridge, MA: MIT Press, 1989).
Krugman, Paul and Lance Taylor, "Contractionary Effects of Devaluation, ” Journal of

International Economics, Vol. 8, Number 3 (August 1978), pp. 445-456.
Kumar, Ramesh and Ravinder Dhawan, "Exchange Rate Volatility and Pakistan's
Exports to the Developed World, 1974-1985,” World Development, Vol. 19, Number 9
(September 1991), pp. 1225-1240.
Kuznets, Simon, "Economic Growth and Income Inequality,” American Economic

Review, Vol. 45, Number 1 (March 1955), pp. 1-28.
Kuznets, Simon, "Quantitative Aspects of the Economic Growth of Nations: I. Levels

and Variability of Rates o f Growth,” Economic Development and Cultural Change,
Vol. 5, Number 1, (October 1956), pp. 5-94.
255
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth o f Nations: II.

Industrial Distribution of National Product and Labor Force,” Economic Development
and Cultural Change, Vol. 5, Number 4, Supplement (July 1957), pp. 3-111.
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth o f Nations: III.

Industrial Distribution of Income and Labor Force by States, United States 1991-1921
to 1955,” Economic Development and Cultural Change, Vol. 6, Number 4, Part II,
(July 1958), pp. 1-128.
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth o f Nations: IV.

Distribution o f National Income by Factor Shares,” Economic Development and
Cultural Change, Vol. 7, Number 3, Part II (April 1959), pp. 1-100.
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth o f Nations: V. Capital

Formation Properties: International Comparisons for Recent Years,” Economic
Development and Cultural Change, Vol. 8, Number 4, Part II (July 1960), pp. 1-96.
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth o f Nations: VI. Long-
Term Trends in Capital Formation Proportions,” Economic Development and Cultural
Change, Vol. 9, Number 4, Part II (July 1961), pp. 3-124.
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth of Nations: VII. The
Share and Structure of Consumption,” Economic Development and Cultural Change,
Vol. 10, Number 2, Part II (January 1962), pp. 3-92.
Kuznets, Simon, "Quantitative Aspects of the Economic Growth o f Nations: VIII.

Distribution of Income by Size,” Economic Development and Cultural Change, Vol.
11, Number 2, Part II (January 1963), pp. 1-80.
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth o f Nations: IX. Level
and Structure of Foreign Trade: Comparisons for Recent Years,” Economic
Development and Cultural Change, Vol. 13, Number 1, Part II (October 1964), pp. 1-
106.
Kuznets, Simon, Modem Economic Growth: Rate, Structure, and Spread, (New
Haven: Yale University Press, 1966).
Kuznets, Simon, "Quantitative Aspects o f the Economic Growth of Nations: X. Level

and Structure o f Foreign Trade: Long-Term Trends,” Economic Development and
Cultural Change, Vol. 15, Number 2, Part II (January 1967), pp. 1-140.
256
Kuznets, Simon, Economic Growth o f Nations: Total Output and Production Structure,
(Cambridge, MA: Belknap Press of Harvard University Press, 1971).
Lanyi, Anthony and Esther C. Suss, "Exchange Rate Variability: Alternative Measures
and Interpretation,” IMF Staff Papers, Vol. 29, Number 4 (December 1982), pp. 527-
560.
Larrain, Felipe and Rodrigo Vergara, "Investment and Macroeconomic Adjustment:

The Case of East Asia,” in Luis Serven and Andres Solimano (eds.), Striving for
Growth After Adjustment: The Role of Capital Formation, (Washington, D.C.: World
Bank, 1993), pp. 229-274.
Lastrapes, William D. "Exchange Rate Volatility and U.S. Monetary Policy: An

ARCH Application,” Journal of Money, Credit and Banking, Vol. 21, Number 1
(February 1989), pp. 66-77.
Lawrence, Denis, “An Aggregator Model of Canadian Export Supply and Import Demand
Responsiveness, Canadian Journal o f Economics, Vol. 22, Number 3 (August 1989), pp.
503-521.
Leahy, John V. and Toni M. Whited, "The Effect of Uncertainty on Investment: Some
Stylized Facts,” Journal o f Money, Credit, and Banking, Vol. 28, Number 1 (February
1996), pp. 64-83.
Learner, Edward E., "Measures of Openness,” in Robert E. Baldwin (ed.), Trade

Policy Issues and Empirical Analysis, (Chicago and London: University of Chicago
Press, 1988), pp. 147-200.
Levine, Ross, "Financial Development and Economic Growth: Views and Agenda,”
Journal o f Economic Literature, Vol. 35, Number 2 (June 1997), pp. 688-726.
Levine, Ross and David Renelt, A Sensitivity Analysis of Cross-Country Growth

Regressions, American Economic Review, Vol. 82, Number 4 (September 1992), pp.
942-963.
Levine, Ross and Sara Zervos, Stock Market Development and Long-Run Growth,
World Bank Research Observer, Vol. 10, Number 2 (May 1996), pp. 323-339.
Lewellen, Wilbur G. and S.G. Badrinath, "On the Measurement o f Tobin's Q ,”

Journal o f Financial Economics, Vol. 44, Number 1 (April 1997), pp. 77-122.
Lewis, W. Arthur, "Economic Development with Unlimited Supplies of Labour,” The

Manchester School, Vol. 22, Number 2 (May 1954), pp. 139-191.
257
Love, James, "Commodity Concentration and Export Earnings Instability: A Shift from
Cross-section to Time Series Analysis,” Journal o f Development Economics, Vol. 24,
Number 2 (December 1986), pp. 239-248.
Love, James, "Export Instability, Imports and Investment in Developing Countries,”

Journal o f Development Studies, Vol. 25, Number 2 (January 1989), pp. 183-191.
Lucas, Robert E. Jr., "Expectations and the Neutrality of Money, Journal o f Economic
Theory, Vol. 4, Number 2 (April 1972), pp. 103-124.
Lucas, Robert E. Jr., "Some International Evidence on Output-Inflation Tradeoffs,”

American Economic Review, Vol. 63, Number 3 (June 1973), pp. 326-334.
Lundborg, Per, “Elasticities of Supply and Demand for Swedish Exports in a

Simultaneous Model,” Scandinavian Journal o f Economics, Vol. 83, Number 3, (1981),
pp. 444-448.
Lutz, Matthias, "The Effects of Volatility in the Terms of Trade on Output Growth:
New Evidence,” World Development, Vol. 22, Number 12 (December 1994), pp.
1959-1975.
Magee, Stephen P., A Theoretical and Empirical Examination o f Supply and Demand
Relationships in U.S. International Trade, (Washington, DC: Council of Economic
Advisers, 1970).
Mankiw, N. Gregory and David Romer (eds.), New Keynesian Economics,

(Cambridge, MA: MIT Press, 1991).
Manove, Michael, Gustav F. Papanek with Harendra K. Dey, "A Theory of Wage
Determination in Less Developed Countries" Asian Center Discussion Paper No. 30.
Marglin, Stephen A., Growth, Distribution, and Prices, (Cambridge, MA: Harvard
University Press, 1984).
Mavrotas, George, "Foreign Aid-Disaggregation and the Macroeconomic Impact of

Aid on Private Investment,” mimeo, June 1996.
McCarthy, F. Desmond, Lance Taylor and Cyrus Talati, "Trade Patterns in

Developing Countries, 1964-82,” Journal of Development Economics, Vol. 27,
Number 1-2 (October 1987), pp. 5-37.
258
Medhora, Rohinton, "The Effect of Exchange Rate Variability on Trade: The Case of
West African Monetary Union's Imports,” World Development, Vol. 18, Number 2
(February 1990), pp. 313-324.
Mendoza, Enrique G ., "Terms-of-Trade Uncertainty and Economic Growth: Are Risk

Indicators Significant in Growth Regressions?,” Board o f Governor o f the Federal
Reserve System, International Finance Discussion Papers, No. 491 (December 1994).
Mendoza, Enrique G., "The Terms-of-Trade, the Real Exchange Rate, and Economic
Fluctuations,” International Economic Review, Vol. 36, Number I (February 1995),
pp. 101-137.
Mendoza, Enrique G., "Terms-of-Trade Uncertainty and Economic Growth,” Journal

of Development Economics, Vol. 54, Number 2 (December 1997), pp. 323-356.
Mezzera, Jaime, "Segmented Labour Markets Without Policy-induced Labour Market

Distortions,” World Development, Vol. 9, Numbers 11/12 (November/December
1981), pp. 1109-1114.
Michaely, Michael, "Concentration of Exports and Imports: An International

Comparison,” Economic Journal, Vol. 68, Number 272 (December 1958), pp. 722-
736.
Michaely, Michael, "The Shares of Countries in World Trade,” Review o f Economics

and Statistics, Vol. 42 Number 3 (August i960), pp. 307-317.
Michaely, Michael, Concentration in International Trade, (Amsterdam: North-

Holland, 1962).
Michaely, Michael, "Exports and Growth: An Empirical Investigation,” Journal of

Development Economics, Vol. 4, Number 1 (March 1977), pp. 49-53.
Michaely, Michael, Trade, Income Levels, and Dependence, (Amsterdam: North-

Holland, 1984).
Mishkin, Frederic S., "Understanding Financial Crises: A Developing Country

Perspectives," in Michael Bruno and Boris Pleskovic (eds.), Annual World Bank
Conference on Development Economics 1996, (Washington, DC: World Bank, 1997),
pp. 29-62.
Mussa, Michael, Morris Goldstein, Peter B. Clark, Donald J. Mathieson, and Tamim
Bayoumi, Improving the International Monetary System: Constraints and Possibilities,
IMF Occasional Paper 116 (Washington, DC: IMF, December 1994).
259
Nelson, Daniel B., "Conditional Heteroskedasticity in Asset Returns: A New

Approach,” Econometrica, Vol. 59, Number 2 (March 1991), pp. 347-370.
Neumann, Manfred, "Real Effects o f Exchange Rate Volatility,” Journal of

International Money and Finance, Vol. 14, Number 3 (June 1995), pp. 417-426.
Ogawa, Kazuo and Kazuyuki Suzuki, "Uncertainty and Investment: Some Evidence
from the Panel Data of Japanese Manufacturing Firms,” Institute of Social and
Economic Research, Osaka University, Japan, Discussion Paper No. 436 (March
1997).
Ojo, Oladeji, and Temitope Oshikoya, "Policy Determinants o f Long-Term Growth:

Some African Results,” African Development Bank, Economic Research Papers, No.
19 (1993).
Oshikoya, Temitope W ., "Macroeconomic Determinants of Domestic Private

Investment in Africa: An Empirical Analysis, Economic Development and Cultural
Change, Vol. 42, Number 3 (April 1994), pp. 573-596.
Pack, Howard, "Endogenous Growth Theory: Intellectual Appeal and Empirical

Shortcomings,” Journal o f Economic Perspectives, Vol. 8, Number 1 (Winter 1994),
pp. 55-72
Pagan, Adrian and Aman Ullah, "The Econometric Analysis o f Models with Risk
Terms,” Journal of Applied Econometrics, Vol. 3 , Number 2 (April 1988), pp. 87-
105.
Pasinetti, Luigi L ., "The Structure of Long-Term Development: Concluding

Comments,” in Luigi L. Pasinetti and Robert M. Solow (eds.), Economic Growth and
the Structure o f Long-Term Development, (New York: St. Martin's Press, 1994), pp.
353-362.
Papanek, Gustav F., "Aid, Foreign Private Investment, Savings, and Growth in Less
Developed Countries,” Journal of Political Economics, Vol. 81, Number 1 (January-
February 1973), pp. 120-130.
Papanek, Gustav F., "The Effect of Aid and Other Resource Transfers on Savings and
Growth in Less Developed Countries,” Economic Journal, Vol. 82, Number 327
(September 1972), pp. 934-950.
Pauly, Louis, W ., "National Financial Structures, Capital Mobility, and International

Economic Rules: The Normative Consequences of East Asian, European, and
260
American Distinctiveness,” Policy Sciences, Vol. 27, Number 4 (November 1994), pp.
343-363.
Perotti, Roberto, "Income Distribution, Democracy, and Growth: An Empirical

Investigation,” (mimeo, March 1995).
Pindyck, R oberts., Irreversibility, Uncertainty and Investment, NBER Working Paper

No. 3307, (Cambridge, MA: NBER, March 1990).
Pindyck, Robert S., "Irreversibility, Uncertainty and Investment,” Journal o f

Economic Literature, Vol. 29, Number 3 (September 1991), pp. 1110-1148.
Pindyck, Robert S., "A Note on Competitive Investment under Uncertainty,” American
Economic Review, Vol. 83, Number 1 (March 1993), pp. 273-277.
Pindyck, Robert S., "Irreversibility, Uncertainty and Investment" in Serven, Luis and
Andres Solimano (eds.), Striving for Growth After Adjustment: The Role o f Capital
Formation, (Washington, D.C.: World Bank, 1993), pp. 31-80.
Pindyck, Robert S. and Andrds Solimano, "Economic Instability and Aggregate

Investment" in Oliver Jean Blanchard and Stanley Fischer (eds.), NBER
Macroeconomic Annual: 1993, (Cambridge, MA: MIT Press, 1993).
Pollin, Robert, "Introduction,” in Robert Pollin (ed.) The Macroeconomics o f Saving,

Finance, and Investment, (Ann Arbor: University o f Michigan Press, 1997).
Poterba, James M. and Lawrence H. Summers (1986), "The Persistence of Volatility

and Stock Market Fluctuations,” American Economic Review, Vol. 76, Number 5
(December 1986), pp. 1142.
Price, Simon, "Aggregate Uncertainty, Capacity Utilization and Manufacturing

Investment,” Applied Economics, Vol. 27, Number 2 (February 1995), pp. 147-154.
Pritchett, Lant, "Measuring Real Exchange Rate Instability in Developing Countries,”

World Bank Policy, Research, External Affairs Working Papers, WPS 791 (October
1991).
Rama, Martin, "Empirical Investment Equations for Developing Countries,” in Luis

Serven and Andr6s Solimano (eds.), Striving for Growth After Adjustment: The Role of
Capital Formation, (Washington, D.C.: World Bank, 1993), pp. 107-143.
261
Ramey, Garey and Valerie A. Ramey, "Cross-Country Evidence on the Link between
Volatility and Growth,” American Economic Review, Vol. 85, Number 5 (December
1995), pp. 1138-1151.
Rodrik, Dani, "Policy Uncertainty and Private Investment in Developing Countries,”

Journal of Development Economics, Vol. 36, Number 2 (October 1991), pp. 229-242.
Rodrik, Dani, "Why Do More Open Economies Have Bigger Governments?,” National
Bureau of Economic Reseach, Working Paper No. 5537 (April 1996); also publsihed as
Centre for Economic Policy Research, Discussion Paper No. 1388 (May 1996).
Sachs, Jeffrey, "Social Conflict and Populist Policies in Latin America,” NBER
Working Papers, No. 2897 (March 1989); also published as International Center for
Economic Growth, Occasional Papers, No. 9, (San Francisco, CA, ICS Press, 1990).
Sachs, Jeffrey and Andrew Warner, "Economic Reform and the Process of Global
Integration,” Brookings Papers on Economic Activity, (1995, Part I), pp. 1-118.
Sandmo, Agnar, "The Effect of Uncertainty on Saving Decisions,” Review cf

Economic Studies, Vol. 27(3), Number 111 (July 1970), pp. 353-360.
Sawides, Andreas, "Investment Slowdown in Developing Countries during the 1980s:

Debt Overhang or Foreign Capital Inflows, Kyklos, Vol. 45, Number 3 (August 1992),
pp. 363-378.
Schmidt-Hebbel, Klaus and Tobias Muller, "Private Investment under Macroeconomic

Adjustment in Morocco,” in Ajay Chhibber, Mansoor Dailami and Nemat Shafik
(eds.), Reviving Private Investment in Developing Countries: Empirical Studies and
Policy Lessons, (Amsterdam:, North Holland, 1992), pp. 125-158.
Schmidt-Hebbel, Klaus, Luis Serven and Andres Solimano, "Saving and Investment:
Paradigms, Puzzles, Policies,” World Bank Research Observer, Vol. 11, Number 1
(February 1996), pp. 87-117.
Schydlowsky, Daniel M ., "Structural Adjustment for the 1990s,” in Daniel M.

Schydlowsky (ed.), Structural Adjustment: Retrospect and Prospect, (Westport:
Praeger, 1995), pp. 220-239.
Schydlowsky, Daniel M. and Moshe Syrquin, "The Estimation of CES Production

Functions and Neutral Efficiency Levels using Effective Rates of Protection as Price
Deflators,” Review o f Economics and Statistics, Vol. 54, Number 1 (February 1972),
pp. 79-83.
262
Seabra, Fernando, "Short-run Exchange Rate Uncertainty in Latin America,” Applied

Economics, Vol. 27, Number 5 (May 1995), pp. 441-450.
Serven, Luis, "Terms-of-Trade Shocks and Optimal Investment: Another Look at the
Laursen-Metzler Effect,” mimeo, 1994.
Serven, Luis, "Uncertainty, Instability, and Irreversible Investment,” World Bank

Policy Research Working Paper, No 1722 (February 1997).
Serven, Luis and Andres Solimano, "Economic Adjustment and Investment

Performance in Developing Countries: The Experience o f the 1980s,” in Vittorio
Corbo, Stanley Fischer, and Steven B. Webb, (eds.), Adjustment Lending Revisited,
(Washington, DC: World Bank, 1992), pp. 117-138.
Serven, Luis and Andres Solimano, "Debt Crisis, Adjustment Policies and Capital
Formation in Developing Countries: Where Do We Stand?,” World Development, Vol.
21, Number 1 (January 1993), pp. 127-140.
Serven, Luis and Andres Solimano, "Adjustment Policies and Investment Performance
in Developing Countries,” in Luis Serven and Andres Solimano (eds.), Striving for
Growth After Adjustment: The Role of Capital Formation, (Washington, D.C.: World
Bank, 1993), pp. 147-179.
Shackle, George L. S., Uncertainty in Economics, and Other Reflections, (Cambridge,

England: University Press, 1968).
Sibert, Anne and Jiming Ha, "Portfolio Substitution and Exchange Rate Volatility,”
Journal of Monetary Economics, Vol. 39, Number 3 (August 1997), pp. 517-534.
Singh, Ajit, "Openness and the Market Friendly Approach to Development: Learning
the Right Lessons from Development Experience, World Development, Vol. 22,
Number 12 (December 1994), pp. 1811-1823.
Solimano, Andres, "How Private Investment Reacts to Changing Macroeconomic

Conditions: the Case of Chile in the 1980s,” in Ajay Chhibber, Mansoor Dailami and
Nemat Shafik (eds.), Reviving Private Investment in Developing Countries: Empirical
Studies and Policy Lessons, (Amsterdam:, North Holland, 1992), pp. 19-42.
Stancanelli, Elena, "An Empirical Model of the Aggregate Investment Function for the
Italian Economy,” Economia Intemazionale, Vol. 43, Number 4 (Novembre 1990),
pp. 342-356.
263
Steindl, Josef, Maturity and Stagnation in American Capitalism, (Oxford: Basil

Blackwell, 1952).
Stiglitz, Joseph E., "Why Financial Structure Matters,” Journal of Economic

Perspectives, Vol. 2, Number 4 (Fall 1988), pp. 121-126.
Summers, Robert and Alan Heston, "The Penn World Table (Mark 5): An Expanded
Set o f International Comparisons, 1950-1988,” Quarterly Journal of Economics, Vol.
106, Number 2 (May 1991), pp. 327-368.
Sundararajan, V ., "The Debt-Equity Ratio of Firms and the Effectiveness of Interest

Rate Policy: Analysis with a Dynamic Model of Saving, Investment and Growth in
Korea,” IMF Staff Papers, Vol. 34, Number 2 (June 1987), pp. 260-310.
Sundrum, R.M ., Economic Growth in Theory and Practice, (New York: St. Martin's
Press, 1991).
Syrquin, Moshe, "Structural Transformation and the New Growth Theory,” in Luigi L.
Pasinetti and Robert M. Solow (eds.), Economic Growth and the Structure of Lang-
Term Development, (New York: St. Martin's Press, 1994), pp. 3-21.
Syrquin, Moshe and Hollis B. Chenery, Patterns of Development, 1950 to 1983, World
Bank Discussion Papers, No. 41, (Washington, D.C.: World Bank, 1989).
Syrquin, Moshe, Lance Taylor and Larry E. Westphal (eds.), Economic Structure and
Performance: Essays in Honor o f Hollis B. Chenery, (Orlando: Academic Press, 1984).
Tanzi, Vito, "Taxation and Economic Structure,” Modified version of the

commemorative lecture delivered at the Congress of the Japan Fiscal Science
Association (Nihon Zaisei Gakkai) held in Fukuoka University (Fukuoka, Japan) on
October 29, 1993.
Tavares, Jose and Romain Wacziarg, "How Democracy Fosters Growth,” (mimeo,
August 1996).
Taylor, Lance, "Development Patterns: A Simulation Study,” Quarterly Journal of

Economics, Vol. 83, Number 1 (May 1969), pp. 220-241.
Taylor, Lance, Structuralist Macroeconomics, (New York: Basic Books, 1983).
Taylor, Lance, "Macro Policy in the Tropics: How Sensible People Stand, World
Development, Vol. 15, Number 12 (December 1987), pp. 1407-1435.
264
Tegene, Abebayehu, “The Supply and Demand for Exports in Some African Countries: A
Simultaneous Equation Model,” Journal o f Quantitative Economics, Vol. 6, Number 2
(July 1990), pp. 395-410.
Tyler, William G ., "Growth and Export Expansion in Developing Countries: Some

Empirical Evidence,” Journal of Development Economics, Vol. 9, Number 1 (August
1981), pp. 121-130.
United Nations, Instability in Export Markets o f Underdeveloped Countries, (New

York: United Nations Secretariat, 1952).
United Nations Conference on Trade and Development (UNCTAD), International

Monetary and Financial Issues for the 1990s , Volume IV, Special Issue, Proceedings
of a Conference sponsored by the Group of Twenty-Four on the Occasion of the
Fiftieth Anniversary of the Bretton Woods Conference (New York: United Nations,
1994).
United Nations Development Programme (UNDP), Human Development Report 1996,

(New York: Oxford University Press, 1996).
Vamvakidis, Athanasios, "Trade Openness and Economic Growth Reconsidered,”

(mimeo, Harvard University, August 1996)
Van Wincoop, Eric, "Terms of Trade Uncertainty, Savings, and the Production
Structure,” Journal of International Economics, Vol. 33, Numbers 3-4 (November
1992), pp, 305-325.
Velasco, Andres, "Animal Spirits, Investment and International Capital Movements,”

Journal of International Money ad Finance, Vol. 15, Number 2 (April 1996), pp. 221-
237.
Vinod, H.D. and B.D. McCullough, “Bootstrapping Demand and Supply Elasticities: the
Indian Case,” Journal o f Asian Economics, Vol. 5, Number 3 (Fall 1994), pp. 367-379.
Warner, Andrew M ., "Did the Debt Crisis Cause the Investment Crisis?" Quarterly
Journal of Economics, Vol. 107, Number 4 (November 1992), pp. 1161-1186.
World Bank, Private Capital Flows to Developing Countries: The Road to Financial
Integration, (New York: Oxford University Press, 1997).
World Bank, World Development Report 1989: Financial Systems and Development,
(Oxford, Oxford University Press, 1989).
265
World Bank, World Development Report 1995: Workers in an Integrating World,

(Oxford, Oxford University Press, 1995).
World Bank, World Development Report 1995: From Plan to Market, (Oxford, Oxford
University Press, 1995).
Zamowitz, Victor and Louis Labros, "Consensus and Uncertainty in Economic

Prediction,” Journal o f Political Economy, Vol. 95, Number 3 (June 1987), pp. 591-
621.
IMAGE EVALUATION
TEST TARGET ( Q A - 3 )
1.0 £12 12
IS 1m 112.2
IS 13.6
£ |i° 2.0
l.l
1.8
1.25 1.4 1.6
150mm
A P P L I E D A IIW1GE . I n c
_----- 1653 East Main Street
v - = ~- Rochester, NY 14609 USA
= Phone: 716/482-0300
- ^ = ~-= = Fax: 716/288-5989
0 1993, Applied Image, Inc.. All Rights Reserved
with permission of the copyright owner. Further reproduction prohibited without permission.

Thesesdissertations 2633 OBJ

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

Thesesdissertations 2633 OBJ

Uploaded by

Copyright:

Available Formats

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the

Oversize materials (e.g., maps, drawings, charts) are reproduced by

Photographs included in the original manuscript have been reproduced

Faculty o f the College of Arts and Sciences of

the Requirements for the Degree of

Prof. Robert A. Blecker

Dean of the College

American U niversity^' •. ST034

THE AMERICAN UV" STY LIBRARY

UMI Microform 9826672

This microform edition is protected against unauthorized

ALL RIGHTS RESERVED

"A good mathematical theorem dealing with economic

hypothesis is very unlikely to be good economics."

Alfred Marshall (1906)

increase macroeconomic uncertainty. One whole chapter is used to define macroeconomic

between investment and development, it is concluded that investment usually increases

are drawn on implications for economic policy and future research.

I am indebted to Prof. William H. Branson at Princeton University and Isabel

related to economic structure and many other suggestions. I am also indebted to

supported me throughout my graduate studies at American University. Last not least, I

O’Donnell, and Eileen Pereira.

H. HOW TO RELATE DEVELOPMENT TO ECONOMIC STRUCTURE 10

1. Outline of Some Major Contributions.................................................. II

3. Defining Economic Structure.................................................................... 39

HI. EMPIRICAL RESULTS OF PATTERNS OF DEVELOPMENT............. 58

1. Sectoral Composition of O utput........................................................... 59

2. Investment Shares in G D P ................................................................... 66

3. Savings and Consumption...................................................................... 69

5. Inflation and Money Supply..................................................................... 75

6. Overall Trade and Import V ariables...................................................... 78

8. Export Product Concentration (E P C ).................................................... 87

9. Market Power in World Export M arkets............................................. 89

10. Financial Market Development.............................................................. 90

11. Summary of Results: Stylized F a c ts..................................................... 92

IV. HOW TO MEASURE MACROECONOMIC UNCERTAINTY.............. 95

1. Uncertainty versus R isk .......................................................................... 96

2. The Different Concepts of Volatility and Uncertainty........................ 100

3. Measures of Volatility and Uncertainty................................................ 106

4. A Preliminary Measure of Uncertainty................................................. 119

5. Areas of Macroeconomic Uncertainty.................................................. 122

6. How to Obtain an Unbiased Uncertainty M easure............................. 123

7. Checking for Consistency...................................................................... 130

8. How Different are Volatility and Uncertainty in Reality?................. 134

9. The Relationship of Uncertainty Across A reas................................. 136

10. The Relationship of Uncertainties AcrossPeriods............................. 138

1. The Relationship between Development and Uncertainty.................. 142

2. Insights from the Sign of the Correlation

3. The Relationship between Economic Structure and

4. Regressing Uncertainty on Economic Structure................................ 147

5. Remaining Stylized F a cts...................................................................... 151

VI. INVESTMENT UNDER UNCERTAINTY.................................................. 154

1. Investment and Risk at the M icro-Level............................................ 155

2. Intuitive Explanations for a Positive Correlation.............................. 158

3. Empirical Analysis of Investment under Uncertainty....................... 160

4. How to Specify the Aggregate Domestic Investment Function 169

5. How Robust are Impacts of Macroeconomic Uncertainty? 179

VH. THE RELATIONSHIP BETWEEN INVESTMENT AND

1. Investment and G row th.......................................................................... 189

2. Growth and Development..................................................................... 192

3. Investment as a Major Factor of Development................................. 195

1. Connecting Major Results................................................................... 197