Strucutreand Growthof Service Sectorin Ethiopia

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/320868688
STRUCTURE AND GROWTH OF SERVICE SECTOR IN ETHIOPIA
Preprint · November 2017

DOI: 10.13140/RG.2.2.15343.28325
CITATIONS READS
0 7,765
1 author:
Zelalem Ejigu Kabeta

Ethiopian Civil Service University
2 PUBLICATIONS 12 CITATIONS
SEE PROFILE
All content following this page was uploaded by Zelalem Ejigu Kabeta on 29 November 2017.
The user has requested enhancement of the downloaded file.

STRUCTURE AND GROWTH OF
SERVICE SECTOR IN ETHIOPIA
A
THESIS
Presented to the Faculty of Social Sciences of the
Punjabi University, Patiala
in Fulfilment of the Requirements for the Degree of
DOCTOR OF PHILOSOPHY
IN
ECONOMICS
Established Under Punjab Act No. 35 of 1961

Department of Economics
November 2017
STRUCTURE AND GROWTH OF
SERVICE SECTOR IN ETHIOPIA

A
THESIS
Presented to the Faculty of Social Sciences of the
in Fulfilment of the Requirements for the Degree of
DOCTOR OF PHILOSOPHY
IN
ECONOMICS
by
Supervisor
Professor Inderjeet Singh
Established Under Punjab Act No. 35 of 1961

Department of Economics
November 2017
CERTIFICATE
This is to certify that this thesis entitled “Structure and

Growth of Service Sector in Ethiopia” embodies the work carried
out by Zelalem Ejigu Kabeta himself under my supervision and
that it is worthy of consideration for the award of the Ph.D. degree.
Supervisor’s Signature,
(Dr. Inderjeet Singh)

Professor,
Economics Department,
Punjabi University,
Patiala (Pb.), INDIA.
Page | i
DECLARATION
I, Zelalem Ejigu Kabeta, certify that the work embodied in
this Ph. D. thesis is my own bonafide work carried out by me under
the supervision of Professor Inderjeet Singh from 13/05/2014 to
10/03/2017, at Department of Economics, Punjabi University,
Patiala. The matter embodied in this Ph. D. thesis has not been
submitted for the award of any other degree/diploma. I declare that
I have faithfully acknowledged, given credit to and referred to the
research workers wherever their works have been cited in the text
and the body of the thesis. I further certify that I have not wilfully
lifted up some other’s work, paragraph, text, data, results, etc.
reported in the journals, books, magazines, reports, dissertations,
theses, etc., or available at web-sites and included them in this Ph.
D. thesis and cited as my own work. I also declare that I have
adhered to all principles of academic honesty and integrity and
have not misrepresented or fabricated or falsified any
idea/data/fact/source in my submission. I understand that any
violation of the above will be cause for disciplinary action by the
University.
Date:_____________ Signature of the Candidate

Place: Patiala
(Zelalem Ejigu Kabeta)
This is to certify that the above statement made by the candidate is

correct to the best of my knowledge.
Date: ______________ Supervisor’s Signature

Place: Patiala
(Dr. Inderjeet Singh)
Professor,
Economics Department,
Punjabi University,
Patiala (Pb.), INDIA.
Page | ii
DEDICATION
To my Parents Abaye and Ejigu, for your care, love, and support
And
To the beloved late mothers W/ro Assefash and W/ro Gizhachew,
And
To my late uncles (Nigussie, Jebessa, and Chombe), whom I lost during

the study
Page | iii
ACKNOWLEDGMENTS
Primarily, I would like to express my gratitude to my
supervisor Professor Inderjeet Singh, whose guidance, enthusiasm
and encouragement are invaluable to the completion of this
study. It was the first conversation I made with him on the topic
that gave me the impetus to make this study.
I am also grateful to the Department of Economics, Punjabi
University Patiala for giving me admission and capacitating me
through a coursework program. I am also grateful to Department
heads Professor Inderjeet Singh and later Professor Lakhwinder
Singh, who were very helpful throughout the course work and the
study. In addition, I appreciate the services provided by the office
assistants and the librarians at the department. Last but not the
least; I would like to thank Wollega University and Ministry of
Education of Ethiopia for permitting the scholarship program and
financing the study.
Special thank also goes to my friend Tasew Tadesse for
enlightening me regarding the time series model in the study as
well as the insightful comments and ideas we shared. My gratitude
also goes to Ermias Engida and my dear friend Henock Adamu for
providing me important ideas and data on Ethiopian Social
Accounting Matrix.
Finally, I would like to thank my parents. My mother
Azalech Ayele (Abaye) and my father EjiguKabeta, for your care,
love, support. Abaye, this should belong to you more than it
belongs to me. My sisters Eske, Sihnu, Hawiye, thank you for your
unconditional love and support. In addition, I am thankful to my
family Engida, Kete, Alye, Bele, Adanu, Getish, Tigist, Sewdegu,
Trhas. Thank you for your encouraging words. If it were not for
the friendship of the following people, my journey would surely
have been more difficult and lonely. My friends Tamru, Tasew,
Dereje, Habtamu, Gahsaw, Teshale, and Tefera thank you for your
friendship. When times get tough, I always turned to Medi who
deserves all the credit for her continuous support and love for
keeping me sane especially at the final stages of my study. I love
you Medi. I would like also to extend my special thanks to my dear
friends Yite, Bety, Get, Heny, Bini, Shumet and Eyob. Thank you so
much for your friendship.
(Zelalem Ejigu Kabeta)
Page | iv
ACRONYMS
BL Backward Linkage
CGE Computable General Equilibrium
EDRI Ethiopian Development Research Institute
EEA Ethiopian Economic Association
ESSP Ethiopian Strategy Support Program
ETB Ethiopian Birr
FL Forward Linkage
GDP Gross Domestic Product
GDPPC Gross Domestic Product per Capita
GERD Grand Ethiopian Renaissance Dam
GGDC Groningen Growth and Development Center
GTP Growth and Transformation Plan
GVA Gross Value Addition
HDI Human Development Index
HDR Human Development Report
ICT Information Communication Technology
IFPRI International Food Policy Research Institute
IMF International Monetary Fund
IO Input Output
ISIC International Standard Industrial Classification
LP Labor Productivity
NBE National Bank of Ethiopia
NLFS National Labor Force Survey
PASDEP Plan for Accelerated and Sustainable Development to End Poverty
PPP Purchasing Power Parity
SAM Social Accounting Matrix
SDPRP Sustainable Development and Poverty Reduction Program
TFP Total Factor Productivity
UN United Nations
UNCTAD United Nations Conference on Trade and Development
UNDP United Nations Development Program
USD United States Dollar
WB World Bank
WTO World Trade Organization
Page | v
TABLE OF CONTENTS
CERTIFICATE .............................................................................................................. i
DECLARATION ......................................................................................................... ii
DEDICATION ............................................................................................................ iii
ACKNOWLEDGMENTS .......................................................................................... iv
ACRONYMS ............................................................................................................... v
TABLE OF CONTENTS ............................................................................................ vi
LIST OF TABLES .................................................................................................... viii
LIST OF FIGURES ..................................................................................................... x
CHAPTER - I: INTRODUCTION .............................................................................. 1
1.1 Background......................................................................................................... 1
1.2 Problem Statement ............................................................................................. 2
1.3 Objectives .......................................................................................................... 5
1.4 Justification ........................................................................................................ 6
1.5 Significance ....................................................................................................... 6
1.6 Scope ................................................................................................................. 7
1.7 Data and Methodology ...................................................................................... 7
1.8 Organization ...................................................................................................... 8
1.9 Main conclusion of the Study ............................................................................ 8
1.10 Main Limitations .............................................................................................. 9
CHAPTER-II: REVIEW OF THEORY AND EMPIRICS ........................................ 10
2.1 Theories on Service Sector Growth and Structure ........................................... 11
2.1.1 Conceptualization of Service Sector ...................................................... 11
2.1.2 Structural Change and the Growth of Service Sector ............................ 17
2.1.3 Determining Factors of the Service Sector Growth ............................... 30
2.1.4 Sustainability of Service-led Growth .................................................... 35
2.1.5 Sectoral Inter-Linkage of the Service Sector ......................................... 37
2.2 Empirical Review of Literature ....................................................................... 38
2.2.1 Growth of the service sector .................................................................. 38
2.2.2 The Sustainability and Viability of Service-Led Growth ..................... 40
2.2.3 Structural Change, Economic Growth, and the Service Sector ............ 44
2.2.4 Determining Factor of Service Sector Growth ..................................... 51
2.2.5 The Sectoral Inter-linkages of the Service Sector ................................. 58
2.2.6 Experience of Developing Countries in Structural Change and Service
Sector Growth ........................................................................................ 61
CHAPTER-III : METHODOLOGY, DATA AND MODEL ..................................... 81
3.1 Growth Decomposition Analysis ..................................................................... 82
3.1.1 Job Generation and Growth Decomposition .......................................... 82
3.1.2 Shift Share Analysis .............................................................................. 84
3.1.3 Structural Bonus and Structural Burden Hypothesis ............................. 87
Page | vi
3.2 Specification of the Autoregressive Distributed Lag Model ............................ 89
3.3 Sectoral Inter-linkage Analysis ....................................................................... 94
3.3.1 Social Accounting Matrix ...................................................................... 95
3.3.2 Input-Output Model Framework ............................................................ 98
3.3.3 Description of Ethiopian 2005/06 SAM and I-O table ......................... 100
3.3.4 Inter-industry linkage analysis of Service sector.................................. 102
3.3.5 Key Sector identification ...................................................................... 107
CHAPTER-IV: STRUCTURAL CHANGE AND PRODUCTIVITY IN ETHIOPIAN
ECONOMY ........................................................................................ 108
4.1 Growth of the Service Sector in Ethiopia ....................................................... 108
4.1.1 Service sector growth in output, employment, and productivity.......... 108
4.1.2 Growth and Composition of Service sub-sectors ................................. 122
4.1.3 Composition of Service sub Sectors in GVA and Employment ........... 130
4.1.4 The Contribution of Service sector in Output Growth and
Employment ........................................................................................ 133
4.2 Role of the Service Sector in Structural Change in Ethiopia ...................... 143
4.2.1 The Shift-share Analysis ...................................................................... 143
CHAPTER- V: DETERMINANTS OF GROWTH OF SERVICE SECTOR AND
ITS SUSTAINABILITY ..................................................................... 158
5.1 Determinants of Service Sector Growth in Ethiopia ..................................... 158
5.1.1 Descriptive Statistics ............................................................................ 159
5.1.2 Unit Root Test ..................................................................................... 160
5.1.3 Bound Test for Co-integration Analysis............................................... 164
5.1.4 Regression Results of ARDL ............................................................... 165
5.2 Sustainability of Service-Led Growth in Ethiopian........................................ 174
CHAPTER- VI: LINKAGE ANALYSIS OF SERVICE SECTOR IN ETHIOPIA . 184
6.1 Service Sector Inter-linkages in Ethiopia ....................................................... 184
6.1.1 Description of Ethiopian Economy from 2005/06 SAM ...................... 184
6.1.2 Structure of Import and Exports ........................................................... 193
6.1.3 Results from SAM and Input-Output Analysis .................................... 195
CHAPTER- VII: MAIN FINDINGS AND POLICY IMPLICATIONS .................. 214

Bibliography ............................................................................................................. 229
Annexure -1 .............................................................................................................. 246
Annexure -2 .............................................................................................................. 247
Annexure -3 .............................................................................................................. 248
Page | vii
LIST OF TABLES
Table 3.1: Basic Structure of SAM 97
Table 3.2: Input-Output Framework of Ethiopia for 10 sectors 101
Table 4.1: GVA Share of Ethiopia across Comparator Countries (1961- 109
2011)
Table 4.2: Contribution of Sectors to Real GDPPC Growth of Ethiopia 110
(2007/08 – 2014/15)
Table 4.3: Employment Share for Ethiopia and Comparator Countries 116
(1961-2011)
Table 4.4: Labour Productivity level of Ethiopia and Comparator 119
Countries (1961-2011)
Table 4.5: GVA Share of Service Sectors, Ethiopia and Comparator 124
Countries (1961-2011)
Table 4.6: Employment Share of Service Sectors, Ethiopia and 127
Comparator Countries (1961-2011)
Table 4.7: Labour Productivity of Service Sectors, Ethiopia, and 128
Table 4.8: Composition of Service Sectors GVA and Employment 132
Share in Ethiopia (1961-2011)
Table 4.9: Decomposition of Growth in PCGDP, Ethiopia (1999 - 2013) 138
Table 4.10: Contribution of Sectors to the Change in Employment rate, 139
Ethiopia (1999 - 2013)
Table 4.11: Contribution of Employment Changes to Change in 140
PCGDP, Ethiopia (1999-2013)
Table 4.12: Growth Decomposition, Contribution to Total Growth in 142
GDPPC, Ethiopia (1999-2013)
Table 4.13: Shift-share Result of Labour Productivity Growth, Ethiopia 145
(1961 - 1991)
Table 4.14: Shift-share Result of LP growth, Ethiopia (1992 - 2011) 146
Table 4.15: Shift-share of LP growth, Role of Sectors in Ethiopia 150
(1961 – 1992)
Table 4.16: Shift-share of LP growth, Role of Sectors in Ethiopia 152
(1992 – 2011)
Table 4.17: Gross Value Added, Employment, and Labor Productivity, 156
Ethiopia (1961 - 2011)
Table 5.1: Pearson Correlation for Service GVA and Employment 159
Table 5.2: Augmented Dicky-Fuller Test for Unit Root 162
Table 5.3: ADF Unit Root Test with Break 163
Table 5.4: ARDL Bound Test for Service GVA and Service 164
Employment
Table 5.5: Estimated Long run and Short run Coefficients using the
ARDL Approach 168
Page | viii
Table 5.6: Diagnostics Tests 173
Table 5.7: Productivity Level and Growth in Ethiopia (1981 – 2011) 175
Table 6.1: Aggregated Ethiopian Macro SAM (2005/06) 186
Table 6.2: Share of Sectors in Total Gross Output and Total Demand 187
Table 6.3: Share of Sectors Value Added at Factor Cost 189
Table 6.4: Households Consumption Expenditure Shares by Sector, 192
Ethiopia (2005/06)
Table 6.5: Export and Import trade Structure of Ethiopia, 2005/06 194
Table 6.6: Production Technology Coefficients, Ethiopia (2005/06) 197
Table 6.7: Backward and Forward Linkages, Ethiopia (2005/06) 200
Table 6.8: Weighted Backward and Forward Linkages, Ethiopia 203
(2005/06)
Table 6.9: Pure Backward and Forward linkages, Ethiopia (2005/06) 205
Page | ix
LIST OF FIGURES
Figure 4.1: Sectoral Shares and GDPPC Growth of Ethiopia (1961- 112
2011)
Figure 4.2: Relationship Between Sectoral Output Shares in GVA and 113
GDPC (2013)
Figure 4.3: Relationship Between Sectoral Shares in Employment and 117
GDP Per Capita (2013)
Figure 4.4: Relative Labor Productivity, Ethiopia and SSA (1961 – 121
2011)
Figure 4.5: Sectoral Labor Productivity Levels of Ethiopia (1961- 122
2011)
Figure 4.6: Labour Productivity of Services and Manufacturing in 130
Ethiopia (1961-2011)
Figure 4.7: Composition of Service Sector in GVA in Ethiopia (1961- 132
2011)
Figure 4.8: Composition of Service Sector Employment in Ethiopia 133
(1961-2011)
Figure 4.9: Contribution of Sectors to Annual GVA growth (1962- 135
2011)
Figure 4.10: Contribution of Sectors to Employment growth (1962- 136
2011)
Figure 4.11: Shift Share Decomposition results, Labor Productivity 155
Growth, Ethiopia (1961 - 2011)
Figure 5.1: Line Graph for Service GVA, Employment, and 161
Determinant Variables
Figure 5.2: Service GVA and Employment (Hodrick-Prescott Filter 164
Decomposition)
Figure 5.3: Hodrick-Prescott Decomposition of GDPPC Income, 172
Figure 5.4: Cumulative Sums of Recursive Residuals for Service GVA 173
Figure 5.5: Cumulative Sums of Recursive Residuals for Service 173
Employment
Figure 5.6: Annual Growths of the Service and Manufacturing Sector, 176
Figure 5.7: Service Convergence in Ethiopia (1981 – 2001) and (2001 177
– 2011)
Figure 5.8: Manufacturing Convergence in Ethiopia (1981 – 2001) 178
and (2001 – 2011)
Figure 5.9: Sectoral Productivity and Change in Employment Share, 180
Ethiopia
(1981–2001)
Figure 5.10: Sectoral Productivity and Change in Employment Share, 181
Ethiopia (2001 – 2011)
Figure 5.11: Share of Service Export Earnings and Manufacturer 182
Imports
in Ethiopia
Figure (2005-2012)
6.1: Relative Factors Share of Ethiopia (2005/06) 189
Figure 6.2: Sectoral Composition of Labour, Ethiopia (2005/06) 191
Page | x
Figure 6.3: Key Sectors Based on Hirschman-Rasmussen Linkages, 207
Ethiopia (2005/06)
Figure 6.4: Key Sectors based on weighted Backward and Forward 209
Linkages, Ethiopia (2005/06)
Figure 6.5: Key Sector based on Pure Backward and Forward 211
Figure 6.6: Economic landscape of Ethiopia (2005/06) 213
Page | xi
CHAPTER-I
INTRODUCTION
1.1 : Background
Economic history shows that the features of structural change process

across countries are similar. As an economy moves from lower to higher
stages of development, output and employment shifts from the primary
agriculture sector to the secondary industrial and tertiary service sectors. Yet,
there exist variations between developed and developing countries as well as
among the developing countries in their process of structural change. The
developed countries have followed the conventional transformation process
involving the shift of resources and labor from the traditional agriculture
sector to the industrial sector and then to the service sector sequentially. This
also involves the transfer of resources from low productive sectors to high
productive sectors. However, the developing countries are deviating from the
traditional transformation process as production and employment shifts from
the agriculture sector directly to the service sector, leapfrogging the
industrialization process. This is becoming the common structural feature of
most of the developing countries in Asia, Latin America and Africa. The
economic structure of Ethiopia in the recent periods is also characterized by
the shift of resources from agriculture to the service sector.
In Ethiopia, the tertiary sector share in GDP exceeds the agriculture

sector share. The service sector accounts for the lion’s share in terms of the
structure of GDP (46.6 percent) in 2014/15 taking the lead from the
agriculture sector. The overgrown tertiary sector has far-reaching implications
for output, employment and sectoral linkages of the economy. In this context,
the work is an attempt to analyze the tertiary sector in accordance with the
structural change process of Ethiopian economy.
Page | 1
1.2 : Problem Statement
Structural change involves changes in the composition of the sectoral

gross value added, employment and productivity growth. There is evidence
that shows the level of Ethiopian structural transformation from the
perspectives of output, employment, and productivity. The gross value
addition in agriculture has indicated that there is a change in the share of
agriculture in GDP from 66 percent in 1990 to 44 percent in 2014, whose role
is overtaken by the service sector in the same year, whereas, the share of the
industry possesses insignificant change. On the employment side, there is no
big change that the Ethiopian economy has experienced within the last 20
years. Still, the larger proportions (78 percent) of Ethiopians are employed
within the agriculture sector. On the productivity side, output per worker has
doubled in the last decade. However, productivity in the manufacturing sector
became lower than service sector and productivity in the agriculture remained
very low, whereas, the service sector has high and increasing productivity
level. Hence, the lag in the structural change process and size of the services
sector in Ethiopia are one of the striking features of the economy.
According to Moller & Wacker (2015), public infrastructure

investment, financed in part by restrained government consumption, was the
key structural driver of growth in Ethiopia. Heterodox policies pursued by the
Ethiopian government, such as financial repression, overvalued real exchange
rate, and monetary policy induced inflation are more controversial. The model
simulations under alternative policy scenarios are indicative that growth is
likely to decelerate in the coming decade making it challenging Ethiopia to
attain its middle-income country target by 2025. Eshete & Kimuyu (2014a)
also claim the unlikely of Ethiopia to cross the middle-income country lower
threshold by 2025. Some productivity gains were infrastructure related,
agricultural productivity due to improved farming practices and input use,
capital accumulation contributed by private investment and shifts of labor out
of low productivity agriculture and into higher productivity construction and
services activities boosted growth.
Page | 2
Agriculture in Ethiopia has been the center of the country’s
development policy, given the largest role and share that the sector has in the
economy. Various studies on Ethiopia have identified that agricultural growth
induces higher overall growth and faster poverty reduction and larger growth
effect than growth in non-agricultural sector (Diao, Hazell, Resnick, &
Thurlow, 2007; Mellor & Dorosh, 2010; Engida, et al., 2011). According to
Diao et al. (2007), the impact of growth on poverty is larger when the
additional growth is driven by agriculture rather than non-agriculture sector. In
addition, according to Dorosh & Thurlow (2014), in Ethiopia, there are larger
linkages between agricultural production and small towns and these shows that
redirecting urban growth toward towns rather than cities leads to broad-based
economic growth and poverty reduction. Within non-agriculture, industry-led
growth has a larger multiplier than services-led growth. Both industry and
services have weak linkages to rural areas. This suggests that, based on current
economic structures, industrialization in Ethiopia will weaken national growth
linkages and widen the rural–urban divide (Dorosh & Thurlow, 2014).
Along with view on the importance of agriculture in Ethiopia, there is

an argument that views the untimely growth of the service sector inhibiting a
healthy structural transformation process. According to EEA (2015 p. 182),
for a healthy structural transformation, either the service sector needs to have
locally produced industrial as well as agricultural products to trade or the
sector need to export part of its import bills. The first justification is that the
contribution of the service sector in GDP growth is not a recent phenomenon
as it has contributed much in the past two decades. The second justification is
associated with the composition of the service sector. The service sector in
Ethiopia is dominated by traditional distributive service activity, which
involves the trading of locally produced agricultural commodities and
imported goods.
The Ethiopian government is also following a policy to attract and

direct private and foreign investment from the construction and service
investment towards manufacturing activities. However, most local private
Page | 3
investments prefer investment in the service sector as the manufacturing sector
is constrained by many factors. These constraints are infrastructure, power
shortage, transportation, and customs clearing and associated services during
import and exports, shortage of foreign exchange to import intermediate
inputs, access to finance, lack of skilled workforce in local labor markets and
other related problems. Therefore, the productivity improvements and efficient
services specifically in the transportation, storage and communication
services, finance and business services, and training institutes in the education
sector can facilitate the gear for the growth of the manufacturing sector.
Various studies have shown that productivity in services plays a critical role as
a strategic driver of competitiveness in the manufacturing sector (World
Economic Forum, 2015). Moreover, opening up of the economy to
international service and further liberalization of the economy will attract FDI
that could be used to finance import bills of the intermediate goods.
Furthermore, the service sector is a driver of the recent economic

growth in contrast to the policy objective of the country. It remained a
question how to sustain the economic growth and create employment
opportunity for the labor that moves out of agriculture. In addition, identifying
how the service sector stimulates growth in the manufacturing sector is also
important for industrialization and structural transformation objectives of
Ethiopia. There is also huge literature gap in Ethiopia in the service sector
growth and performance. Few studies explored the service sector growth and
structural changes. The review of studies is indicative of the fact that most of
the studies done so far, in Ethiopia, are too aggregative or even if
disaggregated, a comprehensive study of the underlying dynamism of the
sector has not been done. Most of the studies have dealt with individual sub-
sectors or the overall economy-wide aggregates and have failed to capture the
underlying structural dynamics of the tertiary sector in terms of time and
composition. A study fortified with disaggregated data going rigorously into
structure and growth of Ethiopian economy is a need of the time.
Page | 4
The role of the service sector in Ethiopia, hence, requires a detailed
analysis of structure and growth of the sector that covers not only composition
and temporal dynamics but also the linkage patterns and the determinants of
tertiarization. In this context, this work is an attempt to analyze the structure
and growth of the tertiary sector in Ethiopia with special reference to
structural change, linkage patterns, and determinants.Therefore, with the
objective of transforming its economy and joining the middle-income country
group, Ethiopia is striving to industrialize its economy. However, with a
higher share of GDP contributed by the service sector and low level of
manufacturing sector growth; industrializing the economy is a daunting task.
So, evaluating the service sector growth option at this time may provide
alternative strategies to move the manufacturing sector and the economy
forward.
1.3 : Study Objectives
The main objective of the study is to analyze the structure and the
growth of the Ethiopian service sector. The specific objectives are as follows:
a) To assess the growth trend of Ethiopian service sector and its sub-
sectors.
b) To delineate the structural change in Ethiopia with special reference to

tertiarization.
c) To determine the factors that explain the growth of the Ethiopian

service sector.
d) To analyze the service sector inter-linkage with agriculture and

industry sectors.
e) To examine the viability and sustainability of service sector growth in

Ethiopia.
f) To prescribe the policy framework for better functioning and

sustainability of the Ethiopian economy.
Page | 5
1.4 : Justification of Study
The justification for this study is the shift of Ethiopian economy

towards the service sector and the problem associated with the economic
structure of the country in the past decades. In addition to this, the economic
development of the country is hampered by various socio-economic problems.
There is stagnant growth in the manufacturing sector and the growth of the
sector is also affected by its low productivity levels. However, some of the
developing countries that failed to industrialize their economy are
experiencing a high service sector growth along with high economic growth.
The success of the South Asian countries like India in service sector growth
and service export has created a glimpse of hope for those developing
countries that missed economic growth through industrialization. Along with
the economic growth, Ethiopia has achieved in the last decades, the service
sector has increased in its contribution to the growth and it overtook the lead
from the agriculture sector in its share of GDP and contribution to the high
economic growth rate. Hence, the major justification for the study on the
Ethiopian service sector growth and structure is the structural shift towards
service sector in the past decade.
1.5 : Significance of Study
This research on the service sector growth and structure focuses on one
of the important macroeconomic issues of the time in developing countries.
The service sector growth is a controversial issue in development economics
as some consider it as unproductive and cannot lead economic growth and
others view it as a dynamic sector with a capacity to be an engine of growth.
Given this controversy and the problem associated with poor performance of
the manufacturing sector and overgrowth of the service sector, this study will
be pertinent to developing countries like Ethiopia. The study has relevance to
sustainable economic development in Ethiopia and it has a multifaceted
contribution to Ethiopian economy and fills the research gaps that exist in
service sector growth and structural change literature. The thesis brings to the
attention of scholars the role of the service sector in output, employment and
Page | 6
structural change in Ethiopia. Further, it tests the various hypotheses that drive
the growth of the service sector. There is also knowledge gap in the linkage of
the Ethiopian service sectors with the manufacturing and agriculture sectors.
The study will be a stepping stone for further studies on the role and
mechanisms of the service sector in stimulating growth in the manufacturing
sector in Ethiopia.
1.6 : Scope of Study
This study generally concentrates on exploring the structure of the

Ethiopian economy, with emphasis on the growth of service sector. It assesses
the growth trend of the service sector in output, employment, and productivity
from 1961 to 2011. In addition, it identifies the composition of the service
sector, the role of the service sector in employment and output growth in
Ethiopia. The scope of this study is also limited to the role of the service sector
in structural change process of Ethiopia since 1961. The structural change in
the Ethiopian economy is estimated using Shift-share decomposition tool. In
this case, labor productivity growth is used to measure productivity growth of
the Ethiopian economy. The study also analyses, the driving factors for the
growth of the service sector since 1981, focusing only on four different
hypotheses employing time series analysis. Finally, the sectoral inter-linkage
of the service sector in Ethiopia is estimated using the Social Accounting
Matrix data collected for 2005/06. The estimation of the sectoral inter-linkage
represents only the situation and patterns of economic structure in Ethiopia in
2005/06.
1.7 : Data and Methodology
The structure and growth of the service sector in Ethiopia is analyzed

by applying different methodologies. The data used for this research are
collected from secondary sources from GGDC 10 sectors database, from
National Bank of Ethiopia (NBE), Central Statistics Agency (CSA), and
World Development Indicator (WDI). Various analytical tools are also used
for each objective. Descriptive statistics and growth decomposition
methodologies have been applied in order to explore the growth trend of the
Page | 7
sector and to identify the role of the sector in Ethiopian output and
employment growth and structural change. The Autoregressive Distributed
Lag (ARDL) model, Social Accounting Matrix (SAM)/Input-Output (I-O)
analysis and sustainability analysis are conducted in order to identify the
determinants, inter-linkage, and sustainability of the service sector
respectively.
1.8: Thesis Organization
The thesis has seven chapters, which are organized in the following
way. The first chapter is an introduction, which introduces the study by
presenting the rationale and objectives of the study. The second chapter
presents the theoretical and empirical reviews on service sector growth and
structural change along with the experience of Asian and African countries.
The third chapter focuses on the methodological issues, describing the growth
decomposition analysis, the time series analysis, and input-output analysis.
The fourth chapter presents analyses and discusses the structural change and
productivity in the Ethiopian economy. Chapter five presents results and
discussion on determinants of service sector growth and its sustainability. The
sixth chapter deals with the linkage analysis of the service sector in Ethiopia.
Finally, the last chapter, chapter seven, gives conclusions and policy
implications of the study.
1.9 : Main Conclusion of the Study

The overall conclusion that emerges from the study is that in Ethiopia
the service sector has witnessed high growth and has contributed to the growth
in output and employment during the recent growth periods. The sector is also
the source of growth-enhancing structural change. In addition, the growth of
the sector is largely driven by growth in income, female economic
participation, and economic reforms. The service sector has also strong inter-
sectoral linkages through the distributive trade service. Furthermore, it has
also the potential to sustain the high economic growth and stimulate growth in
other sectors.
Page | 8
1.10 : Main Limitations
The main limitation of this study is associated with the availability of
recent SAM or IO data for Ethiopia to identify the key sectors and other inter-
linkage analysis for the recent years. Similarly, the other limitation of the
study is that the time included in most of the structural change and time series
analyses are before 2011. Recent data were not available from the same
sources at the time of this research work.
Page | 9
CHAPTER-II
REVIEW OF THEORY AND EMPIRICS
The theoretical and empirical review of the literature on service sector

growth and structural change is presented in this chapter. A recent trend in the
high growth of service sector has created various theories and arguments
among a number of economists. The role of the service sector in the
productivity growth and economic growth at large has been neglected in the
classical economics. But, this time there is a revisited interest in the source of
service-led growth and its role in structural changes of developing economies.
Large numbers of empirical studies have been conducted on the growth of
service sector and structural changes in the developing countries within the
last two decades. A growing number of these researches are conducted in Asia
specifically in India, after the success of the Indian model of service growth.
The high growth performance of India and other developing countries is
closely associated with their service sector growth. This has revitalized the
service sector and it is considered as a path for structural change due to the
failure of some countries to be competitive in the manufacturing sector. In
addition, the sector can promote productivity in manufacturing and agriculture
sectors.
The literature in this section is broadly classified as a theoretical and

empirical review. The theoretical review includes literature on concept,
definitional issues and classifications of service sector, structural change and
service sector growth, sustainability of service-led growth, drivers of service
sector growth and sectoral inter-linkages of the service sector. The empirical
review highlights the major empirical researches on service sector growth, the
role of the service sector in structural changes, drivers of the service sector
growth and sustainability of service-led growth in developing countries.
Finally, the sectoral inter-linkages and the service sector growth experience of
Asian and African countries are presented.
Page | 10
2.1 : Theories on Service Sector Growth and Structure
2.1.1 : Conceptualization of Service Sector
The term service represents and constitutes various issues. It is difficult

to define it as the given definitions fail to represent the diversity of the sectors.
Services cover ‘a wide range of intangible and heterogeneous products and
activities that are difficult to encapsulate within a simple definition’ (WTO,
2010, p. 7). According to UNCTAD (2015), services may be defined as
‘changes in the condition of an economic agent or in the condition of a good
or asset belonging to an economic agent, as a result of some activity of a
second economic agent, carried out with the prior agreement of the first
agent’. However, in early periods, services were considered as unproductive
economic activities that cannot be accounted in the national accounts. For
instance, for classical economists such as Adam Smith, services are considered
as unproductive economic activities compared to commodities. They perish
immediately as they are consumed or performed, rarely leaving a value behind
them for producing an equal quality of service afterward(Smith, 2007). Smith
criticized service sector as offering product that immediately vanishes unlike
outputs from primary and secondary sectors, which are tangible and contribute
to development. In addition, based on the Marxist-Leninist theory of social
production, national accounts statistics of socialist economy considers all
services as part of the non-material production and hence unproductive (Seth,
2008). However, in the present days, the productive capacity of services is
identified and the role of services is reconsidered as services constitute the
world’s largest share in total economic production and employment.
Hill (1977) has made the earliest attempt to define services. He defined
services as a change in the condition of a person or of a good belonging to
some economic unit, performing some activity for the benefit of the other
(Hill, 1977). He also argues that ‘goods and services belong to different
logical categories’ because one cannot store services, rather services must be
consumed as they are produced. This signifies the specific characteristic
feature of services, which is non-storability. Thus, services can be also defined
Page | 11
by looking into the common or specific features, which make them different
from goods and other types of economic activities. Parasuraman, Zeithaml,
and Berry (1985) identified three characteristic features of services. These are
intangibility, heterogeneity, and inseparability. The intangibility feature
implies that services cannot be counted, tested, stored or verified prior to sale.
Heterogeneity of services arises due to variation in the production of same
services from day to day or from producer to producer as service often
depends upon unreliable labor (Parasuraman, Zeithaml, & Berry, 1985).
Services are also inseparable as production and consumption of service are
simultaneous. In addition, services are perishable, which means that ‘once the
time of potential service passes, the opportunity to sell that service perishes’
(Rust & Chung, 2006). However, recent development in the sector has eased
the problems associated with these features of services and the gap between
goods and services is narrowed. Due to fast development in the information
and technology, some services are having components of tangibility, visibility,
storability and may not require face-to-face contact between producers and
consumers. For instance, software programs and digital electronics services
have only limited tangibility but are storable and transferable. Since some of
the definitions given for services have contributed for the negative
connotations and perceptions towards services, it is important to look it by
classifying services.
The WTO (2010) classification of service into two categories may give
the broader view about services. Accordingly, there are change-effecting
services (transformation services) and margin services. The change-effecting
services are ‘activities of producers that change properties of consuming units
or the physical or/and mental conditions of an agent or good’. In this type of
services, ownership cannot be established, as they cannot be separated from
production and the change brought about may be permanent or temporal. For
instance, it includes transforming physical goods through repairing, cleaning
or transporting, or change in the mental condition of persons through services
of education and entertainment. On the other hand, the margin services
facilitate the exchange of goods or assets or services between the user of such
Page | 12
services and another party. These margin services are offered by wholesale
and retail distribution, financial institutions that provide insurance, packaging,
transportation and communication intellectual property rights and distribution
(WTO, 2010, p. 7). In addition to the transformation and margin services,
there are also knowledge-capturing products. These products are hybrid
products, with characteristics of both goods and services such as newspapers,
electronic media, and digital information. They are transferable and can be
accessed more than once. In addition, the production and consumption of these
products may not be necessarily simultaneous due to their physical existence.
However, such products have the features of services as they can effect
changes on the mental condition of users through the information and
knowledge they contain (UNCTAD, 2015).
According to the International Standard Industrial Classification

(ISIC), Rev. 3.1, service activities can be broadly classified into eleven
categories (coded from G to Q). These are: (i) wholesale and retail trade;
repair of motor vehicles, motorcycles and personal and household goods (ii)
hotels and restaurants (iii) transport, storage and communications. (iv)
financial intermediation (v) real estate, renting and business activities (vi)
public administration and defense (vii) education (viii) health and social work
(ix) other community, social and personal service activities (x) private
households activities as employers (xi) extraterritorial organizations and
bodies (ISIC, 2002). In the ISIC hierarchy, the construction sector followed by
electricity, gas, and water, are placed in the wholesale trade and retail trade,
which are sometimes included in the secondary or tertiary (service) sectors.
The definitional characteristics of services have brought different

challenges in the measurement of services output (value added). The first
challenge is measuring the real output of services activities, which is adjusted
for inflation. It is difficult for services to assume constant quantity over time
and standardize as services are heterogeneous and depends on the user as well
as the producer (e.g. in education services). The second challenge of
measurement arises from the need for service inputs in the production of both
Page | 13
goods and services. For instance, input services such as financial management
may be provided by the producer of an output, in which case they are not
measured separately as a services output. The other issue with regard to
measurement is related to the informal sector activities, which is dominated by
services activities, overlap with the unobserved economy in the national
accounts system (UNCTAD, 2015).
Additionally, the diversity in the definition and classification of service

may emanate from lesser economic importance given for services as compared
to goods producing sectors starting from the early economists. Joshi (2008)
argues that lack of definition and lack of research on the tertiary sector
originates from the views of Adam Smith and Karl Marx. They viewed the
tertiary sector as ‘residual sector’ and ‘the services of lawyers, physicians and
other occupations which didn't produce any physical products were classified
as unnecessary part of tertiary sector’ (Joshi, 2008). This was because of their
interest in services as distinct from goods for defining productive labor.
Moreover, most of the time the service-producing sector is defined by
exclusion, i.e., services are defined by what they are not. That is services are
those activities that are not agricultural and industrial manufacturing. This
residual definition has contributed to the negative perception of the value of
the service sector (McLachlan, Clark, & Monday, 2002). This can be shown
from the definitions given to services starting from the Physiocrats. The
Physiocrats (1750) consider ‘all activities other than agricultural production as
services’, to Adam Smith (1723 – 1790) services are ‘all activities that do not
end in tangible products’. According to J.B. Say (1767 – 1832), all non-
manufacturing activities that add utility to goods are services. The western
country economists during (1925 – 1960) consider service as those that do not
lead to a change in the form of a good. The contemporary definition also
generally considers services as an activity that does not lead to a change in the
form of a good (Joshi, 2008, p. 27).
According to Bhagwati (1984a), services can be categorized into those

necessarily requiring the physical proximity of the provider and user and those
Page | 14
that do not require physical proximity (Bhagwati, 1984a). The categorization
is used in the WTO General Agreement on Trade in Services (GATS), which
is the first set of multilaterally negotiated and legally enforceable rules on
international trade in commercial services (excluding government services)
since 1995. The services that do not require physical proximity (the service
only crosses the border) are a cross-border trade (mode 1 in GATS). In this
case, the delivery of the service can take place through telecommunications.
On the other hand, the services that require physical proximity can be further
categorized into three different groups. These are (i) mobile provider and
immobile user – commercial presence (mode 3). This type of services includes
services such as medical services provided by a foreign-owned hospital, and
banking services supplied by a subsidiary of a foreign bank.(ii) mobile user
and immobile provider – consumption abroad (mode 2) that includes hospital
and tourism services, and (iii) mobile user and mobile provider - the presence
of natural persons (mode 4), which requires temporarily migrating of labor to
provide the services such as lectures, consultancy, and similar services.
Even though the recent growth in services trade was accompanied by

significant changes in the structure of services traded, all the peculiar nature
and characteristic features of services made them difficult for trading. The
technological changes and the ‘splintering and disembodiment of services’
have made many services tradable, like goods (Bhagwati, 1984a). The
information and communication technology (ICT) revolution curbed the
obstacles in the international delivery of a number of services and it is driving
the services revolution(Ghani & Kharas, 2010). These activities of the service
sector dominate other economic activities (agriculture and manufacturing)
through time due to the economic progress of nations. Fisher (1939) has also
observed this pattern, the emergence of large service sector due to economic
progress, in early times. He suggested that economic progress would lead to
the emergence of a large service sector, after the growth of agriculture and
industry sectors. Thus, he distinguished the three broad sectors as 'primary',
'secondary', and 'tertiary' sectors, referring the full development of each sector
by stages (Fisher, 1939).
Page | 15
According to Katouzian (1970), the growth pattern of the service
sector can be analyzed by dividing the sector into new services, old services,
and complimentary services. New services are services that have experienced
a great shift in demand due to ‘high mass consumption', which is an increasing
function of per capita income and leisure time. Because of further growth and
automation, the demand for these services will further increase. The old
services had demand that came from a less equitable distribution of income
and the relative lack of alternative employment for the producers of these
services. However, due to economic progress, these services are continuously
substituted by industrial durable consumer goods and the new services.
Finally, the demand for complementary services is mainly associated with the
growth of demand for intermediate goods due to industrial expansions,
urbanization, and unification of home and international markets. These
services have high growth potential as long as the technological revolution
continues (Katouzian, 1970).
Eichengreen and Gupta (2013) also classify service into traditional and
modern services. They argue that the aggregation of services as one broader
activity obscures two distinct ‘waves of service sector growth’. The first wave
of the service sector growth occurs in economies having relatively low level of
per capita GDP in the traditional service sectors (such as personal services).
The second wave occurs at higher per capita incomes in activities (such as
financial, communication, computer, technical, legal, advertising and business
services) that depend largely on information technology, and it is easily
tradable across borders. Even though the share of services has increased since
the 1970s in all countries irrespective of their income levels, growth in the
modern services (second wave) is observed in lower income levels in recent
periods than before. However, the change in the second wave is not equally
observed in all countries. Countries that are ‘open to trade, more democratic,
and relatively close to the global financial centers have increased their share in
modern services’ (Eichengreen & Gupta, 2013).
Page | 16
Therefore, the service sector involves economic activities that have
diverse characteristic features than the goods sector that have limited their
tradability, transferability, and storability. However, in this age due to
advancement in information and communication technology, the sector is
expanding in all economies, specifically; developing countries are flooded
with service activities ahead of their underdeveloped manufacturing and
industrial sectors. The growth of service sector is highly interlinked with
changes in the structure of economies. Based on early period observations
from the stages of growth theories to the recent observation of the waves of
service sector growth, the change in the composition of service sector growth
involves structural change. The next sub-topic discusses the growth of the
service sector and tries to put the sector in the economic growth theories and
structural change literature.
2.1.2 : Structural Change and the Growth of Service Sector
Economic development literature of the Post-Second World War

viewed the process of development as a series of successive stages of
economic growth through which all countries must pass. According to Todaro
and Smith (2012), the four major theories after World War II are (i) the linear-
stages-of-growth model, (ii) patterns of structural change, (iii) international
dependence revolution, and (iv) the neoclassical, free-market
counterrevolution (Todaro & Smith, 2012). A brief review of each of the
theories is presented in the next paragraphs.
The linear-stages-of-growth model, which is also called Rostow’s

Stage of Growth, after the economic historian Walt W. Rostow, assumes ‘the
transition from underdevelopment to development can be described in terms of
a series of steps or stages through which all countries must proceed’ (Todaro
& Smith, 2012, p. 111). This model is complemented by the Harrod-Domar
growth model, which argues the important strategy for takeoff is the
mobilization of saving to generate sufficient investment and accelerate
economic growth. The stage of growth theory is replaced in the 1970s by the
pattern of structural change and the dependence theory. The structural change
Page | 17
theory focuses on the mechanisms by which developing economies transform
their economic structures from the dominant traditional subsistence agriculture
to the modern manufacturing and service economy. The proponent of this
theory used modern economic theory and statistical analysis to identify the
process of structural change to a sustainable rapid economic growth of
developing economies. The well-known examples of this model are the ‘two-
sector surplus labor’ theoretical model of Arthur Lewis Lewis (1954), and
‘patterns of development’ empirical analysis of structural change model of
Chenery & Syrquin (1989).
According to Todaro & Smith (2012), one of the early theoretical

models of development that focused on the structural transformation of a
primarily subsistence economy was formulated by W. Arthur Lewis in the
mid-1950s and later modified, formalized, and extended by Ranis & Fei
(1961). According to the Lewis two-sector model, ‘surplus labor from the
traditional agricultural sector is transferred to the modern industrial sector
promoting industrialization and stimulating sustained development (Lewis,
1954). The Lewis model focused on the transformation of the economic
structure of countries from agriculture, which has low productivity of labor,
towards an industry that has a high productivity of labor(Todaro & Smith,
2012).
In addition to the two-sector model, the patterns-of-development

approach is another type of structural change approach that is developed by
the empirical works of Chenery and his colleagues (Chenery, 1979; Chenery
& Syrquin, 1986; Chenery & Taylor, 1968; Chenery & Watanabe, 1958). The
approach focuses on the sequential transformation of the economic, industrial,
and institutional structure of developing countries over time in order to replace
the traditional agriculture that has been the engine of economic growth. They
examined patterns of development for developing countries with different
levels of per capita income using both cross-sectional and time-series data.
They identified several characteristic features of the development process,
which include the shift from the agriculture sector to the modern industrial
Page | 18
production, physical and human capital accumulation, the change in consumer
demands from food and basic necessities to manufactured goods and services.
In addition, they observed ‘growth of cities and urban industries due to the
migration of people from farms and small towns and the decline in family size
and overall population growth, and population growth first increasing and then
decreasing in the process of development’ (Todaro & Smith, 2012). The
various schools of development economists of 1950s and 1960s ‘emphasized
the importance of structural change and saw structural differences because of
market failures’. In addition, they proposed the use of government
interventions for facilitating structural change by import substitution. Hence,
most developing countries and multilateral development institutions followed
these policy recommendations by using Keynesianism as the main intellectual
foundation. However, results were disappointing as the gap between
developing countries and the industrialized countries widened. According to
Lin (2012), the structural difference among countries at different levels of
development was ignored in the policy recommendations(Lin, 2012).
The other development theory that has followed the structural change
theory was the international dependence model, which gained prominence
during the 1970s among intellectuals of developing countries due to the
dissatisfaction with the previous stages of growth and structural change
models. The model views developing countries as affected by their
institutional, political, and economic relations; and they are caught up in a
dependence and dominance relationship with rich countries. The three major
streams of dependence mode are the neocolonial dependence model, the false-
paradigm model, and the dualistic-development thesis.
In the 1980s and 1990s, some countries of the developed world with
conservative governments started advocating a neoclassical (neoliberal)
counter-revolution economic theory and policy. The main argument of the
theory is that underdevelopment is the result of poor resource allocation due to
incorrect pricing policies and too much government intervention and
regulation of the economy by governments of developing economies. The
Page | 19
theory also favored supply-side macroeconomic policies and privatization of
public corporations in the developed countries, and free markets and the
dismantling of public ownership in developing countries. The traditional
neoclassical growth theory has also revived in the 1990s as the cornerstone of
the neoclassical free-market argument. Accordingly, the assertion is that
liberalization of national markets by the developing nations could draw
additional domestic and foreign investment, which can increase the rate of
capital accumulation. The Solow neoclassical growth model represented this
theory, which argues that there are diminishing returns to each factor of
production but constant returns to scale and the exogenous technological
change generates long-term economic growth (Todaro & Smith, 2012, p. 128).
According to Lin (2012), the emergence of stagflation in the 1970s, the Latin
American debt crisis, and the failure of the socialist planning system in the
1980s also gave rise to the rational expectations theory and helped to refute
the structuralist theoretical foundation for the state’s role in using fiscal,
monetary, and trade policy. The multilateral institutions and development
agencies advocated this thinking and influenced economic policies of
developing countries through their conditionality programs of stabilization and
structural adjustments. The policy advice and programs promoted economic
liberalization, privatization, and the implementation of rigorous stabilization
programs, whose results are mixed and controversial.
Generally, most of the neoclassical economists believe that structural

change is the prerequisite for economic progress and the share of economy
shifts from the traditional agriculture sector to modern manufacturing and
service sector. They also put structural change at the center of economic
development, as sustained economic growth cannot happen without structural
changes (Kuznets, 1966). Furthermore, the recent studies on the history of
economic development show that developing economies that have
successfully converged and led the world growth experienced structural
changes in the composition of output and employment of their agriculture,
industry and service sectors.
Page | 20
Structural Change
The structuralist economics of the Post-Second World War period has

revived in the development economics literature of this time, and it came to be
known as the ‘New Structural Economics’ (Lin, 2012). It emphasizes that
growth has a poverty-reducing effect if it ensures that a country’s labor force
and other resources are directed to the productive activities. According to
Silva and Teixeira (2008), studies on structural change spurt in the 1990s as an
alternative to the rigorous modeling framework of neoclassical economics
(Silva & Teixeira, 2008). Following this, there has been renewed interest in
studies of structural change and economic growth. Some of these are(Dietrich,
2012; Aggarwal & Kumar, 2012; Chen, Jefferson, & Zhang, 2011; Cortuk &
Singh, 2013; Dabla-Norris, Thomas, Garcia-Verdu, & Chen, 2013; Fagerberg,
2000; Fan, Zhang, & Robinson, 2003; de Vries, Timmer, & de Vries, 2013;
McMillan, Rodrik, & Verduzco-Gallo, 2014).
In development economics and in economic history, structural change

is commonly understood as the change in the distribution of economic activity
and productive factors among various sectors of the economy. The most
common use of structural change in development and in economic history
refers to the relative importance of sectors in the economy in terms of output
production, employment, and factor use. Thus, the structural change issue in
the development literature emphasizes on increases in the rates of
accumulation, shifts in the sectoral composition of economic activity that
focuses on the allocation of employment and production and factor use. In
addition, it includes changes in the location of economic activity to
urbanization centers and other concomitant aspects of industrialization that
includes demographic transition and income distribution. Thus, the interrelated
processes of structural change that accompany economic development are
jointly referred to as the structural transformation (Syrquin, 1988).
Thus, the process of economic development is highly associated with

structural change. It is about transforming the productive structure of the
economy and accumulating the capabilities necessary to undertake this process
Page | 21
(Chenery, 1986; Chenery & Syrquin, 1986; Kuznets, 1966). It is also about the
reallocation of economic activity across the three broad economic sectors of
agriculture, manufacturing, and services (Herrendorf, Rogerson, & Valentinyi,
2014). Several stylized facts characterize the outcome of structural change or
transformation process. These are: (i) a declining share of agriculture in total
output and employment; (ii) rapid urbanization; (iii) emergence a modern
industrial and service economy; and (iv) demographic transition that leads to a
spurt in population growth.
On the other hand, Chenery (1986) identified accumulation of human

and physical capital, shifts in the composition of demand, trade, production,
and employment to be at the core of structural transformation. During
structural transformation, there would be economic growth at a higher level of
per capita income, and transition from a low-income agrarian economy to an
industrial urban economy (Syrquin, 1988). Thus, all the successful developed
countries have passed through the process of structural transformation.
Therefore, it is a critical prerequisite for economic and social development.
Through its impact on labor productivity, structural change plays an
instrumental role in sustaining economic growth, generating productive
employment, and raising living standards (Martins, 2014).
The patterns of structural transformation, as it is documented by

Syrquin (1988), include growth patterns, accumulations, change in sectors
proportions and relative prices. A high growth rate in per capita income
characterizes the economies of developed countries. There is also growth
acceleration in the middle-income countries during their transformation
process. During the structural transformation process, the total factor
productivity growth explains much of the growth of the industrial modern
sector than agriculture sector. Accumulations, the other common pattern of
structural transformation, refer to the flow of resources to increase the
productivity capacity of an economy. Saving and capital formation are the
mediums of accumulating physical capital that helps to raise the productive
capacity of an economy. The other most common and noticeable feature of
Page | 22
structural transformation is the change in sectoral proportions. An increase in
per capita income is also usually accompanied by the rise in the share of
industrial and service output, which is associated with Engel’s Law, a falling
share of food demand as income increases (Syrquin, 1988).
Labour productivity is one of the important features in structural

transformation. It shows a different pattern in different sectors during the
process of structural change. Initially, the relative labor productivity in the
agriculture sector declines and it rises as the country develops due to a decline
in labor force and increases the wages of the agricultural sector. The relative
labor productivity in the manufacturing sector rises until it reaches a
maximum level and tends to decline to a stable level after the economy is
developed, following deindustrialization. However, the service sector
generally has declining labor productivity in the process of transformation
(Syrquin, 1988). In addition, (Kuznets, 1973) also observed the sectoral
differences during the process of structural change. He distinguished between
two phases of structural transformation. In the first phase of structural
transformation, which is the beginning of the development process, an
economy allocates most of its resources to the agricultural sector and as the
economy develops, resources are reallocated from agriculture into industry
and services. In the second phase, resources are reallocated from both
agriculture and industry into services.
On the other hand, the process of structural transformation differs from

country to country based on the factors endowments or comparative advantage
and policy decisions about the level of trade and foreign capital inflows. Based
on Syrquin’s analysis, a country with average performance can undergo a
transformation at an income level of about USD 850. Large countries also can
reach at the semi-industrial stage at a per capita income level of USD 550.
However, the transformation may take more time for small countries
specializing in primary exports until the per capita income reaches 1300 USD.
Structural change and early industrialization can happen for large countries
that specialize in manufacturing through the policy of import substitutions
Page | 23
because of the large domestic market (Syrquin, 1988). However, all the
developed countries have followed a similar process of structural
transformation. See (Chenery & Syrquin, 1986; Syrquin, 1988).
Recently, Bah (2010) studied whether developing countries are also

following a similar process of structural transformations like the developed
countries. Accordingly, he finds great differences among paths being followed
by the developing countries. Most of the African countries, which have very
low GDP per capita, have low agricultural output shares and high services
shares. As compared to the developed countries, Latin American countries
moved from the first to the second phase of structural transformation at a
lower level of GDP per capita. On the other hand, Asian countries have
relatively higher industrial shares and comparable service shares. Among the
developing countries, Asian countries are closest to the structural
transformation path of developed countries. In addition, many developing
countries of Africa and Latin America are found to have significant sectoral
changes at the time of economic stagnation and decline, which implies that,
structural transformation, can occur with very little growth in GDP per capita
(Bah, 2010).
Therefore, based on the review of the major literature on structural

change and transformation, structural change can be summarized by a set of
stylized facts. These are; (1) change in the composition of sectors (in terms of
value added, employment and productivity) from agricultural dominance to
industry and service, (2) an increase in capital accumulation, (3) change in the
income distribution and an increase in per capita income, (4) urbanization
(rural–urban migration) and industrialization, and (5) change in the structure
of demand and trade. Structural change may also occur irrespective of the
growth in GDP per capita and significant change in sectoral composition.
Structural Change and Sectoral Growth
Sectoral growth and performance analysis are basically originated from

the dual economy models of (Lewis, 1954)and(Hirschman, 1958). The models
attempt to explain economic growth by examining the role and relationship
Page | 24
between the traditional agricultural sector and modern manufacturing sector.
The sectoral changes in output, which represents structural change, occur as
development proceeds due to the different effect of income elasticity of
demand in each sector. The income elasticity of demand for agriculture sector
is inelastic while manufacturing goods have higher income elasticity.
However, the income elasticity of demand for services is greater than one
(Falvey & Gemmell, 1996). Besides, this implies that at a higher level of
income services will have more demand than the other sectors. Those sectors
with the strongest linkages to the growth will result in the development of a
leading sector and structural change. The balanced and unbalanced growth
debate in development economics is also related with whether the
development would proceed more effectively with the more investment in a
leading sector or in all sectors. The balanced growth approach focuses on
investment in all sectors to grow by supporting those sectors that may not
grow naturally. That means a big push might be needed by a government to
make a balanced growth in the economy. The unbalanced growth approach, as
described by Albert Hirschman, advocates unbalanced sectoral growth for the
overall economic growth (Hirschman, 1958). He suggests that a developing
economy can promote economic growth by initially investing in industries
with high backward and forward linkages.When growth is unbalanced, prices
will rise in those sectors where output growth is relatively slow, which will
call for more investment by private investors or government.
There are various forms of sectoral development approach to bring

economic growth. Agricultural led industrialization approach is the one, which
suggests growth in agriculture stimulates overall economic growth and
structural change as witnessed during the industrial revolution in Europe and
green revolution in Asia. It puts agriculture at the center of the growth process
as it provides food, supplies raw materials, creates more employment and
generates foreign currency. Agriculture by its nature has high sectoral inter-
linkage with the manufacturing sector, which helps to fuel growth in the
manufacturing and service sectors. Thus, based on this approach, agriculture
Page | 25
can ignite industrialization, speed up the process of structural change and lead
the overall economic growth of nations at early stages of development.
On the other hand, the service-led industrialization is a recent view that

arose due to the growth observed in service output, employment, and
productivity. Conventionally, the service sector is perceived as unproductive.
The possibilities for productivity growth in services are limited because of
their labor-intensive nature. According to Baumol’s cost disease hypothesis,
services have lesser productivity than the manufacturing sector and
productivity growth in services is harder to achieve. However, Baumol’s law
has been recently refuted, because there are market services that have higher
productivity growth than the manufacturing sector (Baumol, 1967). The
growth in the productivity of the services is improved due to the introduction
of ICT and innovation. Service sector can bring structural change when factors
accumulations are met with an increase in the productivity growth of the
sector. As a result of low productivity of the service sector, economies that
have very large share of service sectors grow at a slower rate than others with
a smaller share. Thus, advanced economies, which are service-dominated
economies, have slower aggregate growth rates than developing economies.
According to Szirmai (2011, p. 11), this creates ‘new possibilities for
catching-up for developing countries where the manufacturing sector have a
proportionately larger share in output’ The larger share of service sector in
developing countries coupled with large, inefficient and unproductive
government sectors has created structural burden at early lower level of
income. However, in some countries like India, the structural burden observed
in early periods is avoided as the service productivity is improved due to the
dominance of dynamic service activities like information technology and
communication services (Szirmai, 2011).
However, the dominant approach in economic growth is the direct

approach of industrialization through manufacturing sector development. The
manufacturing sector plays the role of engine of growth because of its highest
potential for productivity. An industrial development approach to sustainable
Page | 26
economic growth and structural change through high investment in the
manufacturing sector is the strategy most countries follow. In his recent
research, Szirmai (2011) concludes that the manufacturing sector is important
in accelerating growth and achieving catch-up in developing countries that are
at the lower level of per capita income. However, when he compares to the
past 60 years, the market service sectors will become relatively more
important as potential sources of growth and catch up than manufacturing
sector (Szirmai, 2011).
Therefore, whether the agriculture-led industrialization or service led

industrialization or manufacturing sector development, the important target is
sustainable growth and structural change. Generally, structural change is a key
feature of economic development for perpetual growth, which implies that
sustainable economic growth cannot be possible without structural change
(Kuznets, 1966). Hence, all developed countries followed the same process of
structural change. However, developing countries are following different paths
of structural change that deviate from that of developed countries in different
ways, specifically due to the early growth of the service sector (Bah, 2010).
Service Sector Growth
Theoretical and empirical studies on the growth of service sector

associated with economic growth and sectoral shares in output, consumption,
and employment have evolved starting from the pioneering work of (Fisher,
1935; Clark, 1940; Fuchs, 1965; Baumol, 1967; Chenery & Taylor, 1968).
These studies attribute economic development as a ‘three-stage process’,
wherein primary (agriculture), secondary (manufacturing) and tertiary
(service) sectors dominate the economic activity in a sequence. In the process,
the expansion of the service sector is natural and a prerequisite of
development. It expands in relative terms only after the secondary sector has
already acquired dominance both in terms of value added and in terms of the
workforce (Kuznets, 1966). Based on the model, the following two facts are
observed in the service sector during the process of economic development.
First, there exists a positive relationship between the per capita income and the
Page | 27
share of the service sector. The share of the service sector grows together with
the growth of the per capita income. For countries that have low per capita
incomes, the large proportion of the population spends the greater share of
their income on basic goods, foodstuffs, and agricultural products. But, along
with the increase in incomes the demand for manufactured products also
increases and the society will spend more of their income on services. Second,
the relative price of services when compared to the price of goods is
continuously increasing due to slower productivity growth in the service
sector. An increase in the productivity of industry results in the rise of wages
with a tendency of the leveling of wages, which in turn increases the prices in
sectors relatively lower growth in productivity. In those sectors, the rate of
wage exceeds that of productivity and leads to an increase in the prices more
than the price of those sectors with a rapid growth of productivity. As a result,
more developed countries have a higher relative price for services than
countries at a lower level of development.
When the relative size of industry predominates that of the other

sectors, the tertiary sector then acquires significance in output and
employment because of a rise in per capita income originating from the
commodity-producing sector. The demand for services increases as the
consumption demand for commodities is saturated. The analysis in terms of
such stages of development, however, has been questioned in the recent
literature. According to Kongsamut, Rebelo, & Xie (2001) rising per capita
GDP is associated with an increase in services and as the economy matures,
the sectoral share given up by the agricultural sector shifts more to the services
sector and less to industry (Kongsamut, Rebelo, & Xie, 2001). In other words,
the sectoral share given up by agriculture as the economy matures goes more
to the services sector and less to the industry than the suggestions of Kuznets
and Chenery.
On the other hand, the positive association between economic growth

and the share of services in the distribution of the labor force has been
observed and documented by a number of investigators. Few of them are
Page | 28
(Fisher, 1939; Clark, 1940; Fuchs, 1968; Glasmeier & Howland, 1994;
Katouzian, 1970; Kuznets, 1973). Fisher had suggested that economic
progress would lead to the emergence of a large service sector following the
development of a primary and secondary sector. Fisher (1935) and Clark
(1940) also attributed the dominance of services of advanced economies into
their level of income. They recognized the low productivity in services as a
factor behind the faster employment growth in services than in industry.
Similarly, Fuchs (1965) also observed the same for the US economy in the
1960s. This productivity differential is the basis of the well-known ‘cost
disease’ hypothesis of services propounded by Baumol (1967). Baumol points
out that productivity improvement in services is harder to achieve than in
goods-producing industries. The higher productivity of the industry was
expected to raise wages even in services more than the productivity growth,
which leads to a tendency for costs and prices of services to increase relative
to goods. The unbalanced growth models by Baumol and Fuchs (1968) helped
in promoting the idea that service-sector activities cannot be made more
efficient through capital accumulation, innovation, or economies of scale
because of their labour-intensive nature.
In addition, as pointed by Baumol (1967), it is difficult to mechanize

and increase productivity in services as industrial activities (Baumol, 1967).
Thus, the international differences in the productivity of services are always
smaller as compared to the goods-producing sectors. The productivity of
manufacturing and agricultural activities grows slower in less developed
countries, which also slows the rate of increase of wages in both goods-
producing sectors and the service sector. Therefore, according to the Kravis-
Heston-Summer explanation, services are cheaper in the relative price in
developing countries with low income than in that of rich developed countries
(Bhagwati, 1984b).
The low productivity of services suggested by Baumol has been

questioned later by different researchers like (Griliches, 1992; Triplett &
Bosworth, 2000; Nordhaus, 2006). According to Griliches (1992), the low
Page | 29
level of services productivity is assumed to be a because of mismeasurement
of services output. Besides, factors like technological change, deregulation and
increased competition are set to raise productivity at least in selected services
(Maclean, 1996). In his later article, Baumol himself has recognized the case
of ‘progressive’ services with substantial productivity gains (Baumol,
Blackman, & Wolff, 1985).
There is also a Clark-Fisher hypothesis, which states that development

will lead to the shift of the majority of the labor force to the service sector.
Initially, Allen Fisher in 1935 suggested that economic progress would lead to
the emergence of a large service sector in an economy, following the
development of a primary and secondary sector (Fisher, 1935). In 1940, Colin
Clark developed a similar argument (Clark, 1940), which created the Clark-
Fisher theory. The Clark-Fisher model shares some characteristics of early
linear stage models and the structural change models. In addition, Victor
Fuchs in 1960s and 1970s observed the changing patterns of employment
within the service sector and took this to be a key indicator of economic
progress (Fuchs, 1980). He observed growth in service sector employment
across western economies and he argued that the service sector growth
contributed to the slow-down in economic growth rates in more developed
economies. As the Clark-Fisher model had proposed, productivity growth in
the service sector would tend to be much slower than the manufacturing
sector. This works of Fischer and Clark has brought a total break with the
concept of services represented by Adam Smith and Karl Marx, separating the
productive and non-productive activities (Palocz, 1988). Therefore, the
literature on service sector growth is predominantly associated with income
growth and productivity changes. In addition, other factors may drive service
sector growth, which is reviewed in the next topic.
2.1.3 : Determining Factors of the Service Sector Growth
There have been various theories and arguments regarding the cause of
the shift in the service sector in both output and employment. Here in the
literature, the causes of the shift in services are classified into the following
Page | 30
four categories. These are the ‘hierarchy of needs’ hypothesis, the ‘cost
disease’ hypothesis, the ‘exogenous demand’ hypothesis and
‘deindustrialization’ (Kim, 2006).
The ‘hierarchy of needs’ hypothesis is a ‘demand-side’ explanation of

the shift to service that argues the shift towards services consumption is as a
result of the change in income. It is a traditional view first originated by Fisher
(1935) and Clark (1940) who put that the service sector increases and the
consumption structure of societies change from goods to services due income
growth. Clark argues that ‘services satisfy higher needs than goods’. Thus, as
income grows, a larger share of the income will be used for the consumption
of services. Consequently, due to the shift towards consumption of service, the
demand for services will increase and share of service employment will be
higher than the past. Therefore, employment in services is higher in countries
that have a higher per capita income. In other words, this happens due to
Engel's Law, which states that income elasticity of the demand for goods is
less than one (Nayyar, 2012) whereas the income elasticity of the demand for
services is greater than one. According to this view, the consumption demand
for manufacturing goods will be saturated and the service share in
employment increases to meet the rising demand for services driven by the
income growth. Therefore, services have the high-income elasticity of
demand (Falvey & Gemmell, 1996). This implies that at any relative price of
services the quantity absorbed in services rises more than the quantity
absorbed in commodities as real per capita income increases.
The ‘cost disease’ hypothesis is the ‘supply-side’ interpretation of the

shift towards service sector. According to Baumol (1967) and Baumol,
Blackman, & Wolff (1985), the shift to service employment is not the result of
changing final demand, but it is due to differential productivity growth. He
assumed that the demand for services and goods is not related to income when
it is measured in constant prices, and the share of services and goods in real
output is constant over time and the same across countries. If an economy has
manufacturing sector with high productivity growth and a low productive
Page | 31
service sector, the share of the service sector increases on the current price
basis (Kim, 2006). This is because of the fall in the relative price of the
manufacturing sector as the cost and the relative price of the service sector
increased by an equivalent amount of the productivity gap. According to this
view, if the ratio of services in real value added is more or less constant
irrespective of the income level, the proportion of employment in the service
sector should rise. Therefore, the share of service sector employment is larger
in high-income countries and grows with rising income, because more labor
input is needed due to the low productivity of the sector. Clark (1940) also
puts productivity differences as a major force behind employment shifts.
According to him, labor will be reallocated from manufacturing industries that
experience high rates of productivity growth but stagnating demand, to
services that experience lower rates of productivity growth but rising demand.
Nevertheless, to Kravis, Heston, & Summers (1982), the driving force behind
the expansion of service employment is the evolution of technology rather
than the change in demand due to the rising income. In addition, they observed
that ‘productivity is lower in poor countries relative to the rich countries in
both services and commodity-producing sectors, but it is lower by a larger
margin in commodities than services’ (Kravis, Heston, & Summers, 1982).
The third theory is the ‘exogenous demand shock’ hypothesis,

asserting that the shift to services is the result of structural changes (Kim,
2006). It is associated with an increase in the demand for intermediate service
inputs (producer services) and changes in the household expenditures on
services. These are changes in the structure of demand, like the increase in the
demand for intermediate inputs, the increase in female participation in
economic activities and household consumption expenditure on services.
Producer services, such as finance, insurance, real estate and business services,
have rapidly expanded in most economies as an intermediate input causing an
increase in the share of services in the economy (Fuchs, 1980). The rise of
service employment is also highly associated with an increase in female labor
force participation, as ‘families with working wives tend to spend a higher
proportion of their income on services’ (Fuchs, 1980). When women
Page | 32
participate in paid economic activities, the household's will have more money
income and the household may reallocate expenditures for additional
consumption of services since the women may have less time for production in
the home. Therefore, the change in female economic participation can increase
the share of services in the economy through increased household
consumption on services (Kim, 2006). Other than intermediate input demand
and female economic participation, other factors like economic reform (in the
1990s) have also contributed to the change in the structure of demand.
Economic reform and liberalization can be seen as causes for the growth of
service sector that is observed after the 1990s. Most of developing economies
have conducted economic reforms by deregulating their economy (Jain &
Ninan, 2010).
Economic reforms and liberalization of the services sector involve the

abolition of monopolies, removal of barriers to entry and the privatization of
state-owned enterprises. This will facilitate the entry of new domestic and
foreign providers in the market and hence increase the choice of service
provisions, which in turn may affect the performance of the manufacturing
sectors. According to Arnold, Javorcik, & Mattoo (2011), the quality and
reliability of services may improve because of privatization, competition and
the entry of internationally successful service providers. According to Cali,
Ellis, & Velde (2008), trade in services create opportunities for countries to
create jobs, contribute more to GDP and generate foreign exchange if they
have a comparative advantage in the sectors. Trade in services can be
especially important for those countries that are relatively isolated from
world’s goods market due to poor transport infrastructure or being landlocked.
The entry of foreign service providers may yield better services for domestic
consumers, and it can improve the performance and competitiveness of
domestic firms. It can also serve to bring capital into the country through
foreign direct investment (FDI). Even though the welfare gains from services
liberalization are larger than for goods and agricultural liberalization, there are
risks associated with it. On the other hand, services liberalization carries risks,
and appropriate regulation and other complementary policies are required to
Page | 33
ensure that liberalization delivers the expected benefits (Cali, Ellis, & Velde,
2008). According to Cali, Ellis, & Velde (2008), the risks of liberalization in
service are the following. 1) The crowding out of domestic providers and
instead allowing foreign firms to capture the profits for themselves and taking
the money out of the country. 2) There is a risk that private providers will
cream skim the most profitable clients and cease to serve some other groups of
clients or geographical areas that are deemed likely to be unprofitable
specifically in the financial and energy sectors. 3) There will be also risks of
financial sector instability as the world financial crises are transmittable. 4)
The movement of persons to provide services can aggravate skills shortages in
developing countries and results in brain drain. 5) Finally, service
liberalization can have a risk of environmental degradation due to uncontrolled
tourism development associated with deforestation and erosion; degradation
and depletion of biological diversity; disruption of natural habitats; and over-
consumption of resources like fresh water and energy (Cali, Ellis, & Velde,
2008).
The last theory associated with the shift to service is the

‘deindustrialization’ hypothesis. Deindustrialization refers to a fall in
manufacturing employment of the most advanced economies (Rowthorn &
Ramaswamy, 1999). It is observed in industrialized countries due to the
transfer of labor-intensive manufacturing industries to less developed
countries as the trade between them expands. Here, the hypothesis is that ‘the
service sector's share increases rather passively due to the hollowing out of the
manufacturing sector’ (Kim, 2006). However, Rowthorn and Ramaswamy
(1999) disprove this view, as the effect of trade between the industrialized and
less developed countries on the employment structure is insignificant due to
the very small trade volume between them relative to the economic size of the
industrialized country. Generally, the causes for the shift in the service sector
in output and employment is not only limited to the stated hypothesis. There
are a number of factors that could affect the growth of services that are not
incorporated in this study, such as government consumption expenditure and
trade in services.
Page | 34
2.1.4 : Sustainability of Service-led Growth
A number of economists have questioned the role of the service sector

as an engine of economic growth and its potential to lead growth in the other
sectors and the economy at large. Swan (1985) probed the sustainability of the
service led growth in 1985, long before the revolution of the service sector.
One of argument against service emanates from the term ‘service’ itself. The
‘service sector services’, which means it services the goods sector. Thus,
without the resources or goods sector, there would be no service, which
implies that services are derivative and parasitic so that they cannot be an
engine of growth. However, to Swan (1985) and other economists, the
question of whether services can be an engine of growth is the same with
whether a continuous productivity improvement in the service sector is
possible or not. From the experience of different countries in the past,
productivity has risen significantly. This shows that ‘services can be an engine
of growth, even if the other engine weakens the service sector engine can
continue to pull independently’ (Swan, 1985). Finally, Swan concludes that ‘at
any given growth rate of resources, an economy will grow faster if
productivity increases in the service sector’.
Nevertheless, the popular belief and argument in the development

literature is that manufacturing sector is the engine of economic growth
(Kaldor, 1966). According to Kaldor (1966), there is a positive relation
between manufacturing output growth and average GDP growth; growth of
manufacturing output and productivity; and the growth of manufacturing
output and overall productivity through spillover effects. Kaldor’s theory was
also supported by Baumol (1967) that states productivity growth in the
manufacturing can be increased easily through capital accumulation,
innovation and economies of scale as compared to services. Evidence also
shows that almost all of the growth miracles in the advanced economies have
been the results of rapid industrialization because of its higher productivity
growth, increasing returns to scale, innovations, and learning by doing.
Page | 35
In the recent periods, however, literature that challenges the role of
industrialization and the manufacturing sector as an engine of economic
growth are emerging. This is due to the change in the service sectors as some
of the service sectors are fulfilling Kaldor’s observation. For instance, the
business, financial, and the communication service sectors have the scope of
achieving economies of scale, high contribution to technological progress and
enhancing overall productivity growth and(Dasgupta & Singh, 2005; Dasgupta
& Singh, 2006). In addition, Hansda (2001) shows the expansionary effect of
the service sector on other sectors through its interlinkage. Banga & Goldar
(2004) also find the increasing use of the service sector has contributed to the
productivity growth in the manufacturing sector and claims that the service
sector will be ‘successful in creating its own demand, which enables the
service sector to sustain its own growth’ (Nayyar, 2012). According to Nayyar
(2012), the service-led growth could be sustained through its effect in the
international trade. In the past decades, the tradability of the service sector has
increased in the international market. Specifically, business services, transport,
communication and financial services are boosting productivity growth
through ‘learning-by-exporting’ effects, ‘scale economies’, ‘technological
transfers’ and competition (Nayyar, 2012).
Apart from the productivity side, the sustainability of service can be

evaluated from its employment role. Naturally, services tend to be labor-
intensive so they play a vital role in generating productive employment
opportunities. Therefore, at a broader level, service sector development can
promote inclusive growth by creating jobs in both traditional and modern
services. According to Noland, Park, & Estrada (2012), the Asian service
growth experience shows that services have been a major source of jobs and
the shift in the economic structure toward services help to reduce poverty and
inequality (Noland, Park, & Estrada, 2012).
Therefore, given the continuous productivity improvements in some of

the service sector, the high contribution of modern services to technological
progress and increasing service share in the international trade, service led
Page | 36
economic growth can be sustainable. In addition, the high inter-sectoral inter-
linkage, and service being a major source of employment growth, it could be
important not only for sustainable development but also for poverty reduction
and equitable growth. Thus, in this age, the service-led growth looks more
likely than in the past, three to four decades ago. However, in this case, the
role of the manufacturing and the agriculture sector in sustaining economic
growth cannot be discredited.
2.1.5 : Sectoral Inter-Linkage of the Service Sector
The literature on inter-sectoral growth in developed and developing

countries has been skewed to the agriculture and the industrial sectors. Studies
on the service sector inter-linkage with the agriculture and the industrial
sectors have been limited. Hirschman (1958) was among the first development
theorists that highlight the backward and forward linkages created in the
industrial sector. Thereafter, Johnston and Mellor (1961) extended the concept
of sectoral inter-linkage beyond the industrial sectors to show the interactions
between agricultural and non-agricultural sectors. The inter-sectoral linkages
between the agriculture, industry and service sectors will determine the course
of structural transformation in many developing countries.
According to Glasmeier & Howland (1994), services as inputs to other

industries, enhance productivity and their presence in the economy stimulates
competitiveness of other industries. For instance, producer services enable
adoption of skills, technology, products and processes by manufacturing firms.
In addition, according to Banga’s (2005a) description, the expansion of service
sector has benefited manufacturing sector. The growth of service sector has
also prompted a growth in demand and production of a variety of
manufactured goods, especially electronics like computers, mobile phones,
digital scanners and much more types of apparatus. Therefore, this type of
close connection between the service and the manufacturing sectors will have
spillover effects in the sectors and to the economy. Noland, Park, & Estrada
(2012) also indicates the importance of the service sector in raising the
productivity of the manufacturing sector and other sectors of the economy.
Page | 37
Particularly this applies to business services that provide key intermediate
inputs such as finance, human resource recruitment, marketing, legal services,
and information technology to manufacturing and other sectors. The evidence
for the growth of such services is the increasing trend of outsourcing tasks
related to business services to specialized service providers, which makes the
manufacturing firms more cost efficient. Therefore, sectoral inter-linkages, in
both the production and consumption sides are important for growth and
growth can be enhanced and sustained by promoting those sectors with larger
inter-sectoral linkages.
2.2 : Empirical Review of Literature
In the previous section, the theoretical literature on service sector

growth and structural change were presented. This section deals with the
review of some of the pertinent empirical literature on the growth of service
sector, sustainability of service-led growth, structural change, the drivers of
service sector growth in output and employment and the inter-linkage of the
sector with other sectors.
2.2.1 : Growth of the service sector
A number of studies have tried to examine the service sector and its
productivity growth after the works of Kaldor (1966), who emphasized that
labor in non-manufacturing sector is less productive. In a recent study, Lee &
Mckibbin (2014) investigated the role of an expanding service sector for
structural adjustment and economic growth in Asia. An employment shift
toward the service sector is observed in 11 Asian economies since 1990. A
decomposition result shows that the service sector has made a significantly
positive contribution to aggregate labor productivity growth, both through
within and structural change effects, exceeding the net contribution of the
manufacturing sector. Despite the strong convergence of labor productivity at
both the aggregate economy level and sectoral levels, there remain significant
differences in labor productivity across economies and across sectors.
Nevertheless, modern services including the transportation, storage, and
communications and the financial and business services have higher
Page | 38
productivity growth, playing a role of a second growth engine in Asia. In
addition, the service trade in Asia, specifically inICT-based services has
increased. Ghani & Kharas (2010) also identified that the global trade in
services has grown faster than the global trade in merchandise goods. The
ratio of service trade in service output for developing countries has increased
much faster than it has for the developed countries. This implies that
developing countries in Asia are more focused on the production of services,
which can be traded, rather than used for domestic consumption.
Eichengreen and Gupta (2010) used the cross-country data from the
European Union (EU) to study the productivity levels for different groups of
services. They divided the service sector into three groups. The first groups are
traditional services, which include wholesale and retail trade, transport and
storage, public administration and defense. The second group, a hybrid of
traditional and modern services, includes hotels and restaurants, education,
healthcare and social work, and other community, social and personal
services. The third groups are modern services that consist the business
services, financial intermediation, computer services, communications, legal
and technical services. They find that the productivity growth of the modern
services is the highest and the fastest growing services are business services,
communications, and banking. The traditional services have a low elasticity
of demand and the hybrid group has a cost-disease problem, which leads to
low productivity.
Park & Shin (2012b), in their research on the Asian service sector
growth, identified that ‘the lower the per capita GDP, the greater the scope for
labor productivity growth in the service sector’. Hence, there is still a lot of
room for services productivity growth in Asia, as the income level of much of
the countries remains relatively low. According Park & Shin (2012b), if a
‘relatively underdeveloped service sector contributes significantly to growth,
then a more developed service sector will contribute even more’. Further, they
suggest that the Asian service sector could generate more jobs and growth if
Page | 39
the obstacles of excessive regulation and state monopolies, barriers to services
trade and FDI are removed.
According to Noland, et al. (2012), some Asian countries, most

particularly India, and to a lesser extent the Philippines have succeeded in
leveraging ICT and other new technologies to boost services exports and
growth. The experience of these countries shows that technological progress
allows the countries to leapfrog industrialization and move straight into the
post-industrial phase. The Philippines and India have skipped the
industrialization phase with weak manufacturing sector in the Philippines and
relatively strong manufacturing sector in India. However, according to
Noland, et al. (2012), the leapfrogging hypothesis could be dangerous if it is
misused as an excuse for the failures in the growth of the manufacturing
sector. While ICT and other new technologies have opened up many new
possibilities for the service sector, especially by improving their tradability, a
good balance between services and manufacturing remains the most viable
growth strategy for Asian countries (Noland, Park, & Estrada, 2012).
Singh (2007) also provided an integrated analysis of the role of the

service sector in the Indian economic development. Based on the econometric
and input-output analysis, it appears that India’s manufacturing sector
development may have been constrained by weaknesses in key service sectors
such as transportation and electricity. Singh (2007) also identified the role of
international trade in services, which is of growing importance.
2.2.2 : The Sustainability and Viability of Service-Led Growth
It is evident from the bulk of literature that industrialization occupies a

central place in the development theory and practices. The belief is that
economic growth through industrialization brings high per capita income and
sustains growth. However, given the increasingly difficulty that most
developing countries face to achieve rapid growth through industrialization,
and especially through export-oriented activities, service sector based rapid
economic growth is seen as the way out. In a recent study, Lee & Mckibbin
(2014) assessed whether enhancing productivity in the service sector can make
Page | 40
the service sector the engine for growth and lead a strong and sustainable
growth in Asia. Using simulations of a multi-sectoral general equilibrium
model, they find that ‘faster productivity growth in the service sector in Asia
can significantly benefit all sectors, contributing to more balanced and
sustainable growth of Asian economies’. In Asia, particularly, in India, the
resilience of the economy is due to the resilience of the services sector.
Numerous studies since the 2000s have shown that services induced growth of
GDP of India is feasible.
Accordingly, Hansda (2001) addressed the sustainability of services-

led growth of the Indian economy in terms of the inter-sectoral linkages using
the input-output transactions tables for 1993/94. Based on the input-output
analysis, the Indian economy is found to be predominantly services-intensive
and the service sector has the largest inducing effect on the economy as per the
Rasmussen measure of both backward and forward linkages. In addition, the
service sector is found to have the largest multiplier effect on the rest of the
economy. Therefore, according to Hansda, the services-led growth can sustain
the overall growth of the Indian economy. However, ‘the expansionary
potential of a services-led growth may not be over emphasized unless
accompanied by growth impulses from other sources’ (Hansda, 2001).
Aggarwal & Kumar (2012) also raised the question whether service
oriented growth of India is sustainable or not? According to their observation,
however, the high informality of the Indian economy in general and the
service sector in particular, is considered to inhibit the sustainability of the
service led growth. Despite the increasing global economic integration of the
Indian economy, the informal economy continues in terms of its share in the
sectoral and total economy, which can constrain the growth due to the low
productivity. Besides, the service-led growth model is sustainable not only
from the economic perspective but from social and environmental perspectives
as well. According to Singh (2012b), service-led growth is sustainable as
services are tradable without any harm to the environment and they can be
easily transported via satellite without any carbon dioxide emissions.
Page | 41
In addition, Dasgupta & Singh (2006) examined the role of
manufacturing and services in sustaining growth using a Kaldorian
framework. Their findings indicate that manufacturing continues to be an
important sector in economic development, but services those connected with
ICT, also make a positive contribution in a number of developing countries
such as India. In India, apart from ICT related services, many other have a
faster rate of growth than either manufacturing or GDP. Dasgupta & Singh
(2006) concluded that services fulfill the requirements of dynamic sectors in
the Kaldor sense and thus, they could be regarded as an additional engine of
growth. Therefore, in the Indian case, according to Dasgupta & Singh (2006),
with the fast development of the country achieved in ICT, the new engine
might help India to leapfrog in technological development to catch up with
advanced countries.
The Indian growth experience in manufacturing and service sector is

different from China. China’s economic growth is more associated with
manufacturing than services. However, the initial phase of China’s growth
after 1979 was driven more by agriculture and services than by industry, but
from about 1990 industry has been the major contributor to China’s growth
(Sheehan, 2008). In this sense, China has experienced a process of
industrialization over the past two decades. In addition, in terms of
employment, the predominant shift has been from agriculture to services
between 1978 and 2006, and the share of industrial employment was flat for
nearly two decades until 2004.
Ghani (2010) has also studied the service growth pattern of the South
Asian countries and examined the sustainability of the growth. According to
him and other co-authors, the conventional knowledge that labor-intensive
manufacturing is the only sustainable way to promote rapid and sustained
growth is challenged. This is because of the global services revolution that has
altered the characteristics of services. As a result of the ICT revolution, wider
varieties and types of new products can be produced and exported at low costs,
where the South Asian countries have taken advantage of these possibilities.
Page | 42
India is leading service exporter as the country has established a clear
comparative advantage in ICT based services. There are also changes that the
service revolution has brought in most of the developing countries. The wages
in the service sector have risen faster than wages in the agriculture and
industry sectors. The service sector encourages greater labor force
participation of women. In addition, the services growth also leads to a rapid
urbanization as people join the urban labor market in search of jobs.
However, there has been a lot of debate about the capacity of the
service sector to generate employment. The change in the output structure
from agriculture to services has not been reflected by an equivalent change in
the employment. As a result, service-led growth has been jobless growth
(Banga, 2005b). This is the major concern for among researchers that limit the
viability of service-led growth in developing countries. There seems an
agreement that the service sector has limited potential for employment
opportunities due to its low employment elasticity and non-uniform growth of
disaggregated services. However, according to Singh (2012a), ‘synergies from
service-led growth pattern can be built through expansion of services in micro,
small and medium enterprises and rural non-farm economy by the promotion
of entrepreneurship’. Furthermore, liberalizing the service sub-sectors of
wholesale and retail trade, financial services and software services will boost
employment and output growth in the service sector through FDI (Mukherjee,
2013).
Historically, the manufacturing sector was considered as the fastest

path to a sustainable economic growth and the way out of poverty.
Nonetheless, there is a number of evidence that this path may be closed to the
developing countries in this age. Thus, some scholars suggested that services
might provide a new path to sustain growth, while others expressed skepticism
about the claim for developing countries. Amirapu & Subramanian (2015)
contribute to this debate by using a multi-sector growth framework of five
criteria that any sector must exhibit in order to lead an economy to rapid,
sustained, and inclusive development. The five criteria includes 1) a high level
Page | 43
of productivity, 2) a dynamic productivity growth (high growth rates coupled
with domestic and international convergence), 3) an expansion of the sector in
terms of its use of inputs, 4) comparative advantage (alignment between
resource requirements of the sector and resource endowments of the country)
and 5) exportability. Based on these criteria Amirapu & Subramanian (2015)
compare the performance of manufacturing against specific service subsectors
of India as a case. They find that the manufacturing sector’s high productivity
and international and domestic convergence in productivity are shared by the
finance, insurance, and real estate service sectors. These subsectors, similar
with the manufacturing, are also too skill intensive and hence unaligned with
India’s comparative advantage (Amirapu & Subramanian, 2015).
Moreover, according to Ghani (2010), the services growth has

contributed more than manufacturing and agricultural growth to poverty
reduction globally and regionally. The cross-country evidence from some
developing countries shows that growth in the service sector is more
correlated to poverty reduction than growth in manufacturing sector. For
instance, in India, some states have experienced a significant decrease in urban
poverty that is associated with the increase in the service sector share.
According to Ghani, the South Asia’s experience of service revolution shows
that globalization of services provide more opportunities for developing
countries to find niches, beyond manufacturing. It also shows the latecomer’s
developing countries that industrialization is not the only route to economic
development. Ghani & O’Connell (2014) concludes that service-led growth
can be sustained due to the globalization of services. Even, the high growth
performers of Africa can sustain service-led growth, as there is enormous
space for catching up and convergence.
2.2.3 : Structural Change, Economic Growth, and the Service Sector
Structural change is a key feature of economic development. As

theories implied and economic history shows, the advanced countries have
followed the path of structural change. Developing countries also possess a
deviated form of structural change, which is also different among the
Page | 44
countries. The empirical literature on structural change and structural
transformation generally focus on the shifts towards industrialization in terms
of gross value addition, employment and productivity. This section, however,
deals with empirical literature on structural change, productivity growth,
drivers of structural change and urbanization associated with the service sector
growth.
Causality of Structural Change and Growth
The ‘New Structural Economics’ presented by Lin (2012) in his book,

postulates that ‘the economic structure of an economy is endogenous to its
factor endowment structure and that sustained economic development is
driven by changes in factor endowments and continuous technological
innovation’. The argument is, hence, the best way for a structural change ‘to
develop industries at any specific time according to the comparative
advantages determined by its given endowment structure at that time’. Yet,
there is a consensus in the literature that in economic development, aggregate
economic growth is accompanied by structural change.
Nevertheless, the question whether economic growth causes structural

change or changes in the economic structure cause aggregate growth is
debatable. Dietrich (2012) has examined the causality between structural
change and growth using a Granger causality test in a panel environment for
seven OECD countries from 1960 to 2004. The result shows that aggregate
economic growth causes a structural change in terms of employment or in
terms of real value added especially for those countries with the largest
economies. Structural change decelerated in the very short run and accelerated
with a lag in time. When structural change is measured in terms of
employment, economic growth has a decelerating impact on structural change,
which seems to be driven by the recession. Whereas, if structural change is
measured in terms of real value added, changes in the sectoral structure of
output is driven by changes in demand due to rising income or productivity
growth differences. Therefore, structural change promotes aggregate economic
growth or, at least, does not decelerate it.
Page | 45
Besides, Aggarwal & Kumar (2012) also attempted to identify the
causality of structural change and growth in India. Accordingly, they find no
significant relationship between growth and structural change during the state-
led growth regime although the causality runs from structural change to
growth in the market driven growth regime. In addition, the finding by Cortuk
& Singh (2013)shows that structural change of the Indian economy is
significant in explaining the growth of the economy for the period of 2000 to
2006 but not vice versa, as growth does not seem to lead to structural change.
They identify a one-way causality from structural change to growth in the
period 1988–2007 but there is no evidence for this linkage before 1988.
Structural Change and Productivity Growth
A number of empirical researches are carried out on structural change

and productivity growth at the country level as well as regional levels. Fan,
Zhang, & Robinson (2003) have studied the contribution of reallocation of
resources across sectors to aggregate growth in China’s during 1978–1995.
Their findings show that about 17 percent of the aggregate growth in China
over the period is due to structural change, which is due to the shift of
resources from lower to higher productivity sectors. This finding is also
supported by a recent study by Chen, Jefferson, & Zhang (2011). Using a
decomposition technique, they found that structural change in China has
contributed to TFP and output growth but decreasing over time. In addition,
according to Fan, Zhang, & Robinson (2003), the inter-sectoral labor
movements have increased the efficiency gain as capital reallocation hindered
efficiency due to policy constraints on capital mobility. The TFP growth of
China is also found to be very high by international standard, contributing 4.2
percentage points to the aggregate annual growth rate.
Several studies have shown that the TFP growth in the modern services
is increasing and becoming the source of structural change. Verma (2012)
studied the structural transformation and TFP growth in services sectors in
India during 1980–2005. She found that the TFP growth was fastest in
services contributing to structural transformation.Duarte & Restuccia (2010)
Page | 46
also investigated the role of sectoral labor productivity in explaining the
process of structural transformation across countries. According to their
observation, productivity differences across countries are large in the
agriculture and services but smaller in the manufacturing sectors. Over time,
the productivity gaps have been substantially reduced in agriculture and
industry but not in services. As countries move through the process of
structural transformation, relative aggregate labor productivity can first
increase when labor moves from agriculture to industry and later stagnate or
decline as labor moves from agriculture and industry to services. The larger
productivity gaps in services relative to manufacturing across countries are
because of a tendency for services to be less competitive pressure as services
are less traded than manufacturing (Duarte & Restuccia, 2010).
Maroto-Sánchez & Cuadrado-Roura (2009) studied the impact of

tertiarization on overall productivity growth, using a sample of 37 OECD
countries in the period between 1980 and 2005. Accordingly, structural change
generally had a positive effect on labor productivity even though the labor
productivity growth is dominated by the within productivity effects. That
means most of the growth in labor productivity has been due to the increase in
productivity within each industry, not due to the reallocation of resources
(structural change). Based on the ‘cost disease’ hypothesis, economic growth
and the overall productivity growth of services would lead to a deceleration.
Consequently, the low productivity growth of many advanced economies has
been associated with the low productivity level of services and the high
incidence of services in these economies. Nevertheless, latest empirical works
show that several service activities have registered higher productivity growth
rates than the manufacturing industries, contributing more to the overall
productivity growth (Maroto-Sanchez & Cuadrado-Roura, 2007; Maroto-
Sánchez & Cuadrado-Roura, 2009).
Timmer & de Vries (2009) also studied structural change and growth
acceleration using a modified shift-share method that accounts the
contribution to growth from the expanding sectors for 19 countries in Asia and
Page | 47
Latin America for the period of 1950-2005. They find that growth
accelerations are explained by productivity increases within sectors, but not by
structural change (reallocation of employment to more productive sectors).
One important finding of their study that challenges the conventional wisdom
is that productivity improvement in market services is found to be more
important than productivity growth in the manufacturing sector.
On the other hand, the study by de Vries et al. (2012) on the structural
transformation and its implications for productivity growth in the BRIC
countries (Brazil, Russia, India, and China) shows that reallocation of labor
across sectors is contributing to aggregate productivity growth for China,
India, and Russia, but not in Brazil. In Brazil, however, the increasing
formalization of the economy since 2000 became growth enhancing. The
employment reallocation towards formal activities increases aggregate growth.
Whereas in India the increase in the informality after the reforms is growth
reducing. For example, in India, the informal sector accounts up to 30 percent
of the manufacturing’s value added and 80 percent share in employment,
which indicates large differences in the productivity between the formal and
informal activities.
McMillan & Rodrik (2011) documented ‘productivity reducing

structural change’ for countries in Latin America and sub-Saharan Africa but
for Asian countries, a growth-enhancing structural change is observed as
workers shift from sectors with below-average productivity into sectors with
above-average productivity. Bosworth & Collins (2008) also found a strong
growth-enhancing structural change in China and India. According to
McMillan & Rodrik (2011), the factors contributing to growth-enhancing
structural change are lower share of natural resources in exports, undervalued
real exchange rates, and labor market flexibility. On the other hand,
developing countries tend to have large productivity gaps between sectors.
The agriculture sector is mostly the sector with the lowest productivity levels,
followed by wholesale and retail trade, construction, and ‘community, social,
personal and government services’ (McMillan and Rodrik, 2011). Yet, over
Page | 48
the course of development and income growth, the inter-sectoral productivity
gaps, as measured by the coefficient of variation of the log of sectoral labor
productivities, tend to decline. This implies that structural change contributes
to convergence within and across countries.
The key findings in the study conducted by Kucera & Roncolato

(2012) show that aggregate labor productivity growth in Asia is driven by as
many services as by industry. The within-sector effects on aggregate labor
productivity growth are more important than employment reallocation effects.
However, there is a significant difference between countries in the roles of
sectors, for instance, the industry is dominant in China and service is dominant
in India. Furthermore, at the aggregate level, they identified a stronger
positive relationship between output and employment growth in developed
countries and a stronger negative relationship between labor productivity and
employment growth in developing countries.
Driving Forces of Structural Change
The process of economic growth is usually characterized by the

significant reallocations of resources out of agriculture or structural change.
Most literature shows that structural change has been driven by productivity
increases, but there is no consensus on whether the technological progress in
agriculture, in manufacturing or in the service sector drives structural change.
Alvarez-cuadrado & Poschke (2011) studied the drivers of the structural
change. According to their result, productivity improvements in the non-
agricultural sector were the main driver of structural change before 1960 and
advancement in the non-agricultural productivity is more important in
countries that are in the early stage of structural transformation. Similarly,
Esteban-Pretel & Sawada (2014) finding shows that the rapid growth of TFP
in the non-agricultural sectors is responsible for the Japanese economic
miracle and the structural transformation of Japan in the post-war period. In
addition, According to Breisinger, et al. (2015), the sources of structural
change and transformation are diverse but include the following major
economic changes. These are technology-led productivity growth driven by
Page | 49
innovation and technology adoption, capital investments especially in the
agriculture sector, changes in consumer demand, and institutional changes in
general and market development in particular.
Dabla-norris, et al.(2013) also documented the stylized facts on the

process of structural transformation around the world for a panel of 168
countries over the period 1970-2010. They find that a large proportion of the
cross-country variation in sector shares is accounted by country characteristics
that include real GDP per capita, demographic structure, and population size.
They find also liberalization of the agriculture sector is negatively related to
the real value added share in agriculture, but positively related to
manufacturing and services share. Reforms in the electricity and
telecommunications sectors are also positive and statistically significant for
the services sector share and negative for the agriculture. Additionally, trade
openness is strongly correlated with industry productivity but less with
services productivity. However, measures of the regulatory environment are
strongly correlated with productivity in services. (Duarte & Restuccia, 2010).
On the other hand, structural change and transformation are also

associated with the structure of export trade. Empirical literature, in this case,
shows that structural change of economies depends on the structure and value
of country’s export. Badibanga, et al. (2009)analyzed the dynamics of
structural transformation as reflected by the structure of export trade. They
find that China’s relatively rapid structural transformation is determined by the
high proximity of its export basket to capital goods, consumer durables, and
intermediate inputs. Malaysia also started its transformation process earlier
than China and achieved industrial clusters with sophisticated export profiles,
which resembled those of advanced economies. In contrast, the transformation
of the Ghanaian economy is far behind in new product content and increasing
value. Agricultural and other primary products dominate the export profile and
the economic structure of Ghana (Badibanga, Diao, Roe, & Somwaru, 2009).
Structural change is also mostly associated with urbanization. One of

the important elements of economic transformation is the movement of people
Page | 50
and resources out of agriculture and rural areas into services and industry in
urban areas (Kolavalli, et al., 2012). Jedwab (2013) documents several stylized
facts regarding the processes of urbanization and structural transformation
using the cross and within country evidence. According to Jedwab (2013),
most developing countries, especially in Asia have urbanized with structural
transformation and resource-rich developing countries, especially in Africa
have experienced urbanization without structural transformation. In Africa,
thus, the processes of urbanization and structural transformation are
disconnected. Contrary to the standard theories of structural transformation,
Africa did not urbanize following a green revolution or an industrial
revolution, but Africa’s urbanization is because of natural resource exports
(Gollin, Jedwab, & Vollrath, 2016).
2.2.4 : Determining Factor of Service Sector Growth
In most of the developing countries, the service sector is leapfrogging

the manufacturing sector. This triggered various empirical studies on the
causes of the shift to the service sector. Various factors can be enumerated as
determining factors that drive the growth of the service sector. As it is
discussed in the theoretical review section, they are generally categorized into
the ‘hierarchy of needs’ hypothesis, the ‘cost disease’ hypothesis, the
‘exogenous demand’ hypothesis and ‘deindustrialization’. Moreover, in this
section, the effect of economic reforms and liberalization will be discussed.
The empirical studies show that the relationship between income and
service sector growth is positive as it is postulated in the ‘hierarchy of needs’
hypothesis. In order for the view to be supported, however, there must be a
positive correlation, in a cross-sectional and time series data between income
and the proportion of the service sector. Baumol (1985), using cross-sectional
data, shows that income per capita and the proportion of the service sector
may be positively correlated on the nominal basis, but that they are not related
on the real PPP basis. According to Schettkat & Yocarini (2006), through the
analysis of input-output tables of major OECD countries, they reported that as
Page | 51
per capita income increases, the proportion of services in final demand has
increased on the constant price basis as well as on the current price basis.
Kongsamut, Rebelo, & Xie (2001) extend the Engel’s law by

postulating that as a household’s income increases, he/she not only spends less
on food but also desires to spend more on services. Fuchs (1968) analysis also
shows that the demand elasticity for services is slightly higher than that for
goods. He also presented evidence showing that the income elasticity of the
demand for goods is influenced by the demand for food. But, excluding food
from the analysis gives income elasticity slightly above one, which is similar
to the income elasticity for services (Fuchs 1968). Apart from private
household consumption, government consumption and exports also contribute
towards explaining the shift to services in final demand. Fuchs (1968, p. 42),
for instance, finds that in the United States, the income elasticity for local
government spending is 1.07 (Fuchs, 1968).
The other determining factor of service sector growth is productivity

differential, known as the ‘cost disease’ hypothesis. The Baumol’s ‘cost-
disease’ hypothesis posits that the share of service employment is mainly
increasing because of productivity growth in services is lagging. A positive
correlation between the expansion of productivity gap between manufacturing
and service industries and an increase in the proportion of the service sector
has been supported empirically by many studies (Fuchs, 1980; Rowthorn &
Ramaswamy, 1999). According to Fuchs (1968) estimates, the service sector
productivity growth is below the manufacturing productivity growth mainly
because skill upgrading has been less visible in services. Fuchs (1968) finding
supports the ‘cost disease hypothesis’, but he finds a minor role for the
demand shifts in the growth of service employment. Some studies analyzed
the changes in the structure of final demand and assume that demand patterns
remain unchanged; others assume that shifts in employment just reflects the
changes in demand. But, according to Clark (1940), the demand shifts are the
major cause of expanding service employment.
Page | 52
Eichengreen & Gupta (2013) also examine the relationship between
service sector size, income, and productivity. They find a positive correlation
between output share of services and income per capita. The relationship holds
only for service activities that are a combination of traditional and modern
services consumed by households (education and health) and for modern
services consumed by both households and businesses. Additionally, their
study shows that modern services not only have the highest productivity
growth among the service activities, but their share in output increases rapidly
at high-income levels.
Gani & Clemes (2002) also studied the determinants of per capita
economic growth particularly; the contribution of services to the growth of
real GDP per capita in Asia using data from 1965 to 1994. The result confirms
a strong positive effect of growth in manufacturing and government spending
on service sector expansion. According to them, services have a spillover
effect on manufacturing and the service sector contributes directly to GDP and
per capita income through increased production of services. Further, it also
contributes to GDP and income indirectly through its impact on raising the
productivity of other sectors and enhancing the efficiency of resource
allocation.
The literature on the change in the structure of demand for services,

associated with the ‘exogenous demand shock’ hypothesis, asserts that service
sector growth is derived by the change in the demand for intermediate service
inputs for the manufacturing sector and consumption expenditure for services
as a final demand. In addition, the increase in female participation in economic
activities is also linked with the increase in service employment (Fuchs, 1980).
In the US 31 percent of the increase in the proportion of services in total
employment occurred between 1929 and 1965, which can be explained by
exogenous demand shocks, and the figure rose to 69 percent for the period in
1966-1981. Likewise, according to Eckstein & Heien (1985), the dramatic
increase in the services employment in the U.S. economy since World War II
is primarily due to the dramatic increase in government expenditures at the
Page | 53
state and local levels on services and for Medicare. In addition, the growth in
the private components of final demand is almost equally important with the
increase in government expenditure. Eckstein & Heien (1985) believe the
evidence that services have higher income elasticity than goods is not definite
unless government expenditures as part of the demand are not considered.
However, if government expenditures are included, the income elasticity for
services exceeds unity. In addition, ‘technological progress and the availability
of highly skilled workforce have led to the growth of services in ICT and ICT-
enabled services’ and due to this developed countries are outsourcing services
to developing countries like India, leading to a rise in demand for services
(Bhagwati, 1984a).
On the other hand, services are increasingly integrated into the

manufacturing activities and used as intermediate input facilitating production
and marketing of goods. According to Pilat & Wölfl (2005), the
manufacturing activities are becoming more and more service intensive in
both upstream (design, research, and development) and downstream
(marketing and advertising). Due to the importance of services, the
competitive advantage of firms in this age depends more on providing
specialized services like financing and after sales facilities than on production
as it is reflected by the increased demand for intermediate services. In
developing economies, the rise in service inputs into manufacturing has been
confirmed by some empirical studies. For instance, Gordon and Gupta (2004)
measured the increasing usage of services in other sectors through changes in
the input-output coefficients. Their result shows that the use of services sector
as an input to industry increased by about 40 percent in between 1979/80 to
1993/94 in the Indian economy. In addition, Banga & Goldar (2004) have also
estimated the increased usage of services by the manufacturing sector in the
pre and post-reform periods. They estimated the contribution of services
empirically as an input to the manufacturing output growth using the KLEMS
(capital- labor-energy-materials-services) production function. According to
their result, the increase in the use of services has a ‘significant favorable
effect on the growth of output in Indian manufacturing’ in the 1990s, The
Page | 54
contribution of the service input to output growth in manufacturing was about
1 percent in the 1980s, and it increased to about 25 percent in the 1990s.
The determining factors to the shift in the service sector in both output
and employment are studied more comprehensively by Kim (2006). Kim
studied the factors that cause a shift towards services in Korean economy
using regression analysis. He tested the hypotheses of ‘hierarchy of needs’,
‘cost disease’ and ‘exogenous demand shock’. The result of the estimation
shows that the share of the service sector in employment is significantly
associated with inter-sectoral productivity differential and the share of
producer services. Female economic participation ratio also became
significant for the period 1981-2003. However, the share of producer services
is the most significant factor accounting for the rise in the share of the service
sector in total value addition. On the other hand, per capita real value added is
insignificant and shows negative sign in contrast to the ‘hierarchy of needs’
hypothesis. According to Kim (2006), the increase of the service sector's share
in employment is mainly caused by the productivity gap between the
manufacturing and service sectors.
Apart from the major determining factors of service sector growth that
are discussed in the previous paragraphs, economic reforms and liberalization
are other important factors that a number of empirical literature has studied.
Existing empirical literature shows that liberalization and reforms have
contributed to the growth of the service sector (Jain & Ninan, 2010).Moreover,
service liberalization will have positive effects on performance of
manufacturing sector. Thus, due to the increasing service intensity of
manufacturing, one can conclude that service trade barriers not only affect the
service sector but also manufacturing firms (Falk & Jarocinska, 2010).
The existing empirical literature on trade liberalization and service

sector growth largely focuses on the financial sector. However, some studies
estimate the impact of trade liberalization on the growth of services. For
instance, El Khoury & Savvides (2006) measure the effect of openness in
telecommunication and financial services on economic growth for countries at
Page | 55
different stages of economic development. Their estimates confirm that
telecommunication services openness is positive indicating its positive impact
on growth for low-income economies but the opposite is true for financial
services. According to El Khoury & Savvides (2006), liberalizing
telecommunication services trade by lower income economies frequently
produce FDI that contributes to raising productivity growth in the sector.
Moreover, lower income countries do not have the necessary institutional or
regulatory structures to ensure the effective functioning of the financial sector,
which cannot be solved through the greater openness of the financial sectors.
Gordon and Gupta (2004) estimated sector specific liberalization

dummies in a panel data regressions for trade, hotels, transport,
communications, insurance and other services for the period 1970-2000.
According to Gordon and Gupta (2004), sectors that were open for FDI,
external trade, or private ownership experienced faster growth. Banga &
Goldar (2004) also use multiple regression analysis and show that trade
reforms carried out in the 1990s in India explain the rapid growth of the use of
services in manufacturing. Lower tariff and lower non-tariff barriers were also
found to have led to an increase in the usage of services in the manufacturing
sector. Along with liberalization, improved technology has also led to higher
use of services with the expansion of electronics, information and
communications technologies in developing countries. In addition, Mattoo,
Rathindran, & Subramanian (2006)provides econometric evidence that shows
the relatively strong effect of openness in financial services and less strong
influence of telecommunications sector on long run growth. Their estimates
suggest that countries with fully open telecom and financial sectors grow up to
1.5 percentage points faster than other countries.
Contrary to studies that show positive effect of liberalization on service

growth, Cattaneo (2011) argues that generalized services trade liberalization is
not the appropriate strategy for the services sector either in South Africa or in
developing countries. According to Cattaneo (2011), although efficient and
reliable services are needed for industrialization, services trade liberalization
Page | 56
in GATS or in North–South trade agreements is not necessarily the best way
to improve services sector efficiency and exploit the sector’s growth and
employment potential. However, a number of studies confirm the importance
of liberalization and trade openness as one important channel for improving
services performance through adoption of new technologies and competitive
environment created by foreign suppliers. According to Hoekman & Mattoo
(2008), the competitiveness of firms in open economies is increasingly
determined by access to low-cost and high-quality producer services such as
telecommunications, transport and distribution services, financial
intermediation, etc.
Fernandes (2009) examined the performance of the service sector in

the Eastern European transition economies during the 1997–2004 periods. The
estimates show a positive and significant effect of liberalization on service
labor productivity growth and growth of downstream manufacturing
industries. Fernandes (2009) suggested a potential for service-driven
productivity growth if policy-makers pursue further liberalization efforts by
removing the product market barriers in various service sub-sectors and by
attracting more FDI. Similar to other findings, Fernandes also finds that
services liberalization is likely to benefit the performance of the
manufacturing sector. Furthermore, reductions in the restrictiveness of product
market regulations in services seem to benefit more the traditional service sub-
sectors that are lagging in the technological frontier, which can help the
convergence of the transition economies towards the advanced European
Union countries.
On the other hand, Estrada, et al. (2013) studied the service sector
growth and factors determining the growth in low-income Asian countries.
Accordingly, service sector growth has been supported by strong industrial
growth in some countries while in others the critical factors have been
liberalization, structural reforms, government support, and foreign
investments. According to Estrada, et al. (2013), in Viet Nam, services
liberalization led to the rapid expansion of retail trade, telecommunications,
Page | 57
and transportation industries. In addition, Viet Nam had higher FDI in services
due to the country’s entry into the World Trade Organization in 2007 and
several bilateral trade agreements. On the other hand, in Papua New Guinea,
the removal of the state monopoly in the mobile telephone market, greater
competition in the aviation industry, and structural reforms in the financial
sector contributed to the growth in service output and employment. Generally,
based on the experience of the Asian low-income countries, policy reforms
that ease trade in service will help to achieve a balanced growth in which the
service and industry sectors support and reinforce each other. According to
Estrada, et al. (2013), for most of the developing Asia, services that are more
labor-intensive than industry, can foster inclusive growth and serve as an
engine of growth.
2.2.5 : The Sectoral Inter-linkages of the Service Sector
Linkage analysis, which is used to examine the interdependence in

production structures, has a long history within the field of input-output
analysis. Various studies are also conducted on sectoral inter-linkages of an
economy. For instance, Azad (1999) studied the direct and indirect
contribution made by services in Bangladesh. The result shows that the service
sector contributed most to the growth of GDP in Bangladesh as its share
increased from 39 percent of GDP in 1976/77 to 62 percent in 1997/98.
Services are used widely as inputs in commodity and service production, thus
service contents are the largest per unit of output. This means the industrial
production will not only require increased supply of inputs but a cheaper and
better supply of services, which reduces per unit costs of production.
Furthermore, the service production itself requires services as inputs, which is
the largest user of services as inputs in the Bangladesh economy. The inter-
industry linkages of services have revealed that electricity and gas, transport,
communication, and public administration create greater backward linkages
whereas trade, banking, and insurance create larger forward linkages. A
regression analysis also confirms strong inter-industry linkages of services
with the commodity producing sectors.
Page | 58
Hansda (2001) also used the input-output approach at a much more
disaggregated level with a goal of addressing the sustainability of services-led
growth for the Indian economy. His analysis suggests that the Indian
economy, in general, is services-intensive with the industry being the most
services-intensive sector. The manufacturing sector, construction, and utilities
are all services intensive whereas transport, storage and communication are
industry intensive. However, he finds that the level of linkages of the
agriculture sector, financing, insurance and real estate are quite low.
Banga & Goldar (2004) also used input-output inter-linkage analysis to

estimate the impact of services input on productivity growth of manufacturing
sector in India by constructing a multilateral total factor productivity index for
major industry groups for the period 1980/81-1999/00. According to their
estimation, there is a positive relationship between use of services input and
industrial productivity. This means a higher use of services has led to greater
specialization and improved productivity in the manufacturing sector.
Therefore, according to Banga & Goldar (2004), the dual spill over effects of
growth in the manufacturing and service sectors puts the Indian economy on a
higher growth trajectory. Noland, Park, & Estrada (2012) also confirm the
same result recently. According to Noland, Park, & Estrada (2012), a more
productive service sector like ICT and transportation has a positive spillover
effect on manufacturing and the rest of the economy. In addition, a study by
Pilat & Wölfl (2005) indicates services key contributions to production
through the total output, final demand, and provision of intermediate inputs.
The high interdependence between services and manufacturing also can be
seen in the rising amount of service sector value added, which is embodied in
manufacturing goods.
On the other hand, Falk & Jarocinska (2010) studied the direct and
indirect linkages between manufacturing and services both on the input side
and on the output side for the EU-25 countries and the US. They find that
direct and backward linkages between manufacturing and services increased in
the last few decades with the change more substantial for the backward
Page | 59
linkages. Similarly, there is an increase in service occupations in
manufacturing. On the output side, manufacturing firms generate an increasing
proportion of sales from services activities.
Various studies have also utilized backward and forward interlinkage

analysis to identify key sectors as well as those sectors with high interlinkage
requiring more investment for sustained economic growth. For instance,
Kolavalli, et al. (2012) conducted a quantitative analysis of production
linkages using the Ghana 2007 Social Accounting Matrix (SAM). The result
shows that manufacturing and its subcomponent, such as agro-processing, has
the strongest backward linkages. Ghana’s agricultural sector, on the other
hand, has the weakest backward linkages to the rest of economy and
particularly to the industrial sector. Similarly, Tregenna (2008) investigates
the inter-sectoral linkages in the South African economy using input-output
tables. The estimate shows that manufacturing is found to be important as a
source of demand for the services sector and the rest of the economy through
its strong backward linkages. In addition, the result shows that the share of
manufacturing in GDP has been declining slightly over time in South Africa,
whereas that of services has been growing.
Andreosso-O'Callaghan & Yue (2004) also made a comparative

analysis of traditional and modern methods of inter-linkage analysis to
examine the interdependence in production structures and in identifying key
sectors in the case of China between 1987 and 1997. They find that backward
and forward linkages have generally increased in China, indicating an increase
in inter-sectoral interdependence. According to Andreosso-O'Callaghan & Yue
(2004), with regard to the identification of key sectors, they find agriculture,
textiles, chemicals, building materials, machinery, commerce and other
services as being the key sectors by both the total linkage and pure linkage
methods. Therefore, the various empirical studies have used backward and
forward inter-linkage analysis to show the sectoral inter-linkages of the sector.
In most of the studies, the service sector has shown strong inter-linkage with
the manufacturing sector as it is used an intermediate input in the production
Page | 60
process. The next section discusses the empirical literature on the specific
experiences of Asian and African countries in structural change and service
sector growth.
2.2.6 : Experience of Developing Countries in Structural Change and

Service Sector Growth
In the literature, structural change and structural transformation are
usually linked to development or economic growth. There is a strong
association between changes in the economic structure and level of economic
growth in both developing and developed countries. Recent studies have also
documented that, structural change can also happen during the periods of
economic stagnation and decline. According to Bah (2010), as a result, there is
no systematic link between structural transformation and economic growth. He
explains that the nature of the relationship between GDP per capita and the
sectoral shares is misleading as the average GDP per capita is sometimes
decreasing. For instance, for African countries, GDP per capita in 2000 was
the same as what it was in 1965; yet they experienced structural
transformation during the period 1965-2000, While on average Latin
American countries followed the path of the developed countries. In Asia,
first, there was a steady decline in the agricultural output shares and after GDP
per capita passes $2000, there were steady increases in the shares of industry
and services. On the other hand, in Africa, the share of agriculture in output is
low whereas its share in total employment is very high implying that it is
unproductive. The share of services is also high in both Africa and Latin
America because of its low level of productivity. This is also associated with
the informality of the distributive trade service sub-sector, which is the biggest
service sub-sector. Thus, the high level of informality and the lack of skills in
this service make it unproductive.
Dabla-Norris, et al. (2013) presented the experience of structurally

transformed countries and that of developing countries striving to bring
structural change. They document stylized facts on the process of structural
transformation around the world. On the services side, their result suggests
that reforms in the distribution sectors (telecommunication and electricity),
Page | 61
strengthening human capital and greater labor market flexibility, particularly
in countries with high services shares, can have a significant positive impact
on productivity growth in the sector.
On the other hand, Timmer, de Vries, & de Vries (2014) used the
updated and extended Groningen Growth and Development Centre (GGDC)
10-Sector database to document patterns of structural change in 11 countries
in Asia, 9 in Latin America and 11 in Sub-Saharan Africa from 1950 onwards.
They find that the expansion of manufacturing activities during the early post-
World War II period was related to a growth-enhancing reallocation of
resources in most countries in Asia, Africa, and Latin America. But the
process of structural change stalled in many African and Latin American
countries during the mid-1970s and 1980s. In the 1990s, when growth
improved, workers mainly shifted to the market services specifically to
distributive trade services. However, such services have higher productivity
than much of the agriculture; they are not technologically dynamic and have
been falling behind the world frontier (Timmer, de Vries, & de Vries, 2014).
The Asian Experience
The service sector has been a major source of economic growth in

South Asia, but the extent of its contribution varies widely across the
countries. The share of the service sector in GDP has grown considerably;
particularly it increased in India, Pakistan, and Sri Lanka. The share of
services in total employment also has expanded in the larger economies. The
growth of services in output has been accompanied by improved labor
productivity rather than employment gains. The productivity differences
between industry, services, and agriculture are particularly large for India. In
addition, from the South Asian countries, trade in services is considerable only
for India. India’s exports are dominated by sales of computer and business
services. In recent years, FDI flows to the region have also increased at a rapid
pace, mainly due to increased investment toward services in India (Bosworth
& Maertens, 2010).
Page | 62
According to Noland, Park, & Estrada (2012), services’ contribution to
growth has been higher in the South Asian region than in other regions. In
India, the Maldives, and Sri Lanka, roughly over 60 percent of the growth in
2000-2010 was due to services. The sector also contributed to over half of the
growth in Indonesia, Malaysia, the Philippines, and Singapore. Yet, in East
Asia, particularly the Peoples Republic of China, the Republic of Korea, and
Taipei China, the industry rather than the service sector is driving the overall
growth. In most of the South Asian countries and the Philippines, where the
pace of industrialization has been slow, the service sector especially the
modern service sector has played an important role in driving the overall
growth.
In addition, Estrada, et al. (2013) also confirm that the service sector
has been a huge contributor to growth more than industry and agriculture from
2000 to 2010 across the lower-income economies of Asia including
Bangladesh, Cambodia, Nepal, and Uzbekistan, services contributed. When
compared to the 1990s, growth proceeded more rapidly from 2000 to 2010 in
these countries with the service sector as the key driver. However, a key issue
is whether growth in the service sector has led to significant job creation and
to improvements in productivity. In terms of employment, lower-income
economies are still primarily agricultural, but there is some evidence that
services have contributed substantially to employment growth in the past
decade. In Bangladesh, services contributed to about one-third of the
employment growth from 2000 to 2010, which was more than industry did.
Therefore, services will continue to make substantial contributions in lower-
income Asian countries as the share of the output rises with per capita income.
Since the income level of the developing Asia’s lower-income countries is
low, they have relatively wider possibility for labor productivity growth and
hence growth in the service sector as their income increases through time. In
addition, these countries are still in the early stages of transforming, so that the
share of agriculture in GDP is still relatively high and the shares of industry
and services are correspondingly low. Thus, the sectors have sufficient room
to grow in the future (Estrada, et al., 2013).
Page | 63
There are also country-specific studies in Asia predominantly that
focuses on India, China, and South Korea. To start with India, Ramaswamy &
Agrawal (2012)studied employment growth, structure and job quality
outcomes in urban India between 1999/2000 and 2009/10.Their result did not
indicate any acceleration in employment growth of the service sector relative
to manufacturing in the urban areas of India. Rather they find greater duality
in the services sector in terms of the incidence of informality and wage
inequality. The service sector is found to be relatively more skill demanding
than manufacturing. The service sector is also skill-biased as those with more
skills have received higher increases in real wage than the wages in the
manufacturing. Therefore, the service sector will not be the destination for the
millions of low-skilled job seekers. Thus, they suggested that India needs to
focus on manufacturing sector to provide large-scale employment. However,
according to Eichengreen & Gupta (2011), the growth in the service sector
output and employment in India depends on the continued expansion of
modern services and the application of modern information technology to the
traditional services.
According to Das, et al, (2013), labor productivity in Indian service

sector has been growing substantially over decades, and much of this
productivity gain emanates from the acceleration of labor productivity in the
market services and ICT-intensive services. In addition, they found that the
labor reallocation effect in India is positive during 1980-2009. It has increased
in 2000-2009, suggesting a growth-enhancing structural transformation. They
also examined the dynamics of total factor productivity in the service sector,
using KLEMS growth accounting framework. They find a remarkable
performance of market-basedICT-intensive sectors especially in
telecommunications and financial services.
Among the few studies that investigated the service sector growth and
structural change in China, Qin (2006)explores whether the growing service
sector in China leads to cost disease, which is a likely consequence of
tertiarization according to Baumol’s unbalanced growth model. According to
Page | 64
the model, tertiarization is likely to decelerate overall productivity
improvement of an economy and hence retard economic growth. According to
Qin (2006), the positive contribution of the service sector to growth in China
is largely due to shifts of labor from the primary sector into services.
However, the signs of cost disease are visible from the weak responses to price
signals in the demand for services, in wage determination and labor input
demand of the service sector.
On the other hand, Chen, Jefferson, & Zhang (2011)estimated the

stochastic frontier sectoral production function for China's industry. According
to them, China's industry has experienced robust growth under persistent
structural reform since 1978. The TFP growth has also exceeded the
quantitative growth of inputs since 1992, but the contribution of productivity
to output growth declines after 2001. Using a decomposition technique, they
identified a substantial contribution to structural change to TFP and output
growth though it is decreasing over time. Therefore, the empirical analysis
reveals that the reforms in factor markets and sectoral structure significantly
explain the overall trend and the sectoral heterogeneity of factor allocative
efficiency during the industrial transformation process (Chen, Jefferson, &
Zhang, 2011).
There is also a comparative study conducted by Wu (2007) that

compares service sector growth in China and India. Both countries witnessed a
phenomenal economic growth since the 1990s in the world economy. The
average annual growth for China and India during 1992–2005 was 10.2
percent and 6.2 percent respectively. The comparison of China’s and India’s
economic structures reveals that the role of the service sector is very different.
The service sector has become the dominant contributor to the Indian
economy, accounting for 54.2 percent of GDP in 2004. However, in China, the
service sector has lagged well behind the manufacturing sector, though its role
in the economy improved slightly. From 1990 to 2004, the service sector share
of China’s GDP increased from 34.3 percent in 1990 to 40.7 percent in 2004.
The two countries have taken different paths in developing their service
Page | 65
economies. Wu (2007) found that the role of services in both China and India
has been rising, with China starting from a lower base. Growth has been
driven in both societies mainly by increasing specialization of production,
rising living standards and accelerated urbanization. There are also some non-
economic factors, which are difficult to measure in empirical analysis but have
played an important role in service development including biased development
strategies in China.
South Korea is the other Asian growth miracles that achieved and
sustain high output growth for many years. South Korea fits very well the
structural transformation process accompanying economic development as
described by Kuznets (Bah, 2010). According to Park & Shin (2012a), the
core problem of the country is associated with the underperforming of
productivity growth in the service sector, although the share of services in
output and employment has risen. The analysis by Park & Shin (2012a) clearly
confirms the belief that Korea’s services sector still lags the manufacturing
sector even though deindustrialization already began in the early 1990s.
Therefore, the central challenge for the South Korea in the post-industrial
phase is thus to renovate and upgrade the services sector so that a productive,
high value-added, modern services sector can become an engine of growth.
Park & Shin (2012a) argue that ‘the highly educated workforce of Korea,
which enabled the country to quickly move up the technological ladder, can in
principle also serve as a key ingredient in the leveling up of the services
sector’ (Park & Shin, 2012a).
Singh & Singh (2013) also conducted a comparative study of service

sector growth in India and South Korea. In India, the service sector is an
engine for the GDP growth rate but in South Korea, industry followed by
service sector is leading the GDP growth. In the South Korean economy, the
employment share of the service sector has grown faster than the GDP share in
the past two decades. However, in India, the service sector has failed to play
any significant role in employment generation. In India, in terms of
employment, the primary sector remains the largest employer. The tertiary
Page | 66
sector's share in GDP has increased but it has not been able to displace the
labor from primary sector. Indian primary sector is depicting a greater
backward integration with itself, in comparison to South Korea, whereas, the
South Korean service sector is well integrated with other sectors on the
backward side. On the other hand, the ratio of service exports as a percentage
of total exports in India is not only higher than South Korea but it has also
been growing at a much faster pace from 20.2 percent in 1990 to 35.5 percent
in 2010 (Singh & Singh, 2013).
The African Experience
African economies generally experienced a decreasing share of

agriculture with a stagnant share on manufacturing and increasing the share of
the service sector. However, when it comes to structural change, only a few
African countries have recorded sustained economic growths and structural
changes. Some countries have joined the middle-income country status with a
structural change while others joined without structural change in their
economies. Most of the countries that joined the middle-income group without
securing structural change are resource rich countries like Angola, Botswana,
Zambia, and Equatorial Guinea. Whereas, other countries like Cape Verde has
joined the middle-income countries because of the growth in the tourism
services. The issue of African structural change and transformation has
attracted the attention of a number of researchers and bulk of literature exists
on the structural transformation of Africa. However, the literature on the
service sector growth in Africa is very scanty as compared to the Asia. This is
because of the service revolution in Africa, which is just starting as compared
to the Asian countries. Some of the empirical literature on structural
change/transformation and experience of service sector growth of the African
countries are discussed subsequently.
De Vries, Timmer, & de Vries (2013) studied recently about the

structural transformation in Africa and its implications for productivity growth
for 11 Sub-Saharan African countries during 1960 to 2010 by extending the
work of McMillan & Rodrik (2011). According to de Vries, Timmer, & de
Page | 67
Vries (2013), the expansion of manufacturing activities during the early post-
independence period in Africa led to a growth enhancing reallocation of
resources. The process of structural change held up in the mid-1970s and
1980s, but in the 1990s, labor mainly shifted to the market services. In
addition, the market services activities had above-average productivity levels,
but productivity growth was low and increasingly falling behind the world
frontier. Therefore, the pattern of static gains but dynamic losses of
reallocation since 1990 are observed for many African countries. The situation
is comparable to the patterns observed in Latin America but different from
those in Asia. According to McMillan and Rodrik (2011), the structural
change did not contribute to growth in Africa, despite Africa’s high growth
performance. They also show that during the period from 1960 to 1975, Africa
took a step forward by expanding its manufacturing activities, which was
related to growth-enhancing structural change. However, after 1990 market
services activities expanded and the productivity levels in market services
were above the economy average. The overall effect was a limited role for
structural change post-1990, which compares unfavorably to Africa’s earlier
period of high growth, which is also observed in Latin America, but not in
Asia.
In another similar study, McMillan, Rodrik, & Verduzco-Gallo (2014)

find that on the average structural change in Africa during 1990–99 was
growth reducing but since 2000 labor has moved from low to high
productivity activities, contributing to Africa’s growth by 1 percentage point
per annum. They also identify factors that help determine the extent to which
structural change contributes to overall productivity growth. Accordingly,
countries with a relatively large share of natural resources in exports
experienced growth-reducing structural change. On the other hand,
competitive or undervalued exchange rates and labor market flexibility have
contributed to the growth-enhancing structural change. In addition, rural-to-
urban migration accompanies structural change as relatively poor rural
dwellers leave rural areas for opportunities for income generation in urban
areas (Mcmillan & Headey, 2014).
Page | 68
In this respect, rural-urban migration is one of the necessary
components of the structural transformation and economic development
process. The migration of labor out of agriculture into the manufacturing and
service sectors has been a feature of the growth path of every country that has
developed. According to de Brauw, Mueller, & Lee (2014), the lack of more
rapid rural–urban migration and large shares of the population living in rural
areas may hinder economic growth in Sub-Saharan Africa. The migration rates
in Sub-Saharan Africa are very low due to different factors. Some of these
factors are the explicit policies in some countries designed to inhibit rural-
urban migration, policies that indirectly limit mobility (weak property rights
over land), and lack of education, which appears to be an important
determinant of rural–urban migration (de Brauw, Mueller, & Lee, 2014).
Christiaensen & Todo (2013) attempted to unveil the driving forces for
the rapid reductions in poverty in developing countries given limited growth-
enhancing structural change and weak rural – urban migration using a cross-
country panel data from household surveys for developing countries during
1980–2004. Particularly in Africa, structural transformation and urbanization
patterns are different. In some countries, the migration is out of agriculture
into rural off-farm activities and secondary towns, whereas in other countries
there is rapid agglomeration in mega cities. The evidence from Christiaensen
& Todo (2013) suggests that migration out of agriculture into the missing
middle, which is to the rural nonfarm economy and secondary towns, is
strongly associated with poverty reduction than migration/agglomeration to
megacities.
On the other hand, Dorosh & Thurlow (2014) developed dynamic

economy-wide models for Ethiopia and Uganda that capture the growth
linkages, foreign trade, and benefits from internal migration and urban
agglomeration effects. Their simulation results suggest that urban
agglomeration is an important source of long-term growth and structural
transformation, but investing in cities does not greatly reduce national poverty
over the short-term. In addition, agricultural growth is found to be more
Page | 69
effective, despite slower national growth. Therefore, poverty reduction in the
short-term of necessitates further agricultural investment. Based on these
findings, there is a need to understand the pattern of urbanization in Africa to
exploit the untapped potential of economic gain from the structural
transformation in the long-term and poverty reduction in the short-term.
Barrett, et al. (2017) argues that ending extreme poverty in Africa will require
reallocation of resources from low to high productivity sectors (structural
change). Aryeetey & Moyo (2012) also stressed the huge opportunity of
African countries for significant gains from the structural change in the future,
as the productivity growth due to structural change in 1990 – 2005 was
negative even if there was significant growth within the sectors. However, for
Ajakaiye & Page (2011), breaking into global markets for manufacturing,
agro-processing industry and tradable services are the only viable path to
accelerated structural change in Africa.
When the issue of service trade is raised as a potential to structural

change, Africa is lagging behind Asia. Nevertheless, the service sector is
growing like anything in most of the African countries and most of these
countries skipped industrialization. According to UNCTAD (2015), during the
period 2000–2012, the services sector contributed an increasing share to gross
domestic product (GDP), trade and employment. From the period 2001–2004
to the period 2009–2012, the share of services in African output rose from 45
percent to 49 percent. In addition, 21 African countries had a share of services
in output greater than 50 percent. Among these, Seychelles was the most
service dominant economy (80 percent) followed by Djibouti, Mauritius and
South Africa. The sector accounts for 32.4 percent of total employment in
Africa during the period 2009–2012. African services are predominantly
informal and small in scale. Informal firms dominate the wholesale trade and
retail trade, restaurants, and transportation. The informal sector ranges from 50
to 80 percent of GDP and 60 to 80 percent of total employment and accounts
for 90 percent of new jobs in Africa.
Page | 70
In addition, Africa is a marginal player in global services exports and
imports. In 2012, it accounted for only 2.2 percent of the world’s total exports
of services compared to 3.6 per cent for developing America and 24.3 per cent
for developing Asia. At the country level, Egypt and South Africa are Africa’s
two main global service exporters, accounting for 0.49 percent and 0.34
percent of world services exports. On the other hand, Angola, Nigeria, and
South Africa are the main African global service importers, accounting for
more than 0.40 percent of total world services imports. Even though Comoros,
Ethiopia, and Liberia were classified as services exports-dependent during
2009–2012, the share of services in output in these countries did not exceed 50
percent (UNCTAD, 2015).
According to UNCTAD (2015), during 2001–2004 and 2009–2012, of

the 45 countries in Africa where the share of services in output rose, 30
experienced a contraction in the manufacturing sector. The competitiveness of
African manufacturers is negatively impacted by high indirect costs related to
infrastructure services. The overall level of profitability of African firms,
which is much lower than elsewhere, results from high indirect costs related to
infrastructure and public services such as energy, transport, communications,
water, and security. Therefore, the service sector has the potential to contribute
to structural transformation, economic growth, and development in Africa if
informalities in the service sector are tackled by generating more jobs that are
formal. Thus, African countries may exploit the potential complementarities
and linkages between the services sector and other productive sectors of the
economy.
Recently, Enache, Ghani, & O’Connell (2016) assessed the historical

patterns of structural transformation and productivity convergence in Africa at
disaggregated sectors level as countries across the income distribution are
uniformly increasing the share of labor in service sectors. They also examined
the sector that can bring structural transformation and contributes most to
rapid and sustained growth using a scorecard approach lent from Amirapu &
Subramanian (2015). Their analysis identifies two sectors among African
Page | 71
economies that share at least some of the characteristics to lead a rapid and
sustained growth. According to Enache, Ghani, & O’Connell (2016), these
sectors are the wholesale and retail trade services and the construction sector.
However, both sectors suffer from a lack of tradability and exportability.
Another important finding from the study by Enache, Ghani, &

O’Connell (2016) is that in Africa, at least in recent decades, convergent
pressures in the services industries are stronger than in the manufacturing. In
the decades prior to the 1990s, some countries experienced an early transition
of labor out of agriculture and the manufacturing sector absorbing this labor.
Another group of countries experienced a late transition out of agriculture in
the 1990s and 2000s. In these countries, the services sector played a large role
in labor reallocations. This indicates that African structural change is unlike
the pattern exhibited by East Asian and European economies, as labor is
largely shifting towards the services. Moreover, the manufacturing sector in
Africa is dominated by informal firms, which are less productive than the
formally organized firms (Enache, Ghani, & O’Connell, 2016).
Furthermore, Enache, Ghani, & O’Connell (2016) have also identified

two broad growth patterns among African countries. The first pattern is early
transition or shift of labor out of agriculture in 1970s/1980s. African countries
that experienced this pattern are Botswana, Nigeria, South Africa, Mauritius,
and Egypt. In addition, the pace of structural change in these countries has
slow downed in the 1990s and 2000s. The second pattern involves the late
transition of labor out of agriculture, the pace of structural change not slow
downed in 1990s/2000s and wholesale & retail trade remain the main sector
absorbing labor in the 2000s. These countries are Zambia, Tanzania, Ethiopia,
Senegal, Malawi, Ghana, and Kenya. They also look to identify other
countries that have historical patterns of structural change that are similar to
those of the two groups of African countries. For the first pattern, the early
movers out of agriculture that experienced a relative slowdown in the pace of
structural change in recent decades include Mexico, Taiwan, and China. Other
Page | 72
countries that fit the second pattern, late transitions out of agriculture and high
employment in wholesale & retail trade services, are Thailand and India.
There are also various country level studies on structural

change/transformation and service sector growth in Africa, though studies on
service sector are very limited. Among some of the studies, the study by
Breisinger, Diao, & Thurlow (2015) is one and recent, which evaluated how
accelerated growth in Ghana's service sectors can contribute to the country's
goal of achieving middle-income country status. Wholesale, retail trade and
transport services in Ghana are more labor-intensive sectors whereas the
finance and communications services are capital-intensive sectors. Between
1994 and 2010, Ghana’s service sector grew most rapidly in GDP, with an
annual growth rate of 5.9 percent, which is also dominated by public services
and domestic-oriented private services (Kolavalli, et al., 2012). According to
Breisinger, Diao, & Thurlow (2015), services affect the rest of the economy
by lowering the production costs of other economic sectors through
improvements in service sector productivity. The inter-sectoral growth linkage
result shows that the linkage effect is more evident for the non-agriculture
related manufacturing, in which trade and transport services account for 12.4
percent of total production costs. Based on such intersectoral linkage effects,
the industrial growth accelerates from 5.9 percent to 6.4 percent per year
without an additional exogenous increase in the sector's productivity. On the
other hand, lowered service prices stimulate growth in the non-service sectors,
offsetting the increase in the service sector's size in GDP measured in current
prices. Kolavalli, et al., (2012)also viewed Ghana as a benchmark for other
African countries that are seeking transformation due to the consistent
economic growth in Ghana, (ongoing for almost 30 years), and the rapid
urbanization in recent years.
On the other hand, Uwitonze & Heshmati (2016) attempt to study the
development of the service sector over the years in Rwanda’s economy and
empirically estimate the determinants by using an econometric methodology.
In Rwanda, services are considered as an alternative to manufacturing led
Page | 73
development as the country aims to become a service-based hub to serve
countries in the East African Community following an impressive growth of
the sector. The empirical results are based on micro-data collected from the
Rwanda Enterprise Survey of 2011 and 2014. The results show that the growth
of service firms in innovation, sales, and turnovers is driven by access to
finance, increased labor force, training personnel, ICT applications, embryonic
innovations and the tax system. The size of the loans approved by financial
institutions like banks and cooperatives has had a positive effect on the three
years total annual sales, capital used by the service firms also positively
influenced turnovers of service firms and the acquisition of fixed assets
positively affected service innovativeness.
Empirical Review of Literature on Ethiopian Service sector and structural

change
Although there are a number of studies on the structural change

process of Ethiopia, but the literature on the service sector and structure of the
country are not adequate. In this section, a brief review of empirical literature
on Ethiopian service sector and structural change are reviewed. To start with, a
recent empirical study by Eshete & Kimuyu (2014a) examined the impacts of
alternative sectoral drivers of economy-wide growth using sectoral TFP in the
dynamic CGE model. The finding show that economic openness for the
agriculture sector, imported capital goods and services for the industry, and
service liberalization for services are the positive drivers and enablers of
sectoral total factor productivity in Ethiopia. The TFP induced by the
liberalization of the service sector is also more effective in enhancing the
growth rate of the economy. Industrial TFP induced by imported capital goods
and services also foster a structural change of the economy. Eshete & Kimuyu
(2014a) simulation model shows that ‘for a more sustainable growth, rapid
structural change, and welfare gain, Ethiopia should focus on broad-based
growth options that combine induced TFP growth in all sectors’. The broad-
based growth option could sustain the growth in Ethiopia, though it could not
lead the country to join the middle-income country status by 2025. Therefore,
Page | 74
Eshete & Kimuyu (2014a) suggests the active involvement of the government
in the manufacturing sector, foreign trade openness, service trade
liberalization for driving sectoral total factor productivity.
Another study by martins (2014) on the Ethiopian structural change
from employment perspective revealed that structural change in the
composition of employment in the sectors is absent in the Ethiopian economy
(Martins, 2014). Based on data disaggregated into eight sectors for the period
1996–2011, his analysis suggests that the structure of output has changed
considerably from agriculture to services. Yet, changes in the composition of
employment have lagged behind. In addition, the labor productivity growth in
the period has been strong across most sectors, though mainly driven by the
within-sector productivity improvements. Structural change is also
accelerating in its pace and relative contribution to output growth. On the
other hand, labor productivity growth has been strong, with output per worker
doubling in the past decade. In fact, GVA growth has been mainly driven by
labor productivity growth. Conversely, employment elasticities have fallen but
remain positive – suggesting that productivity growth was accompanied by
employment growth. Moreover, productivity gaps across economic sectors are
very large. While within-sector productivity growth accounts for much of the
aggregate productivity performance, structural change (i.e. between-sector
productivity growth) is playing an increasingly important role. Therefore, in
Ethiopia, economic growth has generated some employment, but the growth-
employment link has been weakened, which is intrinsically linked to
demographic trends. Not only has population growth slowed down, but the
proportion of people outside the labor force has also increased (Martins,
2014).
McMillan and Rodrik (2011) argue that the potential gains in overall
productivity growth from sectoral reallocations can be very large. Based on
their estimate, Ethiopia’s productivity could increase six-fold by assuming that
sectoral productivity levels remain constant and that the inter-sectoral
distribution of employment matches that of advanced economies. The
potential gain of Ethiopia is positive as compared with other countries in
Page | 75
Africa, only lagging behind Senegal and Malawi. The increases in labor
productivity can be accounted by several factors which include increased
efficiency in the use of labor, improved use of other inputs, and labor shifting
to more productive activities (structural change). McMillan and Rodrik (2011)
also observed a significant degree of heterogeneity across countries in the
structural change. Some countries in Africa experienced growth-enhancing
structural change, such as Ethiopia and Ghana. They find that in a sample of
38 developed and developing countries, Ethiopia has the second highest
contribution of structural change to productivity growth.
Strong economic growth may not always help in poverty reduction,
because of its failure to generate sufficient productive employment in terms of
better jobs. Martins (2013) also compares the fast-growing African countries
including Ethiopia, Ghana, Mozambique, and Tanzania. He compares the
different growth paths being pursued, and the policy choices, which might
explain the gaps in key development outcomes. According to him, the pattern
of economic growth matters for poverty reduction. In addition, from the
experience of Mozambique and Tanzania, growth that is driven by capital-
intensive sectors seems to generate limited benefits for the poor. So, to
improve inclusiveness of the growth, Martins (2013) suggests diversifying
production structures into more employment-intensive sectors. From the
experience of Ethiopia, sector specific policies play an important role in
reducing poverty. Particularly, investment in the agriculture is not only crucial
to reducing the incidence of poverty but can also accelerate the pace of
structural transformation. Therefore, according to Martins (2013),
improvement in agricultural productivity and employment in higher-
productivity employment-intensive activities, such as in the manufacturing
and modern services, will be important to sustainably raise living standards in
Africa.
According to Dorosh, Schmidt, & Shiferaw (2012), although
agriculture continues to be the foundation for economic growth and poverty
reduction, in Ethiopia government policy slowed rural-urban migration
through various regulations including prohibiting the sale of land, loss of land
Page | 76
rights for those who leave rural areas, and registration requirements for new
migrants. The modes rate of urbanization and slow development of industrial
sector inhibited structural transformation in Ethiopia. Therefore, allocation of
public investments across sectors along with policies and regulations on land
and labor mobility will be important factors for the growth of Ethiopia’s
economy and poverty reduction in the coming decade (Dorosh, Schmidt, &
Shiferaw, 2012).
An obvious change in the Ethiopian structural change is the high
growth in per capita GDP, rapid accumulation of both physical and human
capitals, and the declining contribution of the agricultural sector and rise of the
service sector. Moreover, the Ethiopian economy since 2005/06 has been
stable and resilient to drought, agricultural price and financial price shocks
implying the sustainability of the growth. (EEA, 2015). The per capita GDP of
Ethiopia in 2014/15 is 567 USD, which has increased from its low level in the
past decades. In terms of accumulation, Ethiopia is a typical poor country that
had an average rate of gross domestic saving of 5.8% during 2005/06 –
2009/10 and gross fixed investment of 26.3%, creating a resource gap of
20.9% that is financed by remittances, FDI, grants, and loans. However,
during the GTP period, the gross domestic saving rate increased to 22.5% and
the rate of gross fixed investment reached 40.3% in 2013/14 due to
government extensive resource mobilization efforts through Great Ethiopian
Renaissance Dam (GERD) bond and condominium housing schemes. The
source of Ethiopian growth from factor decomposition shows the labor has
been the major source of growth exceeding capital until 2004/05 (EEA, 2015).
Since 2004/05, however, capital is contributing more than labor and it has an
increasing trend. During the GTP I (2009/10 – 2013/14) period, the
contribution of capital and labor to the 10% growth was 6.6% and 5.7%
respectively. The residual (‘Solow’s residual), which is 2.8% is the
contribution of the TFP. The TFP to growth in Ethiopia before 2004/05 has
been erratic and mostly negative. However, after 2004/05 it is positive ranging
between 4.5 and 5.4%, contributing much to the high growth observed in the
period. From the demand side in 2010/11 – 2013/14, the decomposition of
Page | 77
GDP growth shows that expansion in the domestic demand driven by high
public investment contributes 85% of the growth as export and import
substitution accounted 5.2% and 3.4% respectively. From the service sub-
sectors, the wholesale and retail trade, the hotel and restaurant and the
business and real estate sub-sectors contribute 16.3%, 9.1% and 7.9% to the
10% GDP growth rate during 2009/10 – 2013/14 (EEA, 2015).
According to WB (2015) report on Ethiopian economy, the rapid
economic growth during the growth periods is concentrated in agriculture and
services and was driven by a considerable investment in the public
infrastructure, which is supported by a conducive external environment. At the
beginning of the take-off, the agriculture sector was the main economic sector.
But in recent years the services sector took over the role, complemented by a
construction boom. Out of an average annual growth rate of 10.9 percent in
2004-14, services contributed by 49.5 percent followed by agriculture 33
percent and industry with 15.6 percent (EEA, 2015). Private consumption
contributed to most growth on the demand side with public investment
becoming increasingly important.
Growth decompositions for Ethiopia reveal relatively high
contributions from total factor productivity and structural change. According
to WB(2015a), Ethiopia had high total factor productivity growth of 3.4
percent per year, which is greater than other non-resource rich fast-growing
Sub-Saharan African countries. Most of the labor productivity growth is due to
the within sectors productivity. The structural change also explains a quarter
of the per capita GDP growth. Using a cross-country regression model,
Ethiopia’s key drivers of economic growth in Ethiopia was driven primarily
by public infrastructure investment. Ethiopia has experienced high economic
growth and some structural change on the output side but from the
employment side, a shift from agriculture to the services is modest. Since
agricultural labor productivity is so small, the shift of labor out of agriculture
gave rise to static efficiency gains in construction and services that have
higher average value added of a worker (WB, 2015a).
Page | 78
For the 1999-2013 periods as a whole, more than 70% of growth is
attributed to within-sector labor productivity gains, especially in agriculture
and commerce. The structural change and demographic effects are noticeable
in (2005-13) period. The employment effect is negative due to a rise in the
student population. Between 1999 and 2013, labor productivity exhibited an
average annual growth rate of 4.5 percent. However, the nature of the
structural change and sectoral growth dynamics in Ethiopia differs from the
government policy and targets. Specifically, economic strategy in Ethiopia
aims to promote the kind of Lewis model-based structural change in which
workers move out of agriculture into manufacturing. World Bank has called
this model as ‘trodden path of development’ only partially as economic
activity (output and jobs) have shifted from agriculture and into construction
and services, bypassing the critical phase of industrialization (WB, 2015a).
Finally, Ferede & Kebede (2015)analyzed the structural change of
Ethiopia using decomposition tool. According to their findings, the within-
sector productivity represents the largest contribution to aggregate labor
productivity growth, accounting for about 66% of total labor productivity
growth. The structural effect, which consists of employment and the
interaction effects, contributed close to a third of the total labor productivity
growth, with the employment effect accounting for about 40% of the total
labor productivity growth. The contribution of the interaction effect to the
overall labor productivity growth is negative, supporting the structural burden
hypothesis. This indicates that sectors with fast-growing labor productivity
cannot maintain their shares in total employment. The negative effect can be
larger if sectors with high productivity growth are faced with declining
employment shares. While agriculture, manufacturing, and electricity and
water sectors exhibited declining employment shares with positive labor
productivity growth, transport and communication and other services
experienced the opposite.
To recap, the role of the service sector in the productivity and

economic growth has been neglected in the early economics literature.
However, in the present days, the service sector is revitalized and it is
Page | 79
considered as a path for structural change, sustainable growth and greater
interlinkage with the manufacturing and agriculture sectors. Theories and
empirics also show that the growth of service in output and employment is
mainly driven by income, productivity difference and changes in the structure
of demand including economic reforms, the participation of female in
economic activities and demand for intermediate service inputs.
Page | 80
CHAPTER-III
METHODOLOGY, DATA AND MODEL
In this chapter, the methodologies followed in analyzing the respective

objectives are presented. The structure and growth of the service sector in
Ethiopia is analyzed from different perspectives by applying different
methodologies. Descriptive statistics is used to identify the growth trend,
composition, and contribution of the service sector in output, employment, and
productivity. The growth decomposition methodologies are applied in order to
explore the growth trend of the sector and to identify the role of the sector in
Ethiopian output and employment growth and structural change. In addition,
the Autoregressive Distributed Lag (ARDL) model is used to identify the
determinants of the growth in the sector and furthermore, the inter-linkage of
the sector with other sectors and how much the service sectors are key in
Ethiopian economy are assessed using Social Accounting Matrix
(SAM)/Input-Output (I-O) analysis. The trend in service sector growth and its
role in structural change are studied for the period of 1961 – 2011. The
determinants of the service sector growth are analyzed using time series data
from 1981 – 2011. Finally, the I-O analysis is conducted using the Ethiopian
SAM data for 2005/06. Thus, in this section, the growth decomposition
methodologies of the Shapley decomposition method using the Job Generation
and Growth/ JoGG decomposition tool is presented first. Then, the Shift-share
decomposition method is presented followed by the time series analysis model
of ARDL. Finally, the methodologies applied in the aggregation of the
Ethiopian SAM and construction of the I-O table is presented. Furthermore,
the types, source, and description of the data used are presented under each
methodology.
Page | 81
3.1 : Growth Decomposition Analysis
3.1.1 : Job Generation and Growth Decomposition
In this study, Shapley decomposition method is applied using the Job

Generation and Growth/JoGG decomposition tool to describe how growth is
reflected in the sectoral pattern of growth and employment generation. Here, it
is used to describe how growth is linked to change in employment at the
aggregate level and specifically with service sector. It shows ‘how growth has
translated into productivity, employment, and demographic changes’ at the
aggregate level or by sectors (WB, 2012a). The Shapely decomposition is a
simple additive method associated with changes in per capita GDP (Per capita
Value Added) by taking into account the relative size of the sector and the
magnitude of change. The methodology decomposes GDP growth using
several consecutive steps. However, in this study, since the focus is on sectoral
contribution on changes in employment rate and per capita GDP, only some of
the steps are considered. First, growth in per capita GDP is decomposed into
employment rate change, change in output per worker and demographic
changes. Then, employment changes are further decomposed into changes in
employment by sectors. The other step identifies the role played by each
sector on the aggregate effect of employment reallocation across sectors (WB,
𝑌
2012a). Thus, Per capita GDP is written as;𝑁
𝑌 𝑌 𝐸 𝐴
= 𝐸 . 𝐴 . 𝑁; (1)
𝑁
𝑌
Where 𝐸 is total output per worker, as a measure of productivity,
𝐸
is employment rate and
𝐴
𝐴
is the share of working-age population measuring demographic changes.
𝑁
Thus, per capita GDP can be decomposed into growth associated with
changes in the size of the working age population. This means that total
change in per capita GDP is the sum of growth attributed to each of the
components 𝜔, 𝑒 𝑎𝑛𝑑 𝑎; each representing output per worker, employment
̅, 𝑒̅ and 𝑎̅as
rate and working-age population respectively. Thus, if we denote 𝜔
Page | 82
a fraction of the growth attributed to each component of 𝜔, 𝑒 𝑎𝑛𝑑 𝑎, then the
total growth rate of an economy can be expressed as;
∆𝑦 = 𝜔
̅ ∗ ∆𝑦 + 𝑒̅ ∗ ∆𝑦 + 𝑎̅ ∗ ∆𝑦 (2)
In the decomposition, each component has the interpretation of a

counterfactual scenario. For instance, the employment rate 𝑒̅ ∗ ∆𝑦, will be the
amount of growth consistent with a scenario in which output per worker 𝜔
and the share of working-age population 𝑎, had remained ‘unchanged’. To
understand the way in which sectors contributed to employment generation
and to total per capita growth it is possible to further decompose employment
(rate) growth (𝑒Δ) by sectors. The easiest way is to express the total growth in
employment as the sum of employment generation in each sector.
∆𝑒 = ∑𝑠𝑖=1 ∆𝑒𝑖 (3)
𝐸𝑖
Where ∆𝑒𝑖 = ∆ 𝐴 - is the changes in employment in sector i as a share of total
working age population. This gives a simple measure of which sector

contributed more to changes in the employment rate.
The other step is the decomposition of output per worker to changes in

output per worker within sectors and changes in the relocation of workers
between sectors. Output per worker can be also decomposed into sectoral
employment shifts and changes in output per worker within sectors.
𝑌 𝑌𝑖 𝐸𝑖
= ∑𝑠 𝐸𝑖 . 𝐸 (4)
𝐸
Where Yi is Value Added of sector i=1…s; Ei is employment in sector I, and E

is total employment. This equation just states that total output per worker is
the weighted sum of output per worker in all sectors, where the weights are
simply the employment share of each sector (WB, 2012).
Data
To perform this decomposition, data was collected on output (Value

Added) and population from the National Bank of Ethiopia report (NBE,
2015). Employment data was collected from the 1999, 2005 and 2014
Page | 83
National Labour Force Survey of the Central Statistical Agency of Ethiopia.
The analysis is made for the period (1999-2013) by classifying the period as
the low growth period of 1999-2005 and the high growth period of 2005-2013
(See (WB, 2015a). Another sort of secondary data for the descriptive analysis
is collected from the Groningen Growth and Development Center (GGDC) 10
sectors database for the periods 1961-2011 (de Vries, de Vries, Gouma, Pahl,
& Timmer, 2014).
3.1.2 : Shift-Share Analysis
There are various types of decomposition analysis, which are used to

analyze the sources of growth and regional growths. A number of studies have
also utilized a growth decomposition technique called shift-share analysis to
study productivity growth in sectors and structural changes in various
economies (Fagerberg, 2000; Timmer & Szirmai, 2000; Van Ark & Timmer,
2003; Singh L. , 2004; Maroto-Sanchez & Cuadrado-Roura, 2007; Peneder,
2003; Castaldi, 2009; Timmer & de Vries, 2009).
In this case, the shift-share technique is used to delineate the structural

change process in Ethiopia with reference to tertiarization. The technique is
‘convenient to research how aggregate growth is linked to the differential
growth of labor productivity and the shift of labor between sectors’ (de Vries,
Timmer, & de Vries, 2013; Maroto-Sánchez & Cuadrado-Roura, 2009). To
measure the contribution of the shift of labor across sectors to growth, various
researchers have utilized this decomposition method, which was originally
used by Fabricant in 1942. The tool decomposes the change in aggregate
productivity into two effects: a within (productivity growth) and a between
effects (structural changes). The within effect captures productivity growth
within sectors, whereas the between effect measures the productivity effect of
labor reallocation across different sectors and it can further be divided into
static and dynamic effects.
The shift-share decomposition can be performed in various ways

depending on the choice of base and end year of the periods. One alternative is
to use base period employment shares and final period productivity levels as in
Page | 84
(Timmer & Szirmai, 2000; de Vries, Timmer, & de Vries, 2013) to decompose
the change in aggregate productivity. The other alternative is using final
period employment shares and base period productivity levels. On the other
hand, Timmer & de Vries (2009) used period averages with the average share
of sectors in overall employment, and the average labor productivity level of
sectors. This decomposition takes a middle ground between the base and end
year choices with respect to the contribution of structural change. It can be
represented as:
∆𝑃 = ∑𝑖(𝑃𝑖𝑇 − 𝑃𝑖𝑜 )𝑆̅𝑖 + ∑𝑖(𝑆𝑖𝑇 − 𝑆𝑖𝑜 )𝑃

̅𝑖 (5)
Where Si is the share of sector i in overall employment, Pi the labor

productivity level of sector i, and superscript 0 and T refer to initial and final
period, 𝑆𝑖̅ 𝑎𝑛𝑑 𝑃̅𝑖 𝑎𝑟𝑒 the average sectoral employment share and labour
productivity level. The change in aggregate productivity (∆𝑃) is decomposed
into within-sector productivity changes and the change in labor reallocation.
The within-sector productivity change, the first term on the right-hand side in
equation 5, is also called the ‘within-effect’ or ‘intra-effect’. The reallocation
effect, second term is also called the “shift-effect” or ‘structural-change
effect’. The within-effect is ‘positive when the weighted change in labour
productivity levels in sectors is positive’. The ‘structural-change effect’
measures the ‘contribution of labour reallocation across sectors, and it will be
positive when labour moves from less productive sectors to sectors that are
more productive’ (Timmer & de Vries, 2009).
However, the structural change term is only a static measure of the

reallocation effect as it depends on differences in productivity levels across
sectors, not growth rates. McMillan and Rodrik (2011) argue that workers
could move to low- productivity growth sectors (McMillan & Rodrik, 2011).
Thus, those sectors that absorb additional workers will have low marginal
productivity for the additional workers, which affect productivity growth rates.
Thus, an alternative decomposition method accounts for the possible joint
effect of employment shift and productivity changes. This introduces a third
Page | 85
term in the decomposition that can be written as follows: (see (Timmer & de
Vries, 2009; de Vries, Timmer, & de Vries, 2013)
∆𝑃 = ∑𝑖(𝑃𝑖𝑇 − 𝑃𝑖𝑜 )𝑆̅𝑖 + ∑𝑖(𝑆𝑖𝑇 − 𝑆𝑖𝑜 )𝑃

̅𝑖 + ∑𝑖(𝑃𝑖𝑇 − 𝑃𝑖𝑜 ) ∗ (𝑆𝑖𝑇 − 𝑆𝑖𝑇 ) (6)
The first term is the within effect, the second term measures whether
workers move to above-average productivity level sectors (static reallocation
effect, which is called the static effect). The third term in equation 6 is known
as the dynamic effect or interaction term (Van Ark & Timmer, 2003; de Vries,
Timmer, & de Vries, 2013). It represents the joint effect of changes in
employment shares and sectoral productivity. It is positive if workers are
moving to sectors that are experiencing positive productivity growth. Hence,
the reallocation term of equation 5 is split into two terms: whether workers
move to above-average productivity level sectors (static reallocation effect)
and whether productivity growth is higher in sectors that expand in terms of
employment shares (dynamic reallocation effect).
In addition, to properly measure the role of sectors in accounting for

growth, the decomposition presented in equation 6 has to be adjusted.
According to (de Vries, Timmer, & de Vries, 2013), the rationale for this
adjustment is as follows. In the decomposition method presented in equation
5, ‘all expanding sectors contribute positively to aggregate productivity, even
when they have below-average productivity levels or growth rates’. de Vries,
Timmer, & de Vries (2013) explain that ‘if productivity growth in non-market
services is below average while manufacturing productivity growth is above
average, the shift in employment shares will result in lower aggregate
productivity growth’. However, ‘based on the measurement of the traditional
method, the contribution to structural change from the expansion of non-
market services is positive’. This implies that the traditional decomposition
methods, in equation 5 and 6, are not suitable to measure the contribution of
sectors to productivity growth. In the modified method de Vries, Timmer, &
de Vries (2013) adjust the static and dynamic reallocation effect of an
expanding sector to take into account its relative productivity level and its
relative productivity change. They divide sectors into expanding and shrinking
Page | 86
ones based on their changes in employment shares and calculate the static
between-effect relative to the average productivity level of the shrinking
sectors and the dynamic between- effects relative to the average productivity
change of the shrinking sectors. The decomposition is modified as follows:
∆𝑃 = ∑𝐽𝑖(𝑃𝑖𝑇 − 𝑃𝑖𝑜 )𝑆̅𝑖 + ∑𝐽𝑖(𝑆𝑖𝑇 − 𝑆𝑖𝑜 )(𝑃𝑗𝑜 − 𝑝𝑜∗ ) + ∑𝐽𝑗 ((𝑃𝑗𝑇 − 𝑃𝑗𝑜 ) − (𝑝𝑇∗ − 𝑃 𝑜∗ )) (𝑆𝑗𝑇 − 𝑆𝑗𝑜 ) (7)
Where J is the set of expanding sectors and K is the set of shrinking sectors
and average labor productivity of shrinking sectors at time O and T is given
by,
∑𝐾 𝑇 𝑂 𝑂
𝑘 (𝑆𝑘 − 𝑆𝐾 )𝑃𝑘
𝑝𝑂∗ = ⁄ 𝐾 𝑇 (8)
∑𝑘 (𝑆𝑘 − 𝑆𝐾𝑂 )
∑𝐾 𝑇 𝑂 𝑇
𝑘 (𝑆𝑘 − 𝑆𝐾 )𝑃𝑘
𝑝𝑇∗ = ⁄ 𝐾 𝑇 (9)
∑𝑘 (𝑆𝑘 − 𝑆𝐾𝑂 )
This adjusted decomposition does not affect the aggregate contributions from
the within and between effects. However, equation (7) is better to examine the
contribution of sectors in accounting for productivity growth.
3.1.3 Structural Bonus and Structural Burden Hypothesis
The structural bonus and structural burden hypothesis are hypotheses

regarding the employment shifts in an economy. The structural bonus
hypothesis, which is calculated as the sum of employment shifts weighted by
their productivity level (static effect), assumes a positive relationship between
structural change and economic growth. On the other hand, the dynamic or
interaction effect shows structural burden hypothesis, which is calculated as
the sum of the interactions between the changes in the weight of employment
and the changes in the labor productivity of each branch of activity. (See
(Timmer & Szirmai, 2000; Peneder, 2003; Fagerberg, 2000; Chen, Jefferson,
& Zhang, 2011; Havlik, 2013).
It is assumed that, in the process of economic development,

‘economies shift from activities with relatively low productivity levels to
industries with a higher value added per labor input’ (Timmer & Szirmai,
2000). Thus, the structural bonus hypothesis for the manufacturing sector, the
Page | 87
static effect, will be positive when the sectors with high productivity levels
attract more labor. It may also have a negative impact on aggregate
productivity growth if labor shifts to sectors with slower productivity growth.
On the other hand, according to Baumol (1967), the structural burden
hypothesis is valid if the value of this is negative due to a shift of labor from
manufacturing to services that have less productivity. Thus, the interaction
term can be used to capture Baumol’s hypothesis of a structural burden of
labor reallocation on aggregate growth, which predicts that employment shares
shift away from progressive industries towards those with lower growth of
labor productivity. See (Peneder, 2003).
Accordingly, the second component of ‘equation 6’ is the static or net

effect, which represents structural bonus. It is calculated as the sum of the
changes in the weight of each sector in the total employment between the final
and initial years, weighted by the value of the labor productivity of the sector
in the initial year (equation 10). It will be positive (negative) when the sectors
with high productivity levels attract more (less) labor and consequently
increase (reduce) their weight in total employment. Thus, the structural bonus
hypothesis of manufacturing sector is written as;
̅𝑖 ≥ 0
∑𝑖(𝑆𝑖𝑇 − 𝑆𝑖𝑜 )𝑃 (10)
The third component in ‘equation 6’, the dynamic or the interaction

effect, can represent the structural burden hypothesis, which is also presented
in ‘equation 11’. It is calculated as the sum of the interactions between the
changes in the weight of employment and the changes in the labor productivity
of each branch of activity. The structural burden hypothesis of services is
written as;
∑𝑖(𝑃𝑖𝑇 − 𝑃𝑖𝑜 ) ∗ (𝑆𝑖𝑇 − 𝑆𝑖𝑇 ) ≤ 0 (11)
The growth decomposition approach, shift-share analysis, which is

described in this section, has its own limitation. Even though based on some of
the limitations it has been modified, there are various criticisms that are not
yet answered. See (Van Ark & Timmer, 2003; Timmer & Szirmai, 2000).
Page | 88
According to Van Ark & Timmer (2003), the methodology is ‘a labor
productivity model and as such non-labour inputs are ignored’. Economic
activities are not solely using labor inputs only, other inputs such as capital,
materials, and service inputs are used along with labor. Therefore, the sectoral
productivity measures should incorporate non-labour inputs. The other
criticism is that the decomposition into shift effects and intra-sectoral effects
depend on the selected price base year of the output series.
Data
The data for the shift-share analysis in this study is collected from
Groningen Growth and Development Center (GGDC) 10 sectors database (de
Vries, de Vries, Gouma, Pahl, & Timmer, 2014). The database covers the ten
main sectors of the economy as defined by the International Standard
Industrial Classification, Revision 3.1 (ISIC, 2002). These ten sectors are;
agriculture, manufacturing, other industry (includes construction and utilities,
distributive trade services (includes wholesale and retail trades, hotels and
restaurants), transport, storage and communication services, finance and
business services, public services (public administration and social services),
and other services (includes other community, personal and household
services). The employment data refers to ‘all persons employed’, including all
paid employees, also self-employed and family workers. The Gross value
added is measured at constant 2005 Ethiopian prices.
3.2 : Specification of the Autoregressive Distributed Lag Model
In this study, the Autoregressive Distributed Lag (ARDL) model,

proposed by Pesaran, Shin, and Smith (2001) is used for the long-run
cointegration relationships between service growth determinant variables. It is
widely used to model the relationship between economic variables in a time-
series analysis (Kripfganz & Schneider, 2016), as it has many advantages over
other co-integration methods.
The ARDL approach can be applied whether the regressors are I(1)
and I(0) (Pesaran, Shin, & Smith, 2001). The Johansen co-integration
Page | 89
techniques require large data samples for validity, but the ARDL procedure
avoids the problem of biases that arise from small sample size and gives a
statistically significant result (Pesaran, Shin, & Smith, 2001); (Narayan, 2005).
In addition, a dummy variable can be included in the co-integration test
process when using the ARDL model. Therefore, the following ARDL model
is specified to identify the long run and short run relationships of service gross
value addition and service employment growths with other explanatory
variables.
In ARDL model, all the dependent and independent variables can be

introduced in the model with lags. "Autoregressive" refers to lags of the
dependent variable and "Distributed" refers to the lags of explanatory
variables. The choice of lag order for the ARDL model is crucial for long-run
analysis and the lag orders have to be selected based on diagnostic tests for
residual serial correlation, functional form misspecification, non-normality
and heteroscedasticity using one of information criteria (Akaike Information
Criterion (AIK), the Schwarz Bayesian Criterion (SBC), or the Hannan-Quinn
Criterion (HQC)). Pesaran, Shin, & Smith (2001) show that the ARDL model
yields consistent estimates of long-run coefficients under asymptotic
normality. This result holds for regressors that are purely I(0), I(1) or mixed.
The ARDL bounds testing procedure is also used to investigating the existence
of a long-run relationship between the variables. According to Pesaran, Shin,
and Smith (2001), the small sample properties of the bounds testing approach
makes it better than that of the traditional Johansen cointegration (Pesaran,
Shin, & Smith, 2001).
A general ARDL (p,q,….q) model for a scalar variable is given as:
𝑦𝑡 = 𝑐0 + ∑𝑝𝑖=1 𝛽𝑖 𝑦𝑡−1−𝑖 + ∑𝑞𝑖=0 𝛽𝑗 𝑥𝑖,𝑡−1 + 𝜇𝑡 (12)
Where, Co constant, 𝑦𝑡 is endogenous variable, 𝑥𝑖,𝑡 is the ith explanatory

variables, p is the maximum lag to be used as the dependent variable and q is
the maximum lag to be used for the explanatory variables, 𝛽𝑖 and 𝛽𝑗 are
parameters, and 𝜇𝑡 is the white noise error.
Page | 90
The variables in (𝑦𝑡 ,𝑥𝑖,𝑡 ) are allowed to be purely I(0), purely I(1), or
co-integrated. The optimal lag orders p and q can be obtained by minimizing a
model selection criterion of the Akaike information criterion (AIC) or the
Schwarz Bayesian information criterion (BIC) (Kripfganz & Schneider, 2016).
Based on the theoretical framework described in the literature section, the
following log-linear type model is specified. The ‘hierarchy of needs’
hypothesis, the ‘cost disease’ hypothesis and the ‘exogenous demand shock’
hypothesis are included. The ‘deindustrialization’ hypothesis is not considered
as it is irrelevant to Ethiopia.
The service sector growth in GVA (LnSGVA) and employment

(LnSEMP) are hypothesized to be determined by the following variables.
Based on the ‘hierarchy of needs hypothesis, income (LnGDPPC) has a
positive effect on both Service GVA and Service employment. Productivity
difference (LnPTVDC) is the expected to have a positive effect based on the
‘cost disease’ hypothesis. Producer service (LnPSERV) captures intermediated
demand for services used as an input by other sectors and LnFMEAC refers
female labor force in the economy. Based on the ‘exogenous demand shock’
hypothesis, changes in the structure of the economy that moves the demand
curve upward increases service sector in the economy. Accordingly,
LnPDERV and LnFMEAC shift the demand for services upward and hence a
positive effect is expected on both LnSGVA and LnSEMP. Finally, DREFO92
refers to liberalization and reform period dummy for Ethiopia. It is also
expected to have a positive effect on service growth.
All the ratio variables are log transformed to reduce the problem of
heteroscedasticity as it compresses the measured scale of the variables
reducing the difference between two values (Gujarati, 2004). Thus, in the
model, all parameter coefficients represent constant elasticities.
Therefore, the ARDL model is specified in the equations below:
LnSGVAt = f(LnGDPPCt, LnPTVDt, LnPSERVt, LnFEACTt, DREFO92) (13)
LnSEMPt=f(LnGDPPCt, LnPTVDt, LnPSERVt, LnFEACTt, DREFO92) (14)
Page | 91
Where
LnSGVAt is Natural Logarithm of Service in Gross Value Addition at time t
LnSEMPt is Natural Logarithm of Service Employment at time t
LnGDPPCt is Natural Logarithm of GDP per Capita at time t
LnPTVDt is Natural Logarithm of Productivity difference at time t
LnPSERVt is Natural Logarithm of Producers service at time t
LnFEACTt is Natural Logarithm of Female Economic Activity at time t
DREFO92 is Dummy variable for Reform Period (1992-2011)
Based on the above equations the functions becomes in the form of:
𝑞 𝑞
𝐿𝑛𝑆𝐺𝑉𝐴𝑡 = 𝛼0 + ∑𝑃𝑖=0 𝛽1 𝐿𝑛𝑆𝐺𝑉𝐴𝑡−1−𝑖 + ∑𝑖=0 𝛽2 𝐿𝑛𝐺𝐷𝑃𝑃𝐶𝑡−𝑖 + ∑𝑖=0 𝛽3 𝐿𝑛𝑃𝑇𝑉𝐷𝑡−𝑖 +
∑𝑞𝑖=0 𝛽4 𝐿𝑛𝑃𝑆𝐸𝑅𝑉𝑡−𝑖 + ∑𝑞𝑖=0 𝛽5 𝐿𝑛𝐹𝐸𝐴𝐶𝑇𝑡−𝑖 + ∑𝑞𝑖=0 𝛽6 𝐷𝑅𝐸𝐹𝑂92𝑡−𝑖 + 𝜇𝑡 (15)
𝑞 𝑞
𝐿𝑛𝑆𝐸𝑀𝑃𝑡 = 𝛼0 + ∑𝑃𝑖=0 𝛽1 𝐿𝑛𝑆𝐸𝑀𝑃𝑡−1−𝑖 + ∑𝑖=0 𝛽2 𝐿𝑛𝐺𝐷𝑃𝑃𝐶𝑡−𝑖 + ∑𝑖=0 𝛽3 𝐿𝑛𝑃𝑇𝑉𝐷𝑡−𝑖 +
∑𝑞𝑖=0 𝛽4 𝐿𝑛𝑃𝑆𝐸𝑅𝑉𝑡−𝑖 + ∑𝑞𝑖=0 𝛽5 𝐿𝑛𝐹𝐸𝐴𝐶𝑇𝑡−𝑖 + ∑𝑞𝑖=0 𝛽6 𝐷𝑅𝐸𝐹𝑂92𝑡−𝑖 + 𝜇𝑡 (16)
According to (Pesaran, Shin, & Smith, 2001), the dependent variable

needs to be (I(1)), but the explanatory variables can be either (I(0)) or (I(1)).
So, the equation 13 and equation 14 can be written in terms of lagged levels
and differences. In addition, vector error correction model (VECM) is added
to separate the short-run and long-run multipliers of the model. Therefore, the
error correction form of the ARDL model will be:
𝑞 𝑞
𝑑𝐿𝑛𝑆𝐺𝑉𝐴𝑡 = 𝛼0 + ∑𝑃𝑖=0 𝛽1 𝑑𝐿𝑛𝑆𝐺𝑉𝐴𝑡−𝑖 + ∑𝑖=0 𝛽2 𝑑𝐿𝑛𝐺𝐷𝑃𝑃𝐶𝑡−𝑖 + ∑𝑖=0 𝛽3 𝑑𝐿𝑛𝑃𝑇𝑉𝐷𝑡−𝑖 +
∑𝑞𝑖=0 𝛽4 𝑑𝐿𝑛𝑃𝑆𝐸𝑅𝑉𝑡−𝑖 + ∑𝑞𝑖=0 𝛽5 𝑑𝐿𝑛𝐹𝐸𝐴𝐶𝑇𝑡−𝑖 + ∑𝑞𝑖=0 𝛽6 𝑑𝐷𝑅𝐸𝐹𝑂92𝑡−𝑖 + 𝜆1 𝐿𝑛𝑆𝐺𝑉𝐴𝑡−1 +
𝜆2 𝐿𝑛𝐺𝐷𝑃𝑃𝐶𝑡−1 + 𝜆3 𝐿𝑛𝑃𝑇𝑉𝐷𝑡−1 + 𝜆4 𝐿𝑛𝑃𝑆𝐸𝑅𝑉𝑡−1 + 𝜆5 𝐿𝑛𝐹𝐸𝐴𝐶𝑇𝑡−1 + 𝜆6 𝐷𝑅𝐸𝐹𝑂92𝑡−1 +
𝛾𝐸𝐶𝑀𝑡−1 + 𝜇𝑡 (17)
𝑞 𝑞
𝑑𝐿𝑛𝑆𝐸𝑀𝑃𝑡 = 𝛼0 + ∑𝑃𝑖=0 𝛽1 𝑑𝐿𝑛𝑆𝐸𝑀𝑃𝑡−𝑖 + ∑𝑖=0 𝛽2 𝑑𝐿𝑛𝐺𝐷𝑃𝑃𝐶𝑡−𝑖 + ∑𝑖=0 𝛽3 𝑑𝐿𝑛𝑃𝑇𝑉𝐷𝑡−𝑖 +
∑𝑞𝑖=0 𝛽4 𝑑𝐿𝑛𝑃𝑆𝐸𝑅𝑉𝑡−𝑖 + ∑𝑞𝑖=0 𝛽5 𝑑𝐿𝑛𝐹𝐸𝐴𝐶𝑇𝑡−𝑖 + ∑𝑞𝑖=0 𝛽6 𝑑𝐷𝑅𝐸𝐹092𝑡−𝑖 + 𝜆1 𝐿𝑛𝑆𝐸𝑀𝑃𝑡−1 +
𝜆2 𝐿𝑛𝐺𝐷𝑃𝑃𝐶𝑡−1 + 𝜆3 𝐿𝑛𝑃𝑇𝑉𝐷𝑡−1 + 𝜆4 𝐿𝑛𝑃𝑆𝐸𝑅𝑉𝑡−1 + 𝜆5 𝐿𝑛𝐹𝐸𝐴𝐶𝑇𝑡−1 + 𝜆6 𝐷𝑅𝐸𝐹092𝑡−1 +
𝛾𝐸𝐶𝑀𝑡−1 + 𝜇𝑡 (18)
Where: d is the first difference of a variable; β1...β6 represent the short-run

coefficients; λ1...λ6, represent to the long-run coefficients. ECM refers to error
correction model, and γ represents the speed of adjustment process of the
ECM. After estimation, the coefficient of the lagged error correction model is
expected to be negative and statistically significant for the existence of a
cointegrating relationship.
Page | 92
In addition, the degree of stationarity is determined by carrying out a
unit root testing through the standard Augmented Dicky-Fuller (ADF) test.
This test was undertaken to check the order of integration of the variables. It
was done for three alternative specifications. First, it is tested with constant,
trend, and then it is tested with neither constant nor trend. The null hypothesis
to test the stationarity is that the time series data have a unit root and the
alternative hypothesis is it has no unit roots. Therefore, if the critical statistic
in absolute value is greater than the t-value, we reject the null hypothesis and
if it is less we accept the null hypothesis.
The long run equilibrium relationship between the variables is tested

using bounds test for co-integration as proposed by (Pesaran, Shin, & Smith,
2001). In the bound test, if the F-statistic lies above the upper bound of critical
value, there is a non-spurious long run relationship of variables for a given
significance level (Pesaran, Shin, & Smith, 2001). On the other side, if the F-
statistic lies below the critical value of the lower bound, we can say that there
is no long-run level relationship between the explanatory variables with the
dependent variable. However, if it lies in-between the lower and the upper
limits, there is inconclusive result. The hypotheses for the bound test are given
below:
Ho: means there is no long-run relationship between the variables.
Ha: means there is a long-run relationship between the variables.
Therefore, For the Service GVA (SGVA) equation:
Ho: 𝜆𝐿𝑛𝑆𝐺𝑉𝐴 = 𝜆𝐿𝑛𝐺𝐷𝑃𝑃𝐶 = 𝜆𝐿𝑛𝑃𝑇𝑉𝐷 = 𝜆𝐿𝑛𝑃𝑆𝐸𝑅𝑉 = 𝜆𝐿𝑛𝐹𝐸𝐴𝐶𝑇 = 𝜆𝐷𝑅𝐸𝐹𝑂92 = 0

Ha:𝜆𝐿𝑛𝑆𝐺𝑉𝐴 ≠ 𝜆𝐿𝑛𝐺𝐷𝑃𝑃𝐶 ≠ 𝜆𝐿𝑛𝑃𝑇𝑉𝐷 ≠ 𝜆𝐿𝑛𝑃𝑆𝐸𝑅𝑉 ≠ 𝜆𝐿𝑛𝐹𝐸𝐴𝐶𝑇 ≠ 𝜆𝐷𝑅𝐸𝐹𝑂92 ≠ 0
For Service Employment (SEMP) equation:
Ho: 𝜆𝐿𝑛𝑆𝐸𝑀𝑃 = 𝜆𝐿𝑛𝐺𝐷𝑃𝑃𝐶 = 𝜆𝐿𝑛𝑃𝑇𝑉𝐷 = 𝜆𝐿𝑛𝑃𝑆𝐸𝑅𝑉 = 𝜆𝐿𝑛𝐹𝐸𝐴𝐶𝑇 = 𝜆𝐷𝑅𝐸𝐹092 = 0

Ha:𝜆𝐿𝑛𝑆𝐸𝑀𝑃 ≠ 𝜆𝐿𝑛𝐺𝐷𝑃𝑃𝐶 ≠ 𝜆𝐿𝑛𝑃𝑇𝑉𝐷 ≠ 𝜆𝐿𝑛𝑃𝑆𝐸𝑅𝑉 ≠ 𝜆𝐿𝑛𝐹𝐸𝐴𝐶𝑇 ≠ 𝜆𝐷𝑅𝐸𝐹092 ≠ 0
The error correction term (ECT) is derived from the corresponding

long-run model whose coefficients are obtained by normalizing the equation.
Various tests are conducted after the estimating the long run and short run
Page | 93
model to check the robustness of the model. These tests are model
specification test, normality test, serial correlation test, heteroscedasticity test
and test for stability of the model. Stata 13 statistical package and the ARDL
model for Stata (Kripfganz & Schneider, 2016)is used in order to estimate the
models specified in equation 15 and equation 16 and to perform the pre-
estimation and post estimation diagnostic tests.
Data
The data for this study is collected from secondary sources from 1981
– 2011. The service sector output measured in Gross Value Addition (GVA),
Service employment (number of the labour force employed in the service
sector), the output level of producer services are collected from GGDC
(Groningen Growth and Development Centre) 10 sector Database (de Vries,
de Vries, Gouma, Pahl, & Timmer, 2014). The Productivity difference
between manufacturing and service sector is also computed using the same
database. Furthermore, female economic activity data is also collected from
GGDC database. The GVA output of service sector and producer services
output are measured in constant 2005 local currency unit (Ethiopian Birr). The
sectors are classified on the basis of the International Standard Industrial
Classification of All Economic Activities, Rev.3.1(ISIC, 2002) code. Thus, the
service sectors aggregated in this study includes distributive trade services,
transport and communication services, finance and business services, public
services and personal services. Finally, the data on GDP per capita income is
measured in constant 2005 USD and it is collected from WDI (2016).
3.3 Sectoral Inter-linkage Analysis
This study used the 2005/06 Ethiopian Input-Output (hereafter I-O)

table and Social Accounting Matrix (hereafter SAM) constructed by EDRI
(EDRI, 2009) to analyze the service sector inter-linkage with other sectors and
to identify the place of service sectors as the key sector in Ethiopia.The year
2005/06 was taken as it is the only recent period that the country has its I-
O/SAM data. The first detailed SAM for Ethiopia, based on economic flows in
EFY 1994 (2001/02), was completed in 2007. This SAM distinguishes 42
Page | 94
production activities, 61 commodity groups, 5 primary factors, 2 household
groups, 17 tax instruments as well as aggregate accounts for trade margins,
transport margins, government, investment, and the rest of the world (Taffesse
& Ferede, 2005). The 2005/06 Ethiopian SAM is the most recent SAM and I-
O table constructed for the Ethiopian financial year of 1998 (2005/06) and it is
completed in 2009. Thus, the 2005/06 Ethiopian SAM is used to extract the I-
O data for the analysis of sectoral inter-linkage using SimSIP SAM software.
3.3.1 : Social Accounting Matrix
Social accounting matrix (SAM) is ‘a comprehensive, economy-wide

set of accounts, which quantify the flows of incomes and expenditures in an
economy for a given period of time, usually one year (EDRI, 2009). The basic
structure of SAM is presented in Table 1. The column accounts represent
expenditure and the row accounts represent income. There are “activities” and
“commodities” accounts in the first column. ‘Activities are entities that
produce goods and services, whereas commodities are those goods and
services produced by activities’ (Breisinger, Thomas, & Thurlow, 2009). Here
activities and commodities are different because an activity may produce more
than one type of commodity and commodities also can be produced by more
than one type of activity. Activities can be divided into agriculture, industry,
and services at aggregated or disaggregated level. They produce these goods
and services by combining factors of production with the intermediate inputs.
Activities also pay wages, rents, and profits they generate for the factors,
which gives the value-added. Similarly, intermediate demand is a payment
from activities to the commodities. When the value-added and intermediate
demand adds together it gives gross output. Thus, the information in the
activity column regarding production technologies refers to the input part of a
typical “input–output table,” which is a factor and intermediate inputs per unit
of output.
Page | 95
Table 3.1: Basic Structure of SAM
Expenditure
Saving & Rest of the
Activities Commodities Factors Households Government Total
Investment World
Domestic Activity
Activities
Supply income
Intermediate Consumption Recurrent Investment Export Total
Commodities
Demand Spending Spending Demand Earnings Demand
Total Factor
Factors Value Added
income
Factor Total
Social Foreign
Households payments to Households
Transfers remittances
Households income
Income
Sales Taxes Foreign

Government
Government and Import Direct taxes grants and
Income
tariffs loans
Current
Saving and Private Total
Fiscal Surplus account
Investment Savings Savings
balance
Foreign
Rest of the Import
exchange
World Payments
outflow
Total Total Foreign
Total factor Government
Total Gross Output Total Supply Household investment exchange
spending expenditure
spending spending inflow
Source; adopted fromBreisinger et al, 2010
Page | 96
On the other hand, commodities can be either supplied domestically or
imported from the rest of the world and various economic entities purchase
commodities. Activities demand commodities as intermediate inputs for
production. A final demand for commodities also comes from household
consumption, government consumption, gross capital formation, and export
demand. These sources of demand make up the commodity row or the total
demand, which are payments by different entities for commodities. Therefore,
the values in the commodity accounts are measured at market prices. SAM
does not only trace the income and expenditure flow of activities and
commodities. It also contains complete information on different institutional
accounts including households and the government (Breisinger, Thomas, &
Thurlow, 2009). As households are the ultimate owners of the factors of
production, they receive the incomes earned by factors during the production
process. They also receive transfer payments from the government and from
the rest of the world, pay taxes directly to the government, and purchase
commodities. The household income that remained after payment of tax and
consumption expenditure can be saved. On the other hand, in the government
account, a government may receive transfer payments from the rest of the
world as foreign grants or/and development assistance. All the different tax
incomes and foreign grants and loans add up to determine the total
government revenues. It uses these revenues to pay for recurrent consumption
spending and transfers to households. There could be a fiscal surplus or deficit
based on the difference between total revenues and expenditures.
Furthermore, SAM shows the investment or gross capital formation account,
which includes changes in stocks or inventories. The total investment must be
equal to total savings (private savings and public savings). The total capital
inflows from abroad or the current account balance is the difference between
total domestic savings and total investment demand. This is equal to the
difference between foreign exchange receipts (from exports and foreign
transfers) and expenditures (for imports and government) (Breisinger,
Thomas, & Thurlow, 2009).
Page | 97
3.3.2 : Input-OutputModel Framework
The Input-output analytical framework was developed by the Nobel

Prize winner Wassily Leontief in the late 1930s. Thus, an I-O model is also
referred to as a Leontief model. The basic use of the I-O framework is to
analyze the interdependence of industries or sectors in an economy. The I-O
analysis is the flow of products from a producing sector to all other sectors
that consume the product. Therefore, it shows the activity of all sectors of an
economy that both produce as well as consumer goods from other sectors in
order to produce their products (Miller & Blair, 2009).
The inter-sectoral transactions table in an input-output model shows

the basic information. The rows of the table describe the distribution of a
producer’s outputs as the columns describe the composition of inputs required
by a particular industry to produce its outputs. These inter-industry exchanges
of goods and services are described as an intermediate demand. There are also
additional columns, final demand, which record the sales of outputs by each
sector to final demanders. On the rows side, there are also additional rows that
show the value added account for factor inputs (labor, land, and capital).
SimSIP SAM Software
SimSIP SAM stands for Simulations for Social Indicators and Poverty
through Social Accounting Matrix. The software is based on a ‘Microsoft
Excel application with MATLAB running in the background’, which is used to
analyze I-O tables and SAM (Parra & Wodon, 2010a). It was developed in
2008 by J. C. Parra and Q. Wodon of the World Bank and version 1.1 became
available in 2009 and the latest version 1.1.2 is available since 2010. The
application is useful to perform several types of analysis such as aggregation,
disaggregation (decomposition), updating and balancing of the I-O tables to
obtain descriptive and analytical results of an economy. In addition, the tool is
useful for interlinkage analysis, multiplier analysis, and different experimental
designs. Various studies have used the software including Parra and Wodon
(2009), (Parra & Wodon, 2010b) for Tanzania; (Fofana, Parra, & Wodon,
Page | 98
2009) for Senegal; and (Khanal, 2011; Khanal, Gan, & Becken, 2014) for Lao
Tourism analysis.
The initial step in using the I-O model with SimSIP SAM software is
to identify the endogenous and exogenous accounts of the economy.
According to Sadoulet and de Janvry (2003) as cited in Parra & Wodon
(2010a), ‘exogenous accounts are those for which the expenditure is set
independently of income and endogenous accounts are those that change the
level of expenditure following any change in income’. Therefore, the software
applications consider the government, capital and rest of the world (exports
and imports) accounts as exogenous accounts and all other accounts as
endogenous. In addition, each cell in the I-O table is expressed in price or
value terms (Parra & Wodon, 2010a). The SimSIP SAM software application
is divided into several analysis components/blocks suitable for I-O and SAM
analyses out of which four analysis blocks were used in this study. These are
the matrix design and balance, GDP and labor multiplier, multiplier analysis
and sectoral impacts.
The following procedures were followed in the extraction of the I-O

table from SAM and in using the SimSIP SAM tool. First, the original
2005/06 SAM data of Ethiopia is arranged and defined to include it in the
input matrix. In doing so, the distinction between activities and commodities is
dropped to make it consistent with the standard input-output table format. The
total margin (trade and transport margins) is also moved to the intermediate
input block by adding to intermediate purchases of trade and transport
services. Then after checking whether the matrix is balanced or not, the
technical coefficients, the Leontief inverse matrix, and the total economic
impact (direct, indirect and induced effects) were computed. The sector impact
of economic activities and different types of linkages and interdependence
analysis such as standard, pure and weighted backward and forward linkages
are also estimated. The key sectors of an economy were computed based on
the inter-industry linkages. In addition, the economic landscape of Ethiopia
Page | 99
during 2005/06 was graphed based on the imposed linkage hierarchies from
the Multiplier Product matrix of the I-O (See Annex).
3.3.3 Description of Ethiopian 2005/06 SAM and I-O table
The 2005/06 Micro SAM of Ethiopia is disaggregated into various

accounts with a total of 255 rows and columns square matrix (EDRI, 2009).
There are 99 activity accounts that are classified into13 agricultural activity
groups, and 1 aggregate forestry and fishing activity, 21 industrial activities
and 12 service activities. The 91 commodities are divided into marketed
commodities and own-consumed commodities. The agricultural sector
produces 25 marketed and 20 own-consumed commodities, whereas the
industry sector produces 30 marketed commodities and the service sector
generates two own-consumed and 14 marketed service commodities. Further,
the Micro SAM has also factor income flows that are recorded in 25 factor
accounts. There are 14 household accounts that distinguish rural households
by their income class as poor and non-poor as well as by their residence as
urban and rural. There are also accounts for marketing margins, the
government, capital accounts, saving-investment flows, and transactions of the
Ethiopian economy with the rest of the world (EDRI, 2009).
For the purpose of this study, the Ethiopian I-O inter-industry flows are
extracted from the 2005/06 SAM and the economy is divided into 10 sectors.
The classification is based on ISIC Rev. 3.1. These sectors are agriculture,
manufacturing, mining and quarrying, utility, construction, distributive
service, transport and communication, business and finance, public services,
and personal and other services. The values for each cell is measured in
Ethiopian Birr (1USD = 8.681 ETB on average in 2005/06).
In Table 2, the columns of the I-O table are the selling sectors and the
rows are the purchasing sectors of the economy.According to Miller & Blair
(2009), the inter-industry sale shows that the total sales of each sector output
to itself, to another sector as intermediate inputs, and to the final demanders.
This flow of product depends on the total output of the sectors and I-O
Page | 100
coefficients. The I-O coefficients can be derived by dividing the inter-industry
flows of each sector by the total output of that sector.
Table 3.2: Input-Output Framework of Ethiopia for 10 sectors

Total
Final Total
Sectors Total Final Demand Demand Output
Intermediate
1 2 … 10 Demand C G I E
1 a1,1 a1,2 … a1,10 A1 C1 G1 I1 E1 Y1 X1
Commodities
2 a2,1 a2,2 … a2,10 A2 C2 G2 I2 E2 Y2 X2

… … … … … … … … … … … …
10 a10,1 a10,2 … a10,10 A10 C10 G10 I10 E10 Y10 X10
Total Intermediate
input A1 A2 … A10
Labor w1 w2 … w10 W
Value Added
Capital r1 r2 … r10 R
Gross Domestic Product
Indirect Taxes t1 t2 … t3 T
Import m1 m2 … m10 M
Total Value Added V1 V2 … V10
Total Inputs X1 X2 … X10 C G I E X
Source; adopted from (Khanal, 2011) and (EDRI, 2009) I – Investment, C – Households
Consumption, E – Exports and G - Government, W – wage, R – return to capital, T – Tax, and
M - import
Apart from the selling and buying sectors (commodities), there is value
addition block which is used in the production process. It includes labor;
capital and different inputs and these factors are the value added block of the
I-O table. They are compensation for employees as a wage for the total 10
sectors (W10); capital (R10); taxes (T10); and imports (M10). Therefore, the
total expenditures on inputs constitute the sum of all domestically produced
and imported goods and services and value added inputs. Thus, total inputs
for each sector (represented as𝑋1 , 𝑋2 and 𝑋10 ) is the sum of inter-industry sale
(intermediate inputs) and value added inputs. It is given as;
𝑋1 = 𝐴1 + 𝑉1 , 𝑋2 = 𝐴2 + 𝑉2 and 𝑋10 = 𝐴10 + 𝑉10 (19)
In addition, the production process requires an economy to have

expenditure from households and government and exports for the 10 sectors in
total. These are included in the final demand block. These are households’
consumption expenditure (C10); government consumption expenditure (G10);
investment (I10); and exports (E10). Therefore, the total final demand for each
sector’s outputs (represented as𝑌1 , 𝑌2 and𝑌10 ) is the sum of expenditures or
Page | 101
purchases of each sector’s outputs. The total final demand, each sector’s
outputs purchased as a final demand is given as;
𝑌1 = 𝐶1 + 𝐺1 + 𝐼1 + 𝐸1 , 𝑌2 = 𝐶2 + 𝐺2 + 𝐼2 + 𝐸2 and 𝑌10 = 𝐶10 + 𝐺10 + 𝐼10 + 𝐸10 (20)
One of the features of I-O table is the balancing of the sum of the total
inputs (row totals) with the sum of the total outputs (column totals). Hence,
the total output column which is the total gross output (X) in each sector is
𝑋 = 𝑋1 + 𝑋2 + ⋯ + 𝑋10 + 𝑊 + 𝑅 + 𝑇 + 𝑀, Similarly, the total input (X) in each
sector is given as 𝑋 = 𝑋1 + 𝑋2 + ⋯ + 𝑋10 + 𝐶 + 𝐺 + 𝐼 + 𝐸 . Therefore,
combining the two equations, we get; (see (EDRI, 2009)).
𝑊+𝑅+𝑇+𝑀 = 𝐶+𝐺+𝐼+𝐸 (21)
This can be written as,
𝑊 + 𝑅 + 𝑇 = 𝐶 + 𝐺 + 𝐼 + (𝐸 − 𝑀) (22)
In the Equation (22) above 𝑊 + 𝑅 + 𝑇 represents gross domestic

income or value-added at market prices which is the total factor payments in
the economy. Whereas, 𝐶 + 𝐺 + 𝐼 + (𝐸 − 𝑀)shows the gross domestic product
or GDP at market prices which is the total value of goods and services spent in
the economy (EDRI, 2009).
3.3.4 Inter-industry linkage analysis of Service sector
The inter-industry relationships among sectors in an economy can be

examined in various ways. Some of these could be estimating sectoral output
growth multiplier, employment multiplier, the elasticity of sectoral output,
forward and backward linkages, etc. There are different methodologies that
have been developed for the estimation of sectoral linkages. These
methodologies are input-output (I-O) analysis, social accounting matrix
(SAM), econometric modeling and computable general equilibrium modeling
(CGE). Other econometric modeling and statistical causality tests estimate
only the partial linkages between the sectors. Whereas, the I-O model is static
in nature and does not include the flow of capital goods (Dilip Saikia, 2011).
However, I-O linkage analysis which is developed by Rasmussen (1956),
Page | 102
Chenery and Watanabe (1958) and Hirschman (1958), are extensively used to
make a comparison of different sectors of an economy in terms of the
interdependence of their production structures (Pratt, 2011).However, the I-O
model is highly useful to show the inter-industry linkages of sectors. I applied
I-O methodology is applied in the study to explore about service sector in
Ethiopia specifically to find out the level of its inter-linkage with other sectors
and its place as a key sector in the economy.
Therefore, the I-O table of this study will represent the

interdependence of service sector with other sectors in Ethiopian economy. It
shows how the output of service sector that goes into another sector as an
input and the input of service sector. Formally, the I-O framework can be
explained as follows; if an economy consists of the following three sectors;
agriculture (1), industry (2) and services (3). Thus, the input-output coefficient
matrix can be written as;
(23)
Where; 𝑎12 in ‘equation 23’ represents the amount of sector 1 (agriculture)

input required for producing 1 unit of sector 2 (industry) sectors output. On the
other hand, the value added in jth sector (𝑉𝑗 ) is given by the relation in
equation (24)
𝑉𝑗 = 1 − ∑3𝑖 𝑎𝑖𝑗 − 𝑎𝑝𝑗 (24)
Where 𝑎𝑝𝑗 is the amount of imported input required per unit of sector j. Now,
given a final demand vector for the three sectors output (F) production and
value added in the three sectors are immediately obtained from equation (23)
and (24) as equation (25)
[𝐼 − 𝐴]𝑋 = 𝐹 (25)
𝑥1
Where, 𝑋 = [𝑥2]gives the column vector of the three sectors’ gross output x1,
𝑥3
x2, and x3 and represents the three sectors final demand, f1 f2 and f3
Page | 103
Multiplying both sides of equation (25) by [𝐼 − 𝐴]−1 we get output of the three
sectors X as a function of F equation (26)
(26)
The corresponding vectors of incomes originating in the three sectors is given

by equation (25),
(27)
Now, the aggregate income of the economy Y is the sum of sectoral incomes
as in equation (28)
𝑌 = ∑3𝑖=1 𝑦𝑖 (28)
Equations (27) and (28) indicate how aggregate and sectoral incomes of the
economy are given by the composition of final demand and the economy’s
input- output coefficient matrix (Chenery & Watanabe, 1958; Miller & Blair,
2009). In the I-O framework, the measurement of linkages has been made
based on the Leontief production matrix (the Matrix A) or the Leontief inverse
matrix [𝐼 − 𝐴]−1
Backward and Forward linkages
In the input-output models, two types of economic effects can be

identified. Production activity of a sector can have backward and forward
linkage effects on other sectors in an economy. A backward linkage shows the
amount of input sector demands from other economic sectors, whereas a
forward linkage shows how much the output of an economic sector is
demanded by the other economic sectors.
Backward Linkages
Based on Miller and Blair (2009) backward linkage serves as an

indicator of an industry’s relative importance as a user of inputs from the
production sector. For backward linkage ‘the direction of causation is in the
usual demand-side model’, indicating ‘interconnection of a particular sector
Page | 104
with those (“upstream”) sectors from which it purchases inputs’ (Miller &
Blair, 2009, p.555). Thus, the measure of the strength of the backward linkage
of sector j (the amount by which sector j production depends on inter-industry
inputs) is given by the sum of the elements in the jth column of the direct input
coefficients matrix, namely 𝑎𝑖𝑗. This is the direct backward linkage since the
coefficients are measures of direct effects only (Miller & Blair, 2009, p.556).
∑𝑖 𝑋𝑖𝑗
𝐵𝐿𝑖 = = ∑𝑖 𝑎𝑖𝑗 (29)
𝑋𝑗
Where; 𝐵𝐿𝑖 is backward linkage of the ith sector

𝑋𝑖𝑗 - is amount of commodity ‘i’ used in the production of commodity ‘j’
𝑋𝑗 - is gross output of ‘j’ sector including total intermediate purchases and
gross value added
𝑎𝑖𝑗 - is I-O coefficient representing amount of ith commodity used in the
production of jth commodity
Rasmussen (1957) has suggested a total measure of backward linkages,

comprising both direct and indirect effects. Total backward linkages represent
the change in economy-wide output in case the final demand for particular
sector increases by one unit (Miller and Blair 2009). The total backward
linkage of sector j is defined as the column sum of the Leontief inverse L = [
] which is thus equivalent to the output multiplier.
𝐵𝐿𝑗𝑅 = ∑𝑛𝑖=1 𝐿𝑖𝑗 (30)
For a reliable comparison of sectoral backward linkages,

normalizations are useful (Miller and Blair 2009). The normalizing of the
indices allows comparisons to be made across economies and periods (Pratt,
2011). Accordingly, in order to calculate the relative strength of the total
backward linkages, Rasmussen (1957) developed the index of power of
dispersion (normalized backward linkages), which reflects the relative extent
to which an increase in the final demand for the products of industry j is
dispersed throughout the total inter-industry system (Drejer, 2002). It is
described as:
1
∑ 𝑙
𝑛 𝑖 𝑖𝑗
∑𝑖 𝑈𝑖𝑗 = 1 (31)
∑ 𝑙
𝑛2 𝑖 𝑖𝑗
Page | 105
With ∑𝑖 𝑙𝑖𝑗 as the column sums of the Leontief inverse, sectors with
above average backward linkage reveal indices greater than one, indicating a
strong integration with the rest of the economy. Those with ‘below average’ or
weaker direct backward linkage will have indices lower than one (Miller and
Blair 2009).
Forward Linkages
On the other hand, for the forward linkage ‘the direction of causation is
in the supply-side model indicating ‘interconnection of a particular sector with
those (“downstream”) sectors to which it sells its output’(Miller & Blair, 2009,
p.555).
∑𝑗 𝑋𝑖𝑗 ∑ 𝑋
𝐹𝐿𝑖 = = ∑𝑛 ∑𝑗 𝑛𝑖𝑗𝑋 (32)
𝑋𝑖 𝑗 𝑗 𝑖𝑗
Where; 𝐹𝐿𝑖 - is forward linkage of the ith sector

𝑋𝑖𝑗 - is amount of commodity ‘i’ used in the production of commodity ‘j’
𝑋𝑖 - is gross output of ‘i’ sector (row vector) including intermediate and final
demands
Jones (1976) as cited in (Parra & Wodon, 2010a), suggests utilizing the
Ghosh inverse G = [gij] for the calculation of the forward linkages. He argues
that using the Leontief inverse twice (which has been suggested by Chenery
and Watanabe and Rasmussen) both as a measure of total backward and total
forward linkages involves a problem of double counting causal linkages, in as
much as sales from sector i to sector j are recorded as i’s forward linkage and
j’s backward linkage. Only one of these linkages can be effectively causal.
Using the Ghosh inverse as a measure of total forward linkages avoids this
problem (Andreosso-O’Callaghan and Yuen, 2004). Accordingly, total
forward linkages are defined as the row sums of G:
𝐹𝐿𝑖 = ∑𝑛𝑗 𝑔𝑖𝑗 (33)
This measure is equivalent to the input multiplier (Miller and Blair,

2009). Following Rasmussen’s sensitivity of dispersion index, forward
linkages based on the Ghosh inverse can be normalized correspondingly:
Page | 106
1
∑ 𝑔
𝑛 𝑗 𝑖𝑗
∑𝑖 𝑈𝑖𝑗 = 1 (34)
∑ 𝑔
𝑛2 𝑖 𝑖𝑗
3.3.5 Key Sector identification
Studies on key sectors identification usually identify key sectors by

calculating from both backward and forward linkage measures normalized
form and select those sectors with a high score on both measures (Miller and
Blair 2009). A key sector means that an increase in the final demand of the
sector’s output will have a large impact on the sectors that supply inputs to the
production of the key sector’s output (Drejer, 2002). Key sector identification
is also an important indicator of inter-industry linkages analysis. A sector is
considered an economy's key-sector if it has at least one of its backward and
forward linkage indices greater than one (Parra &Wodon, 2008). Similarly,
Andreosso-O’Callghan and Yue (2004) reported that if the values of both the
backward and forward linkages indices of a sector are above the
corresponding averages, the sector is called a key sector.
According to (Miller & Blair, 2009, p.559), based on the backward and
forward linkages sectors can be distributed in one over a four-way
classification.
i. generally, independent sector if both linkage measures are less than 1

ii. generally dependent on other sectors if both linkage measures are
greater than 1
iii. dependent on inter-industry supply if only backward linkage is greater
than 1and
iv. dependent on inter-industry demand if only forward linkage is greater
than 1
Wherever required appropriate messaging of data and price
adjustments have been made by using the well-proven approaches.
Page | 107
CHAPTER IV
STRUCTURAL CHANGE AND PRODUCTIVITY
IN ETHIOPIAN ECONOMY
The broad objective of the study is to analyze the structure and the
growth of the Ethiopian service sector. Service sector pattern and growth is
related to the structural change in an economy. Structural change refers to
change in sectoral shares and employment patterns. Structural change is also
manifested by a change in productivity, both at the aggregate and
disaggregates level. In this context, the present chapter explores the broad
structure and structural change in Ethiopian economy and its manifestation in
employment structure. Sector-wise analysis of productivity forms the next
portion of this chapter. Then, the last sub-section assesses the role of the
service sector in structural change using the shift-share analysis.
4.1 : Growth of the Service Sector in Ethiopia
In this section, the growth of service sector in Ethiopia from 1961 up to

2011 has been analyzed. The growth trends, share and composition of the
sector in output, employment and productivity are described. In addition, the
contribution of the service sector in output or gross value added and
employment is analyzed for the period (1961-2011). Finally, the contribution
of the sector in employment generation and per capita growth is estimated
using growth decomposition technique.
4.1.1 : Service sector growth in output, employment, and productivity

4.1.1.1 : Sectoral Output Growth
The main stylized facts observed in the structure of Ethiopian economy

within a half century are the decline of agriculture, the rise of service and the
stagnation of manufacturing sector. Agriculture share in the total output (gross
value added) declined by 43 percentage point from its highest level of 85
percent in 1960 to 42 percent in 2011 (de Vries, de Vries, Gouma, Pahl, &
Timmer, 2014) and declined further to 38 percent in 2014 (NBE, 2015). The
output share in the service sector has increased by 35 percentage points from 9
Page | 108
percent in 1960 to 44 percent in 2011. The manufacturing sector’s output
share has changed slightly from just 2 percent to 5.2 percent within fifty years
(1961-2011). This sectoral structure of the country makes Ethiopia unique
when it is compared with other countries that had the same level of per capita
income at various periods. As it is shown in Table 4.1, in 1961 Ethiopia had
the highest share of output in the agriculture sector (85 percent), which is
incomparable with the other countries' agricultural output share. The service
sector share was also one of the lowest, accounting just only 9 percent of total
output. Furthermore, the manufacturing sector had also share of only 2 percent
of the total output. These figures indicate the country’s level of development,
which is at the initial stage of development dominated by agriculture and
underdeveloped modern sector. In 2011, after 50 years, the Ethiopian
agriculture sector share to total output is still one of the highest (42 percent)
when compared with other countries output share. The manufacturing sector
output share just increased from 2 percent to 5 percent, which is lower than the
SSA average and other developing countries. However, what has changed
significantly is the share of the service sector. It increased from 9 percent to 44
percent, becoming the largest sector in output share and comparable with the
level of service output in other developing countries.
Table 4.1: Gross Value Added Share of Ethiopia across Comparator

Countries, 1961-2011
SSA Average* Ethiopia Ghana
1961 1981 2001 2011 1961 1981 2001 2011 1961 1981 2001 2011
Agriculture 51 18 26 19 85 69 50 42 37 41 33 26
Industry 15 33 21 22 3 5 7 9 20 12 10 19
Manufacturing 6 11 10 9 2 4 5 5 14 13 11 9
Service 29 37 44 50 9 21 38 44 29 34 46 46
China India South Korea
Agriculture 62 40 16 9 51 39 24 16 21 11 4 3
Industry 4 12 13 15 10 13 12 13 5 11 12 10
Manufacturing 6 20 35 37 12 15 17 18 4 18 29 35
Service 28 27 37 40 27 34 47 53 70 60 54 51
Source; Author’s computation using GGDC Database(de Vries, de Vries, Gouma, Pahl, &
Timmer, 2014)
*SSA – Sub-Saharan Africa average of 11 countries. These are Ghana, Kenya, South Africa,
Nigeria, Tanzania, Malawi, Botswana, Mauritius, Ethiopia, Senegal and Zambia
Page | 109
The service sector is an important leading sector in Ethiopian economy
by contributing the largest share to the recent economic growth. As it is
depicted in Table 4.2, the sector had the highest absolute growth rate than
other sectors up until 2010/11. After 2010/11, though the sector continued
with its highest contribution to GDP growth, its growth rate was below the
industry. This could be because of the government attention to the industry
sector under the first GTP (2010/11- 2014/15). Moreover, the contribution of
industry sector to the GDP growth for the first time exceeded the agriculture
sector in 2014/15. During the period, the service sector contributed 46.1
percent as the industry sector become the second important sector by
contributing 29.4 percent to the growth in GDP; while the agriculture sector
contributed 24.5 percent.
Table 4.2: Contribution of Sectors to Real GDPPC Growth of Ethiopia,

(2007/08 – 2014/15)
Year Real Agriculture Industry Services
GDPPC Absolute Contribution Absolute Contribution Absolute Contribution to
growth Growth to GDP growth to GDP growth GDP growth
(%) (%) growth (%) growth (%) (%)
(%) (%)
2007/08 7.3 7.4 33.8 10.3 9.5 16.1 56.7
2008/09 7.1 6.4 31.7 9.6 9.9 14.0 58.4
2009/10 7.5 7.6 34.9 10.8 10.6 13.2 54.4
2010/11 10.6 9.0 31.1 15.8 12.1 12.4 56.8
2011/12 6.1 4.9 25.3 19.7 24.1 9.6 50.6
2012/13 7.2 7.1 31.1 24.0 27.9 9.0 42.0
2013/14 7.5 5.4 22.3 17.1 21.3 13.0 57.1
2014/15 7.6 6.4 24.5 21.6 29.4 10.2 46.1
Source, (NBE, 2015)

The major structural phenomenon in Ethiopia, which is the growth of
the service sector and decline of the agriculture in the output share, is not
accompanied by the desired change in GDP per capita. Based on the 2015
income classification of (WB, 2015a), Ethiopia is one of the low-income
countries in the world with a GNI per capita income of 550 at a current USD
Page | 110
in atlas method. The country is far from the lower middle-income group lower
limit of 1045 USD per capita income. Ethiopian GNI has declined from 280 in
1988 to 110 USD in 2004. During these years Ethiopia was within its longest
civil war which ended in 1991. In 1998, the country once again entered into a
war with the neighboring Eritrea, which was separated from Ethiopia in 1992.
Other than wars, the Ethiopian economy was contained by the recurrent
droughts. It was after the period (2004-2014) that Ethiopia started to grow by
an average of 10.2 percent. This growth of the GDP is accompanied by an
increase in GNI from 110 USD in 2004 to 550 USD in 2014. Nevertheless,
still, Ethiopia is one of the low-income countries in the world ranking 10th
from the bottom. Ethiopia lags behind countries that are in the low-income
group in terms of its sectoral structures. The sectoral GDP shares and
employment share of sectors in Ethiopia are different from other countries.
The sectoral structural and income level of the Ethiopian economy

imply that the country is just in the initial stage of structural change. This can
be elaborated from theoretical as well as from the experience of other
countries. According to the conventional growth and structural change theories
through time as a country improves in its income level the structure of the
economy changes more often from the agriculture sector domination to the
modern sectors of manufacturing and service sectors. The role of the
agriculture sector in the economy will also become insignificant. It is at an
income level of around 1200 USD that a country will have a structural change
[see (Syrquin, 1988)] with the dominance of service and industry sector. The
experience of other countries which structurally transformed their economy
shows that the decline in the agriculture sector is accompanied by a growth in
the manufacturing sector and service sector. Table 4.1shows that countries
which achieved structural change like China and South Korea have
experienced a significant shift in their agriculture and manufacturing sectors.
The agriculture sector GDP share for China declined by 53 percentage points
from 62 percent to 9 percent (from 1960-2010). The Korean agriculture sector
GDP share has also decreased from its initially low level of 21 percent in 1963
to just 3 percent in 2010. Both China and South Korea had very low GDP
Page | 111
share in the manufacturing sector before 50 years. After 50 years their
manufacturing sector increased by 30 percentage points but the Ethiopian
manufacturing sector increased just only by 3 percentage points. Most of the
SSA countries including Ghana experienced a decline in the manufacturing
sector, which is explained by (Rodrik, 2014) as ‘Premature
deindustrialization’. However, what is observed in Ethiopia is the ‘stagnation
of manufacturing’ sector.
Figure 4.1: Sectoral Shares and GDPPC Growth of Ethiopia (1961-2011)
400
.8
GDP Per Capita (2005 Constant US$)

Sectoral Share to GVA
.6
300
.4
200
.2
100
0
1960 1970 1980 1990 2000 2010

Year
Agriculture Industry
Manufacturing Service
GDP Per Capita
Source; Author’s computation using GGDC Database and (WDI, 2016)
The Ethiopian agriculture and service sectors’ GDP share has changed
by larger percentage points as compared to other countries. Agriculture
declined by 43 percentage points and service increased by 35 percentage
points from 1960 to 2010. The change in the percentage share of the
manufacturing and other non-manufacturing industry sectors is very small.
These changes in the sectoral structure of the Ethiopian economy are an
important indicator that there is some sort of sectoral growth dynamics at a
very low level of GDP per capita [See Figure 4.1]. However, there are no
significant changes occurred in the Ethiopian sectoral structure and GDP per
capita within 20 years. Countries that were at the similar level of GDP per
capita and economic structure with Ethiopia are changed. Figure 4.2 below
shows the sectoral GDP share and GDP per capita of Ethiopia and other
countries.
Page | 112
Based on the data from World Development Indicator for 200
countries, a fitted regression line is computed which shows the relationship
between the share of GDP or GVA in sectors and the natural logarithm of
GDP per capita in 1994 for Ethiopia and 2013 for the rest of the countries. As
it is depicted in Figure 4.2, Ethiopia has agriculture sector whose share is far
greater than countries with similar level of income making the country one of
the most agrarian countries in the world. With the increase in the GDP per
capita, the share of agriculture in GDP has declined in 2013 as compared to
the 1994 level. In 2013, similar to the structure in 1994, the agriculture sector
is exceptionally one of the highest in its GDP share. What makes the Ethiopian
case different is that the share of agriculture in GDP is still high, even though
it reduced by 10 percentage point from 55 percent in 1994 to 44 percent in
2013. Based on the fitted regression line, the agriculture sector share in
Ethiopia should have been around 40 percent in 1994 and 30 percent in 2013,
which is the average level at the respective income level in the periods. Even
the lower middle-income countries have agriculture GDP share around 26
percent in 1994 and around 17 percent in 2013.
Figure 4.2: Relationship between Sectoral output shares in GVA and
GDPC (2013)
Agriculture GVA share Vs GDP per capita, 2013 Industry GVA share Vs GDP per capita, 2013
100
60
ETH_1994
Agriculture GVA share (%)
Industry GVA share (%)
20 40 60 80
ETH_2013
40
CHN
BWA KOR
20
IND
LMI LMI
IND
SSA
SSA
CHN ETH_2013
ETH_1994
BWA KOR
0
4 6 8 10 12 4 6 8 10 12
Natural Logarithm of GDP per capita (Constant 2005 USD) Natural Logarithm of GDP per capita (Constant 2005 USD)
Manufacturing GVA Vs GDP per capita, 2013 Service GVA share (%)
100
10 20 30 40 50
Manufacturing GVA share (%)
20 40 60 80
CHN KOR BWA KOR

SSA
LMI
IND
CHN
ETH_2013
IND
LMI ETH_1994
SSA
ETH_1994 BWA
ETH_2013
0
4 6 8 10 12 4 6 8 10 12
Natural Logarithm of GDP per capita (Constant 2005 USD) Natural Logarithm of GDP per capita (Constant 2005 USD)
Source: World Bank (WDI, 2016)

BWA – Botswana, CHN – China, ETH – Ethiopia, IND – India, KOR – Korea, LMI – Lower
Middle-Income countries, SSA – Sub-Saharan African Countries
Note: The Scatter points shows other countries and the line is fitted regression line
Page | 113
On the other hand, Ethiopia also had one of the lowest GDP shares in
the industry and manufacturing sectors. The manufacturing sector share in
Ethiopia is lower than the fitted regression line implying that Ethiopia is far
below from its peer countries that are in the similar level of income in its
manufacturing development. With the change in GDP per capita, the change in
the GDP sectoral share of the manufacturing sector within 19 years is
insignificant. There was only a 1.1 percentage point change from 4.5 percent
in 1994 to 5.6 percent in 2013. Accordingly, given the income level of
Ethiopia during 1994 and 2013, the share of GDP in the manufacturing sector
should have been around 10 percent of its GDP share on the fitted regression
line. In terms of the industry share in GDP Ethiopia is again at the lower level.
Ethiopia also lags behind some low-income countries that have increased their
GDP per capita by utilizing their natural resources. In Ethiopia, the share of
the industry sector that constitutes the mineral, construction and utility sub-
sectors has also small share in GDP. It is also one of the lowest as compared to
countries with same income level with Ethiopia. In 1994 the share of the
industry sector in GDP was 8.6 percent and it increased just only by 3.3
percentage point to 11.9 percent in 2013.
Nevertheless, it is only the service sector of Ethiopia that did not
deviate from the fitted regression line in both 1994 and 2013. The Ethiopian
service sector growth possesses a similar growth pattern with the world
average for respective level of incomes. In 1994, Ethiopia had service GDP
share of 35.6 percent which was more than Ghana and China. In 2013, it
increased to 43 percent, but it was lower than the share in Ghana and China.
Therefore, from the sectoral GDP shares data, we can infer that
Ethiopia is just only in the initial stage of structural change, in which the
country lags behind the world in its sectoral structure and income level.
However, the service sector has shown the relatively similar level of growth
which maintained the average service growth path of countries at the
respective income levels. The other general issue that can be observed from
the figure is that the slope of the fitted regression line for the service sector is
steeper, which implies the share of services in output increases with income.
Page | 114
That means high-income countries have a higher share of service in their total
output than developing countries. In addition, the share of the service sector is
more than 40 percent for most of the countries. This implies that countries are
becoming more and more dominated by the service sector even at the lower
level of income. On the other hand, the regression line for the manufacturing
and industry sectors increases first with the increase in income up to a certain
level. However, after reaching the maximum level, which is not more than 15
percent for manufacturing and not more than 30 percent for the industry, their
share in total output declines. This may imply that the demand for
manufactured and industrial goods gets concentrated and decline after a
country reaches certain high level of income. In other words, there would be
deindustrialization for high-income countries. This type of relationship
between the manufacturing sector and income is also observed using time-
series and cross-sectional data. For agriculture, as income grows the share in
total output declines fast. Thus, most of the countries in the world have
agricultural share below 20 percent accounting very small amount of the total
production.
4.1.1.2 : Sectoral Employment Growth

As it is described in the previous section, Ethiopia is largely an
agrarian economy not only because of its share of GDP is dominated by
agriculture but also mainly due to a large number of its population is
employed in agriculture for centuries. The share of population employed in
agriculture was 96 percent in 1960, and it declined to 73 percent in 2013 [see
Table 4.3]. As the share of agriculture in GDP decreased by 43 percentage
points, the share of employment decreased just only by 23 percent within a
half century (1960-2011). However, because of high population growth, the
number of people employed in agriculture increased from 9 million to 30
million within the same period. This level of employment share in the
agriculture sector is one of the highest in the world and it exceeds the SSA
average of 58 percent in 2011. No countries achieved structural transformation
and joined middle-income countries with a share of agriculture in employment
exceeding 60 percent, manufacturing employment lower than 10 percent, and
Page | 115
service employment lower than 20 percent. The share of employment in the
manufacturing sector in Ethiopia was much lower in the early periods as
compared to SSA average. Ethiopia’s labor employed in manufacturing has
increased from 1 percent in 1961 to 7 percent in 2011 as the SSA average
increased from 4 percent to 7 percent for the same period. In Ethiopia, the
sector employed 119 thousand people in 1960 and it increased to 2.8 million in
2011. Whereas, the service sector employed more than 218 thousand labor in
1960 and it increases to 6.7 million in 2011. This indicates that the shift of
employment is observed in the service sector more than the manufacturing
sector. As the Lewis-type of the dual economy model suggests in the
conventional growth model, workers that move out of subsistence or
traditional agriculture are absorbed in the modern manufacturing sector
(Lewis, 1954). However, in Ethiopia, the sectoral shift of the economy from
agriculture to service sector is opposite to the conventional Lewis model and
the structural change approach.
Table 4.3: Employment share for Ethiopia and Comparator Countries,
1961-2011
SSA average Ethiopia Ghana
Sector 1961 1981 2001 2011 1961 1981 2001 2011 1961 1981 2001 2011
Agriculture 77 61 65 58 96 89 85 73 60 58 53 39
Industry 4 4 3 4 0 0 1 3 6 3 5 6
Manufacturing 4 6 5 7 1 2 3 7 11 12 11 11
Service 15 29 27 31 2 9 11 17 23 27 32 44
Agriculture 77 68 50 37 72 72 59 55 62 33 10 7
Industry 3 4 8 10 2 2 6 8 4 7 8 8
Manufacturing 8 14 14 19 10 9 12 12 8 21 20 18
Service 12 14 28 35 16 17 24 26 26 39 63 67
Source; Author’s computation using GGDC Database

For China, South Korea and India 2010 is used instead of 2011 and for South Korea 1963 is
used instead of 1961
By associating the sectoral employment structure with income level,
we can observe that at the high-income level the share of employment in
agriculture is low and that of the industry is relatively high; and the service
sector is the highest. As figure 4.3 shows the share of employment in
agriculture in Ethiopia was far from the fitted regression line in both periods
Page | 116
of 1994 and 2010, though it declined from 89 percent in 1994 to 73 percent in
2010. Employment in the industry sector is lower than the service sector for all
the countries observed as most of the share in employment in the sector is
between 20-40 percent. Employment share in the industry sector also has
shown a decline in 2010 because of the deindustrialization in the high-income
countries and premature deindustrialization in the low-income countries of
Africa, Asia, and Latin America.
Figure 4.3: Relationship between Sectoral Shares in Employment and

GDP Per Capita (2013)
Agriculture Employment share Vs GDP per capita, 2013 Industry Employment Share Vs GDP Per capita, 2013
Agriculture Emloyment Share (%)
100
60
Industry Employment Share (%)
ETH_1994
80
ETH_2013
40
60
IND CHN
LMI
40
CHN IND
20
LMI
KOR
20
ETH_2013
KOR ETH_1994
0
4 6 8 10 12 4 6 8 10 12
Natural Logarithm of GDP per capita (in constant 2005 USD) Natural Logarithm of GDP per capita (in constant 2005 USD)
Service Employment Share Vs GDP per capita, 2013

Service Employment Share (%)
80
KOR
60
40
LMI CHN
IND
20
ETH_2013
ETH_1994
0
4 6 8 10 12
Natural Logarithm of GDP per capita (in constant 2005 USD)
Note: The Scatter points shows other countries and the line is fitted regression line
Source: World Bank, WDI (2016)
CHN – China, ETH – Ethiopia, IND – India, KOR – Korea, LMI – Lower Middle-Income
countries, SSA – Sub- Saharan African Countries
Figure 4.3 shows that in 2013, the share of employment in Ethiopia in

the service and industry sectors increased while it decreased in the agriculture
sector as compared to 1994. The following relationships are observed from the
figure. Employment in agriculture sector declines while employment in
service sector increases as income increases. Thus, high-income countries
have a low share of agricultural employment and a high share of service sector
Page | 117
employment as compared to low-income countries. However, the employment
share in the industry sector first increases as income increases, and it is
saturated at a certain level of income and later it tends to decline.
Generally, from the theoretical and empirical point of view, it is

expected to observe a high share of agriculture in both GDP and employment
for countries at low GDP per capita but not as high as the share of Ethiopian
agriculture does. It is also expected to observe lower share of manufacturing
and industry in GDP and employment for low-income countries but not as low
as the Ethiopian manufacturing and industry sector does. The share of service
sector in GDP and employment for Ethiopia is not also different from the
average of other countries. Therefore, Ethiopia possesses a different kind of
economic structure that deviates from most of the comparator countries. The
country has very high and declining share agriculture in output and
employment, very low share of manufacturing in output and employment, and
growing share of service sector in output and employment. The sectoral
change observed in the economy is thus the growth of the service sector in
total output and employment.
4.1.1.3 : Sectoral Productivity growth
Productivity growth is an important indicator of sectoral performance

and structural changes. The level of productivity growth is measured as the
ratio of output in each sector per unit of labor employment. As it is depicted in
the Table 4.4, the level of labor productivity of Ethiopian economy is much
lower than the SSA countries average, China, South Korea and India. But it is
higher than Ghana’s level of productivity in all the sectors. The level of
agricultural productivity in Ethiopia prior to 2001 shows a decline from 2.6 in
1961 to 1.6 in 2001, but later it increased from 1.6 to 2.4 in 2011. However,
the agricultural productivity per unit of labor has never declined in all the
years for other countries except for Ethiopia. In addition, the manufacturing
and the industry sectors productivity level have declined in 2011 when
compared to the level in 1961. The industrial sector has shown a rise from
10.9 in 1961 to 14 in 1981 and later on it declined consecutively to 7.7 and 5.8
Page | 118
Table 4.4: Labour Productivity level of Ethiopia and Comparator Countries, 1961-2011
Sector SSA Average Ethiopia Ghana
1961 1981 2001 2011 1961 1981 2001 2011 1961 1981 2001 2011
Agriculture 73.3 114.2 161.7 116.4 2.6 2.3 1.6 2.4 0.9 0.7 1.0 1.4
Industry 291.1 1740.4 1538.3 593.6 10.9 14.0 7.7 5.8 3.2 1.8 2.1 3.2
Manufacturing 156.5 722.9 727.6 762.0 4.8 7.1 4.4 3.2 2.0 1.1 1.6 1.6
Service 209.6 497.3 650.4 268.1 12.1 7.0 9.2 11.3 2.0 1.4 2.3 2.1
Total 111.4 383.6 405.7 214.7 3.0 2.9 2.7 4.2 1.6 1.1 1.5 2.0
China South Korea India
Agriculture 1.6 2.5 5.2 9.9 2038.6 3752.5 11766.9 17019.1 16.5 16.8 23.4 29.9
Industry 1.8 7.5 35.0 71.1 4569.0 11642.8 39981.1 58148.5 42.7 75.6 94.8 161.5
Manufacturing 1.5 6.2 40.2 75.9 2935.1 9775.2 39757.5 65365.6 28.6 49.3 82.6 155.4
Service 4.5 8.6 21.8 45.6 16272.2 17261.7 23258.3 25784.0 37.8 64.5 114.4 208.3
Total 2.0 4.3 16.5 39.8 6082.6 11211.0 26711.9 33761.3 23.1 31.4 57.5 101.7
For China, South Korea and India 2010 is used instead of 2011 and for South Korea 1963 is used instead of 1961
Page | 119
in 2001 and 2011 respectively. Similarly, the manufacturing sector has
increased in its productivity per unit of labor from 4.8 in 1961 to 7.1 in 1981
but later on it declined to 4.4 in 2001 and further to 3.2 in 2011.
When we see the service sector productivity per unit of labor in

Ethiopia, it has declined from 12.1 in 1961 to 7 and in 1981. Nevertheless,
after 1981 it increased to 9.2 and 11.3 in 2001 and 2011 respectively. The
level of productivity in the service sector is relatively higher than the
manufacturing and other sectors for Ethiopia, Ghana, and India. For the SSA
average, the manufacturing productivity per unit of labor was higher than the
service in the 1981s and early 2000; but it was very low in the early 1960s and
in 2011. Similarly, in most of the period (1961-2011), the Ethiopian
manufacturing sector productivity was lower than the service sector. The ratio
of manufacturing labor productivity to service in Ethiopia was above one only
for six consecutive years of 1981-1986. For the rest of the years, the service
sector per unit labor productivity was higher than the manufacturing.
However, for SSA average of eleven countries, the relative productivity of
manufacturing to services was higher than one, implying that in most of the
years the level of labor productivity in the manufacturing was higher than the
service sector. The Ethiopian case of higher productivity of the service sector
over the manufacturing sector is against the theoretical argument that
manufacturing sectors possess higher productivity and employment capability
over the services. Therefore, in Ethiopia, the service sector is an important
sector for the growth of overall productivity growth as it has larger
productivity level than other sectors.
As it is depicted in figure 4.4, the ratio of labor productivity in

manufacturing relative to services is lower than the SSA. However, contrary
the productivity in market service relative to manufacturing sector is a higher
ratio than the SSA averages.
Page | 120
Figure 4.4: Relative Labor Productivity, Ethiopia, and SSA, (1961 – 2011)
2
1.5
SSA
1
.5
Ethiopia
0
1960 1970 1980 1990 2000 2010

Year
Relative Labour Productivity of manufacturing to Service of SSA and Ethiopia
Source; Author’s computation using GGDC database
In another figure (figure 4.5) depicted, the productivity of labor per

unit of output for the manufacturing sector has increased from 1961to its peak
level in 1984. But after 1984 it is decreasing and has crossed its historical
lowest level. On the other hand, the level of productivity in the service sector
has declined from 1961 and reached its lowest level in 1984. Figure 4.5 shows
that as the labor productivity of the manufacturing sector increased the
productivity level for the service sector declined and vice versa. The high level
for the ratio of relative labor productivity may be as a result of the already low
level of productivity in the manufacturing sector. The labor productivity for
service declined to its minimum level of 6.5 in 1984 and in the same year
labor productivity level for manufacturing reached its peak level of 7.5, and it
was only in this year that the productivity of manufacturing sector became
greater than the service sector. The reason for the rise of manufacturing
productivity and decline of service productivity could be associated with the
1984 Ethiopian drought and famine. However, this requires further
investigation to reason out how it contributed to the decline of manufacturing
sector productivity and the rise of service sector productivity after 1984.
Page | 121
Figure 4.5: Sectoral Labor Productivity Levels of Ethiopia, (1961-2011)
3
Market Service
Service
2
Non-Market Service
Manufacturing
1
Agrculture
0
1960 1970 1980 1990 2000 2010

Year
Source, Author’s computation using GGDC Database
When we see the productivity of sectors period wise, we can observe

that the manufacturing productivity was increasing and contributing to growth
until 1984. Even though the manufacturing sector employment and output
increased after 1984, its relative productivity declined. As it is observed in
Figure 4.5, prior to 1984, the relative productivity in all sectors was declining
except for the manufacturing sector. Since then, the relative productivity of
the service sector started to increase mainly driven by the higher level of
relative productivities for the market service and increasing trend of the non-
market service. The productivity growth in the public service sector has a
greater share in the growth of the relative productivity of the non-market
service.
4.1.2 : Growth and Composition of Service sub-sectors
In this study, the Ethiopian service sector is classification based on the

ISIC rev.4(ISIC, 2002) and the 10 sector GGDC database classifications (de
Vries, de Vries, Gouma, Pahl, & Timmer, 2014). Based on this the service
sector is broadly classified as the market service and non-market services. The
market service constitutes the wholesale and retail trade, and hotel and
restaurants (referred as distributive trade services), transport, storage and
communication (aggregated as transport and communication services), and
Page | 122
finance, insurance, real estate and business services (aggregated as business
and finance services). The non-market services include the public service
(government service) and the community, social and personal service
(aggregated as personal service).
4.1.2.1 : Service Sub- sectors Growth in Output (Gross Value Addition)
The GVA share of services in most countries has approached 50

percent of the total economy output. Some countries like India and South
Korea have more than 50 percent share. As depicted in table 4.5, the China’s
share of the service sector is lower than the level of other countries. In
addition, except for China, the market service takes the largest share of
services output as compared to the non-market services. Ethiopia’s market
services and non-market service share in 2011 reached 31 percent and 13
percent of the total GVA respectively. This level is lower than the level
observed in SSA averages, Ghana, India and South Korea. But, the share of
distributive trade service in Ethiopia, which is 19 percent of total output, is
more than the share the Indian and South Korean distributive trade services
share, which are 18 percent and 12 percent of the total GVA output
respectively in 2011. The other market services, transport and communication
and finance and business services share in Ethiopia are lower than the SSA
average and China’s level. Exceptionally, China has a share of Finance and
business, which takes only one percent shares of its GVA for a half century.
However, the size of China’s economy has increased tremendously and
become the world leading economy in its size, implying that the value of the
finance and business services has indeed increased.
On the other hand, the non-market service for Ethiopia is dominated by

the public/government service. The public service increased from just 3
percent in 1981 to 11 percent in 2011, whereas the personal service increased
by 2 percentage points from 1 percent in 1961 to 3 percent in 2011. Contrary
to the Ethiopian case, China, which is a socialist market economy country,
increased its government service share in total output from 5 percent in 1961
to just 6 percent in 2011. And the personal services sector increased from just
Page | 123
Table 4.5: Gross Value Added (GVA) Share of Service Sectors, Ethiopia and ComparatorCountries, (1961-2011)
SSA Average Ethiopia Ghana
Service Sub sectors 1961 1981 2001 2011 1961 1981 2001 2011 1961 1981 2001 2011
Services 29 37 44 50 9 21 38 44 29 34 46 46
Market Services 21 24 33 38 7 16 24 31 23 22 33 33
Distributive Trade 16 15 20 20 5 12 14 19 12 9 11 11
Transport & communication 4 5 6 10 1 3 5 6 9 10 16 16
Finance & business 2 5 8 7 0 2 4 7 1 2 6 6
Non-Market 7 13 11 13 3 5 14 13 6 13 14 14
Government 7 12 9 12 2 3 11 11 5 11 9 9
Personal 1 1 1 1 1 2 3 3 1 2 4 4
Services 12 14 28 35 27 34 47 53 70 60 54 51
Market Services 6 7 12 15 18 22 32 39 22 25 32 29
Distributive Trade 3 4 8 10 11 13 16 18 9 12 14 12
Non-Market 6 7 16 19 8 12 15 14 48 34 22 22
Government 5 4 5 6 6 10 12 12
Personal 2 2 11 14 3 2 2 2
Page | 124
2 percent in 1961 to 14 percent in 2011. From the service sub-sectors, the
personal service sub-sector constitutes the major share of China’s total output.
But in Ethiopia, the personal service sub-sector constitutes the least share in
the total output in 2011.
The growth of Ethiopian service sector is one of the important

phenomena in the structural change of the economy. The sector has grown
from its lowest level below 10 percent share in output to around 45 percent
shares within a half century. The larger increase in the share of the output
(GVA) came from the distributive trade service and the public services.
Immediately after the change in government in 1974, the public service share
in output started to rise because of socialist economic policy favoring
government control in most of the economic spheres. In addition, because of
the war, the country has entered with the then Somali government in 1977 and
because of the civil war until 1991; the military expenditure increased the
share of public service. In 1999/2000, the public service share reached the
maximum 12 percent share of total output when Ethio-Eritrean border war
broke out and in the same year 1999, the military expenditure of Ethiopia has
reached 9.6 percent of GDP (WDI, 2016).
On the other hand, the distributive service sector has grown rapidly
during the Imperial government from 1961-1974. During the Military
government, the distributive sector growth stagnated because of the
discouraging business environment in the country because of nationalization
of business and private properties. After the change in government in 1991,
the sector is increased gradually from 10 percent share in total output in 1991
to 19 percent in 2011. The finance and business, the transport and
communication and personal services are also increasing their share in total
output. However, the growth in these sub-sectors is very slow and the share
for finance and business and transport and communication remained around 6
percent as the share for personal service is still stagnating on 3 percent of the
total output of the country. Therefore, the Ethiopian service sector output
growth can be characterized by the higher and increasing share of the
Page | 125
distributive trade and public service sub-sectors and low share of personal
services.
4.1.2.2 : Service Sub-sectors Growth in Employment
Employment in service sector generally increased in Ethiopia by 15

percentage points within a half century from just 2 percent to 17 percent in
(1961-2011). Even though it manages to employ the growing urban population
and few rural populations, it is far below the share in the SSA and other
countries. This increase in employment is largely accounted by the distributive
service sub-sector, which is followed by the employment in government and
personal services. It is also larger employing sub-sector in all other countries
except in China.
As shown in Table 4.6, employment share in other market services

other than the distributive service is too small in Ethiopia. The distributive
sector employment has increased by 10 percentage points from 1 percent in
1961 to 11 percent in 2011. The transport and communication, and the finance
and business service sub-sectors have only one percentage share each in the
total employment in 2011. Employment in these sectors requires labor that is
more skilled and pays more wages as compared to other service sub-sectors.
As a result, productivity in these sectors is also much greater than other
sectors. Therefore, firms in these sectors may employ few skilled labors in an
attempt to raise productivity and generate more profit. The employment share
in finance and business service in South Korea is very high and it was 14
percent in 2011 implying a well-developed business and financial sector.
However, other countries like China and India have a small share of finance
and business services as compared to South Korea (1 percent for China and 2
percent for India).
In Ethiopia, the share of employment in the service sector increased

from 1961 to 1984. After the 1984 drought and famine, all the service sub-
sectors share to the total employment declined until 1991. After 1991, the
growth in the employment share of the distributive trade service was very fast
and it was followed by the public service. Specifically, the employment share
Page | 126
for the distributive service sector has more than doubled within 7 years
increasing from 5 percent in 2004 to 11 percent in 2011. The distributive
service sector constitutes the largest share of market service and much of the
shift in employment occurs in the distributive service since 1961. From 1961
to 2011, the employment shares in the distributive sector increased by 13 folds
from 0.8 percent to 10 percent of the employment in the service sector. The
output in the distributive sector has also increased by from 5 percent in 1961
to 19 percent in 2011.
Table 4.6: Employment Share of Service Sectors, Ethiopia and

SSA Average Ethiopia Ghana
Service Sub sectors 1961 1981 2001 2011 1961 1981 2001 2011 1961 1981 2001 2011
Services 15 29 27 31 2 9 11 17 23 27 32 44
Market Services 11 16 16 20 1 4 5 12 17 17 22 31
Distributive Trade Service 9 13 13 15 1 4 5 11 14 15 18 25
Non-Market 5 13 11 11 1 5 5 5 6 10 10 13
Government 2 6 6 6 1 2 3 2 4 7 5 6
Personal 3 7 5 5 0 3 3 2 2 3 4 7

Services 12 14 28 35 16 17 24 26 26 39 63 67
Market Services 6 7 12 15 7 8 15 19 16 27 44 44
Distributive Trade Services 3 4 8 10 5 5 10 12 12 20 27 24
Non-Market 6 7 16 19 9 9 8 7 10 11 19 22
Government 5 4 5 6 7 6 5 4
Personal 2 2 11 14 2 3 3 3
Source; Author’s computation usingGGDC Database

Note: For China, South Korea and India 2010 is used instead of 2011 and for South Korea
1963 is used instead of 1961. Values are 0 due to rounding
4.1.2.3 : Productivity of Service Sub-Sectors

The rise in the level of labor productivity in the service sectors is one
of the factors that has contributed to the increasing dominance of service
sector in output and employment both in developing and developed countries.
In Ethiopia, the labor productivity of service at aggregate level has declined
from 12.1 in 1961 to 7.0 in 1981. However, the productivity level has
improved in 2001 as it increased from 7.0 in 1981 to 9.2 in 2001 and 11.3 in
Page | 127
Table 4.7: Labour Productivity of Service Sub-Sectors, Ethiopia, and Comparator Countries (1960-2011)
SSA average Ethiopia Ghana
Service Sub-sectors 1961 1981 2001 2011 1961 1981 2001 2011 1961 1981 2001 2011
Services 209.6 497.3 650.4 268.1 12.1 7.0 9.2 11.3 2.0 1.4 2.3 2.1
Market Services 222.6 579.0 822.3 317.8 18.5 12.0 11.4 11.0 2.1 1.3 2.3 2.1
Distributive Trade 205.2 437.8 636.7 277.3 20.0 9.9 7.8 7.3 1.3 0.7 1.0 0.9
Transport & comm.. 266.3 757.5 1055.9 531.8 18.8 18.3 32.6 45.9 5.5 5.2 8.2 9.3
Finance & business 386.8 2393.6 2058.4 341.5 9.8 65.8 70.0 54.1 7.7 5.1 5.9 4.9
Non-Market 179.7 393.9 394.5 178.7 6.3 2.9 6.9 12.1 1.7 1.4 2.2 2.1
Public service 348.8 793.0 667.7 288.2 6.7 3.8 10.1 18.7 2.1 1.6 2.8 3.0
Personal service 40.6 60.3 92.4 55.0 5.5 2.0 3.4 5.0 0.7 0.9 1.5 1.3
Total Economy 111.4 383.6 405.7 214.7 3.0 2.9 2.7 4.2 1.6 1.1 1.5 2.0
Services 4.5 8.6 21.8 45.6 37.8 64.5 114.4 208.3 16272.2 17261.7 23258.3 25784.0
Market Services 6.3 11.6 35.4 74.1 61.5 84.5 120.5 213.5 8349.8 10342.0 19567.1 21927.4
Distributive Trade 6.5 12.4 22.4 49.2 51.6 74.2 92.9 162.5 4252.4 6618.1 13921.9 17632.1
Transport & comm.. 5.2 9.5 40.2 79.2 44.6 66.9 112.1 199.6 3035.7 12099.4 35514.1 39599.7
Finance & business 8.3 13.1 110.3 229.4 374.5 343.3 358.2 507.9 71261.5 34354.8 24841.6 21072.5
Non-Market 2.7 5.4 11.3 23.5 20.6 45.4 103.1 195.0 28789.0 33933.4 32022.7 33476.6
Public service 2.8 6.6 28.4 62.2 18.2 54.3 146.4 284.9
Personal service 2.2 3.4 3.4 7.3 29.4 27.5 40.6 75.8 28789.0 33933.4 32022.7 33476.6
Total Economy 2.0 4.3 16.5 39.8 23.1 31.4 57.5 101.7 6082.6 11211.0 26711.9 33761.3
Page | 128
2011. Similarly, for China, India, and South Korea the service productivity
level has risen in 2001 and 2011. However, for the SSA average and Ghana
the service productivity level has declined in 2011 as compared to 2001.
In Ethiopia, the finance and business service sector experienced an

increase in productivity until 1991 and declined since then. The finance and
business service sub-sector have the highest labor productivity next to the
utility sector. The productivity level of the finance and service sector has also
started to increase with the introduction of new business ventures and financial
institutions in Ethiopia during the Imperial period and reached its highest level
in 1991. Even though the financial and business output and employment are
increasing after 1991, its productivity level is declining.
As it is shown in Table 4.7, the labor productivity level of the market

service is greater than the non-market services in all the other countries cases
except for South Korea. The Ethiopian market service had also labor
productivity greater than the market service in all the years except 2011. From
all the service sub-sectors, the Ethiopian finance and business service has the
highest productivity level followed by the transport and communication
service sector. This feature is similar to China and India, who have high
productivity level in finance and business followed by transport and
communication services. However, for SSA average, Ghana, and South Korea,
the transportation and communication productivity level was greater than the
finance and business services. From the non-market services, the government
service has higher productivity level than the personal service sub-sector in all
the countries observed.
The labor productivity levels of the market service sectors have been
over the manufacturing sector in all the periods [see Figure 4.6]. However, the
level of productivity in the market services is slightly declining. This situation
is similar to other Sub-Saharan African countries, where workers relocated
mainly to the market services industries. According to de Vries et al (2013),
‘market services activities had above-average productivity levels, but the
productivity growth was low and increasingly falling behind the world
Page | 129
frontier’(de Vries, Timmer, & de Vries, 2013). On the other hand, the non-
market service productivity level has been declining even though it was
greater than the manufacturing sector in the early periods. It started to rise in
1984 as the manufacturing sector productivity level starts to decline. In 2010
and 2011, the non-market service productivity level has surpassed the market
service productivity level due to the higher productivity level observed in the
public service. The aggregate economy-wide productivity level has also shown
improvement since 2004 as a result of high productivity observed in the non-
market services.
Figure 4.6: Labour Productivity of Services and Manufacturing in
3.2
2.8
Productivity level (in natural logarithm)
Market Service
2.4 Service
2.0
Manufacturing
1.6
Non-market Service
1.2
0.8 Aggregate economy
0.4
1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
4.1.3 : Composition of Service sub Sectors in GVA and Employment
The service sector has brought changes in the structure of the

Ethiopian economy. However, there are no significant changes observed in the
GVA of service sub-sectors composition. The larger proportion of service
GVA was constituted by the distributive service sector. As it is described in
Figure 4.7 and Table 4.8, the distributive trade service covered over half of the
total service sector output from 1961 up to 1991. After 1992, its share
decreased and reached 42 percent of the total service sector output in 2011.
Page | 130
However, the share of the public service in the output has increased since 1967
as the share of the distributive service decreased. The share of public service
increased from its lowest share of 10 percent in the early 1970s to 24 percent
in 2011. On the other hand, the share of personal services that includes
community and social services has shown a decline from its highest level of
10 percent in the late 1990s to 6 percent in 2011. The transport and
communication service share in GVA has also shown little change increasing
from 10 percent in 1961 to its highest level of 15 percent in 1992 and then it
declined to 13 percent in 2011. On the other hand, the finance and business
service share has increased steadily from its lowest level of 5 percent in 1961
to 15 percent in 2011.
Figure 4.7: Composition of Service Sector in GVA, Ethiopia (1961-2011)

100%
90%
80%
Percentage Share
70%
60%
50%
40%
30%
20%
10%
0%
1979
1961
1963
1965
1967
1969
1971
1973
1975
1977
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Year
Distributive Trade Transport and communication

Finance and business Public services
Personal Services
From the employment side, the distributive trade service is again the
dominant contributor to the total service employment as shown in Table 4.8
and Figure 4.8. However, the share of the distributive trade service sector in
employment was below 50 percent for most of the period up until 2004 and
exceptionally it was 50 percent in 1970 and 1971. Its contribution to
employment share increased from 44 percent in 1990 to 65 percent in 2011 as
its contribution to GVA share has declined from 50 percent to 42 percent in
Page | 131
the same period. On the other hand, the employment shares in the public
service and personal service have declined since 2000. The public sector has
exhibited a larger decline in its share of employment within the service sector.
It has constituted 35 percent of employment in services in 1961 and declined
steadily to 25 percent in 1990 and 15 percent in 2011. This indicates that the
government limited the public service employment in the country even though
government sizes increases due to its huge public investment and increase in
output. On the other hand, the decline in employment share for the public
service could be associated with the shift of labor from public services to
personal services in the distributive trade services and growing employment
opportunity in the non-governmental organizations (NGOs). In addition, it is
also associated with a change in the economic system from high government
control in the socialist system to a market economy in 1991. Further,
following the economic reforms, privatization of public enterprises may have
contributed to the decline of public service employment share. In recent
periods, there is also growing tendency among the urban population to engage
in self-employment activities, specifically in the distributive trade service.
Table 4.83: Composition of Service Sectors GVA and Employment Share

in Ethiopia (1961-2011)
Service Sub-Sectors GVA Share Employment Share
1961 1981 2001 2011 1961 1981 2001 2011
Market Service 72.6 77.2 62.5 69.8 47.4 45.0 50.2 71.7
Distributive Trade 57.4 55.8 38.5 42.0 34.6 39.4 45.1 65.4
Transport and communication 10.0 11.8 12.7 12.6 6.4 4.5 3.6 3.1
Finance and business 5.1 9.7 11.4 15.2 6.4 1.0 1.5 3.2
Non Market Service 27.4 22.8 37.5 30.2 52.6 55.0 49.8 28.3
Public services 19.3 14.5 28.9 24.1 34.5 26.6 26.4 14.5
Personal Services 8.2 8.3 8.6 6.1 18.0 28.4 23.4 13.8
Service 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
The employment share composition of the personal service has

increased from 18 percent in 1961 to 32 percent in 1992 and declined to 14
percent in 2011. Furthermore, the employment share of both transport and
Page | 132
communication and finance and business services has declined. The
employment share of transport and business service constitutes 6.4 percent of
the service employment in 1961. This level of share has declined continuously
and reached 4.7 percent in 1992 and 3.1 percent in 2011. Similarly, the share
of the finance and business service declined from 6.4 percent in 1961 to 0.8
percent in 1984 and started to rise to 1.1 percent in 1984 and 3.2 percent in
2011. The employment share of the finance and business, and transportation
and communication sub-sectors is very small as both sectors involve highly
skilled labor force and pay higher wages than the other service sub-sectors.
However, they have an output share that is relatively higher, which contributes
to the higher productivity observed in the sectors.
Figure 4.8 Composition of service sector employment in Ethiopia (1961-

2011)
100%
Percentage Share
80%
60%
40%
20%
0% 2001
2009
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2003
2005
2007
2011
Year
Distributive Trade Transport and communication

Finance and business Public services
4.1.4 : The Contribution of Service sector in Output Growth and

Employment
Ethiopia had stable growth in GDP from 2004 to 2011, which at least
indicated a hope to witness structural change and join the middle-income
country. However, the achievement of these goals is affected by the current
growth and structure of the sectors. The current high growth and dominance
observed in the service sector may be considered as unwanted and untimely
growth as the Ethiopian manufacturing sector is still lagging behind in all
Page | 133
measures. At this time when there is no developed industrial sector (including
mining, construction, and manufacturing), the role taken by the service sector
may be used to fill the gap by contributing to GDP and employment growth
and to stimulate growth in the other sectors. Therefore, this section will try to
highlight the contribution of the service sector in GDP (GVA) and
employment for the years 1962-2011 using descriptive statistics and Shapely
growth decomposition.
4.1.4.1 : Descriptive analysis
The Contribution of Service Sector to Annual GVA growth (1962-2011)
In Figure 4.9, the aggregate economy growth of Ethiopia is full of ups

and downs fluctuating from positive to negative growth rates. In between 1974
and 2003, the GVA growth rate recorded negative growths in six periods in
1975, 1978, 1984-85, 1990-92, 1998 and 2003. The 1975 negative growth may
be linked with the Coup d’état and change of the Imperial government by the
Military government and at the same time due to the 1974 drought and famine.
The 1978 and 1998 were a period when Ethiopia was in a war with its
neighbors Somalia and Eritrea respectively. The 1984-85 and the 2003
negative growth in GVA were due to drought in the country. The 1990-92 a
decline in growth was due to the civil war and change of the military
government by the current EPRDF government.
During the positive growth in the GVA during 1962-1973, on average

47 percent of the GVA growth was contributed by the agriculture sector. The
manufacturing sector contributed 8 percent of the growth and the service
sector contributed 39 percent. In 1974-1985, the average GVA growth was
around 1 percent with the minimum of -9.7 percent and maximum 11.5
percent. The larger share of the fall in the growth was attributed to the poor
performance of agriculture because of drought and war. However, on average
the service sector has contributed positively to maintain the positive average
growth rate of 1 percent. The average contribution of agriculture was just only
16 percent growth, manufacturing 12 percent and the service sector 64 percent
of the average total economy growth. The average GVA growth for 1993-2003
Page | 134
was 3.9 percent with the minimum -4.4 percent and maximum 11.9 percent.
Thus, the larger contribution for the rise in the GVA growth was attributed
more to the service sector.
The growth in the economy from 2004-2011 was stable and the
average growth was 11.2 percent with a minimum of 9.9 percent in 2010 and
the maximum of 12.8 percent in 2005. The contribution of agriculture to the
total economy growth rate has declined from 68 percent in 2004 to 35 percent
in 2011 and the period average was 41 percent. The manufacturing sector’s
contribution to the output growth in Ethiopia was 5 percent on average during
2004 -2011 and it increased from 3 percent in 2004 to 6 percent in 2011.
Whereas the service sector contribution has increased from 20 percent in 2004
to 62 percent in 2009 and 45 percent in 2011 with the period (2004-2011)
average of 47 percent, which is more than the agriculture sector. During this
period, the larger contribution from the service sub-sectors is attributed to the
distributive service sector. Therefore, the leading role of agriculture in driving
growth in Ethiopia was overtaken by the service sector since 2007.
Figure 4.9: Contribution of sectors to Annual GVA growth (1962-2011)

100% 15%
80%
Percentage contribution to GVA growth
10%
60%
40%
5% GVA growth (%)
20%
0% 0%
2000
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2002
2004
2006
2008
2010
-20%
-5%
-40%
-60%
-10%
-80%
-100% -15%
Year
Agriculture Manufacturing Services Total Economy Growth
Page | 135
Therefore, we can observe that the role of the service sector to the total
economy growth is increasing from time to time. The market service sector
specifically the distributive service sector takes the largest share of the
contribution to total output growth. After 2004, the service sector leading role
could be attributed to the change within the service sector itself as its output
share and employment share has increased. The other reasons for the growth
of service in Ethiopia, which is associated with income, productivity
difference, and change in demand structure, are studied in chapter 6.
The Contribution of Service Sector to Annual Employment growth
On the employment side, the total employment growth in Ethiopia is

largely affected by the dominant employing sector, agriculture. Since Ethiopia
has huge agricultural labor share, around 80 percent of the employment
growth is contributed by the agriculture sector. Apart from agriculture, the
service sector specifically the non-market service is contributing positively to
the growth in employment since 2004.
Figure 4.10: Contribution of Sectors to Employment Growth (1962-2011)

100% 6.0%
Percentage Contribution to Employment
80%
5.0%
60%
Employment Growth (%)

4.0%
40%
Growth
20% 3.0%
0%
2.0%
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
-20%
1.0%
-40%
-60% 0.0%
Year
Agriculture Manufacturing Services EMP Growth
Figure 4.10 shows that there has been a 2 to 4 percent employment

growth from 1962 -1990. There was no negative employment growth but near
Page | 136
to zero employment growth was observed when Eritrea separated from
Ethiopia and become an independent country in 1993. From 1990 to 1993
employment growth declined below 2 percent as a result of civil war and
succession of Eritrea from Ethiopia. After 1994 the employment growth has
increased though it declined in 1998 and 2002/3 because of Ethio-Eritrean war
and drought respectively. However, since 2004 the employment growth is
increasing along with the GDP growth. The agriculture sector has been
contributing the largest share to the employment growth in all the periods prior
to 2006. In 2006 and 2007, the contribution of agriculture to employment
growth was overtaken by the service sector. Its contribution to employment
growth was only around 40 percent. The dominance of the agriculture sector
contribution in employment growth was reduced since 2006/07 as both the
service sector contribution increased to 40 percent and the manufacturing
sector contribution also increased to about 20 percent.
Therefore, from the descriptive analysis, we can observe that the role
of the service sector to the total economy growth in GVA and employment is
increasing. The leading role of the service sector could be attributed to the
change within the service sector itself as its output share and employment
share has increased along with a rise in its productivity level. The
decomposition analysis can further indicate the sources of this growth.
4.1.4.2 : Growth Decomposition Analysis
This section discusses the Ethiopian service sector contribution to the

change in total employment rate and per capita value added during 1999-2013
by using the Job Generation and Growth/JoGG decomposition tool. It
describes how growth is reflected in the sectoral pattern of growth and
employment generation. The data used for this purpose is collected from the
National Bank of Ethiopia (NBE, 2015), and the National Labour Force
Surveys (NLFS) of 1999, 2005 and 2014 (CSA, 1999), (CSA, 2005), and
(CSA, 2014). The periods are classified based on the growth performance of
the country as the low growth period (1999-2004) with average GVA growth
Page | 137
of 4.8 percent and high growth period (2004-2014) with average growth rate
of 11 percent.
Table 4.94: Decomposition of Growth in PCGDP, Ethiopia (1999 - 2013)

1999-2005 2005-2013
Percentage
Percentage of
Decomposition of Growth in PCGDP of total
ETB* ETB total change
change in
in PCGDP
PCGDP
Total Growth in PCGDP/ per capita GDP 527.2 100.0 3,568.7 100.0
Growth linked to output per worker 173.9 33.0 3,592.3 100.7
Growth linked to ∆ employment rate 255.1 48.4 157.8 4.4
Growth linked to ∆ in the share of population of working-age 98.3 18.6 -181.3 -5.1
Source: Author’s computation using decomposition tool

*ETB-Ethiopian Birr (1USD=14.4 ETB) in 2010/11; Monetary value is in (2010/11 constant ETB)
The growth in per capita GDP (Value Added) decomposition result in

Table 4.9 shows that employment rate growth has accounted the largest share
(48.4 percent) to the total growth in GDP per capita followed by productivity
growth (33 percent) in 1999-2005. However, during 2005-2013 period
productivity growth has much more pronounced contribution to the GDP per
capita growth. The contribution of employment rate has declined to 4.4
percent as the demographic change reduced the growth in per capita value
added by 5.1 percent. The contribution of employment changes to the overall
change in employment rate is decomposed in Table 4.10. It shows how the 5.7
and 2.3 percentage points of growth in the employment rate in 1999-2005 and
2005-2013 respectively were distributed among the different sectors. During
1999-2005, the agriculture sector accounted 85.1 percent of the growth in
employment rate followed by manufacturing, construction and public services
contributing 10.2 percent, 7.8 percent, and 5.8 percent respectively. However,
due to the negative share of the distributive trade service in employment rate,
the service sector at aggregate level had employment rate reducing growth.
Page | 138
Table 4.10: Contribution of Sectors to the Change in Employment Rate,
Ethiopia (1999 - 2013)
1999-2005 2005-2013
Contribution of
Change in total Change in total Contribution of
the sector to
Sector employment employment the sector to
total
rate (percent rate (percent total e growth
employment
points) points) (percent)
growth (percent)
Agriculture 4.87 85.16 -0.50 -21.43
Mining 0.16 2.79 0.19 8.27
Manufacturing 0.59 10.24 0.00 -0.20
Utilities 0.00 0.01 0.10 4.34
Construction 0.45 7.84 0.56 24.24
Service -0.35 -6.05 1.97 84.77
Market Service -0.81 -14.19 0.34 14.55
Distributive Trade -0.85 -14.91 -0.25 -10.56
Transport & Comm. 0.00 0.08 0.41 17.44
Finance & business 0.04 0.64 0.18 7.67
Non Market Service 0.47 8.14 1.63 70.22
Public service 0.13 2.29 0.64 27.61
Personal services 0.33 5.85 0.99 42.61
Total employment rate 5.7 100.0 2.3 100.0
Nevertheless, the service sector contribution to the employment rate
has changed in the high growth periods, accounting 84.7 percent of the total
employment rate growth or job generation. The agriculture sector contributed
negatively (-21.43 percent) to employment rate growth witnessing the shift to
the service sectors. The largest contribution in this period is observed in the
government and personal services (non-market services) accounting 70
percent of the change in the employment rate. The other important
contributors are the construction, and transport and communication sectors
accounting 24.4 percent and 17.4 percent respectively. The manufacturing, the
agriculture, and the distributive trade experienced employment rate reducing
growth. Therefore, the contribution of agriculture to job creation is overtaken
by the service sector. Despite the fact, the government of Ethiopia had
introduced employment generation policies focusing on micro and small scale
manufacturing enterprises, the decomposition result indicates that the
manufacturing and industry sector it is not contributing to job creation, rather
it is the service sector.
Page | 139
Table 4.115: Contribution of Employment Changes to Change in PCGDP,
Ethiopia, (1999 - 2013)
1999-2005 2005-2013
Percentage Percentage
Contribution Contribution
Sector of total of total
to Change in to change in
change in change in
PCGDP PCGDP
PCGDP PCGDP
Agriculture 217.22 41.20 -33.81 -0.95
Mining 7.12 1.35 13.05 0.37
Manufacturing 26.12 4.95 -0.31 -0.01
Utilities 0.02 0.00 6.85 0.19
Construction 20.01 3.79 38.26 1.07
Service -15.42 -2.93 133.76 3.75
Market Services -36.19 -6.86 22.96 0.64
Distributive trade -38.04 -7.21 -16.67 -0.47
Transport & Comm. 0.21 0.04 27.52 0.77
Finance and business 1.64 0.31 12.11 0.34
Non Market Service 20.76 3.94 110.80 3.10
Public service 5.83 1.11 43.57 1.22
Personal Service 14.93 2.83 67.23 1.88
Total contribution 255.1 48.4 157.8 4.4

Note: Monetary value is in (2010/11 constant ETB)
The role of employment changes observed in the service sector on per

capita GDP growth can also be further decomposed as shown in Table 4.11. In
1999-2005, if the growth in which the demographic change, total output per
worker, and employment in all sectors other than agriculture sector remain
unchanged, the larger change in total per capita GDP would have increased by
217.2 ETB. This is due to the increase in employment in agriculture observed
during 1999-2005. In the previous Table 4.10, the contribution of agriculture
to total employment rate growth was 85.16 percent. Hence, due to this change
assuming unchanged working age population, total output per worker, and
employment in all sectors; the total per capita output would have increased by
217.2 birr. This amount takes 41.2 percent of the total change (48.4 percent) in
per capita GDP growth linked to employment rate. Other than agriculture, the
manufacturing, construction, and personal service sectors have contributed
more to per capita GDP growth by 26.12, 20.01 and 14.93 birr respectively.
However, the distributive service trade sector would have reduced per capita
Page | 140
growth by 38.04 birr. This undermines the contribution of service sector at the
aggregate level, reducing per capita GDP by 15.42 birr. The positive and high
contribution of the non-market service (20.76 birr) could not offset the
negative contribution that the distributive service sector experienced during
1999-2005. In addition, during 2005-2013, the share of agriculture in
employment rate growth is overtaken by the service sector, specifically by the
non-market service accounting 3.1 percent of 4.4 percent. On the other hand,
agriculture, manufacturing, and distributive service had a negative
contribution to the change in per capita GDP. This decomposition analysis of
employment changes on per capita GDP identifies the importance of the non-
market service sector for GDP per capita growth through employment
changes.
Finally, in Table 4.12, sectoral contributions are decomposed into the

contribution of changes in employment, contribution to productivity (within
sectors productivity change and the inter-sectoral employment shifts) and the
total effect of the sector. During 1999-2005, the agriculture sector took the
biggest role in per capita growth followed by the service sector that originated
from market services. This is due to the higher within productivity changes in
the market services observed than other sectors. Furthermore, the finance and
business service sector experienced the highest contribution in inter-sectoral
shifts. Hence, the service sector has a greater contribution to productivity than
other sectors but it has a negative contribution to employment changes.
In 2005-2013, the service sector dominated the contribution to changes

in GDP per capita growth from both employment and productivity sides. It
accounted 56.2 percent of the change in GDP per capita growth, the larger
share originating from the distributive trade (26.8 percent). One important
observation in the distributive service is its negative contribution in
employment rate and inter-sectoral shifts in both periods. Even though the
distributive sector is the dominant service sub-sector in Ethiopia (in output and
employment) with higher productivity level, it has lower employment
generation capability and inter-sectoral shifts.
Page | 141
Table 4.12: Growth Decomposition, Contribution to Total Growth in GDPPC, Ethiopia (1999-2013)
1999 -2005 2005-2013
Percentage Output per Worker Total Percent Change in Output per Worker Total (%)
Change in Percentage Inter-sectoral (Percent) Employment Percent within Inter-sectoral
Sector
Employment within Sector Shifts (%) Sector Change Shifts (%)
Change in w* in w*
Agriculture 41.2 3.9 -0.8 44.3 -0.9 30.2 1.9 31.2
Mining 1.4 -11.8 10.7 0.2 0.4 0.5 0.7 1.6
Manufacturing 5.0 -2.2 -0.3 2.5 0.0 5.4 0.0 5.4
Utilities 0.0 1.2 -0.5 0.7 0.2 -0.6 1.2 0.8
Construction 3.8 -1.3 6.4 8.8 1.1 6.7 2.1 9.9
Service -2.9 18.2 9.6 24.9 3.7 34.1 18.4 56.2
Market Service -6.9 27.7 9.3 30.1 0.6 26.3 16.8 43.7
Distributive -7.2 27.1 -9.6 10.3 -0.5 28.5 -1.3 26.8
Transport 0.0 9.7 -1.4 8.3 0.8 0.0 4.1 4.9
Finance 0.3 -9.1 20.3 11.5 0.3 -2.3 14.0 12.1
Non-Market 3.9 -9.5 0.3 -5.3 3.1 7.8 1.6 12.5
Public 1.1 -7.4 0.0 -6.3 1.2 6.9 1.8 10.0
Personal 2.8 -2.1 0.4 1.1 1.9 0.9 -0.3 2.5
Subtotals 48.4 8.0 25.0 81.4 4.4 76.4 24.3 105.1
Demographic component 18.6 -5.1
Total 100 100
Total % change in GDP per capita 16.7 97.0
*w - refers to productivity
Source: Own Computation using decomposition tool
Page | 142
The inter-sectoral shifts with positive contribution refer that on average
labor moved from lower than average productivity sectors to above average
productivity sectors. Hence, in the period 2005-2013 except for distributive
trade and personal service sector, all the other sectors has a positive
contribution implying that on average labor has moved from low productivity
sectors of agriculture and manufacturing sectors to high productivity sectors of
the service sector.
Therefore, the decomposition result identified the sectors and factors

that are most linked to per capita GDP growth. The service sector has become
the major contributor in Ethiopia for the growth of per capita GDP during
2005-2013. In addition, it has contributed more than other sectors to the
productivity growth and inter-sectoral shifts. The major contribution to per
capita GDP growth originates from the distributive sector due to its higher
within sectors productivity change. Whereas the contribution of service sector
employment to the GDP per capita growth emanates from the non-market
service sectors. On the other hand, the contribution of the service sector in
inter-sectoral shifts also comes from the higher inter-sectoral shifts occurred in
the finance and business service sector.
4.2 : Role of the Service Sector in Structural Change in Ethiopia

4.2.1 : The Shift-share Analysis
In the previous analysis, the growth trends of the service sector in
terms of output, employment and productivity are discussed. The service
sector has become an important sector in Ethiopian economy contributing
much to the growth in output and employment. Now, in this section, the role
of the sector in the structural change process of Ethiopia is analyzed and
discussed. Here again, the period considered in the analysis is from 1961-
2011. The period is sub-classified based on different growth regimes and
policy changes. These are the early industrialization period of the 1960s to late
1980s (1961-1981), a pre-reform period (1981-1992), and post-reform period
(1992-2011). These periods also coincides with important periods of political
and socio-economic changes occurred in the country. The post-reform periods
Page | 143
are subdivided into two periods based on the economic growth performance of
the country (WB, 2015a). These are the low growth period (1992-2004) and
high growth periods (2004-2011). Structural change in this study refers to the
shifts in the sectoral composition of economic activity from sectors of low
productivity to sectors of high productivity. In other words, it can be expressed
as the reallocation of labor from low-productivity sectors to higher-
productivity sectors. Therefore, the structural change process in Ethiopia is
first tested using the structural bonus and structural burden hypothesis for the
different periods. Then, results and discussions on productivity growth
decomposition into structural change effects, within-sector productivity effect
and the role of the service sector are presented.
4.2.1.1 : Structural Bonus and Structural Burden Hypothesis
The structural burden and structural bonus hypothesis are evaluated

using the shift-share decomposition analysis by using equation 6 in chapter
three, which takes the initial period for the weight of the productivity. The
‘structural bonus and burden’ hypotheses were estimated based on the works
of Timmer & Szirmai (2000) on Asian economies, Peneder (2003) for 28
OECD countries, Fagerberg (2000) on sample of OECD and developing
countries, Chen, Jefferson, & Zhang (2011) on China, and Havlik (2013) on
new EU member states.
The structural bonus hypothesis, which is calculated as the sum of

employment shifts weighted by their productivity level, assumes a positive
relationship between structural change and economic growth. In the process of
economic development, ‘economies shift from activities with relatively low
productivity levels to industries with a higher value added per labor input’
(Timmer & Szirmai, 2000). Thus, in the structural bonus hypothesis for the
manufacturing sector, the static effect will be positive when the sectors with
high productivity levels attract more labor. It may also have a negative impact
on aggregate productivity growth if labor shifts to sectors with slower
productivity growth.
Page | 144
On the other hand, the dynamic or interaction effect shows structural
burden hypothesis, which is calculated as the sum of the interactions between
the changes in the weight of employment and the changes in the labor
productivity of each branch of activity. The more shifts that take place towards
highly productive sectors, the greater will this effect. This effect can be used
to check the hypothesis of the structural burden proposed by Baumol (1967).
The hypothesis is valid if the value of this is negative due to a shift of labor
from manufacturing to services that have less productivity. Thus, the
interaction term can be used to capture Baumol’s hypothesis of a structural
burden of labor reallocation on aggregate growth, which predicts that
employment shares shift away from progressive industries towards those with
lower growth of labor productivity (see Peneder (2003)).
Table 4.13: Shift-share Result of Labour Productivity Growth, Ethiopia

(1961 - 1991)
1961-1981 1981-1992
Economic Sectors Structural Structural
change Average change Average
Within Within
LP LP
Shift Between Shift Between
Agriculture -0.31 -0.19 0.02 -0.47 -0.47 0.03 -0.01 -0.44

Manufacturing 0.04 0.02 0.01 0.06 -0.02 -0.01 0.00 -0.03
Service -0.45 0.80 -0.34 0.01 0.07 -0.10 -0.01 -0.05
Market Service -0.26 0.54 -0.19 0.09 0.00 -0.05 0.00 -0.05
Non-Market Service -0.17 0.23 -0.13 -0.06 0.04 -0.03 -0.01 0.00
Aggregate LP Growth -0.70 0.67 -0.29 -0.33 -0.46 -0.06 -0.02 -0.54
Percentage
Contribution to 212 -201 89 100 84 12 4 100
Aggregate LP
In Ethiopia, we can observe the existence of structural bonus and

structural burden for the various periods. Accordingly, from Table 4.13 and
4.14 we can observe a structural bonus, the presence of positive shift effect for
the manufacturing sector, in all the periods except for 1981 – 1992. In 1981 –
1992, the shift effect in the labor productivity growth is negative for the
manufacturing sector (-0.01 percentage point) and for the aggregate level (-
0.06 percentage point). This means, labor has shifted from high productivity
manufacturing activities, which are more capital intensive and use more
intermediate inputs, to low productivity level sectors. During the pre-reform
Page | 145
period, the average productivity level of the manufacturing sector was one of
the highest. From the earlier analysis of the productivity growth in Ethiopia,
the manufacturing sector has been increasing until 1984 (see Figure 4.6). On
the other hand, the employment growth for the manufacturing sector was more
than the service sector but it was less than agricultural employment growth.
Thus, the shift effect in 1981 – 1992 for the agriculture sector was positive and
it was negative for all the other sectors. This means there was a shift of labor
from the manufacturing sector and other service sectors to the agriculture
sector. However, in the early industrialization period (1961- 1981), and post-
reform periods (1992 – 2004) and (2004 – 2011) the static shift effect for the
manufacturing sector was positive implying that workers moved out of the low
productive sectors of the agriculture and manufacturing sectors, and joined the
service sectors. Therefore, the structural bonus hypothesis was confirmed in
Ethiopia in all the periods considered except in 1981-1992.
This result is also consistent with the finding by de Vries, Timmer, &
de Vries (2013), which states that most of the African countries have de-
industrialized in the 1970s and 1980s due to a decline in manufacturing share
in aggregate GDP. In addition, they confirmed that Africa had static structural
gain (structural bonus) and dynamic losses (structural burden), which is
comparable to the situation in Latin America but different in Asian countries.
Timmer & Szirmai (2000) also found no evidence on structural-bonus
hypothesis for the manufacturing sector in four Asian countries in 1963–1993.
Table 4.146: Shift-share Result of LP Growth, Ethiopia (1992 - 2011)

1992-2004 2004-2011
Structural Structural
Economic Sectors Change Average Change Average
Within
LP LP
Shift Between Within Shift Between
Agriculture -0.21 -0.08 0.01 -0.28 0.65 -0.17 -0.10 0.39
Manufacturing -0.06 0.12 -0.03 0.02 -0.05 0.13 -0.03 0.05
Service 0.25 0.19 0.06 0.51 0.22 0.60 0.08 0.90
Market Service -0.04 0.31 -0.02 0.26 -0.07 0.69 -0.04 0.58
Non-Market 0.17 0.00 0.00 0.18 0.20 0.01 0.00 0.21
Service
Aggregate LP
-0.11 0.41 -0.01 0.29 0.69 0.81 -0.11 1.39
Growth
% Contribution to
-39 143 -4 100 49 58 -8 100
Aggregate LP
Page | 146
Structural burden, a negative between effects for the service sector,
was detected in 1961-1981 and 1982 – 1991, while no structural burden was
observed in the post-reform periods and growth periods (1992–2004) and
(2004 – 2011). In 1961-1981, the between effect was -0.34 percentage point
and in 1981 – 1992 it was -0.01 percentage point. According to the structural
burden hypothesis, the negative between effects is due to a shift of labor from
the manufacturing sector to the services sector that has less productivity. That
means the employment shares shift away from progressive industries towards
those with lower growth of labor productivity (Peneder, 2003). Thus, during
(1961-1981) and (1981 – 1992), the negative interaction term could be due to
the shift of labor from the manufacturing sector to the service, whose
productivity level was declining. We have observed that the productivity level
in the service sector has declined from 1961 to 1984 while the productivity
level of the manufacturing sector increased until 1984.
In the periods 1992 – 2004 and 2004 – 2011, the productivity level of
the manufacturing sector has declined, though labor has continued to move
into the sector. On the other hand, labor productivity has increased in the
service sector as more labor also shifted into the sector from the agriculture
sector. Thus, no structural burden was observed in the periods due to a higher
productivity level of the service sector than the manufacturing sector. Based
on Baumol’s ‘Cost Disease Hypotheses’; an increasing share of services
results in a productivity slowdown due to the inherent labor-intensive nature
of service production. However, recently high productivity in the service
sectors was observed because of improvements in the productivity of market
service sectors such as the financial sector and transport and communication
services. The high productivity level for most of the market services in
developing countries tend to reduce or eliminate the structural change burden
as it was the case in India. Hence, developing countries are expected to ‘suffer
from a structural change burden at early stages of development’ (Szirmai,
2011). However, the within productivity level of the market service in
Ethiopia for the periods 1992–2004 and 2004–2011 was negative. Rather the
non-market services had a high level of within productivity level (see Table
Page | 147
4.14). From the early descriptive analysis in Figure 4.6, the non-market
service productivity level was increasing due to high productivity level
observed in the public service sector, while the market service productivity
level has slightly declined in recent periods. Therefore, even though Ethiopia
had high growth and expansion of service sector, the structural burden effect
of the service sector was eliminated due to productivity growth in the non-
market services as oppose to the case in other developing countries. This could
be due to the growth in the size of the Ethiopian public service output in GVA
and decline of employment share in the public sector.
In addition to the structural bonus and structural burden observed in
Table 4.13 and 4.14, the aggregate labor productivity growth and the
contribution of the within productivity effect and structural change effect can
be identified. The aggregate labor productivity growth for Ethiopia in the early
industrialization period (1961–1981) and pre-reform period (1982–1991) was
negative, (-0.33) and (-0.54) respectively. It becomes positive (0.29) and
(1.39) in 1992-2003 and 2004-2011 respectively. The positive labor
productivity growth (0.29 percentage point) in 1992-2003 was mainly
attributed to the structural shift effect accounting 143 percent of the total
change, which originated from the market service sector. The within
productivity growth effect had productivity reducing effects due to small
productivity level of the agriculture sector. On the other hand, the highest
productivity growth (1.39) was observed in 2004-2011 as 58 percent of the
growth is contributed by the shift effect because of the labor movement to the
market services. The within productivity effect has also contributed 49 percent
to the growth. In this case, the source for the within productivity growth
emanates from the high labor productivity observed in the agriculture sector
(0.65 percentage point) and non-market services (0.20 percentage points).
Further analysis, focusing on service sectors on the movement of labor to
above-average productivity level sectors and productivity growth level for
expanding and shrinking sectors, can be conducted by considering the
shrinking and expanding sector in the decomposition tool.
Page | 148
4.2.1.2 : Role of the Service Sector in Structural Change in Ethiopia
The contribution of sectors in the structural change effect can be either
positive or negative, depending on whether a sector is expanding or shrinking.
The earlier decomposition methods or the ‘traditional decomposition methods’
are not good enough to measure the contribution of sectors to productivity
growth (de Vries, Timmer, & de Vries, 2013). By dividing sectors into
expanding and shrinking sectors based on the changes in employment shares,
they calculated the static between-effect relative to the average productivity
level of the shrinking sectors and the dynamic between- effects relative to the
average productivity change of the shrinking sectors. Here the structural
change or the reallocation effect is split into two terms: static reallocation
effect and dynamic reallocation effect. The static reallocation effect captures if
workers move to above-average productivity level sectors and the dynamic
reallocation effect identifies if productivity growth is higher in sectors that
expand in terms of employment shares (see (de Vries, Timmer, & de Vries,
2013); (Timmer & de Vries, 2009).
The output of the decomposition in Table 4.15 and 4.16 show the role
of sectors in a structural change in Ethiopia during (1961 – 1992) and in (1992
– 2011) respectively. According to result in Table 4.15, the aggregate labor
productivity growth in 1961 – 1981 was -0.04 and in 1981 – 1992 it was -0.52.
The negative productivity growth implies the slowdown of the growth in the
country. However, in 1992 – 2004, the aggregate labor productivity growth
improved and became 0.3. It further increased in 2004 – 2011 to 1.5
contributing to the high growth observed in the period. Looking at the effects,
in all the periods except in 1992–2004, the within effect dominates the
between effects of structural changes. As shown in Table 4.16, in (1992–2004)
the between effect dominates the within effect, which means, the within
sectors productivity growth have little contribution to the improvement in the
productivity growth as compared to the structural change effect. However, in
the later period (2004–2011) the within sectors productivity growth become
the source of the high productivity growth observed in the country.
Page | 149
Table 4.15: Shift-share of LP growth, Role ofSectors in Ethiopia (1961 – 1992)
Economic Sectors 1961-1981 1981-1992
Within Between effect Average LP Within Between effect Change Average LP
effect Static Dynamic effect Static Dynamic
Agriculture -0.34 0.03 0.00 -0.30 -0.46 -0.62 -0.05 -0.37
Industry 0.05 0.02 0.04 0.10 -0.06 -0.02 0.00 -0.02
Mining 0.00 -0.02 0.02 0.00 0.00 -0.08 0.01 -0.02
Manufacturing 0.03 0.01 0.01 0.05 -0.02 0.06 0.00 0.01
Other Industry 0.00 0.00 0.05 0.06 -0.03 0.00 0.00 -0.01
Utilities 0.00 -0.01 0.03 0.02 0.00 0.00 0.01 0.00
Construction 0.00 0.00 0.03 0.03 -0.04 0.01 -0.01 -0.01
Service -0.12 -0.34 0.62 0.16 0.08 0.61 -0.02 0.23
Market Service -0.07 -0.19 0.46 0.20 0.00 0.17 -0.03 0.05
Distributive Trade -0.08 -0.28 0.47 0.11 -0.04 0.15 -0.01 0.03
Transport & Comm. 0.00 0.00 0.04 0.04 0.02 0.03 -0.01 0.01
Finance and business 0.08 -0.03 0.00 0.05 0.01 0.00 0.00 0.01
Non Market Service -0.04 -0.13 0.13 -0.04 0.05 0.45 0.03 0.18
Public services -0.02 -0.05 0.06 -0.01 0.04 0.18 0.01 0.07
Personal Services -0.01 -0.07 0.06 -0.03 0.01 0.27 0.02 0.10
Aggregate LP growth -0.41 -0.29 0.67 -0.04 -0.43 -0.02 -0.06 -0.52
Percentage Contribution to
11.25 8.12 -18.37 1.00 84 4 12 100
Aggregate LP growth
Page | 150
Now, let us see the contribution of the service sector and other sectors
in the structural change process of the country in the respective periods.
During 1961-1981, the negative growth of the aggregate labor productivity (-
0.04) was mainly due to the negative within sectors productivity level of the
agriculture and service sectors (distributive service, public service, and
personal services). However, the manufacturing sector had a positive within
productivity level, which implies that the country was in the early
industrialization process due to favorable growth in the manufacturing sector.
Even though, the manufacturing sector had a higher productivity growth it
could not offset the decline in the productivity of the service and agriculture
sector, which contributed to the negative productivity growth observed in the
period. The static effect was also negative contributing to the decline in
aggregate labor productivity growth. The negative static between effects
implies that the shift of labor in the period was not towards sectors that have
above average productivity level. Rather labor shifted from below average
productivity sectors to above average productivity sectors. That means the
average productivity level of the shrinking sectors was more than the
productivity level of the expanding sectors. In table 4.17, we can observe that
the agriculture sector was the shrinking sector as all other sectors expand in
terms of their employment share. In addition, productivity has declined in the
agriculture and service sectors. Therefore, labor has moved out of the
agriculture but the productivity level of the agriculture was more than other
sectors, resulting in negative structural change effect. On the other hand, the
dynamic between effects is positively contributing to the overall structural
change effect because of the labor movement to sectors that have above the
average productivity growth. All the sectors had a positive dynamic effect;
however, the larger effect is generated from the service sector, specifically
from the distributive service trade. Therefore, in the early industrialization
period in Ethiopia, even though aggregate productivity growth was negative,
there was productivity gain from the structural change effect due to positive
dynamic effects.
Page | 151
Table 4.16 Shift-share of LP growth, role of sectors in Ethiopia (1992 – 2011)
1992-2004 2004-2011
Within Between effect Change Average Between effect Average
Economic Sectors Within effect
effect Static Dynamic LP Static Dynamic LP
Agriculture -0.23 0.00 0.00 -0.07 0.75 0.00 0.00 0.25
Industry -0.06 -0.08 0.26 0.04 -0.09 -0.13 0.30 0.03
Mining 0.00 0.00 0.01 0.00 0.00 -0.01 0.02 0.00
Manufacturing -0.02 -0.03 0.08 0.01 -0.03 -0.05 0.08 0.00
Other Industry -0.05 -0.12 0.25 0.03 -0.07 -0.10 0.25 0.02
Utilities 0.00 0.00 0.01 0.00 0.00 0.00 0.02 0.01
Construction -0.03 -0.09 0.19 0.02 -0.05 -0.08 0.19 0.02
Service 0.19 0.07 0.15 0.14 0.14 0.03 0.51 0.22
Market Service -0.02 -0.01 0.27 0.08 -0.04 -0.09 0.60 0.16
Distributive Trade -0.02 -0.01 0.16 0.04 -0.02 -0.07 0.34 0.08
Transport &Comm 0.04 0.00 0.01 0.02 0.04 0.01 0.03 0.03
Finance and business -0.02 -0.02 0.09 0.02 -0.02 -0.03 0.19 0.05
Non Market Service 0.16 0.00 0.00 0.06 0.19 0.00 0.00 0.07
Public services 0.12 0.01 0.01 0.05 0.15 0.01 0.02 0.06
Personal Services 0.03 0.00 0.00 0.01 0.03 0.00 0.00 0.01
Aggregate LP growth -0.10 -0.01 0.41 0.30 0.80 -0.11 0.81 1.50
Percentage Contribution to
-33 -4 137 100 53 -7 54 100
Aggregate LP growth
Page | 152
In 1981-1992, as table 4.15 shows, all the within effect, static and
dynamic effects were negative leading the country into a sluggish productivity
growth. About 84 percent of the negative productivity growth was contributed
by the decline in the within sectors productivity growth mainly caused by the
decline in the productivity of the agriculture sector. In addition, the sector has
contributed to the negative static and dynamic effects. Therefore, in the pre-
reform period in Ethiopia, productivity growth was stalled because of negative
within sectors productivity growth coupled with negative structural change
effects.
Table 4.16 shows the contribution of sectors to the productivity growth

in the post-reform Ethiopia (1992-2011). In the first growth periods (1992-
2004), productivity growth has shown improvement and it became positive
(0.3). The between effects or the structural change component has contributed
much to the improvement in the productivity growth. Specifically, the
dynamic effect has contributed 137 percent and the static effect has reduced
the growth only by (-4 percent). The positive dynamic effect implies a shift of
labor to sectors that have above-average productivity growth. In the period
labor shifted out of the agriculture and personal service sector that have below
average productivity growth and joined the service sectors and the industry
sector that have above-average productivity growth. Here, the market service
sector (with 0.27 percentage points), especially, the distributive trade service
(0.16 percentage points) along with the construction sector (0.19 percentage
points) has contributed to the structural change effect.
In 2004-2011, the high aggregate productivity growth was due to the

improvement in the within sectors productivity growth (53 percent) and the
dynamic structural change effect (54 percent) caused by the movement of
labor to above average productivity growth sectors. The main causes for the
positive contribution of the within sectors productivity growth are the
improvement in the productivity of the agriculture sector (with 0.75
percentage points) and the non-market service (0.19 percentage points),
specifically from the public service sector (with 0.15 percent percentage
Page | 153
points). Productivity in the agriculture sector has improved in Ethiopia since
2004 as the use of improved agricultural inputs increased. The public service
sector productivity has also increased mainly due to huge public investment in
the public sector. As it is shown in table 4.17, productivity has improved in in
the public service sector, agriculture and transport and communication
services in (2004-2011). However, the within productivity growth of the
manufacturing sector, construction, distributive service trade, and finance and
business services were negative, limiting further growth in the aggregate
productivity growth. By their nature, these sectors are expected to show high
productivity growth and drive the growth as they have high potential to
growth. Thus, this implies that the country has much potential for high
aggregate productivity growth if policies are laid to improve the productivity
of these sectors.
The positive dynamic effect in 2004-2011 is again caused by the

market service sector (0.60 percentage points), mainly due to the distributive
service (0.34 percentage points) and the finance and business service sectors
(0.19 percentage points). In addition, the construction sector has also
contributed (0.19 percentage points) to the structural change effect as labor
moved to above average productivity growth sectors. However, the static
effect, which indicates the shift of labor to sectors with above average
productivity level, is negative. This implies, even though it has little effect (-7
percent) on the aggregate productivity growth, labor has shifted to sectors that
have below average productivity level. This means when productivity level
(shift effect) is considered labor is shifting to sectors like manufacturing and
mining, which have below average productivity level. Whereas, when
productivity growth or change (dynamic effect) is considered labor has moved
to sectors that have high productivity growth as the non-market services
(public service and personal services), transport and communication, and
agriculture sector as these sectors experienced growth or change in their
productivity in 2004-2011 [see table 4.17].
Page | 154
Figure 4.11: Shift Share Decomposition Results, Labor Productivity
Growth, Ethiopia (1961 - 2011)
2004- 2011
1992-2004
1981-1992
1961-1981
-1.00 -0.50 0.00 0.50 1.00 1.50 2.00

1961-1981 1981-1992 1992-2004 2004- 2011
Within -0.41 -0.43 -0.10 0.80
Shift -0.29 -0.02 -0.01 -0.11
between 0.67 -0.06 0.41 0.81
Average annual labor productivity growth
Generally, as described in Figure 4.11, the aggregate productivity

growth in Ethiopia in the early periods 1961 – 1981 and 1981 - 1992 was
negative and it has improved in the post-reform periods (1992 – 2011). The
structural change effect had a considerable effect on the improvement in
aggregate productivity growth. Nevertheless, the within sectors productivity
growth has contributed to the growth in 2004 – 2011 due to the productivity
growth observed in the agriculture and non-market service sectors. The market
service sector, specifically the distributive trade service has contributed to the
structural change effect for the growth of aggregate productivity observed in
1961 – 1981 and 1992 – 2011. The structural change effect (dynamic effect)
has been positive in these periods indicating a shift of labor to sectors with
above average productivity growth. Whereas, the static effect is negative in all
the periods implying the shift of labor to sectors that have below average
productivity levels. Hence, this result is against the findings of (de Vries,
Timmer, & de Vries, 2013) in Sub-Saharan African average including
Ethiopia. According to their result, there was a pattern of static gains but
dynamic losses in the structural change for most African and Latin American
countries, but not for Asia. However, if we take Ethiopia separately, according
to this finding in this research, there was a static loss but dynamic gain, which
Page | 155
Table 4.17: Gross Value Added, Employment, and Labor Productivity, Ethiopia (1961 - 2011)
Gross Value Added Share (%) Employment Share (%) Labour Productivity Level
Economic Sectors 1961 1981 1992 2004 2011 1961 1981 1992 2004 2011 1961 1981 1992 2004 2011
Agriculture 85.1 69.2 64.7 47.1 41.8 96.2 89.1 90.3 84.4 73.4 2.6 2.3 1.6 1.5 2.4
Industry 5.4 9.5 9.2 14.1 13.8 1.5 2.0 1.8 5.0 10.0 10.9 14.0 10.9 7.6 5.8
Mining 0.1 0.1 0.5 0.6 0.8 0.0 0.0 0.2 0.3 0.7 72.3 11.1 6.2 5.2 4.7
Manufacturing 2.0 4.0 3.8 5.4 5.2 1.3 1.7 1.4 3.6 6.9 4.8 7.1 5.8 4.0 3.2
Other Industry 3.2 5.4 5.0 8.2 7.8 0.2 0.3 0.3 1.1 2.4 50.0 52.3 41.0 20.2 13.5
Utilities 0.4 1.3 2.2 2.3 2.0 0.0 0.1 0.1 0.1 0.1 77.5 62.6 64.1 69.7 73.2
Construction 2.8 4.1 2.7 5.9 5.8 0.2 0.2 0.2 1.0 2.3 47.6 49.8 31.6 15.9 10.6
Services 9.5 21.4 26.1 38.8 44.4 2.3 9.0 7.8 10.5 16.5 12.1 7.0 7.3 10.0 11.3
Market Service 6.9 16.5 17.9 25.4 31.0 1.1 4.0 3.2 5.9 11.8 18.5 12.0 12.2 11.6 11.0
Distributive Trade 5.4 11.9 10.5 15.0 18.7 0.8 3.5 2.7 5.3 10.8 20.0 9.9 8.4 7.7 7.3
Transport &comm 1.0 2.5 4.0 5.6 5.6 0.2 0.4 0.4 0.4 0.5 18.8 18.3 23.9 36.1 45.9
Finance and business 0.5 2.1 3.3 4.9 6.7 0.1 0.1 0.1 0.2 0.5 9.8 65.8 82.7 63.1 54.1
Non Market Services 2.6 4.9 8.3 13.3 13.4 1.2 4.9 4.6 4.6 4.7 6.3 2.9 3.9 7.8 12.1
Public services 1.8 3.1 5.5 10.0 10.7 0.8 2.4 2.1 2.2 2.4 6.7 3.8 5.8 12.1 18.7
Personal Services 0.8 1.8 2.8 3.3 2.7 0.4 2.5 2.5 2.4 2.3 5.5 2.0 2.4 3.8 5.0
Total Economy 100 100 100 100 100 100 100 100 100 100 3.0 2.9 2.2 2.7 4.2
Page | 156
puts the country in similar patterns of structural change and service sector
growth.
To sum up, the service sector in Ethiopia is dominating the Ethiopian

economy in output. There is also a shift in employment specifically to the
distributive service sector. These increments in output and employment in the
service sector have contributed to the growth in GDP per capita and
employment in Ethiopia. In addition, the service sector, specifically the
distributive service sector has contributed much to the productivity growth and
structural change in Ethiopia.
The other section describes and discusses the results of the inter-
sectoral linkage effect of the service sector with other sector estimated using
input-output analysis. The results and discussion on service sector growth
determinant estimated using Autoregressive Distributed Lag Model are also
presented. Finally, the sustainability of service-led growth is evaluated using
scorecards.
Page | 157
CHAPTER-V
DETERMINANTS OF GROWTH OF SERVICE
SECTOR AND ITS SUSTAINABILITY
Economic development is a complex phenomenon. A huge body of

economic theory and empirics has tried to explore its dynamism. The set
pattern of economic transformation identified by the theory is the path of the
primary, secondary and tertiary sector. Recently some of the developing
economies have shown the experience of bypassing the secondary sector; it
has been primary sector directly leading the way to service sector. Economists
are trying to delineate the underlying growth dynamics of the tertiary sector.
In this setting of argument, this chapter is an attempt to identify the
determinants of growth of service sector and the analysis of the sustainability
of overgrown service sector.
5.1 : Determinants of Service Sector Growth in Ethiopia

Using the ARDL model the determinants of service sector growth in
Ethiopia is estimated for the periods 1981-2011. Based on the model specified
in section three, the service sector growth in Gross Value Addition (LnSGVA)
and Service sector growth in employment (LnSEMP) are estimated separately
as Model 1 and 2 respectively. The variable LnSGVA is the dependent
variable representing the natural logarithm of Service Gross Value Added;
LnSEMP is also the other dependent variable representing employment in the
service sector. LnGDPPC is the natural logarithm for GDP per capita income
in 2005 constant USD. LnPTVD is the variable capturing the productivity
differential between services and manufacturing sector (The difference of
manufacturing sector labor productivity and service sector labor productivity).
LnPSERV is the values of producer’s service sector which capture the
intermediate demand for service including business service, finance, insurance
and real estate services. LnFEACT is female economic activity measuring the
participation of female in the labor force. The dummy variable DREF092
represents post-reform periods taking value 1 for the reform period (1992 and
after 1992) and 0 otherwise. The period is crucial breaks that brought major
Page | 158
policy and economic changes in the country as result of the government
change and liberalization policy reform since 1991. The coefficient for the
variables in natural logarithm represents elasticity and the coefficient for the
dummy variable is transformed to exponential values to represent percentage
changes.
Based on the theoretical and empirical findings the study identifies the
determinant of service sector growth. Hence, the descriptive statistics,
Augmented Dicky Fuller (ADF) test, bounds Test for cointegration, regression
results, and diagnostic tests are presented as follows. The model is applied to
annual data for the period of 1981 to 2011 to explore the existence of a long-
run relationship between service growth (in gross value addition and in
employment) with its determinants.
5.1.1 : Descriptive Statistics

The descriptive statistics provides the general picture of the time series
data and nature of the relationship between the variables. The Pearson’s
correlation in Table5.1 provides the correlation between two variables to
indicate the direction of the relationship between variables.
Table 5.1: Pearson Correlation Matrix for Service GVA and Employment
Variable LnSGVA LnSEMP LnGDPPC LnPTVDC LnPSERV LnFMECA
LnSGVA 1.000
LnSEMP 0.996 1.000
LnGDPPC 0.693 0.714 1.000
LnPTVDC -0.989 -0.980 -0.725 1.000
LnPSERV 0.995 0.990 0.677 -0.991 1.000
LnFMECA 0.976 0.967 0.557 -0.961 0.980 1.000
Source, Authors’ Computation
The correlation result indicates that both the service sector GVA and
employment have positive and strong relationships with all other variables
except LnPTVDC. The positive correlation between income and service GVA
and service employment supports the ‘hierarchy of needs’ hypothesis. Driven
by the income growth, the share of service in GVA and employment may
increase to meet the increasing demand for service by channeling resources
Page | 159
into the sector. In addition, the positive correlation of service GVA and
employment with LnPSERV and LnFMECA is also in line with the
‘exogenous demand shock’ hypothesis. Whereas, the Labour Productivity
difference has a negative correlation with service gross value addition and
employment. This indicates that productivity difference in Ethiopia diverges
from the expected relationship.For instance, Fuchs (1980) presumes a positive
correlation between the productivity gap of manufacturing and service sectors
and the service sector share.
Figure 5.1 also shows the time series trend and direction of the
relationship between service growth and determinant variables over the years
(1981-2011).It shows that service gross value addition and employment have
an increasing trend over the periods. Similarly, producer service and female
economic participation variables have increasing trends over 1981-2011.
However, the per capita GDP variable has declined from 1981 to 1992. After
1992 it has an increasing trend with fluctuation but starting from 2004 it is
increasing constantly. On the other hand, productivity differential variable
declined from 1986 onwards. This implies that since 1986 service productivity
is higher that manufacturing productivity. This is a unique experience to
Ethiopia as the manufacturing sector stagnated and has shown a lower
productivity than other sectors.
5.1.2: Unit Root Test

All time series data must be stationary with constant mean and
variance over time in order to conduct regression model. If it is not stationary,
the regression result becomes spurious. Therefore, the unit root test is helpful
to identify the stationarity of the variables. It could be also used in deciding
whether the ARDL model should be used. Hence, before testing for the
existence of co-integration relationship of our model, we need to conduct a
test of the order of integration for each variable by using the standard
Augmented Dickey-Fuller (ADF).
Page | 160
Figure 5.1: Line Graphs for Service GVA, Employment, and Determinant
Variables
lnsgva lnsemp
11.6 9.2
11.2
8.8
10.8
8.4
10.4
8.0
10.0
7.6
9.6
9.2 7.2
82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10
lnptvdc
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
82 84 86 88 90 92 94 96 98 00 02 04 06 08 10
lnpserv lnfmeca
9.5 9.8
9.0 9.6
8.5 9.4
8.0 9.2
7.5 9.0
7.0 8.8
6.5 8.6
82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10
Source; Author’s computation
Augmented Dicky-Fuller Unit Root Test

In order to determine the degree of stationarity, a unit root testing is
carried out through the standard Augmented Dicky-Fuller (ADF) test. This test
was undertaken to check the order of integration of the variables. The test was
done for three alternative specifications of with constant, trend and constant,
and then with no constant and trend.
The results of this test in Table 5.2 show that except LnPTVD and
LnPSERV with constant, all others variables are not stationary at their levels.
Page | 161
In their first differences, some of the variables are stationary at least for one of
the specifications. The results indicate that some of the variables are I (1) and
others are I (0). LnSEMP is not stationary in its level and it becomes
stationary in its first difference only with no constant and trend. Such results
of stationarity test would not allow us to apply the Johansen approach of co-
integration approach. This is one of the main justifications for using the
ARDL approach (bounds test approach of cointegration) developed by
(Pesaran, Shin, & Smith, 2001).
Table 5.2: Augmented Dicky-Fuller Test for Unit Root

Variable Constant Trend and constant No constant and trend
LnSGVA 3.386 -1.118 2.577
D_ LnSGVA -2.908* -3.748** -1.488
LnSEMP 1.287 -0.784 1.909
D_ LnSEMP -2.406 -2.912 -1.727*
LnGDPPC 1.096 -0.13 1.03
D_LnGDPPC -3.86*** -4.796*** -3.73***
LnPVTD 7.38*** 0.133 -1.188
D_LnPVTD -0.777 -4.618*** 0.958
LnPSERV 5.076*** 1.245 1.132
D_LnPSERV -0.418 -7.737*** 1.037
LnFEACT -1.069 -2.919 2.923
D_ LnFEACT -3.286** -3.089 -1.453
Source; Authors’ estimation
***, ** and * sign indicates the rejection of the null hypothesis of non-stationary at 1 percent,
5 percent and 10 percent significant level respectively
In addition to unit root test, it is important to test for the existence of

structural break that affects the pattern of time series of the variables. The
ADF test may fail to reject the unit root hypothesis if the series have a
structural break. The political and economic changes (wars, drought, and
economic reforms) occurred in Ethiopia in 1981-2011 cause the economy to
have a structural break that affects the pattern of time series and the variables
of interest. Hence, the test of structural breaks is conducted using unit root
test with break allowing for a break in intercept only. The break type selected
is outlier innovation and break point.
Table 5.3 for ADF unit root test for break shows, the structural breaks
year for each variable is different. All the break years are in the post-reform
period on 1992 and after 1992. In Ethiopia, some of the structural adjustment
Page | 162
policies are adopted stepwise since 1991 following the downfall of Derg
regime and seize power of the new government. The structural break year for
LnSGVA is 1992 that coincides with the reform period. LnGDPPC and
LnPSERV have breaks during 2003, on the year Ethiopia faced a severe
drought. All other variables break at different periods of time but during the
structural reform periods. The test points out that LnSGVA, LnGDPPC, and
LnPSERV variables are stationary with the existence of one structural break as
all the rest variables are non-stationary at the respective break periods.
Therefore, based on this result and with the intention of identifying the effect
of reform on service growth, structural break period 1992 is included in the
model as a dummy variable taking 1 for the reform periods (1992-2010) and 0
otherwise (1981-1991).
Table 5.3: ADF Unit Root Test with Break

Variables Break Year ADF min t-stat
LnSGVA 1992 0.383
D_LnSGVA 1992 -4.408*
LnSEMP 2004 -0.621
D_LnSEMP 2000 -3.681
LnGDPPC 2005 -1.846
D_LnGDPPC 2003 -5.224***
LnPSERV 2003 -3.665
D_LnPSERV 2003 -7.693***
LnPVTD 1999 3.035
D_LnPVTD 1999 -3.279
LnFEACT 2004 -3.408
D_ LnFEACT 1993 -3.588
Source: - Authors’ Estimation
The Hodrick-Prescott decomposition of the growth in Service GVA
and service employment in Figure 5.2 shows the cyclical and trend growth
decomposition. Both service GVA and employment have increasing growth
trend. Nevertheless, the cyclical growth indicates that both variables have
similar cyclical growth. The growth in both service GVA and service
employment declined during 1991/92 and 2003/04. As it is mentioned before,
in 1991/92 Ethiopia has changed its government and conducted series
economic reforms starting from 1992. On the other hand, the 2003 year was a
period of drought followed by high and consecutive growth since then.
Page | 163
Therefore, we can observe that service sector growth is also affected by
different shocks observed in the periods (1981-2011).
Figure 5.2: Service GVA and Employment (Hodrick-Prescott Filter

Decomposition)
(a) Service GVA (b) Service Employment

Source: Author’s computation, using (de Vries et al, 2014)
5.1.3: Bound Test for Co-integration Analysis

After conducting unit root test the next important step is the bound test
for co-integration. It is used to examine if there exist the long-run relationship
among the variables. Co-integration of the variables shows the movement of
the variables together due to a certain long run equilibrium relationships. The
Schwarz-Bayesian Criterion (SBC) was used in the co-integration analysis.
The ARDL bound test approach in testing the existence of the cointegrating
relationship is applied by comparing the F-statistics with the bound critical
values of Pesaran, Shin, and Smith (2001).
Table 5.4: ARDL Bound Test for Service GVA and Service Employment
Model 1 Service GVA Model 2 Service Employment
Significance
Critical Value bounds for k_4 Critical Value bounds for k_5
Level
Lower I(0) Upper bound I(1) Lower I(0) Upper bound I(1)
1 percent 3.41 4.68 3.41 4.68
5 percent 2.62 3.79 2.62 3.79
10 percent 2.26 3.35 2.26 3.35
F- statistics F-statistics = 21.13*** F-statistics = 8.82***
K is the number of non-deterministic regressors in the long run relationship
Critical values from Pesaran, Shin, and Smith (2001)
Source; Authors’ estimation using ARDL Bounds Test
Page | 164
ARDL bound tests the null hypothesis of no co-integration against the
alternative using the F-test with critical values. The null hypothesis will be
rejected if the calculated F-statistic is greater than the upper bound critical
value. If the computed F-statistics is less than the lower bound critical value,
then we cannot reject the null hypothesis. The result is inconclusive if the
computed F-statistic falls within the lower and upper bound critical values.
The bound test for Service GVA and Service Employment are presented in
Table5.4. The calculated F- statistics for Model 1 is 21.1, which is greater than
the upper and lower bounds at 1 percent significance level. Therefore, there is
long run co-integration between service GVA and its determinant variables at
1 percent significance level. In Model 2 for service employment, the F-
statistics is 8.8 is again significant at 1 percent significance level. Therefore,
the null hypothesis of no long-run co-integration between service
employments and determining variables is rejected. Hence, we can say that
there is co-integration among the set of (I(0) & I(1)) variables and there can be
at least long run or short run relation among these variables. The estimated
ARDL model and the regression result for both LnSGVA and LnSEMP
models are presented below.
5.1.4 Regression Results of ARDL

5.1.4.1 Long run estimation result:
The empirical results of a long-run and short-run ARDL model for
service GVA (Model 1) and service employment (Model 2) are presented in
Table5.5. The long run and short run results for Model 1 and Model 2 are
discussed separately.
In Model 1, LnGDPPC, LnFMECA, and DREFO92 are significant

variables that are cointegrated with the service sector growth in GVA in the
long run. GDP per capita income determines the service GVA in the long run
at 1 percent significance level. Since the model is specified in natural
logarithm form, the coefficients of the dependent variable are interpreted as
elasticity. For a percentage increment in GDP per capita service GVA will
increase by 0.85 percent. It is assumed that as income increases, people will
Page | 165
shift consumption from goods to services. Services have an income elasticity
of demand greater than one, but that of demand for goods is less than one
(Engel’s Law). So, it is expected to have a positive correlation between service
and income. This finding hence confirms the ‘hierarchy of needs’ hypothesis
and the result for Ethiopia is similar to other empirical findings(Singh & Kaur,
2014; Kim, 2006).
Female economic participation is also significant at 1 percent

significance level. For a single percentage change in female economic
participation, service GVA changes by 0.87 percent. This means, the more
participation of female in economic activity in Ethiopia, there will be more
growth in the service sector. According to Fuchs (1980), more labor market
participation of female has income effect and substitution effect, substitution
of home activity with the market (Fuchs, Economic Growth and the Rise of
Service Employment, 1980). Further, the more economic participation of
female results in an increase in household consumption of services (Kim,
2006). Therefore, female economic participation signifies the change in the
structure of demand in Ethiopia, which moved the demand for services
upward.
The coefficient for post-reform dummy is 0.336. By taking the

exponential value of the coefficient, the result indicates that 39.9 percent of
the growth in service is due to the changes in the post-reform period as
compared to the pre-reform period. The value for the constant term (-1.3)
represents the coefficient for a pre-reform dummy. Though it is not
significant, the negative value for the pre-reform period shows a lower growth
of services as compared to post-reform period. Literature shows that
liberalization and reforms have contributed to the growth of the sector (Jain &
Ninan, 2010). Thus, in Ethiopia, the economic reforms undertaken since 1992
have significantly contributed to the growth of service sector.
In Model 1, it is only producer service (LnPSERV) and productivity

difference (LnPTVDC) that are not significant in the long run. The producer
service captures the change in the structure of the demand due to the change in
Page | 166
the demand for intermediate service inputs. However, this is not the case in
Ethiopia as the variable is not significant. Similarly, the productivity
differential, which is the difference of manufacturing and service productivity,
is insignificant and negative. This implies that a service GVA change in
Ethiopia is not only significant but also in contrast to the Cost Disease
hypothesis. The hypothesis is that the shift towards services is due to resource
transfer from manufacturing to services because of the productivity gap
between the two sectors. According to Baumol’s ‘Cost Disease hypothesis’,
‘the service sector rises on the current price basis if there is a manufacturing
sector with high productivity growth and a stagnant service sector with low
productivity’ (Baumol, 1967). However, in Ethiopia, the service sector has
higher productivity level than the manufacturing sector resulting in a negative
productivity gap. Hence, the negative sign is due to the negative productivity
gap observed. Therefore, there could be no transfer of resources from the
manufacturing to the service sector as the Cost Disease Hypotheses argue.
Therefore, we cannot say that service growth in Ethiopia is related with
productivity difference between manufacturing and service sector. Rather,
growth in service GVA is due to the change in income (hierarchy of income
hypothesis) and change in the structure of demand (specifically due to more
female labor force participation). Female labor force participation in the
economy has a greater effect on service GVA than income and reform effects.
Moreover, the coefficient for the long run co-integration equation in

Model 1 is -0.487 and significant at 1 percent significance level. This indicates
a 48.7 percent convergence to equilibrium from the deviation. In other words,
the long-run co-integration equilibrium is corrected by 48.7 percent over each
year. Thus, it will take about 2 years for the co-integration equation to come to
the equilibrium.
Page | 167
Table 5.5: Estimated Long Run and Short Run Coefficients Using the
ARDL Approach
Model 1 Service GVA Model 2 Service Employment
Variables
ARDL(1 0 0 0 0 1) ARDL(1 2 0 01 1)
Coef. t-ratio Coef t-ratio
Long Run
Dependent Variable LnSGVA LnSEMP
LnGDPPC 0.853*** 4.92 1.208*** 6.27
LnPTVD -0.199 -1.07 0.427* 2.03
LnPSERV 0.068 0.35 0.157 0.78
LnFEACT 0.875*** 3.14 1.073*** 3.31
DRFOR92 0.336*** 4.66 0.342*** 4.51
C -1.338 -1.02 -5.495** -2.66
Short Run
D(LNGDPPC) 0.416*** 5.41 0.513*** 5.83
D(LNGDPPC(-1)) -0.220** -2.32
D(LNPTVDC) -0.097 -1.06 0.226* 1.96
D(LNPSERV) 0.033 0.35 0.083 0.79
D(LNFMECA) 0.427** 2.79 1.320** 2.19
DRFOR92 0.031 0.99 0.078** 2.33
CointEq(-1) -0.487*** -6.18 -0.529** -5.31
R-squared 0.961 0.890
Adjusted R-squared 0.858 0.829
Log likelihood 84.79 77.40
F-statistic 9.31(0.0028) 14.63 (0.000)
Durbin-Watson stat 1.574 1.67
Source: Authors’ estimation

Model 1- Co-integration equation = LnSGVA - (0.853*LnGDPPC - 0.199*LnPTVDC + 0.068*LnPSERV+
0.875*LnFMECA + 0.336*DREFO92 -1.338)
Model 2 - Co-integration equation = LnSEMP - (1.208*LnGDPPC + 0.427*LnPTVDC + 0.157*LnPSERV +
1.073*LnFMECA + 0.342*DREFO92 - 5.495 )
In Model 2 for service employment equation, all variables are

significant except LnPSRV in the long run. The coefficient for LnGDPPC is
1.208 is significant at 1 percent significance level. Here also, income is a
significant factor in determining service employment in Ethiopia. A one
percent growth in income will increase employment in the service sector by
1.2 percent. This implies that the service sector growth in Ethiopia is driven by
the rise in income as it is hypothesized. Productivity difference between the
productivity of manufacturing sector and service sector is also found to be
positive and significant at 10 percent significance level. The coefficient for
LnPTVDC, (0.427) implies a 0.42 percent change in service employment due
to productivity gap. Though this result is in line with the ‘Cost Disease
Page | 168
hypotheses, the productivity gap in Ethiopia is negative. The negative
productivity gap between manufacturing and service, due to higher
productivity level of service than manufacturing, contrasts the hypothesis.
Therefore, the growth of employment in the service sector is not due to the
channel of more labor to the service sector due to its productivity level, which
is expected to be lower than manufacturing. Thus, the reason for a significant
and positive relationship between service employment and a negative
productivity gap in Ethiopia requires further investigation by considering the
relative price of services and manufacturing.
Labor participation of female (LnFMECA) is significant at 1 percent

significance level. The coefficient for the LnFMECA (1.073) shows a 1.07
percent growth in service employment because of a 1 percent growth in female
employment in the labor force. According to Fuchs (1980), female labor force
participation drives the rise of service employment through the spending of a
higher proportion of their income on services. This finding is also consistent
with empirical findings of (D’Agostino, Serafini, & Ward-Warmedinger,
2006; Kim, 2006) and others.
In addition, the reform period dummy is significant at 1 percent

significance level. After converting the coefficient of reform period dummy
(0.342) into exponential value, we get a value of 40.7, which indicates that
service employment in Ethiopia has increased by 40.7 percent as compared to
the pre-reform period. Further, the value for the constant term is -5.495,
representing service employment in the pre-reform period. Hence, the level of
service employment in Ethiopia during the pre-reform period is lower that the
post-reform period by 5.49 percent. The long run cointegration equation for
model 2, service employment equation is -0.529, which is significant at 1
percent significance level. Hence, the deviation of the cointegration equation
from the long run equilibrium due to certain shocks is adjusted by 53 percent
over the next year. It will take about 1.9 years for the complete adjustment of
the long run equilibrium.
Page | 169
5.1.4.2 : Short run Error Correction Estimates:
In Model 1, both GDP per Capita and female economic activity are
significant at 1 percent and 10 percent significance level respectively. The
directions of the relationship for all the significant variables are positive as it
is hypothesized. The coefficient for GDP per capita is 0.467 and significant at
1 percent significance level. For a one-percentage change in GDP per capita,
service GVA changes by 0.46 percent. LnFMECA has a coefficient of 0.753,
which is significant at 10 percent significance level. For a percentage change
in the female economic activity, service GVA changes by 0.75 percent. In the
short run, the post-reform period dummy failed to be significant unlike its
significant impact in the long run. The R-squared value for service GVA
explains the 96 percent of the variation in the service GVA. This variation is
hence explained by the explanatory variables that determine service sector
growth.
In Model 2, all the variables except producer service (LnPSERV) are

significant. GDP per capita is significant at 1 percent significance level with
the value of 0.513. There is a 0.51 percent change in service employment for a
1 percent change in income in the short run. The long run effect of income
(GDP per capita) is greater than its short-run effects. However, a percentage
change in one lag period of GDP per capita results in a decline of service
employment by 0.22 percent. This result is significant at 5 percent significance
level and it is in contrast to the ‘hierarchy of needs’ hypothesis. There could
be negative correlation or no relationship between income and service growth
on the real basis or Purchasing Power Parity basis (Baumol, 1967; Baumol,
Blackman, & Wolff, 1985).Negative and significant effect of income on
services is also observed in Korean empirical study (See Kim, 2006). On the
other hand, LnPTVDC has a positive coefficient of 0.226, significant at 10
percent level. Hence, a one percent changes in the productivity difference,
service employment changes by 0.22 percent. LnFMECA has the highest
effect on service employment as compared to other determinant variables. For
a single percentage change in the employment of female labor in the economy,
service employment changes by 1.32 percent. This result is significant at 5
Page | 170
percent significance level. Finally, reform period has also a significant effect
on service employment as it results in 8.1 percent higher service employment
than the pre-reform period. The R-squared value for service employment
model is 89percent, referring 89 percent of the variation in service
employment is explained by the service determining variables.
Generally, these findings are consistent with the hypothesis of

‘hierarchy of needs’ and partially with ‘exogenous demand shock’ hypothesis.
The high growth in GDP per capita income and its significant effect as shown
in the regression results confirm the validation of ‘hierarchy of needs’
hypothesis in Ethiopia. The growth trend and cyclical movement of income in
Ethiopia are depicted using the Hodrick-Prescott Filter decomposition method
in Figure 5.3. The growth in GDP per capita income has a declining trend with
frequent ups and down in the 1980s. The average growth rate of GDP per
capita income in 1981-1992 was -3 percent. It has shown an increasing trend
since 1996 after a government change and introduction of series economic
reforms. The GDP per capita growth rate in the reform period (1992-2011)
was 5 percent and it has a more pronounced growth after 2003. During this
period (2004 to 2011), income has grown by 8 percent. Thus, a relatively
higher growth trend in income is the major factor for service sector growth in
Ethiopia.
However, the insignificant regression results for producer services and

significant results for female economic participation leads us to partially
support the ‘exogenous demand shock’ hypothesis in Ethiopia. The two
variables are proxy to change in the structure of demand. It is only from the
female economic participation side that structure of demand changed in
Ethiopia. As theories and empirics show, more participation of female in the
economy drives more consumption of services and pushes the demand for
services upward. However, producer services, which are used as an
intermediate input to other sectors, do not change the structure of demand for
services. That means service sector growth in Ethiopia is not related to
changes in the demand for service inputs, like finance and business services.
Page | 171
Therefore, the determining factors for service growth are income and changes
in the structure of demand due to changes in the female economic
participation. However, the cost disease hypothesis does not hold in Ethiopia
because of the negative productivity gap.
Figure 5.3: Hodrick-Prescott Decomposition of GDPPC Income, Ethiopia

(1981-2011)
Hodrick-Prescott Filter (lambda=100)
5.6
5.4
5.2
5.0
.10
4.8
.05
4.6
.00
-.05
-.10
-.15
82 84 86 88 90 92 94 96 98 00 02 04 06 08 10
lngdppc Trend Cycle
Source; Author’s computation using WDI (2016)

5.1.4.3 :DiagnosticTesting
The verifiability of the estimated long run ARDL model for both
Service GVA and employment are tested using various diagnostic tests. As it
is indicated in Table 5.6, we have conducted tests for autocorrelation,
heteroscedasticity and normal distribution of errors. The Breusch-Godfrey test
for autocorrelation shows that there is no serial correlation. The Breusch-
Pagan test also shows that there is no heteroscedasticity. The Ramsey RESET
test using also show that there are no omitted variables and the test confirms
that the model is specified well. Hence, the relationship between the variables
is verifiable or valid.
Page | 172
Table 5.6: Diagnostics Tests
Model 1 Model 2
Tests
Service GVA Service Employment
Chi2 Prob. Chi2 Prob.
Serial Correlation LM test (Breusch Godfrey) 1.302 0.253 0.759 0.383
Heteroscedasticity Test (Breusch Pagan) 0.39 0.532 3.64 0.056
Ramsey RESET F(3,19) = 0.33 0.800 F(3,15) = 1.02 0.409
ARCH LM 0.031 0.800 0.933 0.334
Stability Test* CUMSUM Stable Stable
CUMSUMSQ Stable Stable
Source, Authors’ estimation

*See the cumulative sum of recursive residuals plot in Figure 5.4 and 5.5
The stability of long-run estimates is tested using the cumulative sum

of recursive residuals (CUSUM) and the cumulative sum of squares of
recursive residuals (CUSUMSQ) test. The test statistics is graphed in Figure
5.4 and 5.5. The graph shows that the models are stable within 5 percent
significance level since the estimated coefficients move between the critical
bounds of 5 percent significance. The graph also shows a possible instability
or structural breaks within the given periods.
Figure 5.4: Cumulative Sums of Recursive Residuals for Service GVA

15 1.6
10
1.2
5
0.8
0
0.4
-5
0.0
-10
-15 -0.4
1994 1996 1998 2000 2002 2004 2006 2008 2010 1994 1996 1998 2000 2002 2004 2006 2008 2010
CUSUM 5% Significance CUSUM of Squares 5% Significance
Figure 5.5: Cumulative Sums of Recursive Residuals for Service

Employment
Source, Author’s computation
Page | 173
Generally, the determinants of the shift in the service sector in Ethiopia
are estimated using ARDL model. Based on the estimation result, the Service
sector growth in GVA is co-integrated in the long run with GDP per capita,
female economic participation and reform period. Service sector growth in
employment is also co-integrated in the long run with GDP per capita, female
economic participation and reform period and productivity difference. These
findings are consistent with the hypothesis of ‘hierarchy of needs’ and
partially with ‘exogenous demand shock’ hypothesis. However, service sector
growth in Ethiopia is not related to changes in the demand for intermediate
service inputs from the producer service sector. Therefore, the determining
factors for service growth in both employment and output are per capita
income, changes in the structure of demand due to changes in the female
economic participation and economic reforms. Thus, further study is
recommended on the negative productivity gap observed which could have an
implication on manufacturing sector growth.
5.2 : Sustainability of Service-Led Growth in Ethiopian
In this section, the sustainability of the recent growth in Ethiopian that is
driven by the service sector is evaluated. Even though there are arguments
against the viability of service sector led growth, most of the literature from
South Asian and India’s experience indicates the viability of service-led
growth for developing countries as most of these countries have a weak
manufacturing sector. The analysis of the sustainability of Ethiopian service
sector is analyzed borrowing the methodological approach of Amirapu &
Subramanian (2015), who used multi-sector growth framework to establish
five important criteria that any sector must exhibit in order to lead an economy
to rapid, sustained and inclusive development. According to Amirapu &
Subramanian (2015), the five criteria are the existence of high level of
productivity, dynamic productivity growth, expansion of the sector in terms of
its use of inputs, comparative advantage, and exportability. Thus, the
Ethiopian service sector is evaluated based on these scorecards.
Page | 174
High Level of Productivity
The productivity level of the service sector in Ethiopia in 1981, 2001

and 2011 are calculated and presented using GGDC 10 sectors database in
Table 5.7. The aggregate productivity level for the service sector is rising from
7 in 1981 to 9.2 in 2001 and further to 11.3 in 2011, whereas, it has reduced
for the agriculture and the industry sector. The productivity level of the service
sector is not only increasing but also it is higher than both the agriculture and
manufacturing sector. In 2011, the highest productivity levels are observed in
the utilities (73.2) followed by the finance and business services (54.1),
transport and communication (45.9), and public service sectors (18.7).
Table 5.7: Productivity Level and Growth in Ethiopia (1981-2011)

Productivity level Productivity growth
Year 1981 2001 2011 1981-2001 2001-2011
Agriculture 2.3 1.6 2.4 -0.31 0.52
Industry 14.0 7.7 5.8 -0.45 -0.24
Mining 11.1 5.4 4.7 -0.51 -0.14
Manufacturing 7.1 4.4 3.2 -0.38 -0.27
Other Industry 52.3 24.9 13.5 -0.52 -0.45
Utilities 62.6 68.3 73.2 0.09 0.07
Construction 49.8 18.9 10.6 -0.62 -0.44
Service 7.0 9.2 11.3 0.31 0.23
Market Service 12.0 11.4 11.0 -0.05 -0.04
Distributive Trade 9.9 7.8 7.3 -0.21 -0.07
Transport & com 18.3 32.6 45.9 0.78 0.41
Finance and business 65.8 70.0 54.1 0.06 -0.23
Non Market Service 2.9 6.9 12.1 1.39 0.74
Public services 3.8 10.1 18.7 1.64 0.86
Personal Services 2.0 3.4 5.0 0.67 0.48
Total Economy 2.9 2.7 4.2 -0.09 0.58
From the service sectors, the market services have a slight decline in
the productivity level as compared to the non-market services. As shown in
Table 5.7, productivity level in the non-market services increased from 2.9 to
6.9 and furthermore to 12.1 in 1981, 2001 and 2011 respectively. Productivity
growth in the public service sector is the highest growth witnessed in both
1981-2001 and 2001-2011 periods. It increased from 3.8 in 1981 to 10.1 in
2001 with a growth rate of 1.64. In 2001 to 2011, productivity in the public
sector increased by 86 percent from 10.1 to 18.7. However, from the market
Page | 175
services, productivity has declined in distributive trade service and finance and
business services, while the transport and communication service sub-sector
witness an increase in productivity level. Therefore, in Ethiopia, high service
productivity is observed in the non-market services than the market services.
Even though productivity of the market services has a declining tendency, the
level of productivity is higher than the manufacturing sector, which has a
declining productivity level. Therefore, due to a weak manufacturing sector,
the service sector in Ethiopia has a high level of productivity score, which
enables it to lead and sustain the economic growth of the country.
Dynamic Productivity Growth
Dynamic productivity growth refers to the high productivity growth

rates combined with domestic and international convergence (Ghani &
O’Connell, 2014). As it is shown in table 5.7, during 1981-2001, the service
sector at aggregate level has a positive productivity growth as both the
agriculture and the industry sector shows negative productivity growth. The
highest productivity growth is observed in the non-market services due to a
high productivity growth of the public service sub-sector.
Figure 5.6: Annual Growth of the Service and Manufacturing Sector,

40
30
Annual growth (%)
20
10
0
-10
-20
-30
-40
Year
Service annual growth Manufacturing annual growth

In 2001-2011 periods, productivity growth in the agriculture sector
becomes positive. Productivity in the industry particularly in the
Page | 176
manufacturing sector continues to be negative, as the service’s productivity
growth remains positive. If we compare the productivity growth of the
manufacturing and the service sector as shown in Figure 5.6, both have a
similar trend of rise and fall, though the service sector has a little bit higher
productivity growth than the manufacturing sector. However, in the recent
years, 2012-2014, the productivity growth was reversed as the manufacturing
sector witnessed a higher productivity growth than the services. The change in
the productivity growth of the manufacturing sector is due to high investment
in the manufacturing sector following the GTP plan, which gives due
emphasis to the sector.
Figure 5.7: Service Convergence in Ethiopia (1981–2001) and (2001–2011)
.4
BWA
CHN
.4
CHN
log of change in service productivity (2001-2011)
IND
.2
IND
BWA
GHA
.2
ETH
SKR SKR
ETH SSA
BRZ KNY
0
GHA
TNZ
0
-.2
BRZ TNZ
-.2
SSA
-.4
KNY
0 1 2 3 4 0 1 2 3 4
log of service productivity 2001 log of service productivity 2011

A sector that plays an important role in the growth convergence could
have the potential to sustain economic growth. Here, the potential of the
service sector in growth convergence in Ethiopia is evaluated. According to
Ghani & O’Connell (2014), the high performing African countries including
Ethiopia are experiencing faster growth than developed economies. But, the
growth in these African countries is not supported by high growth in the
manufacturing sector, rather it is the service sector that gave the impetus to
high growth. Figure 5.7 and 5.8 shows the labor productivity in the service
Page | 177
sector and manufacturing sector respectively. The productivity level of 9
countries is included in the analysis including China, India, South Korea,
Brazil, Ghana, Botswana, Tanzania, Kenya, Ethiopia and SSA average. In
Figure 5.7, the fitted values for the change in service productivity and
logarithm value of service productivity is downward sloping. It implies that
countries experiencing a high change in service productivity during the
periods 1981-2001 and 2001-2011 have low service productivity level. During
1981-2001, Ethiopia’s value is below the fitted values of the countries
considered. However, in 2001-2011, Ethiopia witnessed a faster catch up and
growth in service labor productivity. This shows that Ethiopia has more rooms
to catch up more quickly to the countries above the fitted line. In addition, this
indicates the convergence of Ethiopia in the service sector to countries
enjoying high productivity levels, which has an important implication for
sustainability of service-led economic growth.
Figure 5.8: Manufacturing Convergence in Ethiopia (1981–2001) and

(2001–2011)
1
CHN
IND
SKR
.2
log of change in manufacturing productivity (2001-2011)
CHN
BWA
SKR
GHA
0
BRZ
.5
KNY
TNZ
ETH
IND
-.2
GHA
BRZ
SSA
0
TNZ
-.4
ETH
SSA
BWA
-.5
-.6
KNY
0 1 2 3 4 5 0 1 2 3 4 5
log of manufacturing productivity 2001 log of manufacturing productivity 2011
Page | 178
On the other hand, in Figure 5.8, the fitted values for the
manufacturing sector productivity level and change in the productivity has an
upward sloping. It shows that countries with high productivity level in the
manufacturing sector have the larger change in the productivity growth.
Unfortunately, Ethiopia is below the fitted line in both (1981-2001) and
(2001-2011) periods indicating a sluggish progress in the manufacturing
sector. Further, the productivity growth of Ethiopia in the manufacturing
sector is negative. The negative productivity growth of the manufacturing
sector limits the capacity of the sector in growth convergence and sustaining
economic growth. Therefore, the convergence in the manufacturing sector in
Ethiopia requires much more productivity growth in the sector. Thus, with this
level, it has limited capacity to sustain the growth. Similar to this result,
(Ghani & O’Connell, 2014) also find that late comers to the development such
as Ethiopia have a more strong catch-up and growth acceleration in the service
sector than the manufacturing sector.
Expansion of Services
The other important score to evaluate the sustainability of service-led

growth is the expansion of the service sector in employment in Ethiopia. From
the shift-share analysis, the finding shows that there is growth-enhancing
structural change. In addition, in Figure 5.9 and 5.10, the positive correlation
of the shift of labor with the individual sector’s labor productivity indicates
growth-enhancing structural change, the movement of labor to the high
productivity sectors. During 1981-2001, the agriculture sector, which is the
largest employer have negative employment growth and low productivity
level. (The size of the circle in Figure 5.9 and 5.10 indicates the employment
share of the sectors)Thus, labor moves out of the agriculture sector and joins
the service and manufacturing sectors. The high productivity growth sectors
finance and business, utilities and transport and communication service sectors
have a smaller share in employment and the shift of labor to these sectors is
very small as compared to the public service, distributive service, and
manufacturing sector. Therefore, there is structural change enhancing
Page | 179
productivity and labor shift in Ethiopia as labor is shifting from low to high
productive sectors. In addition, the service sub-sectors are growing fast in
productivity growth and absorbing labor fast resulting in structural change,
which can sustain the service, led growth in Ethiopia.
Figure 5.9: Sectoral Productivity and Change in Employment Share,

Ethiopia (1981–2001)
1.5
utility finance and business
transport and communication

1
construction
public service
.5
fitted valuesy=0.62 +12.19x distributive service

mining
manufacturing
personal service
0
agriculture
-.5
-.04 -.02 0 .02

change in employment share 1981-2001
Source: Author’s computation using GGDC database
Similarly, during 2001-2011, the productivity level of agriculture

sector is negative and labor is moving out of the agriculture sector and shifting
to high productivity growth sectors of the services and manufacturing. During
2001-2011, even though there is a shift of labor to the manufacturing sector,
labor productivity, however, is negative limiting structural change and
productivity growth. In addition, productivity in the public service has
increased but employment has declined.
Page | 180
Figure 5.10: Sectoral Productivity and Change in Employment Share,
Ethiopia (2001 – 2011)
1.5
utility
finance and business
transport and communication
1 public service
.5
fitted values y=0.44+ 2.58x construction
Personal service distributive service

mining
0
manufacturing
agriculture
-.5
-.1 -.05 0 .05

change in employment share 2001 - 2011

Comparative Advantage
For a sustained growth and structural change, the alignment between
resource endowments and resource requirements of the economy is important.
That means for developing countries like Ethiopia with abundant and low skill
labor has to use the dominant resources for sustainability of growth. From the
2005/06 SAM analysis, the service sector in Ethiopia is found to be highly
capital intensive and the highly productive sectors of the finance and business
and transportation and communication are skill intensive, employing highly
skilled labor. In addition, the dominant employing distributive trade service
employs low skilled labor. However, the comparative advantage of Ethiopia is
labor, which requires labor-intensive and low-skill sector. Except for the
distributive service sector, most of the sectors are skill intensive. This limits
employability of the labor that shifts from the low productive agriculture to the
high productive service sector.
Exportability
Ethiopia has a comparative advantage in transportation and travel

services export. Ethiopia’s service export in 2009-2012 shows that 22.3
percent is travel export, 55.7 percent is transportation export and 22.0 percent
is other service export. The major transportation export of Ethiopia is
Page | 181
generated by the Ethiopian Airline transportation service. Out of total import
in same period 5.5 percent is travel import and 65.5 percent transportation
import and 29 percent in other services. Since Ethiopia is a landlocked
country, it is a net importer of transportation services and other services too.
From the other services, Ethiopia exports constitute 44.5 percent of
government service (Peace and security in the region, as Ethiopia, is one of the
highest contributors of personnel to the UN peacekeeping missions), 24.9
percent communication service export and 24.2 other business exports. In
addition, from other imports Ethiopian service import, 28 percent is other
business service and 13 percent insurance service (UNCTAD, 2015).
The sustainability of service sector from the export side can be

evaluated by observing how the country covers its bills of the goods import
specifically the manufacturing goods imports. The service and goods export of
Ethiopia moving similarly and there is a small difference for exports. This
indicates that the service export value is as important as the goods export in
Ethiopia. Figure 5.11 shows that Ethiopia’s service sector export covers as
much as 50 percent of the import bill for manufacturers’ inputs. This indicates
the potential of the service sector to finance import bills and support the
growth of the manufacturing sector. Whereas, the foreign exchange that
Ethiopia generates from the manufacturers export is too small.
Figure 5.11: Share of Service Export Earnings and Manufacturer Imports

in Ethiopia (2005-2012)
10 60%
manfacturer import (in current $)
Service export/manufacturer
y =9.5 40%
log of service export and
0.0115x + 0.3003
9 20%
import (%)
8.5 0%
2005 2006 2007 2008 2009 2010 2011 2012
Year
Service exports
Manufacturer imports
service exports/manufacturer imports
Linear (service exports/manufacturer imports)
Page | 182
Therefore, the service sectors’ high productivity level and the high
productivity growth, its potential in growth convergence and greater expansion
of the sector in employment indicates the possibility of the sustainability of
service-led growth in Ethiopia. However, the service sector in Ethiopia lacks a
comparative advantage over other sectors as it is capital intensive and requires
skilled labor. Similarly, Amirapu & Subramanian (2015) found that finance
and business service in India is skill intensive similar with the manufacturing
sector. Thus, it is not aligned with India’s comparative advantage. On the
other hand, a study by Enache, Ghani, & O’Connell (2016) used the
sustainability scorecard to identify sectors that can bring structural change and
rapid sustained growth. Accordingly, they find the distributive service trade
and construction sector fulfilling at least some of the characteristics except
tradability and exportability.
Generally, the service sector growth in Ethiopia in output and

employment is largely driven by income, female economic participation, and
economic reform. However, productivity gap and producer services or
intermediate service inputs are not the major drivers of the service growth. On
the other hand, the sustainability of the service-led growth in Ethiopia is also
feasible when evaluated against the productivity level and growth, the
potential for convergence, expansion, and exportability. However, the sector is
not aligned with the comparative advantage of the country.
Page | 183
CHAPTER-VI
LINKAGE ANALYSIS OF SERVICE SECTOR
IN ETHIOPIA
Overgrown service sector in an economy has got its own implications

for the growth, employment and sustainability of an economic system. We
have already identified the determinants of service sector growth and the
sustainability aspects. In the recent years, there has been a debate on whether
this overgrowth of the service sector has been a stand-alone growth or it has
deep inter-linkages with other sectors of the economy. In this context, using
the input-output formalism and some other well-proven methodologies, this
chapter is an attempt to analyze the backward and forward linkages of the
service sector. It also covers an exercise to identify the key-sectors for
economic growth the system.
6.1 : Service Sector Inter-linkages in Ethiopia

Results and discussion of the analysis on the Ethiopian service sector
inter-linkage are presented and discussed in this section. The descriptive
results are presented first on the major economic description of Ethiopian
economy from the 2005/06 SAM and Input-Output data. The sector wise
employment, income, trade, consumption, and expenditure structure of
Ethiopian economy are presented. Finally, the inter-linkage analysis of
backward and forward linkage along with key sector identification is presented
in the latter section.
6.1.1 : Description of Ethiopian Economy from 2005/06 SAM

The aggregated macroeconomic data of Ethiopia during 2005/06 is
presented in Table 6.1 as a Macro SAM. All the cell entries represent
expenditure from the column account and income to the row account.
Commodities row account represents the total demand of commodities, which
is also equal to the total supply of commodities amounting 267.5 billion
Ethiopian birr (ETB). Of the total supply of commodities, 47 billion birr (17.6
percent) is covered by imports. Whereas from the total demand of the
Page | 184
commodities 42.9 percent is household consumption demand, 11.9 percent
investment demand, 5.9 percent government consumption demand and 6.3
percent are export earnings. From this, we can observe huge trade deficit or
negative trade balance of Ethiopia, which is 16.1 percent import and 6.3
percent export. On the other hand, the 187.3 billion birr in the activity account
represents the gross output from expenditure side and activity income in the
domestic supply. It is composed 34. percent of commodity input, 32.2 percent
of labor and 33 percent of capital inputs.
The factor income from labor and capital is 60.3 billion birr and 62.4
billion birr respectively. From the payment side, out of the total 62.4 billion
birr factor payment of capital, 89 percent is paid to households, 10.7 percent is
payment to public enterprises and the rest 0.3 percent is to the rest of the
world. In addition, from the total household income of 133.1 billion birr, the
larger proportion (45.3 percent) is generated from wage and 41.7 percent from
capital, 1.12 percent from the government transfer, and 11.8 percent from
remittance.From the expenditure side, commodity consumption expenditure
accounts (114.8 billion birr) or 86.3 percent of household total expenditure. In
addition, 15.5 billion birr or 11.64 percent goes to household savings and 2.7
billion birr or 2.02 percent to tax payment.
The Ethiopian government total expenditure during 2005/06 period

was 23.3 billion birr of which 15.9 billion birr or 68.3 percent is spent on
government public service activities (public administration, education, and
health services) and 1.5 billion birr (6.5 percent) on household social transfers.
A larger proportion of government revenue, which is 60.6 percent, is
generated from tax, and the rest 22.8 percent from public enterprises and 16
percent is earned from foreign grants and loans.
Page | 185
Table 6.1: Aggregated Ethiopian Macro SAM (2005/06)
Code Receipts \Payments 1 2 3 4 5 6 7 8 9 10 11 12
1 Activities 187.3 187.3
2 Commodities 65.0 23.1 114.8 15.9 31.9 16.8 267.5
3 Margin 23.1 23.1
4 Labor 60.3 60.3
5 Capital 61.9 0.5 62.4
6 Public Enterprises 6.7 6.7
7 Households 60.3 55.5 1.5 15.7 133.1
8 Government 5.4 14.1 3.7 23.3
9 Tax 10.1 1.3 2.7 14.1
10 Capital account 15.5 5.3 11.0 31.9
11 Rest of the World 47.0 0.2 0.1 0.4 47.7
12 Total 187.3 267.5 23.1 60.3 62.4 6.7 133.1 23.3 14.1 31.9 47.7
Source; Ethiopian Macro SAM (EDRI, 2009)
Billion Ethiopian Birr, 2005/06
Page | 186
6.1.1.1 : Sectoral Contribution to GDP
The sectoral structure of Ethiopian economy during 2005/06 can be
generated from the Social Accounting Matrix. The contribution of each sector
in total output, total demand and total value added at factor cost is computed
in the following way. The total output generated by each sector in 2005/06, as
shown in Table 6.2, is 187.3 billion Ethiopian birr. The total output
incorporates the intermediate demand and value added costs. Out of the total
output produced, the service sector constitutes the largest share, which is 42.2
percent. The largest contribution within the service sector also comes from the
distributive service sector, accounting 18 percent followed by the public
service sub-sector with 9.1 percent of the total output. Here, the share of
agriculture in total output is lower than service sector, accounting 34.7
percent, and the manufacturing sector accounts only 9.6 percent. On the other
hand, the total demand refers to the intermediate demand and the final demand
of the commodities. The value of the total demand during 2005/06 was 267.5
billion Ethiopian birr. Here also the service sector has the largest share in total
demand along with the manufacturing sector. The demand for service and
manufacturing sector accounts for 34.3 percent each; which is more than the
agriculture sector (21.9 percent). Specifically, the huge demand for services
comes from the distributive trade service-sub sector (13 percent), which
incorporates all the demands for wholesale and retail trade services and trade
margins (transaction costs), repair services, and hotel and restaurants.
Table 6.2: Share of Sectors in Total Gross Output and Total Demand
Gross Output/GDP Total DD
Sectors
ETB Percent ETB Percent
Agriculture 65.0 34.7 58.5 21.9
Manufacturing 18.0 9.6 91.9 34.3
Other Industries 25.2 13.5 25.5 9.5
Mining 0.7 0.4 1.1 0.4
Utility 3.3 1.7 3.1 1.2
Construction 21.2 11.3 21.3 7.9
Service 79.1 42.2 91.6 34.3
Distributive Service 33.8 18.0 34.8 13.0
Transport and Com 10.3 5.5 18.9 7.0
Business & Finance 14.3 7.6 17.5 6.5
Public Service 17.1 9.1 17.2 6.4
Personal and Other Service 3.6 1.9 3.2 1.2
Total 187.3 100.0 267.5 100.0
Source, Ethiopian Macro SAM (EDRI, 2009)
ETB - Billion Ethiopian Birr, 2005/06
Page | 187
The value added share of the sectors is presented in Table 6.3. During
2005/06, Ethiopia’s largest factor cost is spent on the agriculture sector
amounting 48.1 percent of the total GDP at factor cost. Next to agriculture,
40.4 percent of the total value at factor cost went to the service sector. The
distributive service sub-sector constitutes the largest proportion (13.4 percent)
of all the sub-sectors. However, the contribution of manufacturing to GDP at
factor cost is 4.7 percent, which is lower than all the service sub-sectors
contribution except the personal services. In addition to the sectoral
contribution to GDP, sectoral factor intensity can be also compared. A sector
with a larger share of labor than capital is labor-intensive sector and a sector
that has larger capital than labor is a capital-intensive sector. Thus, the most
labor-intensive sector is agriculture with 73.5 percent of the sectors’ value-
added is paid to labor. In addition, agriculture is also the least capital-intensive
sector taking only 23.4 percent of the sector’s value-added to capital. On the
other hand, services are the most capital intensive sectors which paid 61.4
billion birr of the sectors’ value added to capital and only 18.9 percent is paid
to labor. Among the service sub-sectors, distributive service; and business and
finance services are the most capital-intensive sectors. However, public
services and distributive services are more labor intensive than other service
sub-sectors. Here, we can observe that services are more capital-intensive
sector at the aggregate level but more labor intensive than the manufacturing
and industry sectors.
Generally, agriculture is a very important sector in the Ethiopian

economy as it is a source of livelihood for the larger section of Ethiopian
population due to its labor-intensive characteristics. Whereas, the service
sector constitutes the largest capital-intensive sub-sectors implying that the
availability of capital is an important factor to service growth especially for
the distributive trade service and business and finance service sub-sectors.
Moreover, the distributive trade service and the public service sector is also an
important source of labor employment, as they are more labor intensive than
the manufacturing and other industry sectors.
Page | 188
Table 6.3: Share of Sector’s Value Added at Factor Cost
Labour Capital Total Value at factor Cost
Sector
ETB % ETB % ETB %
Agriculture 44.3 73.5 14.46 23.4 58.78 48.1
Manufacturing 2.38 4.0 3.38 5.5 5.77 4.7
Other Industries 2.21 3.7 6.06 9.8 8.28 6.8
Mining 0.18 0.3 0.47 0.8 0.67 0.5
Utility 0.66 1.1 1.63 2.6 2.29 1.9
Construction 1.36 2.3 3.95 6.4 5.32 4.4
Service 11.37 18.9 38.01 61.4 49.39 40.4
Distributive Service 4.12 6.8 12.23 19.7 16.35 13.4
Transport &Comm 0.61 1.0 5.74 9.3 6.35 5.2
Business & Finance 0.39 0.7 11.54 18.6 11.94 9.8
Public Service 4.39 7.3 6.92 11.2 11.31 9.3
Personal & Other Service 1.85 3.1 1.57 2.5 3.43 2.8
Total 60.29 100.0 61.92 100.0 122.22 100.0
ETB – in billions birr (2005/06)
Source; Authors computation from Ethiopian (2005/06) SAM
In addition, the relative factor shares of each sector at aggregated and
disaggregated level are depicted inFigure6.1. In 2005/06, for Ethiopia, 50.7
percent of GDP is generated by capital, while 49.3 percent is generated by
labor. Sector-wise, there are significant differences between agriculture and
service sector in their factor use. The graph clearly indicates that agriculture is
labor intensive while service is capital intensive. Within the service sector,
distributive service, business and finance, and transport and communication
services are largely capital-intensive sectors.
Figure 6.1: Relative Factors Share of Ethiopia (2005/06)

80.0%
70.0%
Factor Share
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
Labour
Sectors
Source, Authors computation using 2005/06 SAM
Page | 189
6.1.1.2 : Sectoral Composition of Labour
The factor account of the 2005/06 Ethiopian SAM can also inform the
type of labor force employed in each sector. In this study, the labor factor
account is aggregated into four types of workers. These are agricultural
worker, administrative and professional worker, skilled workers and unskilled
workers. According to the 2005/06 Ethiopian SAM, administrative workers
refer to legislation, senior officials and managers; professional workers
include technical and associate professionals. Unskilled workers refer to
workers who are engaged in jobs which only require little skill or lowest level
of education (workers engaged in elementary occupation). Skilled workers are
workers that have some skill obtained through formal education and training,
experience and informal training (EDRI, 2009).
Figure 6.2 shows the composition of workers employed in each sector

in Ethiopia during 2005/06. As it is depicted in the figure agriculture sector
fully employed agricultural workers that operate as an employed worker in
others farm. The manufacturing sector employed a larger proportion of skilled
workers than the service sector at aggregate. About 86 percent of labor
employed in the manufacturing sectors is skilled workers. On the other hand,
the service sector’s share in employing skilled labor is 51.7 percent. However,
at sub-sector level, the distributive sector employed the largest share of skilled
labor (91.6 percent) followed by transport and communication service sub-
sector (79.0 percent).The business and finance service sub-sector, on the other
hand, employed more of administrative and professional workers than skilled
workers. That means there are more of senior officials, managers, technical
and associate professionals in the business and finance service. In addition, the
largest proportion of administrative and professional workers are employed in
the public service sector than other sectors
Page | 190
Figure 6.2: Sectoral Composition of Labour, Ethiopia (2005/06)
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
Agricultural Admin and Professional Unskilled Skilled
Source, Authors estimation from 2005/06 Ethiopian SAM
6.1.1.3 : Sectoral Consumption Expenditure by Sectors
The sectoral consumption expenditure share in Table 6.4 shows the

percentage share of households (poor and non-poor households) spending on
different sectors. As it is depicted in the table, households spend a large
proportion of their income on manufacturing goods (38.7 percent) and
agricultural products (37.4 percent). The service sector consumption
expenditure share accounts 22.5 percent of total household consumption
expenditure. From the service sub-sectors, households spending on business
service is the largest household consumption expenditure accounting 9 percent
followed by the distributive service sector (7.8 percent). On the other hand, the
least consumed service by households is the public service sector, as
households only spent only 0.9 percent or 1 billion birr of their income.
Page | 191
Table 6.4: Households Consumption Expenditure Shares by sector,
Ethiopia (2005/06)
Poor Households Non-poor Households Total
Economic Sectors ETB Percent ETB Percent ETB Percent
Agriculture 11.9 46.8 31.0 34.7 42.9 37.4
Manufacturing 8.4 33.1 36.0 40.2 44.4 38.7
Mining 0.0 0.0 0.3 0.3 0.3 0.2
Utility 0.3 1.2 1.1 1.3 1.4 1.2
Construction 0.0 0.0 0.0 0.0 0.0 0.0
Service 4.8 18.9 21.0 23.5 25.8 22.5
Transport &Comm. 0.2 0.9 2.5 2.8 2.8 2.4
Public Service 0.2 0.9 0.8 0.9 1.0 0.9
Total 25.4 100.0 89.4 100.0 114.8 100.0
As theory and empirical evidence show, when compared to poor

households, services were more consumed by the non-poor households. Non-
poor households spent 23.5 percent of their income on consumption of service
as the poor households spent only 18.9 percent. From the service sectors, both
the poor and the non-poor households spent more on the business and finance
services followed by the distributive trade service. However, relative to the
poor households the non-poor households consume more of the distributive
service, transport and communication services, and personal services. The
poor households spent more on business and financial services as compared to
the non-poor households. The proportion of income spent on public service by
both poor and non-poor households is same. On the other hand, in Table 6.4,
the consumption expenditure of government on commodities is 15.9 billion
birr. From these amount of government expenditure, 64 percent is spent on
public administration services, 29 percent on education and 7 percent on
health services.
Page | 192
6.1.2 : Structure of Import and Exports
The SAM of Ethiopia for the year 2005/06 can indicate the structure of
imports and exports. As it is indicated earlier, Ethiopia’s export is well below
the total imports of the country leaving the country in a huge trade deficit. The
total import value of Ethiopia in 2005/06 is 47 billion birr whereas the export
value is only 16.8 billion birr. When we see what constitutes the total exports,
the export of agricultural commodities takes 40.9 percent of the total export
share. This is similar to several developing countries; where the majority of
their export relies on agricultural commodities. However, one important
feature in the Ethiopian export structure is the share of the service sector,
which is almost equal to the agriculture sector accounting 40.1 percent of the
total export. This implies that service sector export is as crucial as the
agriculture sector for export earnings. The source of the increasing role of the
service sector in export is the transport and communication service sub-sector,
which originates from the dominant transportation service of the government-
owned Ethiopian Airlines and Shipping Line. It accounts 30.7 percent of the
total exports of the country in 2005/06. The export of the manufacturing sector
is only 18.7 percent of the total exports.
With reference to import, Ethiopia’s import overwhelmingly

concentrates on manufacturing goods (70.8 percent). Ethiopia imports major
manufactured products that can be used as intermediate input like petroleum
and coal, machinery, chemicals, metal, vehicles, electronics equipment and the
like. Next to the manufacturing sector, the second largest import of Ethiopia is
the transport and communication service, which is 17.8 percent. The service is
mainly related to the port services, as Ethiopia is a landlocked country. It pays
for port services to the neighboring countries for its import and export
activities.
Another important feature of the export and import of Ethiopia can be

described using import penetration ratios and export intensities. These
measures show ‘the relative importance of commodities in the overall
economy’ (EDRI, 2009). The import penetration ratio is the share of imports
Page | 193
in the value of total demands (ratio of total import to total demand), and export
intensity refers to the share of exports in the value of gross output (ratio of
exports to gross output). The import penetration ratio of Ethiopia in 2005/06,
which is shown in table 6.5 indicates that the Ethiopian manufacturing sector
faces the high import competition as 34.7 percent of total demand is imported.
The transport and communication service sub-sector is also one of the most
tradable service activity with 44.5 percent of the service demand is supplied
by foreigners. This is due to the port services import of the country from
Djibouti. According to UNCTAD (2015), the transportation, storage and
communication service sub-sector accounted for more than 25 percent of
output in Djibouti, reflecting the economic importance of its port services to
the landlocked country Ethiopia.
Table 6.5: Export and Import Trade Structure of Ethiopia, 2005/06

Share
Share in Export Import
Export Import in
Sector export intensity Penetration
in ETB in ETB Import
(%) (%) (%)
(%)
Agriculture 6.9 40.9 10.6 2.2 4.7 3.2
Manufacturing 3.1 18.7 17.4 33.3 70.8 40.8
Mining 0.0 0.0 0.2 0.1 0.1 6.2
Utility 0.1 0.3 1.6 0.0 0.0 0.0
Construction 0.0 0.0 0.0 0.0 0.0 0.0
Service 6.7 40.1 8.5 11.4 24.3 10.0
Transport and Com. 5.1 30.7 49.8 8.4 17.8 44.5

Public Service 0.1 0.3 0.3 0.1 0.2 0.4
Total 16.8 100 9.0 47.0 100.0 16.2

On the other hand, the export intensity shows that the transport and
communication sector is the most export-intensive sector than the agriculture
sector. About 49.8 percent of the transport service output is sold abroad, while
only 10.6 percent of the agricultural outputs are exported. Therefore, in
2005/06 the service sector in general, and the transport sector, in particular, is
the most dynamic service sector with high intensity of import and export
Page | 194
trades. This means, Ethiopia exports transport services through the state-
owned Ethiopian Airlines and gets transport services (port services) for its
foreign trade from Djibouti and Somaliland. This made the country to be one
of the few large and land-locked countries in the world that exports more
services than goods (WB, 2012b).
The overall total output export of Ethiopia is only 9 percent of the

country’s total production whereas; the total import is 16.2 percent. The big
difference between the import and export of Ethiopia resulted in the total trade
deficit, which was 30.2 billion birr, accounting for 22.8 percent of the GDP
(EDRI, 2009). In addition, the ratio of total trade to GDP implies that Ethiopia
was a moderately an open economy where the value of its total trade (48.2
percent) was almost half of its gross domestic product at market prices.
6.1.3 : Results from SAM and Input-Output Analysis
6.1.3.1 : Intermediate Demand and Factor Inputs
The production technology coefficients represent the intermediate

demand and factor inputs, which show the goods and services, and factors
used by each sector in the production processes. It shows the share of input
required to produce a unit of each sector’s outputs. The information on sectors
production technologies is taken from the input-output table. The differences
in the production technologies across sectors can be observed from this
technology matrix. From Table 6.6, we can observe each sector’s ratio of
spending on factor and non-factor inputs.
The total factor inputs for agriculture, mining and personal service
account over 90 percent of the total input spending implying their reliance on
factor inputs than commodity inputs. About 68.1 percent of the total factor
input spending (90.5 percent) in agriculture is spent on agricultural workers
and only 21.5 percent is spent on agricultural capital. The mining sector input
requirement is also composed of 65.3 percent of input of the capital and 16.3
percent of skilled labor and 5.2 percent from the utility sector. Whereas for
the personal service sector, out of its total factor input requirement
Page | 195
(95.1percent), 43.7 percent is non-agricultural capital input and 51.5 percent is
labor input.
On the other hand, the manufacturing sector, construction sector, and

the distributive trade service sector required more of intermediate inputs from
the commodity sectors than the factor inputs. The manufacturing sector input
composition shows that 67.9 percent and 32.1 percent of the input is required
from commodity inputs and factor inputs respectively. Above half of the
commodity input for the manufacturing sector is derived from its own output.
From the service sectors, the manufacturing sector is more linked to the
finance and business service sub-sector than other services, which is about 6.4
percent. This means that for 100 birr amount of manufacturing production, 6.4
birr is spent on finance and business services. The sector requires financing
services from financial institutions and various types of business services like
auditing, consulting, renting and the like services. 16.2 percent of the
commodity input for the manufacturing sector is a raw material from the
agriculture sector.
The input requirement for the construction sector also shows that 75
percent of the input is commodity input and 25 percent is factor input. From
the commodity inputs, the sector consumes huge share (52.9 percent) from the
manufacturing sector, 7.3 percent from business and finance service, and 3
percent from transport and communication services. Here also the business
and finance service, and the transport and communication service sectors are
more interlinked than the other service sub-sectors. The distributive trade
service sector input requirement is also satisfied by the commodity input (51.6
percent) and from factor inputs (48.4 percent). This service sub-sector is
highly dependent on transport and communication services because 26.1
percent of the commodity input is covered by the transport and service inputs.
It also purchases 13.1 percent of its inputs from the manufacturing sector.
Page | 196
Table 6.6: Production Technology Coefficients, Ethiopia (2005/06)
Commodity and Factor inputs Sectoral Production Technology Coefficients (%)
1 2 3 4 5 6 7 8 9 10 11 12 13
1 Agriculture 4.3 16.2 4.7 0.0 0.0 5.5 1.7 3.8 0.0 0.0 0.2 0.0 4.4
2 Manufacturing 3.7 38.9 47.5 1.7 22.8 52.9 12.8 13.1 26.1 2.4 15.3 1.3 16.8
3 Other Industries 0.0 3.6 6.0 5.2 3.9 6.3 3.9 2.5 0.6 9.0 5.1 0.5 2.8
4 Mining 0.0 1.5 4.2 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7
5 Utility 0.0 2.1 1.6 5.2 2.8 1.3 1.1 1.4 0.6 0.3 1.7 0.2 0.9
6 Construction 0.0 0.0 0.1 0.0 1.1 0.0 2.8 1.1 0.0 8.7 3.4 0.3 1.2
7 Service 1.5 9.1 9.0 1.9 2.9 10.2 19.2 32.2 11.8 5.1 13.1 3.0 10.7
8 Distributive Service 0.8 0.8 0.0 0.0 0.0 0.0 1.7 0.2 7.4 0.0 3.2 0.0 1.1
9 Transport and Communication 0.4 1.9 2.6 1.1 0.4 3.0 12.2 26.1 2.2 1.4 2.2 0.8 5.8
10 Business & Finance 0.2 6.4 6.2 0.8 1.1 7.2 4.5 5.9 1.2 2.5 6.2 1.3 3.4
11 Public Service 0.0 0.0 0.0 0.0 0.0 0.0 0.3 0.0 0.0 0.0 1.5 0.2 0.2
12 Personal and Other Service 0.0 0.0 0.2 0.0 1.4 0.0 0.4 0.0 1.0 1.2 0.0 0.7 0.2
13 Total Commodity inputs 9.5 67.9 67.2 8.8 29.6 75.0 37.6 51.6 38.5 16.6 33.8 4.9 34.7
Agricultural workers 68.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 23.6
Admin.& Professional labor 0.1 0.4 0.9 2.0 1.3 0.8 5.7 0.9 1.1 1.3 19.8 15.3 2.6
Unskilled workers 0.0 1.4 3.3 7.5 12.3 1.8 1.2 0.2 0.2 0.2 0.9 19.8 1.1
Skilled workers 0.0 11.4 4.6 16.3 6.7 3.8 7.4 11.2 4.7 1.2 5.0 16.4 4.9
Agricultural Capital 21.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 7.4
Non-agricultural Capital 0.8 18.8 24.0 65.3 50.1 18.6 48.1 36.2 55.6 80.7 40.5 43.7 25.6
Factor inputs 90.5 32.1 32.8 91.2 70.4 25.0 62.4 48.4 61.5 83.4 66.2 95.1 65.3
Total input= Gross Output 100 100 100 100 100 100 100 100 100 100 100 100 100
Sector Number Economic Sectors Sector Number Economic Sectors

1 Agriculture 6 Distributive Trade service
2 Mining 7 Transport & Communication Service
3 Manufacturing 8 Business & Finance Service
4 Utility 9 Public Service
5 Construction 10 Personal & other Service
Source, Ethiopian 2005/06 SAM using SimSIP SAM Software
Page | 197
The intermediate input demand is only 34.7 percent while the factor
input accounts 65.3 percent of the total input requirement for Ethiopian
economy in 2005/06. From the intermediate inputs, the manufactured goods
are the most important as it accounts 16.8 percent of the total inputs. The
service sector at the aggregate level is the next important source of input for
Ethiopian economy by contributing 10.7 percent of the intermediate input
requirement. From the specific service sub-sectors, the transport and
communication sector takes the largest share as it is used widely by all sectors
for the purpose of transporting and storing goods (transportation margin as
transaction costs) in the process of marketing. The business and service sector
follows the transportation and communication services in its role as an
intermediate input in the production process.
6.1.3.2 : Sectoral Inter-linkage Analysis
The Sectoral inter-linkage analyses in this section are conducted using

three different approaches to identify the level of service sector inter-linkage
in Ethiopia in 2005/06. These are the Hirschman-Rasmussen approach, the
weighted backward and forward linkage and pure backward and forward
linkage. The forward linkage of service sector measures the change in income
in the sector relative to the average change in the economy, which is caused by
‘a unitary injection in the final demand of all sectors’. According to (Parra &
Wodon, 2010a), ‘if the forward linkage for the sector I is greater than one (or
100 percent in percentage terms), the change in sector i’s income is higher
than the average income change in the economy after a unitary injection in all
sectors’. The higher the value of the forward linkages of a given sector, the
greater will be the impact of a price increase in this sector on the price levels
in the economy (Drejer, 2002). Therefore, Jones (1976) as cited in (Khanal,
2011) termed forward linkage as the input multiplier.
On the other hand, the backward linkage of sector j quantifies the

change in economy-wide income, relative to the average change in the
economy, which is caused by a unitary injection in the final demand of sector j
(Parra & Wodon, 2010a). A unit change in the final demand of sector j will
Page | 198
generate an increase in output above the economy’s average if the backward
linkage is greater than one (or 100 percent in percentage terms). The sector
with the higher value in backward linkages will have the greater effect on the
demand for domestic production (Drejer, 2002). Therefore, the backward
linkage can be termed as the output multiplier (Reis & Rua, 2006).
Backward and Forward Linkages (Hirschman-Rasmussen Approach)
The first inter-linkage analysis described here is the Hirschman-

Rasmussen approach. In this approach, the backward linkage is simply the
column sums of the Leontief inverse matrix and the forward linkage is the row
sum of the inverse matrix. The backward linkage ‘measures the extent to
which a unit change in the demand for the product of sector j causes
production increases in all sectors’ (Cuello, Mansouri, & Hewings, 1992). The
forward linkage ‘measures the magnitude of output increase in sector i, if the
final demand in each sector were to increase by one unit’. This means it
measures the extent to which sector iis affected by an expansion of one unit in
all sectors. However, the backward and forward linkages used here do not take
into account the relative importance of each sector's final demand or level of
sectoral output. According to (Cuello, Mansouri, & Hewings, 1992), ‘the
Hirschman-Rasmussen indices are computed by assuming that ‘all sectors are
equally important in terms of the weighting scheme used so that the scaling
vector is equal to unity for all sectors’. Table 6.7 shows the inter-industry
linkages, backward and forward linkages, using the Hirschman-Rasmussen
approach for Ethiopia in 2005/06.
As it is shown in Table 6.7, all the service sub-sectors, except the

distributive trade service sector, have relatively stronger backward linkage
than the forward linkages indicating the inter-linkage of the service sectors
from the demand side. In the backward linkages, the agriculture sector stood
first followed by the distributive service trade and the public service sector.
The backward linkage for agriculture is 1.46.It indicates that a one-unit change
in the demand for agriculture commodities causes production increases in all
sectors by 1.46 units. Similarly, a unitary change in the distributive trade
Page | 199
service results in a 1.15 unit change increase in the demand for all the sectors.
In the backward linkage analysis, we found that the manufacturing sector is
the least backward oriented sector followed by the transportation and
communication sector and personal service sectors. For the manufacturing
sector, the backward linkage value of 0.672 indicates that for a unit change in
the output of the manufacturing sector, changes the demand for all the sectors
by a lesser proportion of 0.672 amounts. This indicates that the sector’s weak
interconnection in stimulating production in the other sectors. That means the
manufacturing sector in Ethiopia is underdeveloped. Similar to the
manufacturing sector, the service sub-sectors, transport and communication,
and personal services have weaker interconnection and the small multiplier
effect on output growth in other sectors.
Table 6.7: Backward and Forward Linkages, Ethiopia (2005/06)

Backward Forward Total Linkage
Economic Sectors BL Linkage
Rank* FL Linkage
Rank TL Rank
1 Agriculture 1.461 1 2.844 1 4.304 1
2 Mining 1.085 4 0.161 10 1.246 9
3 Manufacturing 0.672 10 2.372 2 3.044 3
4 Utility 1.066 7 0.246 7 1.311 8
5 Construction 1.079 6 0.236 9 1.315 7
6 Distributive Trade services 1.150 2 2.295 3 3.445 2
7 Transport & Comm. 0.694 9 0.650 5 1.344 6
8 Business & Finance 1.080 5 0.748 4 1.828 4
9 Public Service 1.109 3 0.521 6 1.630 5
10 Personal & other Service 0.995 8 0.239 8 1.234 10
Maximum 1.461 2.844 4.304
Minimum 0.672 0.161 1.234
*The rank is based on values for each type of linkages; the highest value sector is ranked 1
and the least as 10.
BL =Backward linkages
FL = Forward linkages
TL = Total linkage (sum of backward and forward linkage)
On the other hand, the agriculture sector is also again the most forward
oriented sector followed by the manufacturing, distributive trade service and
business and finance service sectors. The forward linkage index for the
agriculture sector is 2.84. If the final demand in each sector were to increase
by one unit, the agriculture sector could be affected by 2.84 units for a one-
Page | 200
unit expansion in all other sectors. Whereas, the mining and the construction
sectors are the least ‘forward-oriented linkage’ sectors. A sector with weak
forward linkage implies their weak interconnection with other sectors in
supplying their output as a final demand for other sectors.
The total linkages show that the agriculture sector is ranked first
(4.304); distributive trade service ranked second (3.445); manufacturing sector
ranked third (3.044); and the business and finance, public service and transport
and communication service sector ranked from fourth to six respectively. The
personal service and other services ranked last showing its weak inter-linkage
with other sectors. All the service sectors, except the personal services, have
strong inter-linkage with other sectors as compare to the mining, utility and
construction sectors. In addition, the result shows a stronger inter-linkage of
the distributive trade service more than the dynamic manufacturing sector,
which implies the importance of the sector in the Ethiopian economy. Since
the value of the inter-industry linkage shows the strength of the sectors among
all the economic sectors, the strongest sector is agriculture followed by the
distributive service trade and manufacturing sector.
Weighted Backward and Forward Linkages
In addition to the Hirschman-Rasmussen backward and forward

linkages, the weighted backward and forward linkages are also computed here
to ‘account the relative importance of each sector in terms of GDP, final
demand, or total production’ (Parra & Wodon, 2010a). Table 6.8 shows the
weighted backward and forward linkages of economic sectors for Ethiopia in
2005/06 taking into account the weight of each sector in their inter-linkages.
What is different here from the previous results using Hirschman-Rasmussen
approach is the backward linkage value of the construction sector. When the
weighted backward linkage is considered, the construction sector becomes the
third important sector exceeding the public service sector and the business and
finance services. This is due to the higher weight of the construction sector in
total demand than the public and business finance services. However, both the
agriculture and distributive trade service sectors stood first and second
Page | 201
respectively. This shows that the weight of the two sectors does not alter their
rank, which was computed based on the Hirschman-Rasmussen backward and
forward linkages. Note that the weighted total linkage indices for the
agriculture sector, distributive trade service, and the manufacturing sector are
greater than the values for the Hirschman-Rasmussen backward and forward
linkages.
The weighted backward linkage indices for the transport,

communication and personal service sectors show that they are the weakest
sectors in their backward orientation. Generally, the mining, manufacturing,
utility, transport and communication and personal service sectors have
backward linkage indices below one. For instance, the value for transport and
communication sector, which is 0.625, implies that for a one birr increase in
the final demand for the transport and communication sector increases the
total output of each sector by 0.625 birr only. The weighted backward linkage
value for the manufacturing sector is also below one (0.819). This means, the
total output of each sector in Ethiopia during 2005/06 increased by 0.819 birr
for a 1 birr increment in the final demand of manufacturing sector. These
sectors, with small backward linkages values in the economy, have a lower
dependency on the domestic intermediate goods, and they are more dependent
on import. Except for personal services, the import penetration ratio for the
manufacturing, transport and communication sector, utility and mining sectors
are very high (See Table 6.8). The manufacturing sector requires a larger
proportion of its input from commodity inputs, specifically from its self and
from agriculture. However, these inputs for the manufacturing sector are more
of imported than domestically produced inputs. This hampers the growth of
the manufacturing sector, which results in weak inter-linkage with the service
sector and constrains the growth of service sector as well. Therefore, the
growth of the manufacturing sector as a final demand does not induce growth
in the other sectors. Whereas, the high weighted background linkage values
for the agriculture, distributive trade service, and construction service sector
indicates that these sectors have a higher dependency on the domestic
intermediate inputs.
Page | 202
Table 6.8: Weighted Backward and Forward Linkages, Ethiopia
(2005/06)
Backward Forward Total Linkage
Economic Sectors Linkages
WBL Rank Linkages
WFL Rank WTL Rank
1 Agriculture 1.744 1 3.253 1 4.997 1
2 Mining 0.928 6 0.035 10 0.964 9
3 Manufacturing 0.819 8 2.609 3 3.428 3
4 Utility 0.899 7 0.113 9 1.012 8
5 Construction 1.040 3 0.201 7 1.241 7
6 Distributive Trade services 1.227 2 2.698 2 3.925 2
7 Transport & Comm. 0.625 10 0.675 5 1.300 6
8 Business & Finance 1.011 4 0.718 4 1.728 4
9 Public Service 1.008 5 0.391 6 1.398 5
10 Personal & other Service 0.771 9 0.127 8 0.897 10
Maximum 1.744 3.253 4.997
Minimum 0.625 0.035 0.897
and the least as 10.
WBL = Weighted Backward linkages
WFL = Weighted Forward linkages
WTL = Weighted Total linkage (sum of Weighted backward and forward linkage)
The weighted forward linkage value for agriculture is again very high,
which is 3.253, implying the sector’s high production multiplier effect. That
means, for a one birr increase in the agriculture as primary inputs, the total
output of each sector increases by 3.253 birr. The distributive trade service
sector is second, next to agriculture in its forward orientation. This indicates
the importance of the distributive service sector in Ethiopia, as it has higher
interconnection with other sectors inducing production in other sectors. For a
birr increment in the distributive trade services, the total output of all the
sector (including its self) increases by 2.698 birr. This means the distributive
trade service is highly interconnected with other sectors by providing its
service as an intermediate input. This includes all the transaction costs
(margins) and wholesale and retail trade services that can be used as an input
to other sectors, stimulating production in other sectors. With regard to the
manufacturing sector, it has a weighted forward linkage value of 2.609, which
indicates that the sector has strong forward linkage but weaker backward
linkage (0.819). The total output of all the sectors in the economy would
increase by 2.609 birr if the manufacturing sector as an intermediate input
Page | 203
increased by one birr. This indicates that the manufacturing sector is an
important supplier of input for other sectors. More of the manufacturing sector
outputs do not go to the final consumptions rather it is mainly used as an
intermediate input to other sectors. On the other hand, most of the economic
sectors in Ethiopia during 2005/06 are meant for final consumption.
According to Table 6.8, the mining, utility, personal services, construction,
public service, transport and communication sector, and business and finance
service sectors have a weighted forward linkage indices lower than one. These
sectors are weakly interlinked with other sectors, as the output of all these
sectors goes mainly for final consumption.
Pure Backward and Forward Linkages
The pure linkage measures the backward and forward linkages in

monetary terms, by quantifying the pure impacts of the production of a sector
on the overall economy discounting the demand for inputs by that sector itself
and the feedback from the economy to the sector (Sonnis, Guilhoto, Hewings,
& Martins, 1995). Based on this, the pure backward linkage for the
distributive trade service estimates the pure impact of production in the
distributive trade service sector on the overall economy by excluding the
demand for inputs of the sector from itself, and the feedback effects from and
to the sector. On the other hand, the pure forward linkage for the distributive
trade service sector quantifies the pure impact of the production of the rest of
the economy on the sector (Parra & Wodon, 2010a). In addition, according to
Parra and Wodon (2010a), pure linkages can be used for a direct comparison
of the relative importance of the economic production sectors in monetary
values. The pure inter-industry linkages of Ethiopia in 2005/06 is computed
and presented in Table 6.9.
Page | 204
Table 6.9: Pure Backward and Forward Linkages, Ethiopia (2005/06)
Pure Backward Pure Forward Pure Total
Rank
Economic Sectors Linkages Linkages Linkage
1 Agriculture 308.42 310.25 618.66 1

2 Mining 7.87 5.99 13.86 10
3 Manufacturing 238.75 246.03 484.79 2
4 Utility 21.85 18.59 40.44 8
5 Construction 154.39 107.82 262.21 4
6 Distributive Trade Service 188.33 199.01 387.34 3
7 Transport & Comm. 75.69 21.56 97.25 7
8 Business & Finance 110.62 55.27 165.89 6
9 Public Service 120.73 106.71 227.44 5
10 Personal & other Service 21.06 18.29 39.34 9
and the least as 10. The rank for pure backward and forward linkages is the same with the total
pure linkage.
In terms of pure linkages, the distributive trade service sector ranked

third in both backward and forward linkages. As compared to the previous
inter-linkage types, the manufacturing sector effect has increased in the pure
linkage effects, exceeding the effects of the distributive service sector.
Whereas similar to the case in the Hirschman-Rasmussen and weighted inter-
linkage analysis, the agriculture sector has a leading pure interlinkage effect
generating a total pure linkage effect of 618.6 billion birr with a pure
backward linkage effect of 308.4 billion birr and pure forward linkage effect
of 310.2 billion birr.
The manufacturing sector becomes second generating 238.7 billion birr

of pure backward linkage effects and 246 billion birr forward linkage effects.
This means production in the manufacturing sector has a 238.7 billion birr
pure impact on the overall economy, excluding its own demand for inputs.
When the demand for inputs from the manufacturing sector itself is included
in the previous backward linkage analysis, the manufacturing sector tends to
be the weakest sector in its backward orientation. This is because the fact that
the larger proportion of the intermediate input for the manufacturing sector
comes from itself and in Ethiopia, this is obtained mainly from imports.
Therefore, in the pure backward linkages, the manufacturing sector has higher
Page | 205
backward linkage effects. On the other hand, the pure forward linkage effect
of the manufacturing sector shows that production changes in the other sectors
have a 246 billion birr pure impact on the manufacturing sector.
Generally, the findings in this study are in conformity with various

studies on the level of backward and forward linkages of most of the sectors,
except for the manufacturing sector and distributive trade service in Ethiopia.
In the Hirschman-Rasmussen and the weighted inter-linkage types, we can
observe that both the agriculture and distributive trade service sectors are
leading in all the backward and forward linkage indices. This implies that the
sectors are dynamic as they have high inter-sectoral linkage with other sectors
and with themselves. However, when pure linkage effect is singled out by
excluding the linkage of the sectors with themselves, we find the
manufacturing sector more interlinked with other sectors as compared to the
distributive trade service sector.
Key Sector Identification and Analysis
Key sectors, which have strong inter-sectoral inter-linkage, are usually

sectors with both backward and forward linkages greater than one. ‘A sector
with backward (forward) linkages greater than one, and forward (backward)
linkages less than one, is called backward (forward) oriented’ and ‘if none of
the linkages is greater than 1, the sector is called weak’ (Parra & Wodon,
2010a). The key sector analysis classifies activities based on the size of their
forward and backward linkages in a graphical presentation. The sectors are
classified into four categories or quadrants. These are key sectors, backward
oriented sectors, forward oriented sectors and weak sectors. Activities that
have greater than one backward and forward linkage are key sectors. The
second group of key sectors is those with high backward linkage (backward
linkage coefficient greater than one) and low forward linkage (forward linkage
coefficient less than one). (Hirschman, 1958)give priority to the backward
linkages due to their more powerful pressure on linkages than forward
linkages, especially in developing countries. The third group of sectors in
terms of priority is those having backward linkage coefficient less than one
Page | 206
and high forward linkage coefficient greater than one. While the last group
consists of low forward as well as backward linkages, that is, both backward
and forward linkage coefficients are less than one. Thus, in this section, the
key economic sectors in Ethiopia during 2005/06 are identified from the
backward and forward linkages of the Hirschman-Rasmussen approach, the
weighted inter-linkages, and pure linkages. Figure 6.3 shows the key sectors
identified using the Hirschman-Rasmussen approach while Figure 6.4 shows
the key sectors based on the weighted linkages. Finally, Figure 6.5 depicts the
key sectors based on pure linkages.
The upper right quadrant in Figure 6.3 shows that the agriculture and
the distributive service sectors were the key sectors in 2005/06. This means
both sectors have backward and forward linkages greater than one. The other
service sub-sectors, business and finance and the public sectors, exhibited
strong backward linkages along with the mining, utility and construction
sectors. The sectors’ high backward linkage indices show the interconnection
of the various sectors from which these sectors purchased greater amounts of
intermediate inputs from the domestic economy.
Figure 6.3: Key Sectors based on Hirschman-Rasmussen Linkages,

Ethiopia (2005/06)
1
2.56 Forward Oriented
3 Key
6
2.06
1.56
Forward Linkages
1.06
7 8
0.56 9
452 Backward Oriented
Weak Sectors 10
0.06
0.57 0.77 0.97 1.17 1.37
Backward Linkages
Sectors Economic Sectors Sectors Number Economic Sectors

Number1 Agriculture 6 Distributive Trade service
2 Mining 7 Transport & Communication
3 Manufacturing 8 Service
Business & Finance Service
Page | 207
The manufacturing sector is the only forward oriented sector with
forward linkages greater than one and backward linkages less than one. The
sector’s higher forward linkages indicate the interconnections of the
manufacturing sector as a direct and indirect output provider for all other
sectors. However, the transport and communication sector is the weakest
sector in the economy with lower indices of both backward and forward
linkage. This implies that the sector has weak backward and forward linkages,
poorly integrated into the economy, and heavily dependent on imports. The
transport and the communication sector in Ethiopia had a high export intensity
and high import dependency. Furthermore, the personal service sector is just
on the line between the weaker sector and backward oriented sector, as it has
backward linkage coefficient is near to one (0.995). The distributive trade
service sector, along with agriculture, is the key sector. They have strong
backward and forward linkage implying their importance in the economy and
stimulating more production of outputs in the economy. Finally, we can
observe that six of the economic sectors in Ethiopia during 2005/06 are
concentrated just on the right half of the backward oriented quadrant. This
indicates that most of the sectors in Ethiopia in 2005/06 as measured by the
Hirschman-Rasmussen method are backward oriented. That means the
economy is more of demand-oriented as most of these sectors provide most of
their output as an intermediate input to other sectors.
Page | 208
Figure 6.4: Key Sectors Based on Weighted Backward and Forward
1
3.00 Forward Oriented Key
3 6
2.50
2.00
Forward Linkages
1.50
1.00
7 8
0.50 Weak Sectors 9 Backward Oriented
10 42 5
0.00
0.53 0.73 0.93 1.13 1.33 1.53 1.73
Backward Linkages
Sector Sector
Economic Sectors Economic Sectors
Number Number
Figure 6.4 shows the key sector identification using weighted
backward and forward linkages. Here, the linkages for each sector are
computed taking into account their weight or share in total aggregate
production. The agriculture and the distributive service sectors are again the
key sectors in Ethiopia during 2005/06. The weighted backward and forward
linkage values for both agriculture and distributive trade service are greater
than one. Similarly, the manufacturing sector remained a forward oriented
sector. The difference here in the weighted inter-linkage as compared to the
Hirschman- Rasmussen approach is observed in the mining and utility sectors.
These sectors have been a backward oriented sector in Hirschman- Rasmussen
inter-linkage, but they become weak sectors when their share in total
production is accounted. The backward linkage coefficient for the other
sectors, which are construction, business and finance, public service and
personal services, has also decreased due to their smaller share in total
production as compared to the rest sectors. Therefore, in the weighted key
sector analysis, we observed that sectors that were concentrated in the
Page | 209
backward oriented quadrant have shifted to the weak sector. This shows the
weak inter-linkage for most of the services and utility and mining sector in
Ethiopia in 2005/06.
Finally, the key sectors using the pure backward and forward linkage
analysis is computed and presented in Figure 6.5. According to (Parra &
Wodon, 2010a), pure linkages can be used for a direct comparison of the
relative importance of the economic sectors measured in monetary values. The
sectors were defined as a key, backward oriented, forward oriented and weak
sectors based on their linkage strengths in generating pure linkage effects in
the country’s economy. The average values of the economic sectors for each
backward and forward linkage are used to categorize the sectors into the four
categories. Thus, the average effect of the pure backward linkage and the
forward linkages are 125 and 109 billion birr respectively. If a sector has
above average pure forward and backward linkage, it will be a key sector. The
backward oriented sectors have above average pure backward linkage and
below average pure forward linkage, whereas forward oriented sectors have
above average pure forward linkage and below average pure backward
linkage. The weak sector has below average backward and forward linkages.
Accordingly, the key sectors in Ethiopia in 2005/06 based on the pure

linkages are agriculture, distributive trade service, and manufacturing sector.
In the previous approaches, the manufacturing sector was not a key sector;
rather it was the forward oriented sector with a low level of backward linkage.
Nevertheless, when the pure linkage effect is computed, the manufacturing
sector has higher backward linkage. As explained earlier in the pure linkage
analysis, the manufacturing sector in Ethiopia mainly uses imported
manufactured intermediate inputs. Thus, when the inter-linkage of the sector
with itself is excluded in the pure linkage analysis, we get a higher backward
inter-linkage effect for the manufacturing sector.
Page | 210
Figure 6.5: Key Sector based on Pure Backward and Forward Linkages,
Ethiopia (2005/06)
350 Forward Oriented
Key
1
300
250 3
Pure Forward Linkage
200 6
150
5
100
Weak sector 9
50 8
4
7
0 Backward Oriented
0 2 1050 100 150 200 250 300 350
Pure Backward Linkage
Sector Number Economic Sectors Sector Number Economic Sectors

Apart from the three key sectors, the other important sector in the key
sector analysis is the backward oriented sector. As it is displayed in Figure 6.5,
the construction sector has above average pure backward linkage, which
makes it the only backward oriented sector. All the rest sectors are
concentrated in the weaker sector quadrant, having below average pure
backward and forward linkages. This implies that when the pure inter-sectoral
linkage is computed by discounting the inter-linkage of sectors with
themselves, most of the sectors will have weak inter-linkage with other
sectors. Specifically, except for the distributive service trade, all the service
sectors are considered to have small inter-linkage with other sectors.
Page | 211
Specifically, the personal service and transport and communication have very
weak pure forward linkages.
The Economic Landscape of Ethiopia, 2005/06
Figure 6.6 shows the economic landscape of Ethiopia in 2005/06,

which indicates the hierarchies of backward and forward linkages of the
economy. It is constructed using the I-O multiplier product matrix to depict
graphically the economic sectors imposing the backward and forward linkage
hierarchies (Sonnis, Guilhoto, Hewings, & Martins, 1995). The economic
sectors are ranked based on the size of their hierarchies of the backward and
forward linkages. The row represents the hierarchy of forward linkages while
the column provides the details of the backward linkages. The sectors in the
first column of the figure are those with higher backward linkages in the
economy whereas the sectors in the first row have greater forward linkages.
According to Parra &Wodon (2010), the cell of the I-O matrix denotes the
first-order change in the sum of all elements of the inverse matrix caused by
the change in the technical coefficients.
Therefore, the results show that the distributive trade service, business
and finance services, transport and communication service and the public
service sectors ranked from third to sixth respectively. In the economic
landscape, the agriculture sector leads as it has high backward and forward
linkages. Then, the manufacturing sector follows the agriculture sector due to
its greater forward linkages as compared to the service sectors. The remaining
sectors, which are the utility, construction, personal services, and mining
sectors have very weak backward and forward linkages. Therefore, the
economic landscape of Ethiopia is skewed as most of the economic sectors
have weak inter-linkage with other sectors.
Page | 212
Figure 6.6: Economic Landscape of Ethiopia (2005/06)
3.5
3
1
2.5
3
2 6
1.5 8
7
1
9
0.5 4
0 10
1
1
3 9
6 5
6 2
8
7 5
8 2
9 4
4
10 10
7
Row hierarchy of forward linkages 5 2 3
Column hierarchy of backward linkages
Sector Number Economic Sectors Sector Economic Sectors

Number
To Sum up, the service sector linkage analysis shows that even though
the sector is important in the production of manufacturing and other sectors, it
is in general not strongly interlinked with the other sectors. When the sub-
sectors are analyzed, particularly, the distributive trade service is an important
sector that has high backward and forward inter-linkage. This implies that the
distributive trade service is a key sector along with the agriculture sector.
However, other service sub-sectors have weak forward and backward inter-
linkages with other sectors.
Page | 213
CHAPTER-VII
MAIN FINDINGS AND POLICY IMPLICATIONS
Economic history shows that the features of structural change process

across countries can be generalized. As an economy moves from lower to
higher stages of development, output and employment shifts from the primary
agriculture sector to the secondary industrial and tertiary service sectors. Yet,
there exist variations between developed and developing countries as well as
among the developing countries in their process of structural change. The
developing countries are deviating from the traditional transformation process
as production and employment shifts from the agriculture sector directly to the
service sector, leapfrogging the industrialization process. This is becoming the
common structural feature of most of the developing countries in Asia, Latin
America and Africa. The economic structure of Ethiopia in the recent periods
is also characterized by the shift of resources from agriculture to the service
sector. In Ethiopia, the tertiary sector share in GDP exceeds the agriculture
sector share. The service sector accounts for the lion’s share in terms of the
structure of GDP (46.6 percent) in 2014-15 taking the lead from the
agriculture sector. This higher share has its own implications for employment,
output, and growth of the economy. In this context, the main thrust of the
study has been to analyze the structure and growth of the tertiary sector in
Ethiopia.
An elaborate review of studies indicated of the fact that most of the

studies done so far, on Ethiopian, are too aggregative or even if the
disaggregation has been achieved, a comprehensive study of the underlying
dynamism of the sector has not been done. Most of the studies have dealt with
individual sub-sectors or the overall economy-wide aggregates and have failed
to capture the underlying structural dynamics of the tertiary sector in terms of
space and time. A study fortified with disaggregated data going rigorously into
structure and growth of Ethiopian economy is a need of the time.
Page | 214
The main objective of the study has been to analyze the structure and
the growth of the Ethiopian service sector. The specific objectives have been
as follows: (a)to assess the growth trend of Ethiopian service sector and sub-
sectors; (b) to delineate the structural change in Ethiopia with special
reference to tertiarization; (c) to determine the factors that explains the growth
of the Ethiopian service sector; (d) to analyse the service sector inter-linkage
with agriculture and industry sectors; (e) to examine the viability and
sustainability of service sector growth in Ethiopia; and (f) to prescribe the
policy framework for better functioning and sustainability of the Ethiopian
economy
On the research methodology plane, the structure and growth of the

service sector in Ethiopia has been analyzed from different perspectives by
applying different methodologies. Different data sources and methodologies
have been applied to the respective objectives. For analysis of structure and
structural change, employment and output data have been used for growth
decomposition analysis. Inter-linkage of the service sector with other sectors
has been analyzed using input-output analysis by identifying the backward and
forward linkage, production technology coefficients and key sectors
employing SimSIP SAM software. Finally, the sustainability of the service led
growth is evaluated against five characteristic features using a scorecard
approach lent from Amirapu & Subramanian (2015).
This study investigates the structure and growth of the service sector in
Ethiopia. It attempts to identify the growth trend and composition of the
service sector, the service sector role in structural change, its determinants,
sectoral inter-linkage of the sector, and sustainability of the service led
growth. The study uses secondary data from the GGDC 10 sectors database,
World Development Indicators, and statistical information from Ethiopian
government official publications as well as the 2005/06 SAM of Ethiopia.
Main Findings
Based on the analysis, the study argues that the service sector is
contributing to the growth in output and employment and structural change in
Page | 215
Ethiopia with a potential to sustain the growth and stimulate the growth of the
manufacturing sector. In the light of above defined objectives and using the
methodology outlined, following are the main findings of the study:
a) Analysis of Growth of Service sector in Ethiopia shows that Ethiopia

possesses a unique sectoral structure as compared to other countries
that had the same level of per capita income at various periods, which
is the high share of agriculture and service sector and low share of
manufacturing in output. The high share of agriculture in employment
exceeding 70 percent, manufacturing employment lower than 10
percent, and service employment around 15 percent gave the country a
unique structural feature, which is far from the structural feature of
countries that achieved structural transformation and joined the
middle-income group.
b) The shift of employment observed in the last decade from agriculture is

towards the service sector as compared to the manufacturing sector,
which is contrary to the conventional structural change models. This
implies that Ethiopia is just only at the initial stage of structural
change. In terms of productivity, the labor productivity in
manufacturing relative to services is lower than the average of SSA
and the productivity of labor per unit of output for the manufacturing
sector is decreasing as the level of productivity in the service sector is
increasing since 1984. The output growth of the service sector can be
characterized by the higher and increasing share of the distributive
trade and public service sub-sectors and low share of personal services.
In addition, employment in service sector generally increased in
Ethiopia from just 2 percent to 17 percent in a half century. This
increase in employment is largely accounted by the distributive service
sub-sector, which is followed by the employment in government
services, despite its large decline associated with the gradual changes
in employability out of the public service.
Page | 216
c) Analysis of the contribution of each sector to output and employment
growth has been done since the 1960s. During 1962-1973 growth
periods, on an average 39 percent of the growth in output was
contributed by the service sector, which is lower than agriculture. In
1974-1985, even though there was very weak economic growth, the
service sector has contributed 64 percent to the average total output
growth. In 1993-2003, the larger contribution for the rise in the output
growth relative to the previous period is attributed to the service sector.
In the recent high growth periods of 2004-2011, the service sector
contribution has increased from 20 percent in 2004 to 45 percent in
2011 with the period average of 47 percent, which is also more than
the agriculture sector. On the employment side, the total employment
growth in Ethiopia is largely affected by the dominant employing
sector, agriculture. Apart from agriculture, the service sector
specifically the non-market service is contributing positively to the
growth in employment since 2004.
d) In addition, the decomposition analysis identified the importance of the

service sector in general for the employment and GDP per capita
growth. The sector has played the main role in productivity growth and
intersectoral shifts. The high contribution of the service sector to per
capita GDP growth originates from the distributive sector due to its
higher within sectors productivity change. Whereas the contribution of
service sector employment to the GDP per capita growth emanates
from the non-market service sectors. On the other hand, the
contribution of the service sector in inter-sectoral shifts also comes
from the higher inter-sectoral shifts occurred in the finance and
business sector.
e) The role of the service sector in structural change has been analyzed.
Results from the shift share decomposition show the presence of
structural bonus and structural burden in Ethiopia at various periods.
The structural bonus, the presence of positive shift effect for the
Page | 217
manufacturing sector is observed in all the periods except in 1981-
1992. In the early industrialization period (1961-1981) and post-reform
periods (1992-2004) and (2004-2011), the static shift effect for the
manufacturing sector was positive implying that workers moved out of
the low productive sectors of the agriculture, and joined the
manufacturing sectors and mainly the service sectors, as the sectors
have high productivity level than the agriculture sector. On the other
hand, structural burden, a negative between effects for the service
sector, was detected in 1961-1981 and 1981-1992. However, no
structural burden was observed in the post-reform periods (1992-2004)
and (2004-2011) due to a higher productivity level of the service sector
than the manufacturing sector. The high productivity level for most of
the market services in developing countries tend to reduce or eliminate
the structural change burden. However, the within productivity level of
the market service in Ethiopia for the periods 1992 – 2004 and 2004 –
2011 was negative.
f) Regarding the aggregate productivity growth in Ethiopia, it was

negative in the early periods of 1961-1981 and 1981-1992. But it has
improved in the post-reform periods (1992-2011). The structural
change effect had a considerable effect on the improvement in
aggregate productivity growth. Nevertheless, the within sectors
productivity growth has contributed to the growth in 2004-2011
basically as a result of the productivity growth observed in the
agriculture and non-market service sectors. The structural change
effect (dynamic effect) has been positive in these periods indicating a
shift of labor to sectors with above average productivity growth.
Whereas, the static effect is negative in all the periods implying the
shift of labor to sectors that have below average productivity levels.
Hence, for Ethiopia according to this finding, there is a static loss but
dynamic gain that is a different structural change from the rest of SSA
but more similar with the Asian feature of structural change.
Page | 218
g) Analysis of determinants of the service sector growth has been carried
out using the ARDL model. Based on the estimation result, the service
sector growth in GVA is co-integrated in the long run with GDP per
capita, female economic participation and reform period. Service
sector growth in employment is also co-integrated in the long run with
GDP per capita, female economic participation, reform period and
productivity difference. For a percentage increment in GDP per capita,
service GVA will increase by 0.85 percent confirming the ‘hierarchy
of needs’ hypothesis for Ethiopia. As it assumed, when income
increases, there is the consumption from goods to services. In addition,
more participation of female in economic activity in Ethiopia will
increase the service sector output by increasing household
consumption of services. For a one-percentage change in female
participation, service output will increase by 0.87 percent.
Furthermore, 39.9 percent of the growth in the service sector is due to
the changes in the post-reform period as compared to the pre-reform
period. Hence, economic reform and participation of female in the
labor force cause a change in the demand structure in the economy
confirming the ‘exogenous demand shock’ hypothesis.
h) Analysis of determinants of the service sector growth, further shows

that the intermediate input demand from the service sector is not
significant implying the weak inter-linkage between the producer
services and manufacturing sector or the low-level of development in
the manufacturing activity. In addition, the productivity differential is
insignificant and negative. This implies that the service GVA change
in Ethiopia is not only significant but also it is in contrast to the cost
disease hypothesis as the service sector in Ethiopia has higher
productivity level than the manufacturing sector resulting in a negative
productivity gap.
i) The coefficient for the long run co-integration equation of service

GVA indicates a 48.7 percent convergence to equilibrium from the
Page | 219
deviation and it will take about 2 years for the co-integration equation
to come to the equilibrium. For service employment, the result shows
that the service sector growth in employment is driven by the rise in
income and growth in female employment. Employment in the service
sector increases by 1.2 percent for a 1 percent growth in income. There
could be also a 1.07 percent growth in service employment because of
a 1 percent increase in female labor in the economy. In addition, a 0.42
percent change in the service employment can be observed for a 1
percent change in the productivity gap of the manufacturing and
services. Service employment in Ethiopia has also increased by 40.7
percent as compared to the pre-reform period. In the service
employment equation, the deviation of the co-integration equation
from the long run equilibrium due to certain shocks is adjusted by 53
percent over the next year and it will take about 1.9 years for the
complete adjustment of the long run equilibrium.
j) Using the Social Accounting Matrix (SAM), an elaborate service sector

inter-linkage analysis of the service sector with other sectors has been
carried out. For this, Social Accounting Matrix, 2006-06 has been
used. From the descriptive analysis of the SAM, during 2005/06,
Ethiopia’s largest factor cost is spent on the agriculture sector followed
by the service sector. But, the distributive service sub-sector
constitutes the largest proportion of all the sub-sectors in its factor cost.
In addition, agriculture is also the least capital-intensive sector and
services are the most capital-intensive sectors, whereas, from the
service sub-sectors, the distributive service and business and finance
services are the most capital-intensive sectors.
k) Regarding the composition of labor, the manufacturing sector

employed a larger proportion of skilled workers than the service sector
at aggregate. However, at sub-sector level, the distributive sector
employed the largest share of skilled labor followed by transport and
communication service sub-sector. On the other hand, the business and
Page | 220
finance service sub-sector employed more of administrative and
professional workers than the skilled workers. The largest proportions
of administrative and professional workers are also employed in the
public service sector than other sectors.
l) Furthermore, the descriptive result from SAM also shows that the
service sector consumption expenditure share accounts 22.5 percent of
total household consumption expenditure. From the service sub-
sectors, households spending on business service takes the largest
share followed by the distributive service sector while the least
consumed service by households is the public service sector. Besides,
the non-poor households spend more of their income on consumption
of service as compared to poor households and relative to the poor
households the non-poor households also consume more of the
distributive service, transport and communication services, and
personal services.
m) The export and import structure of Ethiopia in 2005/06 shows that

Ethiopia’s export is well below the total imports of the country leaving
the country in a huge trade deficit. One important feature in the
Ethiopian export structure is the share of the service sector, which is
almost equal to the agriculture sector accounting about 40 percent of
the total export each. This implies that service sector export is as
crucial as the agriculture sector for export earnings. The source of the
increasing role of the service sector in export is the transport and
communication service sub-sector, which is dominated by the
transportation service of the government-owned Ethiopian Airlines and
Shipping Line. Besides, Ethiopia’s major imports are industrial goods,
which are used as intermediate input for the manufacturing sector, and
the transport and communication service. The import penetration ratio
of Ethiopia shows that the Ethiopian manufacturing sector faces the
high import competition, as a quarter of total demand is imported. On
the other hand, the export intensity shows that the transport and
Page | 221
communication sector is the most export-intensive sector as half of the
transport service output is sold abroad. Therefore, in 2005/06 the
service sector in general, and the transport sector, in particular, is the
most dynamic service sector with high intensity of import and export
trades.
n) The sectoral inter-linkage analysis results show that the agriculture

sector, mining and personal service have total input spending that
largely relies on factor inputs than commodity inputs, while, the
manufacturing, construction and the distributive trade service sector
require more of intermediate inputs from the commodity sectors than
the factor inputs. The manufacturing sector input composition also
indicates that over half of the input is required from commodity inputs,
which is largely derived from the manufacturing sector itself. The
manufacturing sector is also more linked to the finance and business
service sub-sector more than the other services. The sector requires
financing services from financial institutions and various types of
business services like auditing, consulting, renting and the like
services. The input requirement for the construction sector also shows
that the sector uses more of business and finance service and transport
and communication services. The distributive trade service sector input
requirement is also satisfied largely by transport and communication
services. At the aggregate, the service sector is the most important
source of input for Ethiopian economy by contributing 10.7 percent of
the intermediate input requirement next to manufacturing which
constitutes 16.8 percent. From the specific service sub-sectors, the
transport and communication sector contributes the largest share,
which is 5.8 percent, more than other sectors due to its wider linkage
with all the sectors. The business and service sector follows the
transportation and communication services in its role as an
intermediate input in the production process by contributing 3.4
percent of the intermediate input.
Page | 222
o) The sectoral linkages analysis using Hirschman-Rasmussen, weighted
and pure inter-linkages show that agriculture has the highest backward
and forward linkages. All the service sub-sectors, except the
distributive trade service sector, have relatively stronger backward
linkage than the forward linkages indicating the inter-linkage of the
service sectors from the demand side. In the backward linkages, the
agriculture sector stood first followed by the distributive service trade
and the public service sector. The manufacturing sector is the least
backward oriented sector followed by the transportation and
communication sector and personal service sectors indicating the
sector’s weak interconnection in stimulating production in the other
sectors. On the other hand, the agriculture sector is the most forward
oriented sector followed by the manufacturing, distributive trade
service and business and finance service sectors. A sector with weak
forward linkage implies their weak interconnection with other sectors
in supplying their output as a final demand for other sectors. All the
service sectors, except the personal services, have strong inter-linkage
with other sectors as compared to the mining, utility and construction
sectors. In addition, the result shows a stronger inter-linkage of the
distributive trade service more than the dynamic manufacturing sector,
which implies the importance of the sector in the Ethiopian economy.
The distributive trade service sector is second, next to agriculture in its
forward orientation. This indicates the importance of the distributive
service sector in Ethiopia, as it has higher interconnection with other
sectors inducing production in other sectors.
p) Furthermore, the key sector identification using the Hirschman-

Rasmussen and weighted linkage analysis shows that the agriculture
and the distributive service sectors were the key sectors in Ethiopia
while the pure linkage analysis indicates the introduction of the
manufacturing sector along with the agriculture and the distributive
service sectors as a key sector in 2005/06. This means both sectors
have backward and forward linkages greater than one. The other
Page | 223
service sub-sectors, business and finance and the public sectors,
exhibited strong backward linkages along with the mining, utility and
construction sectors. The sectors’ high backward linkage indices show
the interconnection of the various sectors from which these sectors
purchased greater amounts of intermediate inputs from the domestic
economy. The result further shows that apart from the key sectors, the
other important issue in the key sector analysis is the backward
oriented sector. The construction sector has above average pure
backward linkage, which makes it the only backward oriented sector.
All the rest sectors are concentrated in the weaker sector quadrant,
having below average pure backward and forward linkages.
q) This implies that when the pure inter-sectoral linkage is computed by

discounting the inter-linkage of sectors with themselves, most of the
sectors will have weak inter-linkage with other sectors. Specifically,
except for the distributive service trade, all the service sectors are
considered to have small inter-linkage with other sectors. Specifically,
the personal service and transport and communication have very weak
pure forward linkages. Finally, the economic landscape of Ethiopia in
2005/06 show that the distributive trade service, business and finance
services, transport and communication service and the public service
sectors ranked from third to sixth respectively. In the economic
landscape, the agriculture sector leads as it has high backward and
forward linkages. Then, the manufacturing sector follows the
agriculture sector due to its greater forward linkages as compared to
the service sectors. The remaining sectors, which are the utility,
construction, personal services, and mining sectors have very weak
backward and forward linkages. Therefore, the economic landscape of
Ethiopia is skewed as most of the economic sectors have weak inter-
linkage with other sectors.
r) The assessment on the sustainability of the service-led growth shows

that the service sector in Ethiopia possesses the major characteristics
Page | 224
features that enable it to sustain economic growth. The sector has high
productivity level and high productivity growth as compared to the
manufacturing and the agriculture sector. There is also growth
convergence for Ethiopia in services than the manufacturing sector.
Additionally, Ethiopia has high export potential in the transportation
services. Nevertheless, the resources endowment for Ethiopia and the
comparative advantage of the country is not aligned with service
sector.
The overall conclusion that emerges from the study is that in Ethiopia
the service sector has witnessed high growth and contributed to the growth in
output and employment during the recent growth periods. The sector is also
the source of growth-enhancing structural change. In addition, it has a strong
inter-sectoral linkage through the distributive trade service. Furthermore, the
service sector has also the potential to sustain the high economic growth and
stimulate growth in other sectors.
Policy Implications
Based on the results on the service sector growth and structure of

Ethiopia, the following substantial policy implications are drawn.
 Adequate policy and strategic focus needs to be given for the service
sector in the Ethiopian development plans as the contribution of the
service sector to the total output growth or GDP growth and to
employment growth are increasing and the sector is becoming an
important sector to drive growth. In addition, the development policies
in Ethiopia should not undermine the role played by the service sector.
The current growth of the service sector is seen as untimely and
unwanted growth from Ethiopian policies perspective. Nevertheless,
the growth in GDP and employment in Ethiopia is the result of the
growth in the service sector, especially from the employment growth
of the non-market service sector and productivity growth of the market
service sector. Without the massive growth in the service sectors, the
much-required high GDP growth would not be achieved.
Page | 225
 To avoid any structural burden to the economy, productivity level in
the market services needs to be improved through the application of
new technologies and opening up the most of the trade restrictions in
the market services. Even though Ethiopia had high growth and
expansion of service sector, the structural burden effect of the service
sector was eliminated due to productivity growth in the non-market
services as opposed to the case in other developing countries. This
could be due to the growth in the size of the Ethiopian public service
output in GVA and decline of employment share in the public sector.
By their nature, these sectors are expected to show high productivity
growth and drive the growth as they have high potential. This implies
that the country has much potential for high aggregate productivity
growth if works policies are laid to improve the productivity of these
sectors.
 Further niche market opportunities should be identified to the exports

of the transport and travel services of Ethiopia. The transport and
communication service sub-sector is the most export-intensive sector
as half of the transport service output is sold abroad. In addition, it is
the most dynamic service sector with high intensity of import and
export trades.
 Policies need to identify alternative intermediate inputs for the

manufacturing sector apart from imports and the service sector
exports need to be strengthened at least to finance the import bills of
the intermediate manufacturer inputs. The manufacturing sector
requires a larger proportion of its input from commodity inputs,
specifically from its self and from the agriculture. However, these
inputs for the manufacturing sector are more of imported than
domestically produced inputs. This hampers the growth of the
manufacturing sector, which results in weak inter-linkage with the
service sector and constrains the growth of service sector as well.
Therefore, the growths of the manufacturing sector as a final demand
Page | 226
do not induce growth in the other sectors. Whereas, the high weighted
background linkage values for the agriculture, distributive trade
service, and construction service sector indicates that these sectors
have a higher dependency on the domestic intermediate inputs.
 Since the resource endowment for Ethiopia and the comparative

advantage of the country is not aligned with service sector, further
investment in human capital building through education and skill
development needs to be conducted to absorb labor in the service
sector from the agriculture sector.
The overall policy brief that emerges from the study is that
development plans and sector specific policies in Ethiopia needs to give due
considerations on strategies to make use of the service sector growth.
Specifically, strategies on economic reforms in the service sector, niche export
markets for service export and specialized services for the manufacturing
sector should be identified and implemented for a balanced and sustainable
economic growth of Ethiopia.
Main Limitations of the Study
The limitation of this study is associated with the availability of recent

SAM or IO data to Ethiopia to identify the key sectors analysis and other inter-
linkage analysis in the recent years. Similarly, the other limitation of the study
is that the time included in most of the structural change and time series
analyses are before 2011. Recent data was not available from the same
sources at the time of this research work.
Future Research Issues
Further research needs to be conducted on Ethiopian service sector

inter-linkage with the manufacturing sector and on the reasons for the decline
of the manufacturing sector productivity as the service sector productivity is
rising. In addition, a macroeconomic simulation can be done using CGE
models to identify the effect of service sub-sectors that have a high multiplier
effect on the manufacturing sector. In addition, by using recent SAM data for
Page | 227
Ethiopia a comparison of the changes in the service inter-linkage with the
2005/06 SAM can be conducted. Furthermore, the negative productivity gap
between service sector and manufacturing sector requires further investigation
by considering the relative price of services and manufacturing. Further
analysis of the structural change process and role of the service sector in
Ethiopia needs to be studied in the most recent periods after, especially after
2011.
Page | 228
Bibliography
Aggarwal, A., & Kumar, N. (2012, November). Structural transformation,
industrialization and poverty reduction: The case of India. United
Nations Economic and Social Commission for Asia and the Pacific, pp.
1-75.
Ajakaiye, O., & Page, J. (2011). Industrialization and Economic

Transformation in Africa: Introduction and Overview. Journal of
African Economies, 21(AERC Supplement 2), ii3–ii18.
doi:10.1093/jae/ejr049
Alvarez-Cuadrado, F., & Poschke, M. (2011). Structural Change Out of

Agriculture: Labor Push versus Labor Pull. American Economic
Journal: Macroeconomics, 3(3), 127–158.
Amirapu, A., & Subramanian, A. (2015, June). Manufacturing or Services?

An Indian Illustration of a Development Dilemma. Center for Global
Development Working Paper 409.
Andreosso-O'Callaghan, B., & Yue, G. (2004). Intersectoral Linkages and Key

Sectors in China, 1987-1997. Asian Economic Journal, 18(2), 165-183.
Arnold, J. M., Javorcik, B. S., & Mattoo, A. (2011). Does services

liberalization benefit manufacturing firms? Evidence from the Czech
Republic. Journal of International Economics, 85(1), 136-146.
doi:10.1016/j.jinteco.2011.05.002
Aryeetey, E., & Moyo, N. (2012). Industrialization for Structural

Transformation in Africa: Appropriate Roles for the State. Journal of
African Economies, 21(AERC Supplement 2), ii55–ii85.
doi:10.1093/jae/ejr043
Azad, A. K. (1999). Interindustry Linkages of Services in the Bangladesh

Economy (With a Case Study of the Ready-made Garments Industry)
and Potential Service Trade. WTO 2000 South Asia Workshop,
organized by the NCAER and the World Bank. New Delhi.
Badibanga, T., Diao, X., Roe, T., & Somwaru, A. (2009, April). Dynamics of
Structural Transformation: An Empirical Characterization in the Case
of China, Malaysia, and Ghana. International Food Policy Research
Institute Discussion Paper 00856, pp. 1-40.
Bah, E.-h. (2010). Structural Transformation Paths Across Countries. The

University of Auckland Working Paper.
Page | 229
Banga, R. (2005a, March). Role of Services in the Growth Process. Indian
Council for Research on International Economic Relations Working
Paper No. 159, pp. 1-36.
Banga, R. (2005b, October). Critical Issues in India's Service-Led Growth.

Indian Council for Research on International Economic Relations
Working paper No. 171, pp. 1-42.
Banga, R. (2005c). FDI in Services: Implications for Developing Countries.

Asia-Pacific Trade and Investment Review, 1(2), 55-72.
Banga, R., & Goldar, B. (2004). Contribution of services to output growth and
productivity in Indian manufacturing: Pre and post-reforms. Indian
Council for Research on International Economic Relations Working
paper (139), pp. 1-33.
Barrett, C. B., Christiaensen, L., Sheahan, M., & Shimeles, A. (2017,

January). On the Structural Transformation of Rural Africa. World
Bank Group Policy Research Working Paper 7938.
Baumol, W. J. (1967). Macroeconomics of Unbalanced Growth: The Anatomy

of Urban Crisis. American Economic Review, 415-26.
Baumol, W. J., Blackman, S. A., & Wolff, E. N. (1985, September).

Unbalanced Growth Revisited: Asymptotic Stagnancy and New
Evidence. The American Economic Review, 85(4), 806-817.
Bhagwati, J. N. (1984a, June). Splintering and Disembodiment of Services and

Developing Nations. World-Economy, 7(2), 133-43.
doi:10.1111/j.1467-9701.1984.tb00265.x
Bhagwati, J. N. (1984b, June). Why are Services Cheaper in the Poor

Countries? The Economic Journal, 94(374), 279-286.
Bhalla, G. S., Chadha, G. K., Kashyap, S. P., & Sharma, R. K. (1990, August).
Agricultural growth and structural changes in the Punjab economy: an
input-output analysis. International Food Policy Research Institute
Research Report 82.
Bhattacharya, T., & Rajeev, M. (2014). Identifying the High Linked Sectors
for India: An Application of Import-Adjusted Domestic Input-Output
Matrix. The Institute for Social and Economic Change Working Paper
329.
Page | 230
Bosworth, B. P., & Triplett, J. E. (2003). Services Productivity in the United
States: Griliches Services Volume Revisited. In E. R. Berndt, & C. R.
Hulten, Hard-to-Measure Goods and Services: Essays in Honor of Zvi
Griliches (pp. 413-447). University of Chicago Press.
Bosworth, B., & Collins, S. M. (2008). Accounting for Growth: Comparing

China and India. Journal of Economic Perspectives, 22(1), 45-66.
Bosworth, B., & Maertens, A. (2010). Economic Growth and Employment

Generation: The Role of the Service Sector. In E. Ghani, The Service
Revolution in South Asia (pp. 103-146). New Delhi: Oxford University
Press.
Breisinger, C., & Diao, X. (2008, September ). Economic Transformation in

Theory and Practice: What are the Messages for Africa? IFPRI
Discussion Paper 00797.
Breisinger, C., Diao, X., & Thurlow, J. (2015). Modeling growth options and
structural change to reach middle-income country status: The case of
Ghana. Economic Modelling, 26(2), 514-525.
doi:10.1016/j.econmod.2008.10.007
Breisinger, C., Thomas, M., & Thurlow, J. (2009). Social accounting matrices
and multiplier analysis: An introduction with exercises. Food Security
in Practice technical guide 5, International Food Policy Research
Institute. doi:10.2499/9780896297838fsp5
Cali, M., Ellis, K., & Velde, D. W. (2008, December). The contribution of
services to development and the role of trade liberalization and
regulation. Overseas Development Institute Working Paper 298.
Castaldi, C. (2009). The relative weight of manufacturing and services in

Europe: An innovation perspective. Technological Forecasting and
Social Change, 76(6), 709-722. doi:10.1016/j.techfore.2008.08.001
Cattaneo, N. (2011). Services Trade Liberalisation and the Role of the

Services Sector in South African Development. South African Institute
of International Affairs Occasional Paper 94, pp. 1-25.
Chen, S., Jefferson, G. H., & Zhang, J. (2011, March ). Structural change,
productivity growth and industrial transformation in China. China
Economic Review, 22(1), 133–150. doi:10.1016/j.chieco.2010.10.003
Page | 231
Chen, S., Jefferson, G. H., & Zhang, J. (2011). Structural change, productivity
growth and industrial transformation in China. China Economic
Review, 22(1), 133-150.
Chenery, H. B. (1979). Structural change and development Policy.

Washington, D.C.: Oxford University Press.
Chenery, H. B. (1986). Growth and Transformation. In H. B. Chenery, S.

Robinson, & M. Syrquin, Industrialization and Growth a Comparative
Study (pp. 13-36). Oxford University Press.
Chenery, H. B., & Syrquin, M. (1986). Typical Patterns of Transformation. In

H. B. Chenery, S. Robinson, & M. Syrquin, Industrialization and
Growth a comparative study (pp. 37-83). Oxford University Press.
Chenery, H. B., & Syrquin, M. (1989). Patterns of Development, 1950 to

1983. World Bank Discussion Papers.
Chenery, H. B., & Taylor, L. (1968). Development Patterns: among countries

and over time. Harvard University, Center for International Affairs.
Cambridge, Massachusetts: Harvard University.
Chenery, H. B., & Watanabe, T. (1958, October). International Comparisons

of the Structure of Production. Econometrica, 26(4), 487-521.
Retrieved from http://www.jstor.org/stable/1907514
Christiaensen, L., & Todo, Y. (2013, May). Poverty Reduction during the
Rural-Urban Transformation: The Role of the Missing Middle. The
World Bank Policy Research Working Paper 6445.
Clark, C. (1940). The conditions of economic Progress. London: McMilan.
Cortuk, O., & Singh, N. (2013). Analyzing the structural change and growth
relationship in India: State-level evidence. UC Santa Cruz Economics
Department Working Papers No 712, pp. 1-21.
CSA . (1999). The 1999 National Labour Force Survey Report. Addis Ababa:
Central Statistical Agency.
CSA . (2014). The 2013 National Labour Force Survey Report. Addis Ababa:
Central Statistical Agency.
CSA. (2005). The 2005 National Labour Force Survey Report. Addis Ababa,
Ethiopia.: Central Statistical Agency.
Page | 232
Cuello, F. A., Mansouri, F., & Hewings, G. J. (1992). The Identification of
Structure at the Sectoral Level: a Reformulation of the Hirschman–
Rasmussen Key Sector Indices. Economic Systems Research, 4(4),
285-296.
D’Agostino, A., Serafini, R., & Ward-Warmedinger, M. (2006). Sectoral

explanations of employment in Europe: the role of services. IZA
Discussion Paper (625), pp. 1-44.
Dabla-norris, E., Thomas, A., Garcia-verdu, R., & Chen, Y. (2013, July).
Benchmarking Structural Transformation across the World.
International Monetary Fund Working Paper WP/13/176, pp. 1-44.
Das, K. D., Erumban, A. A., Aggarwal, S., & Sengupta, S. (2013). Revisiting
the Service-led Growth in India: Understanding India’s service sector
productivity growth. IARIW-UNSW Conference on Productivity:
Measurement, Drivers, and Trends, (pp. 1-34). Sydney.
Dasgupta, S., & Singh, A. (2005). Will Services be the New Engine of Indian
Economic Growth? Development and Change, 36(6), 1035-1057.
Dasgupta, S., & Singh, A. (2006, May). Manufacturing, Services and

Premature Deindustrialization in Developing Countries: A Kaldorian
Analysis. UNU-WIDER Research Paper No. 2006/49.
de Brauw, A., Mueller, V., & Lee, H. L. (2014). The Role of Rural–Urban
Migration in the Structural Transformation of Sub-Saharan Africa.
World Development, 63, 33-42. doi:10.1016/j.worlddev.2013.10.013
de Vries, G. J., Erumban, A. A., Timmer, M. P., Voskoboynikov, I., & Wu, H.
X. (2012). Deconstructing the BRICs: Structural transformation and
aggregate productivity growth. Journal of Comparative Economics,
40(2), 211-227.
de Vries, G., de Vries, K., Gouma, R., Pahl, S., & Timmer, M. ( 2014). GGDC
10-Sector Database: Contents, Sources, and Methods. Groningen
Growth and Development Centre.
de Vries, G., Timmer, M., & de Vries, K. (2013, October). Structural

transformation in Africa: Static gains, dynamic losses. Groningen
Growth and Development Centre Research Memorandum 136, pp. 1-
52.
Page | 233
Debnath, A., & Roy, N. (2012, February 11). Structural Change and
Intersectoral Linkages: The Case of North-east India. Economic &
Political Weekly, XLVII (6), 72-76.
Diao, X., Hazell, P., Resnick, D., & Thurlow, J. (2007). The Role of
Agriculture in Development: Implications for Sub-Saharan Africa.
IFPRI Research Report 153.
Dietrich, A. (2012). Does growth cause structural change, or is it the other

way round? A dynamic panel data analysis for seven OECD countries.
Empirical Economics, 43, 915-944.
Dorosh, P., & Thurlow, J. (2014). Can Cities or Towns Drive African
Development? Economywide Analysis for Ethiopia and Uganda.
World Development, 63, 113–123.
doi:http://dx.doi.org/10.1016/j.worlddev.2013.10.014
Dorosh, P., Schmidt, E., & Shiferaw, A. (2012, Fall). Economic Growth
without Structural Transformation: The Case of Ethiopia. Journal of
African Development, 14(2), 7-40.
Drejer, I. (2002). Input-Output Based Measures of Interindustry Linkages

Revisited - A Survey and Discussion. 14th International Conference
on Input-Output Techniques, (pp. 1-35).
Duarte, M., & Restuccia, D. (2009, February). The Role of the Structural
Transformation in Aggregate Productivity. University of Toronto,
Department of Economics Working paper 348, pp. 1-57.
Duarte, M., & Restuccia, D. (2010, February). Role of the Structural

Transformation in Aggregate Productivity. The Quarterly Journal of
Economics, 125(1), 129-173. Retrieved from
http://www.jstor.org/stable/40506279
Easterly, W., & Levine, R. (2001). It's Not Factor Accumulation: Stylized
Facts and Growth Models. The World Bank Economic Review, 15(2),
177-219. Retrieved from http://www.jstor.org/stable/3990260
Eckstein, A. J., & Heien, D. M. (1985). The U.S. Experience Causes and
Consequences of Service Sector Growth. Growth and Change, 12-17.
EDRI. (2009, December). Ethiopia: Input-Output Table & Social Accounting

Matrix. Ethiopian Development Research Institute.
Page | 234
EEA. (2015). Report on the Ethiopian Economy. Addis Ababa, Ethiopia:
Ethiopian Economic Association.
Eichengreen, B., & Gupta, P. (2011, February). The Service Sector as India’s
Road to Economic Growth? National Bureau of Economic Research
Working Paper 16757, pp. 1-51.
Eichengreen, B., & Gupta, P. (2013). The two waves of service-sector growth.
Oxford Economic papers, 65(1), 96-123. doi:doi:10.1093/oep/gpr059
Enache, M., Ghani, E., & O’Connell, S. (2016, July). Structural

Transformation in Africa: A Historical View. World Bank Group
Policy Research Working Paper No. 7743.
Engida, E., Tamru, S., Tsehaye, E., Debowicz, D., Dorosh, P., & Robinson, S.
(2011, November ). Ethiopia’s Growth and Transformation Plan: A
CGE Analysis of Alternative Financing Options. ESSP Working Paper
30.
Eshete, Z. S. (2013, November). Economic Transformation and Sectoral

Growth Options in Ethiopia: Considerations for Achieving Middle
Income Country Status. Ph.D. Dissertation. School of Economics,
University of Nairobi, Kenya.
Eshete, Z. S. (2014c). Public spending composition and efficiency: It’s

implications for growth, structural change and household welfare in
Ethiopia (using recursive dynamic CGE model). International
Academic Journals, 1(2), 34-56. Retrieved from
http://www.iajournals.org/articles/iajournals_v1_i2_34_56.pdf
Eshete, Z. S., & Kimuyu, P. K. (2014a). In Search of Structural Change and

Perpetual Growth in Ethiopia: Examining Alternative Sectoral Growth
Drivers for Reaching Middle Income Country Status. European
Journal of Business Management, 2(1), 336-341.
Eshete, Z. S., & Kimuyu, P. K. (2014b). Modeling the dynamics of sectoral

TFP growth in Ethiopia: Explaining persistent economic debacles.
American Journal of Business, Economics and Management, 2(5),
144-162.
Esteban-Pretel, J., & Sawada, Y. (2014). On the role of policy interventions in

structural change and economic development: The case of postwar
Japan. Journal of Economic Dynamics and Control, 40, 67-83.
doi:10.1016/j.jedc.2013.12.009
Page | 235
Estrada, G., Acharya, Y., Batten, A., Brimble, P., Doung, P., Gulamov, M. I., .
. . Rahman, S. (2013, April). The Service Sector in Lower-Income
Asian Economies. Asian Development Bank Economics Working
Paper Series 347, pp. 1-25.
Fagerberg, J. (2000). Technological progress, structural change, and

productivity growth: a comparative study. Structural Change and
Economic Dynamics, 11, 393-411.
Falk, M., & Jarocinska, E. (2010, November). Linkages between Services and
Manufacturing in EU countries. SERVICEGAP Review Paper 1, pp. 1-
29.
Falvey, R. E., & Gemmell, N. (1996, September). Are Services income-

elastic? Some new evidence. Review of Income and Wealth, 42(3),
257-269.
Fan, S., Zhang, X., & Robinson, S. (2003). Structural Change and Economic
Growth in China. Review of Development Economics, 7(3), 360-377.
Ferede, T., & Kebede, S. (2015). Economic growth and employment patterns,
dominant sector, and firm profiles in Ethiopia: Opportunities,
challenges, and Prospects. R4D Working Paper 2015/2.
Fernandes, A. M. (2009). Structure and performance of the service sector in

transition. Economics of Transition, 17(3), 467–501.
Fisher, A. G. (1935). The clash of progress and security. London: McMillan.
Fisher, A. G. (1939, June). Primary, Secondary and Tertiary Production.

Economic Records, 15(1), 24–38. doi:10.1111/j.1475-
4932.1939.tb01015.x
Fofana, I., Parra, J. C., & Wodon, Q. (2009). Exports and labor income by
gender: a social accounting matrix analysis for Senegal. Munich
Personal RePEc Archive Paper No. 28473.
Fuchs, V. R. (1965). Reasons for the relative growth of service employment.

In V. R. Fuchs, The growing importance of the service industries (Vol.
I, p. 8). Retrieved from http://www.nber.org/chapters/c1693
Fuchs, V. R. (1968). The Service Economy. New York and London: National
Bureau of Economic Research. Retrieved from
http://www.nber.org/books/fuch68-1
Page | 236
Fuchs, V. R. (1980). Economic Growth and the Rise of Service Employment.
National Bureau of Economic Research Working Paper 486, pp. 1-32.
Gani, A., & Clemes, M. D. (2002). Services and Economic Growth in ASEAN
Economies. ASEAN Economic Bulletin, 19(2), 155-169. Retrieved
from http://www.jstor.org/stable/25773718 .
Gemmell, N., Lloyd, T., & Mathew, M. (1998). Dynamic sectoral linkages and
structural change in a developing economy. Centre for Research in
Economic Development and International Trade, Department of
Economics, University of Nottingham, pp. 1-25.
Ghani, E. (2010). Is Service-led Growth a Miracle for South Asia? In E.

Ghani, The Service Revolution in South Asia (pp. 35-102). New Delhi:
Oxford University Press.
Ghani, E., & Kharas, H. (2010). The Service Revolution in South Asia: An
Overview. In E. Ghani, The Service Revolution in South Asia (pp. 1-
34). New Delhi: Oxford University Press.
Ghani, E., & O’Connell, S. D. (2014, July). Can Service Be a Growth

Escalator in Low-Income Countries? The World Bank Policy Research
Working Paper 6971.
Glasmeier, A., & Howland, M. (1994). Service-led Rural Development:

Definitions, Theories, and Empirical Evidence. International Regional
Science Review, 16(1 & 2), 197-229.
Gollin, D., Jedwab, R., & Vollrath, D. (2016, march). Urbanization with and
without industrialization. Journal of Economic Growth, 21(1), 35-70.
doi:10.1007/s10887-015-9121-4
Gordon, J., & Gupta, P. (2004). Understanding India's services revolution.

IMF Working Paper WP/04/171, pp. 1-36.
Griliches, Z. (1992). Introduction to Output Measurement in the Service

Sectors. In Z. Griliches, Output Measurement in the Service Sectors
(pp. 1-22). University of Chicago Press. Retrieved from
http://www.nber.org/books/gril92-1
Gujarati, D. (2004). Basic Econometrics (4th ed.). McGraw-Hill Education.
Hansda, S. K. (2001). Sustainability of Services-Led Growth: An Input-Output

Analysis of the Indian Economy. Reserve Bank of India Occasional
Papers, 22(1,2 & 3), pp. 73-118.
Page | 237
Hartwig, J. (2010). Structural change, aggregate demand and employment
dynamics in the OECD, 1970-2010. Zurich: KOF Swiss Economic
Institute ETH.
Havlik, P. (2013). Patterns of Structural Change in the New Member States.

GRINCOH Working Paper Series, Paper No. 1.04, pp. 1-25.
Herrendorf, B., Rogerson, R., & Valentinyi, A. (2014). Growth and Structural
Transformation. In Handbook of Economic Growth (Vol. 2B, pp. 855-
941). North Holland, Amsterdam and NewYork: Elsevier B.V.
Hill, T. (1977). On Goods and Services. Review of Income Wealth, 23(4), 315-
338.
Hirschman, A. O. (1958). The Strategy of Economic Development. New

Haven: Yale University Press.
Hoekman, B., & Mattoo, A. (2008). Services Trade and Growth. The World
Bank Development Research Group Policy Research Working Paper
4461.
ISIC. (2002). International Standard Industrial Classification of All Economic

Activities (ISIC) Revision 3.1. New York: United Nations Publication.
Jain, S., & Ninan, T. N. (2010). Servicing India’s GDP growth. In A. Shankar,
& M. Rakesh, India’s economy performance and challenges, Essays in
honor of Montek Singh Ahluwalia (pp. 328–365). New Delhi: Oxford
University Press.
Jedwab, R. (2013). Urbanization without Structural Transformation: Evidence

from Consumption Cities in Africa. Unpublished manuscript,
Department of Economics, George Washington University.
Jorgenson, D. W., & Timmer, M. P. (2011). Structural Change in Advanced

Nations: A New Set of Stylised Facts. Scandinavian Journal of
Economics, 113(1), 1-29. doi:10.1111/j.1467-9442.2010.01637.x
Joshi, S. (2008). Growth and Structure of Tertiary Sector in Developing

Economies. New Delhi: Academic Foundation.
Kaldor, N. (1966). Causes of Slow Rate of Growth in the United Kingdom.

Cambridge: Cambridge University Press.
Katouzian, M. A. (1970, November). The Development of the Service Sector:

A New Approach. Oxford Economic Papers, New Series, 22(3), 362-
382. Retrieved 12 02, 2015, from http://www.jstor.org/stable/2662538
Page | 238
Khanal, B. R. (2011). An Economic Analysis of the Lao PDR Tourism
Industry. Ph.D. Dissertation, Lincoln University.
Khanal, B. R., Gan, C., & Becken, S. (2014). Tourism inter-industry linkages
in the Lao PDR economy: An input-output analysis. Tourism
Economics, 20(1), 171-194.
Kim, H.-J. (2006). The shift to the service economy: causes and effects.
Institute for Monetary and Economic Research, the Bank of Korea.
Kolavalli, S., Robinson, E., Diao, X., Alpuerto, V., Folledo, R., Slavova, M., .
. . Asante, F. (2012, Fall). Economic Transformation in Ghana: Where
Will the Path Lead? Journal of African Development, 14(2), 41-78.
Kongsamut, P., Rebelo, S., & Xie, D. (2001). Beyond Balanced Growth. IMF
Working Paper WP/01/85.
Kravis, I., Heston, A., & Summers, R. (1982). The share of services in
economic growth. In F. G. Adams, & B. Hickman, In Global
Econometrics: Essays in Honor of Lawrence R. Klein (pp. 188-218).
Cambridge: MIT Press.
Kripfganz, S., & Schneider, D. C. (2016). ARDL: Stata module to estimate

autoregressive distributed lag models. Stata Conference. Chicago.
Retrieved from www.kripfganz.de/stata/
Kucera, D., & Roncolato, L. (2012). Structure Matters: Sectoral drivers of

growth and the labor productivity-employment relationship.
International Labour Office 2012 Research Paper No.3, pp. 1-57.
Kuznets, S. (1966). Modern Economic Growth. The American Economic

Review, 63(3), 247–258. doi:10.2307/1910352
Kuznets, S. (1973). Modern Economic Growth: Findings and Reflections. The

American Economic Review, 63(3), 247-258. Retrieved 10 21, 2015,
from http://www.jstor.org/stable/1914358
Lee, C. H. (1996, Spring). The Service Sector and Economic Growth.

ReFRESH 22, pp. 1-8.
Lee, J.-w., & Mckibbin, W. J. (2014, July). Service Sector Productivity and
Economic Growth in Asia. Asian Development Bank Institute Working
Paper Series No. 490, pp. 1-49.
Page | 239
Lewis, A. W. (1954). Economic Development with Unlimited Supplies of
Labor. Manchester School of Economic and Social Studies, 22, 139–
91.
Lin, J. Y. (2012). New structural economics: a framework for rethinking

development. In J. Y. Lin, New Structural Economics: A Framework
for Rethinking Development (pp. 11-48). Washington D.C: The World
Bank.
Maclean, D. (1996, Autumn). Productivity growth in the commercial service

sector. Bank of Canada Review, pp. 3-18.
Maddison, A. (2001). The World Economy: A Millennial Perspective. Paris:

Development Centre Studies OECD.
Maroto-Sanchez, A., & Cuadrado-Roura, J. R. (2007). Productivity and

Tertiarization in Industrialized Countries. A Comparative Analysis.
Universidad de Oviedo, Departamento de Economía, Efficiency Series
Paper 05/2007, pp. 1-28.
Maroto-Sánchez, A., & Cuadrado-Roura, J. R. (2009). Is growth of services an

obstacle to productivity growth? A comparative analysis. Structural
Change and Economic Dynamics, 20(4), 254-265.
doi:10.1016/j.strueco.2009.09.002
Maroto-Sanchez, A., & Cuadrado-Roura, J. R. (2011). Analyzing the role of

service sector on productivity growth across European regions.
Institute of Social and Economic Analysis Working paper 04/2011, pp.
1-24.
Martins, P. (2013). Growth, Employment and Poverty in Africa: Tales of

Lions and Cheetahs. Background Paper prepared for the World
Development Report 2013, pp. 1-56.
Martins, P. (2014, January). Structural Change in Ethiopia: An Employment

Perspective. The World Bank, Policy Research Working Paper, pp. 1-
42.
Mattoo, A., Rathindran, R., & Subramanian, A. (2006, March). Measuring

Services Trade Liberalization and Its Impact on Economic Growth: An
Illustration. Journal of Economic Integration, 21(1), 64-98.
McLachlan, R., Clark, C., & Monday, I. (2002). Australia’s Service Sector: A
Study in Diversity. Productivity Commission Staff Research Paper.
Page | 240
Mcmillan, M., & Headey, D. (2014). Introduction – Understanding Structural
Transformation in Africa. World Development, 63, 1-10.
doi:10.1016/j.worlddev.2014.02.007
McMillan, M., & Rodrik, D. (2011). Globalization, structural change and

productivity growth. In M. Bacchetta, & M. Jansen, Making
Globalization Socially Sustainable (pp. 49-85). International Labour
Organization and World Trade Organization.
McMillan, M., Rodrik, D., & Verduzco-Gallo, Í. (2014). Globalization,

Structural Change, and Productivity Growth, with an Update on Africa.
World Development, 63, 11-32. doi:10.1016/j.worlddev.2013.10.012
Mellor, J. W., & Dorosh, P. (2010, April). Agriculture and Ethiopia’s

Economic Transformation. ESSP II Working Paper 10.
Messina, J. (2004). Institutions and Service Employment: a panel study for

OECD countries. European Central Bank Working Paper Series (320),
pp. 1-42.
Miller, R. E., & Blair, P. D. (2009). Input – Output Analysis Foundations and
Extensions (Second ed.). Cambridge University Press.
Mitra, A. (2011). Trade in services: Impact on employment in India. The

Social Science Journal, 48(1), 72-93.
Moller, L. C., & Wacker, K. M. (2015, June). Ethiopia’s Growth Acceleration

and How to Sustain It: Insights from a Cross-Country Regression
Model. World Bank Group Policy Research Working Paper 7292.
Mukherjee, A. (2013, June). The Service Sector in India. Asian Development

Bank Economics Working Paper Series 352.
Narayan, P. K. (2005). The saving, and investment nexus for China: evidence
from cointegration tests. Applied Economics, 37(17), 1979–1990.
Nassif, A., Feijó, C., & Araújo, E. (2013). Structural change and economic
development: Is Brazil catching up or falling behind? UNCTAD
Discussion Papers No. 211.
Nayyar, G. (2012). The service sector in India’s Development. New Delhi:

Cambridge University Press.
NBE. (2015). Annual report on Ethiopian economy, macroeconomic and

social indicators 2013/14. Addis Ababa: National Bank of Ethiopia.
Page | 241
Ngai, L. R., & Pissarides, C. A. (2007). Structural change in a multisector
model of growth. American Economic Review, 97(1), 429-443.
Noland, M., Park, D., & Estrada, G. B. (2012, November). Developing the
Service Sector as Engine of Growth for Asia: An Overview. Asian
Development Bank Economics Working Paper Series No. 320, pp. 1-
48.
Nordhaus, W. D. (2006, May). Baumol’s Diseases: A Macroeconomic

Perspective. National Bureau of Economic Research Working Paper
12218. Retrieved from http://www.nber.org/papers/w12218
NPC. (2015). The Federal Democratic Republic of Ethiopia: the Second

Growth and Transformation Plan (GTP II) (2015/16-2019/20). Addis
Ababa, Ethiopia: National Planning Commission. Retrieved October
21, 2016, from ttps://www.africaintelligence.com/c/dc/LOI/1415/GTP-
II.pdf
Palocz, E. (1988). Services in the World Economy. Acta Oeconomica, 39(1),

171-183.
Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1985). A Conceptual Model

of Service Quality and Its Implications for Future Research. The
Journal of marketing, 49(4).
Park, D., & Shin, K. (2012a, October). Performance of the Services Sector in
Korea: An Empirical Investigation. Asian Development Bank WP 12-
20, pp. 1-34.
Park, D., & Shin, K. (2012b, December). The service sector in Asia: Is it an
engine of growth? Asian Development Bank Working No. 322, pp. 1-
37.
Parra, J. C., & Wodon, Q. (2010a, October). SimSIP SAM: A Tool for the
Analysis of Input-Output Tables and Social Accounting Matrices
Version 1.1.2. Washington, DC: The World Bank.
Parra, J. C., & Wodon, Q. (2010b). How Does Growth Affect Labor Income
by Gender? A Structural Path Analysis for Tanzania. In J. S. Arbache,
A. Kolev, & E. Filipiak, Gender Disparities in Africa's Labor Market
(pp. 273-296). Washington DC: The World Bank.
Peneder, M. (2003). Industrial structure and aggregate growth. Structural

Page | 242
Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to
the analysis of level relationships. Journal of Applied Econometrics,
16(3), 289-326.
Pilat, D., & Wölfl, A. (2005). Measuring the Interaction Between

Manufacturing and Services. OECD Science, Technology, and Industry
Working Papers, 2005/05. doi:10.1787/882376471514
Pratt, S. (2011). Economic linkages and impacts across the TALC. Annals of
Tourism Research, 38(2), 630-650. doi:10.1016/j.annals.2010.11.014
Ramaswamy, K., & Agrawal, T. (2012, March). Services-led Growth,

Employment and Job Quality: A Study of Manufacturing and Service
sector in Urban India. Indira Gandhi Institute of Development
Research WP-2012-007. Retrieved from
http://www.igidr.ac.in/pdf/publication/WP-2012-007.pdf
Ranis, G., & Fei, C. J. (1961). A Theory of Economic Development. The

American Economic Review, 51(4), 533-565.
Reis, H., & Rua, A. (2006, November). An input-output analysis: linkages vs

leakages. Banco de Portugal Estudos e Documentos de Trabalho
Working Papers 17/2006, pp. 1-31.
Rodrik, D. (2014, October 13). Are Services the New Manufactures? Project
Syndicate. Retrieved from https://www.project-
syndicate.org/commentary/are-services-the-new- manufactures-by-
dani-rodrik-2014-10
Roncolato, L., & Kucera, D. (2014). Structural drivers of productivity and

employment growth: A decomposition analysis for 81 countries.
Cambridge Journal of Economics, 38, 399-424. doi:10.1093/cje/bet044
Rowthorn, R., & Ramaswamy, R. (1999, March). Growth, Trade, and

Deindustrialization. International Monetary Fund Staff Papers 46 (1),
pp. 18-41.
Rust, R. T., & Chung, T. S. (2006). Marketing Models of Service and

Relationships. Marketing Science, 25(6), 560-580.
Schettkat, R., & Yocarini, L. (2006). The shift to services employment : a

review of the literature. Structural Change and Economic Dynamics,
17, 127–147. doi:10.1016/j.strueco.2005.04.002
Page | 243
Seid, Y., Taffesse, A. S., & Ali, S. N. (2015). Ethiopia — an agrarian
economy in transition. WIDER Working Paper 2015/154.
Seth, V. K. (2008). Economics of Services: Story of transformation of

Cinderella into queen of hearts. New Delhi: Ane Books India.
Sheehan, P. (2008, May). Beyond industrialization: new approaches to

development strategy based on the service sector. UNU-WIDER
Working Paper 2008/60, pp. 1-20.
Silva, E. G., & Teixeira, A. A. (2008). Surveying structural change: Seminal

contributions and a bibliometric account. Structural Change and
Economic Dynamics, 19, 273-300. doi:10.1016/j.strueco.2008.02.001
Singh, B. (2012a). Challenges of Service-led Economic Development in India.

2nd International Conference on Economics, Trade and
Development.36, pp. 152-156. Singapore: IACSIT Press.
Singh, B. (2012b, August). Is the Service-Led Growth of India Sustainable?

International Journal of Trade, Economics and Finance, 3(4), 316-
322.
Singh, I., & Singh, L. (2013, October). Services Sector as an Engine of

Economic Growth: Implications for India-South Korea Economic
Cooperation. Centre for Development Economics and Innovation
Studies (CDEIS) Discussion Paper No. 3.
Singh, L. (2004). Technological Progress, Structural Change and Productivity

Growth in Manufacturing Sector of South Korea. World Review of
Science, Technology and Sustainable Development, 1(1), 37-49.
Singh, M., & Kaur, K. (2014). India’s services sector and its determinants : an
empirical investigation. Journal of Economics and Development
Studies, 2(2), 385–406.
Singh, N. (2007). Services-led industrialization in India: Assessment and

Lessons. In U. N. Affairs., & U. N. Development, Industrial
development for the 21st century: sustainable development
perspectives (pp. 235-291). New York: United Nations.
Smith, A. (2007). An Inquiry into the Nature and Causes of the Wealth of
Nations. In S. Soares. MetaLibri Digital Library.
Page | 244
Sonnis, M., Guilhoto, J. J., Hewings, G. J., & Martins, E. B. (1995,
September). Linkages, key sectors, and structural change: some new
perspectives. The Developing Economies, 33(3), 233-270.
Swan, N. M. (1985, July). The Service Sector: Engine of Growth? Canadian

Public Policy, 11, 344-350. Retrieved from
http://www.jstor.org/stable/3551061
Syrquin, M. (1988). Patterns of Structural Change. In H. B. Chenery, & T. N.

Srinivasan, Handbook of Development Economics (Vol. I, pp. 205-
273). B.V: Elsevier Science Publishers.
Szirmai, A. (2011). Manufacturing and Economic Development. UNU-

WIDER Working Paper No. 2011/75.
Szirmai, A. (2012). Industrialization as an engine of growth in developing

countries, 1950–2005. Structural Change and Economic Dynamics, 23,
406-420. doi:10.1016/j.strueco.2011.01.005
Taffesse, A. S., & Ferede, T. (2005, January). The Structure of the Ethiopian
Economy - A SAM-based Characterisation. A Country Economic
Memorandum, the World Bank background paper Report No. 29383-
ET.
Timmer, M. P., & Szirmai, A. (2000). Productivity growth in Asian

manufacturing: the structural bonus hypothesis examined. Structural
Timmer, M. P., de Vries, G., & de Vries, K. (2014, July). Patterns of

Structural Change in Developing Countries. Groningen Growth and
Development Centre Research Memorandum 149, pp. 1-31.
Timmer, M., & de Vries, G. (2009). Structural change and growth

accelerations in Asia and Latin America: a new sectoral data set.
Cliometrica, 3(2), 165-190. doi:10.1007/s11698-008-0029-5
Todaro, M. P., & Smith, S. C. (2012). Economic Development (11th ed.).

Addison-Wesley.
Tregenna, F. (2008, November). Sectoral Engines of Growth in South Africa:

An Analysis of Services and Manufacturing . UNU-WIDER Research
Paper No. 2008/98, pp. 1-42.
Triplett, J. E., & Bosworth, B. P. (2000). Productivity in the Services Sector.

Washington, D.C: Brookings Institution.
Page | 245
Triplett, J. E., & Bosworth, B. P. (2003, September). Productivity
Measurement Issues in Services Industries: “Baumol’s Disease” Has
Been Cured. Federal Reserve Bank of New York provides Economic
Policy Review.
UNCTAD. (2015). Economic Development in Africa Report 2015: Unlocking

the Potential of Africa's Services Trade for Growth and Development.
New York and Geneva: United Nations.
Uwitonze, E., & Heshmati, A. (2016, August). Service Sector Development

and its Determinants in Rwanda. IZA Discussion Paper No. 10117.
Van Ark, B., & Timmer, M. (2003). Asia’s Productivity Performance and
Potential: The Contribution of Sectors and Structural Change.
University of Groningen & Conference Board.
Verma, R. (2012). Can total factor productivity explain value added growth in
services? Journal of Development Economics, 99(1), 163-177.
WB. (2012a). Job Generation and Growth Decomposition Tool.

Understanding the Sectoral Pattern of Growth and its Employment and
Productivity Intensity Reference Manual and User’s Guide Version
1.0. The World Bank PREM.
WB. (2012b). Ethiopia Economic Update: Overcoming inflation, Raising

Competitiveness. Washington, DC: The World Bank.
WB. (2015a). Ethiopia’s Great Run: The Growth Acceleration and How to
Pace It Report No. 99399-ET. World Bank Group.
WB. (2015b). Ethiopia Poverty Assessment 2014. Washington, DC: World

Bank Group.
WDI. (2016). World Development Indicators. The World Bank.
World Economic Forum. (2015). The Africa Competitiveness Report 2015.

Geneva: World Economic Forum.
WTO. (2010, November). Measuring trade in services, a training module for

the World Bank. World Trade Organization.
Page | 246
1. Annexure-1
Economic activities distinguished (ISIC rev. 3.1 code)
1. Agriculture, hunting, forestry and fishing (AtB);

2. Mining and quarrying (C);
3. Manufacturing (D);
4. Electricity, gas and water supply (E);
5. Construction (F);
6. Wholesale and retail trade, hotels and restaurants (GtH);
7. Transport, storage, and communication (I);
8. Finance, insurance, real estate and business services (JtK);
9. Government services (LtN);
10. Community, social and personal services (OtP)
Page | 247
Annexure-2
Technical coefficients for the macro SAM of Ethiopia, 2005/06

1 2 3 4 5 6 7 8 9
1 Commodities 0.1590 0.3475 0.8473 0.5353 0.8957 0.3513
2 Activities 0.6445
3 Labor 0.3219
4 Capital 0.3306 0.0095
5 Enterprises 0.1073
6 Households 0.8130 0.7420 0.8024 0.0064 0.3196 0.2221
7 Government 0.0348 0.1870 0.1473 0.1976 0.0070 0.0585 0.1868
8 Capital account 0.1359 0.0846 0.1043 0.2303
9 RoW 0.1618 0.0034 0.0034 0.0020

Source: SimSIP SAM results from 2005/06 Ethiopian SAM
Page | 248
Annexure-3
Multiplier Product Matrix for Ethiopia, 2005/06
Activities 1 2 3 4 5 6 7 8 9 10
1 Agriculture 3.14 2.33 1.45 2.29 2.32 2.47 1.49 2.32 2.39 2.14
2 Mining 0.18 0.13 0.08 0.13 0.13 0.14 0.08 0.13 0.13 0.12
3 Manufacturing 2.62 1.95 1.21 1.91 1.94 2.06 1.25 1.94 1.99 1.79
4 Utility 0.27 0.20 0.13 0.20 0.20 0.21 0.13 0.20 0.21 0.19
5 Construction 0.26 0.19 0.12 0.19 0.19 0.21 0.12 0.19 0.20 0.18
6 Distributive trade 2.54 1.88 1.17 1.85 1.87 2.00 1.21 1.88 1.93 1.73
7 Transport &commun. 0.72 0.53 0.33 0.52 0.53 0.57 0.34 0.53 0.55 0.49
8 Business and Finance 0.83 0.61 0.38 0.60 0.61 0.65 0.39 0.61 0.63 0.56
9 Public Service 0.58 0.43 0.27 0.42 0.43 0.45 0.27 0.43 0.44 0.39
10 Personal Service 0.26 0.20 0.12 0.19 0.19 0.21 0.13 0.20 0.20 0.18
Page | 249
Page | 1
View publication stats

Strucutreand Growthof Service Sectorin Ethiopia

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

Strucutreand Growthof Service Sectorin Ethiopia

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

STRUCTURE AND GROWTH OF SERVICE SECTOR IN ETHIOPIA

Preprint · November 2017

Zelalem Ejigu Kabeta

The user has requested enhancement of the downloaded file.

Zelalem Ejigu Kabeta

Established Under Punjab Act No. 35 of 1961

SERVICE SECTOR IN ETHIOPIA

Established Under Punjab Act No. 35 of 1961

This is to certify that this thesis entitled “Structure and

(Dr. Inderjeet Singh)

Date:_____________ Signature of the Candidate

This is to certify that the above statement made by the candidate is

Date: ______________ Supervisor’s Signature

To the beloved late mothers W/ro Assefash and W/ro Gizhachew,

To my late uncles (Nigussie, Jebessa, and Chombe), whom I lost during

(Zelalem Ejigu Kabeta)

CHAPTER- VII: MAIN FINDINGS AND POLICY IMPLICATIONS .................. 214

Economic history shows that the features of structural change process

In Ethiopia, the tertiary sector share in GDP exceeds the agriculture

Structural change involves changes in the composition of the sectoral

According to Moller & Wacker (2015), public infrastructure

Along with view on the importance of agriculture in Ethiopia, there is

The Ethiopian government is also following a policy to attract and

Furthermore, the service sector is a driver of the recent economic

1.3 : Study Objectives

b) To delineate the structural change in Ethiopia with special reference to

c) To determine the factors that explain the growth of the Ethiopian

d) To analyze the service sector inter-linkage with agriculture and

e) To examine the viability and sustainability of service sector growth in

f) To prescribe the policy framework for better functioning and

The justification for this study is the shift of Ethiopian economy

1.5 : Significance of Study

1.6 : Scope of Study

This study generally concentrates on exploring the structure of the

1.7 : Data and Methodology

The structure and growth of the service sector in Ethiopia is analyzed

1.8: Thesis Organization

1.9 : Main Conclusion of the Study

The theoretical and empirical review of the literature on service sector

The literature in this section is broadly classified as a theoretical and

2.1.1 : Conceptualization of Service Sector

The term service represents and constitutes various issues. It is difficult

According to the International Standard Industrial Classification

The definitional characteristics of services have brought different

Additionally, the diversity in the definition and classification of service

According to Bhagwati (1984a), services can be categorized into those

Even though the recent growth in services trade was accompanied by

2.1.2 : Structural Change and the Growth of Service Sector

Economic development literature of the Post-Second World War

The linear-stages-of-growth model, which is also called Rostow’s

According to Todaro & Smith (2012), one of the early theoretical

In addition to the two-sector model, the patterns-of-development

Generally, most of the neoclassical economists believe that structural

The structuralist economics of the Post-Second World War period has

In development economics and in economic history, structural change

Thus, the process of economic development is highly associated with

On the other hand, Chenery (1986) identified accumulation of human

The patterns of structural transformation, as it is documented by

Labour productivity is one of the important features in structural

On the other hand, the process of structural transformation differs from

Recently, Bah (2010) studied whether developing countries are also

Therefore, based on the review of the major literature on structural