University of Southern California Dissertations and Theses

Growth, trade and structural change in low income industrializing economies

(USC Thesis Other)

Growth, trade and structural change in low income industrializing economies

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Growth, Trade and Structural Change in Low Income Industrializing Economies
by
Rubina Verma
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
Doctor of Philosophy
(ECONOMICS)
August 2008
Copyright 2008 Rubina Verma
Dedication
For my father, M.N. Verma - my friend, philosopher and guide,
& my mother Rita Verma - my eternal love.
ii
Acknowledgments
My interest to pursue a doctorate degree in economics arose when I was in college.
The Department of Economics at The University of Southern California gave me an
opportunity to do so. I would like to thank all the faculty and staff for giving me an
opportunity to live my dream.
All through my graduate school life, my advisor, Prof. Caroline Betts has been a
wonderful mentor to me. I am eternally grateful for her guidance and support. She has
always been very encouraging and her words and belief were important in motivating me
to write this piece.
My sincere thanks and gratitude to Prof. Guillaume Vandenbroucke whose invaluable
time, knowledge and support were crucial ingredients in completing this dissertation.
I wish to acknowledge helpful comments from Prof. Vincenzo Quadrini, Prof. Yong
Kim, Prof. Robert Dekle, Prof. Jefferey Nugent, Prof Hyeok Jeong and Prof. Ayse
Imrohoroglu.
I truly appreciate the help I received from Rahul Giri without whose support com-
pleting this dissertation would be an impossible and implausible task.
My work has immensely benefited with my conversations with Murat Ungor and I
extend my sincere thanks to him.
I extend my thanks to my family, Avina Verma, Karan Bajaj and Vikrant Bajaj as
well as close friends, Subha Mani, Engin Volkan, Abhijit Chaudhari, Ashish Agarwal,
Rajini Parameswaran and Utteeyo Dasgupta. Their presence, support and words have
not only helped me complete my degree but have also made it a memorable experience.
Lastly, my heartfelt thanks to my parents and grand parents for their belief in me.
iii
Table of Contents
Dedication ii
Acknowledgments iii
List Of Tables vi
List Of Figures vii
Abstract viii
Chapter 1: Introduction 1
Chapter 2: The Indian Economy 8
2.1 Sectoral Data Facts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 Sectoral Growth Accounting. . . . . . . . . . . . . . . . . . . . . . 10
2.1.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Chapter 3: Steady State Analysis 20
3.1 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.1 Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.2 Preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.4 Counterfactual Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Chapter 4: Dynamic Analysis 29
4.1 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.1.1 Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.1.2 Preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.1.3 Competitive Equilibrium . . . . . . . . . . . . . . . . . . . . . . . 32
4.1.4 Model of Unbalanced Growth with Structural Change . . . . . . . 32
4.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
iv
4.2.2 Data for Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.4 Effect of Liberalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.5 Explaining the Rapid Growth of Share of Services in Indian GDP . . . . . 39
4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Reference List 48
List of Appendices
Appendix A
Data Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Appendix B
Numerical Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
B.0.1 Solving the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
B.0.2 Steady State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
B.0.3 Model Properties: Unbalanced Growth and Structural Change . . 53
B.1 Numerical Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
v
List Of Tables
1.1 Growth Rates of GDP Per Capita in Low Income Countries . . . . . . . . 6
2.1 Growth Accounting - Baseline Results . . . . . . . . . . . . . . . . . . . . 15
2.2 Growth Accounting - GTAP Factor Shares. . . . . . . . . . . . . . . . . . 17
2.3 Growth Accounting - Capital Share of One-third . . . . . . . . . . . . . . 19
3.1 Calibrated Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Results for the Two Steady States . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Counterfactual Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1 Calibrated Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.2 Average Annual Growth Rates (%), 1980-2003 . . . . . . . . . . . . . . . 37
4.3 Pre and Post Liberalization TFP Growth Rates . . . . . . . . . . . . . . 38
4.4 Average Annual Growth Rates during 1991-2003 . . . . . . . . . . . . . . 38
4.5 FDI and GDP in Indian Services . . . . . . . . . . . . . . . . . . . . . . . 41
vi
List Of Figures
1.1 A Comparison of the Indian Economy relative to the U.S. Economy . . . 4
2.1 Shares of Sectoral Output in GDP, 1980-2003 . . . . . . . . . . . . . . . . 9
2.2 Shares of Sectoral Employment, 1980-2003 . . . . . . . . . . . . . . . . . . 10
2.3 Shares of Sectoral Exports, 1980-2003 . . . . . . . . . . . . . . . . . . . . 11
2.4 Shares of Sectoral Imports, 1980-2003 . . . . . . . . . . . . . . . . . . . . 11
2.5 Sectoral TFP Levels, 1980-2003 . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1 Share of Output of Services Subsectors in Aggregate GDP . . . . . . . . . 42
4.2 Shares of Sectoral Output, 1980-2003 . . . . . . . . . . . . . . . . . . . . . 45
4.3 Shares of Sectoral Employment, 1980-2003 . . . . . . . . . . . . . . . . . . 46
4.4 Effect of Liberalization - Shares of Sectoral Output . . . . . . . . . . . . . 47
vii
Abstract
The objective of this dissertation is to explain the rapid growth of value added in
the service sector in India, and, to examine the factors driving this services’ led growth
in the economy. In the first chapter of this dissertation, I record the empirical data facts
on output, employment and trade, for the three principal sectors of the economy, namely,
agriculture, industry and services. This is followed by a sectoral growth accounting ex-
ercise which I conduct for the 1980-2003 period. Two empirical facts emerge from the
analysis: changes in total factor productivity (TFP) were the largest source of service
sector value added growth in India and a sharp acceleration in industrial and services’
tradeoccurredfollowingliberalizationin1991. Motivatedbythesefindings, inthesecond
chapter of this dissertation, I build a simple three sector growth model with two main in-
puts: growth in sectoral TFP and trade in industry and services. The model is calibrated
to Indian sectoral data across two steady state years during which trade is balanced. The
results from this chapter indicate that productivity growth versus trade has a more im-
portantroleincapturingthesectoralcompositionofGDPinIndia. Inthesecondchapter
of my dissertation, I extend the steady state analysis and develop a three sector growth
model to evaluate the quantitative performance of differential sectoral TFP growth in
accounting for the structural transformation of India during the 1980-2003 period. The
model is calibrated to Indian data using average growth rates of sectoral TFP as primary
inputs, and performs well in accounting for the evolution of sectoral value added shares
and the growth rates of these shares over the period 1980-2003. Moreover, the perfor-
mance of the model improves significantly when the post 1991 increase in service sector
TFP growth is accounted for. I find that the liberalization policies adopted by India
from 1991, especially the deregulation and privatization of business and communications
services, and quite possibly technological progress, explain the improvement in service
sector TFP, and hence the dominance of service sector activity in India’s recent GDP
growth.
viii
Chapter 1
Introduction
An empirical comparison of the historical growth experiences of contemporary de-
velopedcountrieswiththecurrentgrowthexperiencesofsomefastgrowingcontemporary
developingnations,revealssomesignificantdifferencesintheirgrowthpatterns. Formost
industrialized nations, such as, the United Kingdom, France, United States, historical
data show that at low levels of per capita income, the agricultural sector dominated the
composition of output and employment. As these nations embarked on a path of rapid
and sustained economic growth, resources were transferred from the agricultural sector
to the manufacturing and service sectors. Only when the economy matured and reached
the status of a high income nation, did the role of the service sector become more domi-
nant. Today, for some low income, rapid growing industrializing nations, this process of
sectoral reallocation of economic activity, also known as structural transformation, looks
different. In these countries, even at low levels of per capita income, the service sector
accounts for a significant amount of the economy’s output as measured by its share in
Gross Domestic Product (GDP). Moreover, in these economies the share of services in
GDP has been increasing at a rapid rate, much greater than the corresponding growth
rate witnessed by the service sector in the GDP of contemporary developed economies
when they were at equivalent stages of development. In today’s low income economies,
the role of the service sector has become more prominent at relatively early stages of
economic development.
In order to examine the set of low income, rapid growing economies, which exhibit
service sector led growth, I conduct the following empirical exercise. First, I define a ‘low
income country’ as a country with a level of GDP per capita less than 825 US $ in 1980
1
.
Following this criterion, I identify 42 low income countries in 1980 and calculate their
1
In 2004, The World Bank defined a low income country as a country which had a level of Gross National
Income per capita less than 825 US $.
1
averagegrowthratesofGDPpercapitaduringtheperiod1980-2004. Table1.1liststhese
countries in descending order of their growth rates, together with their respective GDP
per capita in 1980. The average growth rate for the entire sample is 0.51 percent, owing
to a large number of countries which witnessed negative growth rates during this time
period. Amongst these countries, 17 countries experienced negative growth rates, while
11 countries grew at an average rate of 0-1 percent and 3 countries witnessed growth
rates between 1-2 percent. My interest lies in choosing the rapid growing countries which
witnessed average annual growth rates of GDP per capita in excess of 2 percent, which
was the secular growth rate of the U.S. economy in the twentieth century
2
. The U.S.
economywastheindustrialleaderthroughoutthetwentiethcenturyandhencethegrowth
performance of the rapid growers is measured relative to the U.S. economy. I refer to
these 11 countries as Rapid Growers. These countries include China, Thailand, Bhutan,
India, Indonesia, Sri Lanka, Chad, Lesotho, Pakistan, Bangladesh and Nepal.
Next, I examine the performance of the three sectors, namely, agriculture, industry
and services, in contributing to aggregate growth of output in these economies. It is well
recognized that as an economy grows and witnesses structural transformation, growth
proceeds at an uneven rate from sector to sector. Following (Syr88), I examine the
relation between aggregate and sectoral growth by differentiating with respect to time
the definition of total output, V =
P
V
i
and expressing the result in growth terms:
g
V
=
X
i
ρ
i
g
V
i
where g
V
and g
V
i
are the growth rates of aggregate output, V, and sectoral output, V
i
,
respectively, and the weights are sectoral output shares, ρ
i
=V
i
/V. The above equation
expresses the contribution of each sector to aggregate GDP growth measured in terms of
the average share of total GDP accounted by this sector, weighted by the growth rate of
GDP in this sector.
For each of the 11 Rapid Growers, I decompose the growth rate of aggregate GDP
using growth rates of sectoral output and shares of the sectoral output in GDP. Follow-
ing this decomposition, I identify those low income, fast growing, countries which have
witnessedservicesectordrivengrowth. Specifically, intheseeconomies, theservicesector
has made the highest average contribution to aggregate growth during the 1980-2004 pe-
riod. I classify them as service sector dominated countries. This set of countries includes
2
Following(KP02);theycalculatetheaveragegrowthrateofoutputperworking-agepersonintheU.S.economy
to be 2 percent in the twentieth century.
2
India, Sri Lanka, Pakistan, Bangladesh and Nepal. Amongst all these service sector led
countries, India has witnessed the most rapid growth in GDP and in GDP per capita
during the 1980-2004 period.
Figure 1.1 presents an empirical comparison of the current growth experience of India
with the historical growth experience of the United States (U.S.), a contemporary ad-
vanced nation. During the 1980-2003 period, the average annual growth rate of the total
output of the Indian economy was 5.6 percent while the growth rate of the output of the
servicesectorexceededtheaggregategrowthrateat7percent. Inotherwords,theservice
sector’s share in GDP grew at an average annual rate of 1.3 percent for the 1980-2003
period. This growth rate is much higher than the corresponding growth rate witnessed
by the U.S. economy, when the U.S. was at an equivalent stage of development. In the
upper panel of Figure 1.1, the relative Indian/U.S. GDP per capita during the 1980-2003
period is graphed against the services’ share of output in Indian GDP. From this figure,
it is evident that in 1980 when India’s GDP per capita was 5.2 percent of the U.S. GDP
per capita, the share of services in Indian GDP was about 38 percent. By 2002, Indian
GDP per capita had grown to 7.2 percent of U.S. GDP per capita, at which date the
share of services in Indian GDP was 49 percent. By 2003, Indian GDP per capita had
increased to about 7.6 percent of U.S. GDP per capita, and the corresponding share of
services accounted for 52 percent of Indian GDP.
The lower panel depicts how the share of services in U.S. GDP
3
evolved during the
period 1839-1899. In 1839, the U.S. GDP per capita relative to its average 1980-2003
value was similar in magnitude to the 1980 Indian/U.S. GDP per capita ratio. In this
year, the U.S. GDP per capita was 5.3 percent of its average 1980-2003 GDP per capita
value, and services accounted for 38 percent of aggregate GDP. When U.S. had grown to
7.3 percent of its average 1980-2003 GDP per capita value, the output share of services
was 41 percent. By 1899, U.S. GDP per capita had grown to 13.5 percent of its average
1980-2003 GDP per capita value, and the output share of services in GDP had risen only
toabout47percent. Theshareofservices’outputinU.S.GDPgrewatanaveragerateof
0.36 percent during the 1839-1899 period. In comparison, theaverage annual growth rate
of the output share of services in Indian GDP during the 1980-2003 period was one full
percentagepointhigherthanwhattheU.S.witnessed,whentheU.S.wasatanequivalent
stage of development.
The objective of this dissertation is to explain the rapid growth of value added in the
service sector in India, and, to examine the factors driving this services led growth in the
3
These data are obtained from (WG69).
3
Figure 1.1: A Comparison of the Indian Economy relative to the U.S. Economy
1980 1985 1990 1995 2000 2003
30
35
40
45
50
55
India−US Comparison
Year
Services’ share in Indian GDP
1840 1850 1860 1870 1880 1890 1899
30
35
40
45
50
55
US−US Comparison
Year
Services’ share in US GDP
India/ U.S. GDP per capita 5.2%
1980
India/ U.S. GDP per capita 7.2%
2002
India/ U.S. GDP per capita 7.6%
2003
U.S./ U.S. GDP per capita 5.3%
1839
U.S./ U.S. GDP per capita 7.3%
1859
U.S./ U.S. GDP per capita 13.5%
1899
4
economy. In the first chapter of this dissertation, I record the empirical data facts on
output, employment and trade, for the three principal sectors of the economy, namely,
agriculture, industry and services. This is followed by a sectoral growth accounting
exercise which I conduct for the 1980-2003 period. Two empirical facts emerge from the
analysis: changes in total factor productivity (TFP) were the largest source of service
sector value added growth in India and a sharp acceleration in services’ trade occurred
followingliberalizationin1991. Motivatedbythesefindings, inthesecondchapterofthis
dissertation, I build a simple three sector growth model with two main inputs: growth in
sectoral TFP and trade in industry and services. The model is calibrated to the Indian
sectoral data across two steady state years during which trade is balanced. The results
from this chapter indicate that productivity growth versus trade has a more important
role in capturing the sectoral composition of GDP in India. In the second chapter of my
dissertation, Iextendthesteadystateanalysistoatwentythreeyearperiodrangingfrom
1980 to 2003. Here, I develop a three sector growth model to evaluate the quantitative
performance of differential TFP growth across sectors in accounting for the structural
transformation of India. A version of the model that is carefully calibrated to Indian
data, and in which average rates of TFP growth by sector from India are the primary
inputs, performs well in accounting for the evolution of value added shares of the three
major sectors of economic activity over the period 1980-2003. It also accounts well for
the growth rates of the GDP shares of all three major sectors of economic activity over
this period - for the structural transformation of GDP. Moreover, the performance of the
model improves significantly when the post 1991 increase in service sector TFP growth
is accounted for. I find that the liberalization policies adopted by India from 1991, and
especiallythederegulationandprivatizationofbusinessandcommunicationsservices,and
quitepossiblytechnologicalprogress, explaintheimprovementinservicesectorTFP,and
hence the dominance of service sector activity in India’s recent GDP growth.
5
Table 1.1: Growth Rates of GDP Per Capita in Low Income Countries
Countries 1980 GDP per capita Average annual growth rate (%)
less than 825 constant 2000 U.S.$ of GDP per capita
1980-2004
Rapid Growers: growth rate greater than 2%
China 186.44 8.51
Thailand 804.48 4.58
Bhutan 263.65 4.12
India 222.05 3.76
Indonesia 396.63 3.50
Sri Lanka 441.86 3.29
Chad 147.26 2.34
Lesotho 309.65 2.34
Pakistan 327.43 2.31
Bangladesh 240.51 2.16
Nepal 140.08 2.11
Countries with growth rate greater than 1% but less than 2%
Sudan 274.22 1.93
Mozambique 179.01 1.80
Burkina Faso 191.69 1.08
Countries with growth rate greater than 0% but less than 1%
Kiribati 435.41 0.84
Mauritania 361.80 0.79
Guyana 819.41 0.79
Ghana 233.56 0.74
Senegal 405.53 0.53
Benin 292.44 0.47
Mali 220.22 0.302
Solomon Islands 597.09 0.26
Cameroon 638.19 0.15
Papua New Guinea 582.54 0.15
Gambia 327.21 0.12
Countries with growth rate less than 0%
Kenya 435.24 -0.08
Malawi 161.70 -0.23
Guinea-Bissau 144.44 -0.23
Nigeria 425.32 -0.24
Comoros 404.63 -0.29
Rwanda 280.35 -0.48
Burundi 126.36 -0.78
Zimbabwe 598.68 -1.12
6
Table 1.1: (continued)
Countries 1980 GDP per capita Average annual growth rate (%)
less than 825 constant 2000 U.S.$ of GDP per capita
1980-2004
Zambia 450.51 -1.21
Central African Republic 313.57 -1.37
Togo 346.28 -1.45
Madagascar 341.81 -1.66
Niger 245.50 -1.87
Haiti 802.62 -2.57
Sierra Leone 310.40 -2.82
Congo, Dem. Rep. 251.12 -4.29
Liberia 744.48 -7.02
All Countries: average annual growth rate 0.51%
7
Chapter 2
The Indian Economy
2.1 Sectoral Data Facts
During the period 1980-2003, real value added in agriculture, industry and services
grew at an average annual rate of 3.1, 6.1 and 7 per cent, respectively. Figure 2.1 depicts
the evolution of the shares of value added in agriculture, industry and services during
the 1980-2003 period for India. Between 1980 and 2003, the share of value added in
agriculture declined from 37 percent to about 21 percent, the share of industry increased
from 24 to 27 percent, while the share of services grew from 38 percent to 52 percent.
It is evident that the decline in agriculture’s share of value added has been mirrored in
an increase in services’ share of value added, while industry’s share of value added has
increased only modestly over the time period. In terms of growth rates, the share of
agriculture in GDP declined at an average annual rate of 2.4 percent over the 1980-2003
period. During the same period, the share of services in Indian GDP grew at 1.3 percent
per year, while industry’s share in GDP showed a small increase of 0.5 percent per year.
Thisdifferentialbetweenthegrowthratesofsharesofindustrialandservices’valueadded
becomes sharper after 1991. For the 1991-2003 period, share of industry’s value added
grew at a meager rate of 0.04 percent per annum while the share of services’ value added
grew at a much higher rate of 1.95 percent per annum. The share of agriculture’s value
added was declining in both the sub periods - at an average annual rate of 1.48 percent
during 1980-1990 and at 3.31 percent during the 1991-2003 period.
While the value added data show significant growth in the share of services in ag-
gregate output, the share of employment in this sector is still small. This observation
of a high share of service output in aggregate output and low share of service employ-
ment in aggregate employment has been termed as ‘jobless’ growth in services ((BS04),
8
(Ban06b)). Thetrendsintheshareofemploymentinservicesandintheothertwosectors
are presented in figure 2.2.
The employment graph reveals that sectoral reallocation of employment out of agri-
culture and into industry and services has been slow. In 1980, the share of employment
in agriculture was 65 percent, and the corresponding shares in industry and services were
15and20percentrespectively. Evenby2003, theshareofemploymentinagriculturewas
the highest among the three sectors, at 54 percent, whereas in industry and services, it
was 18 and 28 percent, respectively. Clearly, the shares of sectoral employment have not
kept pace with the shares of sectoral value added. Some authors have tried to rationalize
the slow movement of labor from agriculture into industry and services in India. (Pan06)
discusseshowthegrowthofunskilledlaborintheorganizedsectorhasbeenveryslowdue
to stringent labor regulations in that sector. He argues that the formal sector in India
has witnessed increasing wages, and has a lot of potential to absorb unskilled labor. In
India, employment in the informal sector has been rising. However, since the wage dif-
ferential between the non-agricultural informal sectors and the agricultural sector (which
is predominantly informal in nature) is not very large, there does not exist a big enough
incentive for labor to move out of agriculture and into industry and services. Moreover,
inter-state migration has been extremely slow in India due to linguistic differences, lack
of any social protections such as mutual insurance provided to members of the same sub-
caste networks, making it dangerous to travel outside the reach of one’s social network
((MR04)). Additionally, (Ban06a) discusses how the lack of cheap urban housing and
Figure 2.1: Shares of Sectoral Output in GDP, 1980-2003
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Agriculture
Industry
Services
Pre liberalization Post liberalization
9
Figure 2.2: Shares of Sectoral Employment, 1980-2003
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
0.6
Agriculture
Industry
Services
poor planning in urban areas has served as a barrier to migration. Since most of the
industrial and service firms are located in the urban areas in India, the slow rural-urban
migration has some merit in explaining the slow movement of labor across the sectors.
Figure 2.3 and 2.4 depict the evolution of agricultural, manufacturing and services
exportsandimportsrespectively, asmeasuredbytheirshareinGDP.Alookatthegraph
reveals that both agricultural exports and imports were small shares of GDP and were
never in excess of 2 percent of GDP. Industrial exports show a clear increasing trend,
growing from 3 percent of GDP in 1980 to about 9 percent of GDP in 2003. Similarly,
industrialimportsalsoincreasedfromabout3percentofGDPin1980toabout7percent
of GDP by 2003. In the service sector, exports were growing slowly till about mid 1990s,
but following 1995 one observes a clear upward trend, increasing sharply from about 2
percent of GDP in 1995 to about 4 percent of GDP in 2003. Service imports also rose
steadily until early 1990s, thereafter which their growth accelerated during the late 1990s
and started slowing a little by 2000. Notably, all through the sample period, the level of
service imports exceeded the level of service exports in GDP by a small margin.
2.1.1 Sectoral Growth Accounting
To gain further insight into the sources of growth in service sector value added, I
conduct a growth accounting of value added for each of the sectors - agriculture, industry
and services, for the 1980-2003 period. This exercise involves decomposing changes in
10
Figure 2.3: Shares of Sectoral Exports, 1980-2003
1980 1985 1990 1995 2000
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Year
Agriculture
Industry
Services
Pre liberalization Post liberalization
Figure 2.4: Shares of Sectoral Imports, 1980-2003
1980 1985 1990 1995 2000
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Year
Agriculture
Industry
Services
Pre liberalization Post liberalization
value added into the portions due to changes in factor inputs and the portion due to
changes in efficiency with which these factors are used, measured as total factor produc-
tivity (TFP) of a sector. To summarize, the results, indicate that changes in TFP are
significant in accounting for value added growth in the service sector. Also, the growth
of agricultural value added is largely accounted for by TFP growth. By contrast, the
growth of industrial output is largely driven by the growth of factor inputs, primarily due
to growth in capital. Additionally, I find that TFP growth rate in the service sector is
11
the highest across the three sectors for the entire time period, primarily because it has
grown at a very rapid rate after economic liberalization in 1991
1
.
2.1.2 Methodology
This section describes the model of value added by sector used in the growth ac-
counting procedure. The methodology for constructing the factor shares is described in
the following sub section. I follow the standard methodology of growth accounting which
involves decomposing output growth into TFP growth, capital growth and labor growth.
The production function in each sector is assumed to be Cobb-Douglas with constant
returns to scale. In particular, the function is described by
Y
jt
=A
jt
K
ν
j
jt
N
1−ν
j
jt
j ∈{industry,services}
where ν
j
and 1−ν
j
represent the share of rental payments to capital and share of wage
payments to labor in the total income of sector j, respectively. The agricultural produc-
tion function has an additional input of land. The production function is accordingly
modified as
Y
at
=A
at
K
νa
at
L
γa
at
N
1−νa−γa
at
The factor income shares in this sector areν
a
- capital income share,γ
a
- share of rental
income from land and (1−ν
a
−γ
a
) - labor income share.
By differentiating the production function with respect to time,t, and dividing byY
j
,
thegrowthrateoftotalfactorproductivitygrowthinsectorj ={a,i,s}canbeestimated
as
dA
j
/dt
A
j
=
dY
j
/dt
Y
j
−ν
j
dK
j
/dt
K
j
−(1−ν
j
)
dN
j
/dt
N
j
−γ
j
dL
j
/dt
L
j
(2.1)
In industry and services, the last term disappears since land is not a factor of production
in these sectors.
2.1.3 Data
In order to conduct growth accounting, data are collected for the three sectors -
agriculture, industry and services - for the 1980-2003 period.
Real GDP: Data for sectoral real GDP are taken from the Business Beacon, Cen-
ter for Monitoring Indian Economy (CMIE). Agriculture includes forestry, logging and
1
Gross output by sector would also be analyzed, but data is unavailable.
12
fishing; Industry consists of manufacturing, mining, electricity, gas and water supply,
and construction, while Services include trade, hotel, transport, communication, finance,
insurance, real estate, business services and social and personal services. All data are
measured in constant 1994 Indian Rupees.
Capital Stock: The capital stock series are constructed using the Perpetual Inven-
toryMethod(PIM),whereinvestmentismeasuredusingthegrossfixedcapitalformation
series and a constant depreciation rate of 5 percent. In each sector, the initial capital
stock is the sectoral gross fixed capital stock in 1952. Using the PIM, the entire cap-
ital stock series for all sectors are constructed from 1952 to 2003. For my purpose, I
use the capital stock series for the 1980-2003 period. All sectoral capital stock data are
measured in constant 1994 Indian Rupees and are obtained from the Central Statistical
Organization (CSO) of India.
Employment: Indiadoesnotreportthenumberoflaborhoursworkedineachsector.
Hence, I measure employment as the number of people working in each sector. Sectoral
employmentnumbersarecalculatedusingthedefinitionofemploymentonacurrentdaily
status (cds) basis
2
. These data are constructed with the help of annualized growth rates
of sectoral employment reported by (Gup02). In particular, this report presents sectoral
employment numbers for the years 1983, 1987-88, 1993-94, 1999-2000 and 2001-2002 as
well as the average annual growth rates for the intermittent years. Using these growth
rates, I construct sectoral employment series for the 1980-2003 period.
Land: An estimate of land used in the agricultural sector is needed. Data series
on gross sown area are used for this purpose. Gross sown area is defined as the sum of
area covered by all individual crops including the area sown under crops more than once
during a given year. It is also referred to as gross cropped area. These data are obtained
from Business Beacon, (CMIE) from 1980 to 2001. For 2002 and 2003, gross sown area
data have been taken from the Statistical Pocket Book, 2005 available from CSO, India
3
.
FactorIncomeShares: Ifollow(Gol02)andcalculatefactorsharesbyadjustingfor
income of the self employed. For the 1980-2003 period, CSO reports factor incomes from
differentsubsectorswhichcompriseofCompensationofEmployees(COE)andOperating
Surplus (OS). In each sub sector, the COE and OS are further divided into two compo-
nents, one part accruing from the organized sector and the second part as originating in
the unorganized sector. I consider OS of the unorganized sector as Operating Surplus of
2
Details of the cds approach are provided in the data appendix.
3
Note that this is incomplete - land is also used for cattle and large animals etc. but no estimates of these data
are available. Not accounting for these in land estimates probably overestimates TFP growth in agriculture.
13
privateunincorporatedenterprises(OSPUE).Then, usingthesecondadjustmentmethod
followed by Gollin,
4
I compute labor income shares for different sub sectors. Using the
share of each sub sector’s output in the output of the agricultural, industrial and service
sectors’ as weights, I construct weighted labor shares for these three sectors. The share
of capital income in the industrial and service sectors are deduced as residuals.
Theshareofrentalincomefromlandinagriculturalincomeistakentobe0.2(average
over the period 1980-1999) as reported by (Siv04). Consequently, the labor and capital
shares are rescaled to sum to 1 minus the share of land.
I also conduct a sensitivity analysis of the growth accounting results by using two
alternate sets of factor shares. The first set consists of sectoral labor shares computed
using the Global Trade Analysis Project (GTAP) data, and is reported by Terry Roe.
The second set assigns the customary value of one-third as the share of capital income
and treats the residual as the share of labor income in the industrial and service sectors.
For the agricultural sector, the capital income and labor income shares of one-third and
two-thirds, are rescaled so that they sum to 1 minus the share of land, where the share
of land is taken as 0.2.
2.1.4 Results
Table 2.1 reports the decomposition of average annual growth in real value added
due to change in capital, labor, land and TFP in each sector. These results have been
obtained using ‘baseline’ factor shares, calibrated from the CSO data. I refer to these
results as ‘baseline’ results.
For the agricultural sector, the labor income share is 0.58, the share of land is 0.2 and
the share of capital is determined residually as 0.22. The contribution of each factor is
measured as the product of the factor share with the growth rate of the factor. During
the 1980-2003 period, agricultural real value added grew at an average annual rate of
3.06 percent. The contributions of capital, labor and TFP were 19, 24 and 56 percent,
respectively. Land made a negligible contribution of 1 percent during the entire period.
In the pre liberalization period 1980-1990, real value added was growing at 4.23 percent,
of which TFP growth accounted for 52 percent. After TFP, the contribution of labor
was next largest at 29 percent, followed by capital which accounted for about 16 percent.
Land made a small contribution of 3 percent. In the post liberalization period, growth in
real value added decreased to about 2.02 percent and the contribution of TFP increased
4
Labor income share= Compensation of Employees/(Compensation of Employees+Operating Surplus of Incor-
porated Enterprise+Consumption of Fixed Capital)
14
to account for 66 percent of real value added growth. Capital and labor accounted for 26
and 13 percent of growth, respectively, whereas the contribution of land was small and
negative at -4 percent.
Table 2.1: Growth Accounting - Baseline Results
Agriculture Industry Services
Factor share
capital 0.22 0.55 0.44
labor 0.58 0.45 0.56
land 0.2
Decomposition of average annual changes in real value added (%)
Entire period 1980-2003
Growth in real value added 3.06 6.14 6.95
due to capital 0.58 3.77 1.92
(19.0) (61.4) (27.6)
due to labor 0.73 1.40 1.92
(23.7) (22.8) (27.6)
due to land 0.03
(1.1)
due to TFP 1.73 0.98 3.11
(56.3) (15.9) (44.8)
Pre liberalization 1980-1990
Growth in real value added 4.23 6.75 6.53
due to capital 0.68 4.08 1.49
(16.0) (60.5) (22.9)
due to labor 1.22 1.93 2.31
(28.8) (28.6) (35.3)
due to land 0.14
(3.4)
due to TFP 2.19 0.74 2.73
(51.8) (10.9) (41.8)
Post liberalization 1991-2003
Growth in real value added 2.02 5.52 7.44
due to capital 0.52 3.51 2.26
(25.9) (63.6) (30.3)
due to labor 0.26 1.05 1.58
(12.7) (19.0) (21.3)
due to land -0.09
(-4.4)
due to TFP 1.33 0.96 3.60
(65.9) (17.3) (48.4)
The number in parenthesis is the % contribution of the factor to real value added growth.
15
With respect to the industrial sector, the calibrated capital and labor shares are 0.55
and 0.45, respectively. Real value added in industry grew at 6.14 percent during the
entire 1980-2003 period. The contribution of capital was the largest at 61 percent while
that of laborwas about 23 percent. TFP in industry made a relativelysmallcontribution
of 16 percent during this period. In the pre liberalization period, real value added was
growing at 6.75 percent, of which capital made a significant contribution of 61 percent.
The contribution made by labor was 29 percent, followed by TFP which accounted for
a relatively small proportion, 11 percent. In the post liberalization period, growth of
industrialrealvalueaddedsloweddownto5.52percent. Again,thecontributionofcapital
was largest, accounting for about 64 percent of growth in real value added, followed by
labor which made a contribution of 19 percent. In this period, the contribution of TFP
increased to account for about 17 percent of real value added growth in this sector.
For the service sector, the shares of capital and labor income are calculated to be
0.44 and 0.56, respectively. During the 1980-2003 period, real value added grew at 6.95
percent, ofwhichTFPaccountedfor45percent, followedbycapitalandlaborwhicheach
accounted for about 28 percent of services’ value added growth, respectively. In the pre
liberalization period, real value added grew at 6.53 percent. The contributions of capital
and labor were 23 and 35 percent, respectively, while that of TFP was about 42 percent.
Inthepostliberalizationperiod,servicesectorrealvalueaddedgrewat7.44percent. The
contribution of capital increased to 30 percent while the contribution of labor decreased
to about 21 percent in this period. TFP’s contribution increased and TFP growth alone,
in this period accounted for 48 percent of real value added growth.
(BCV07) conduct sectoral growth accounting for the Indian economy and find similar
sectoral TFP growth rates for the 1980-2004 period. Their estimates of TFP growth
rates in agriculture, industry and services are 1.1, 1, and 2.9 percent respectively. They
do not calibrate factor shares but assume a capital share of 0.4 in industry and services.
For agriculture, the factor shares are 0.5, 0.25, and 0.25 for labor, capital and land
respectively. They have another factor input of human capital (education) in each sector.
In spite of this additional input, my estimates of TFP growth rates are similar to their
numbers, suggesting that education has not played a very significant role in contributing
to the growth of sectoral real value added.
From Table 2.1 one observes that the service sector in India has witnessed very rapid
TFP growth which exceeds TFP growth in the agricultural and industrial sectors for the
1980-2003 period, primarily because of the high growth it experienced in the 1991-2003
period. Figure 2.5 depicts the evolution of sectoral TFP from the initial time period,
16
1980, (the levels in all sectors have been normalized to unity) to 2003. It is further
evident from the graph that the service sector witnessed the fastest rate of TFP growth
throughout the sample period. In addition, the rate of TFP growth in services increased
after 1991.
Figure 2.5: Sectoral TFP Levels, 1980-2003
1980 1985 1990 1995 2000
0.8
1
1.2
1.4
1.6
1.8
2
2.2
Year
Agriculture
Industry
Services
Pre liberalization Post liberalization
In the Indian case, the finding of high TFP growth in services does not depend on
the values of factor shares. I report results using two other sets of factor shares. Table
2.2 reports the results using the GTAP computed sectoral factor shares and table 2.3
presents the results using capital share values of one-third in the sectors. These results
validate the finding that amongst the three sectors, TFP growth is highest in the service
sectorfortheentiresampleperiod, especiallyduetothehighgrowthobservedinthepost
liberalization period.
Table 2.2: Growth Accounting - GTAP Factor Shares
Agriculture Industry Services
Factor share
capital 0.21 0.61 0.5
labor 0.41 0.39 0.5
land 0.38
Decomposition of average annual changes in real value added (%)
Entire period 1980-2003
Growth in real value added 3.06 6.14 6.95
due to capital 0.55 4.18 2.18
(18.1) (68.0) (31.3)
17
Table 2.2: (continued)
Agriculture Industry Services
due to labor 0.51 1.20 1.72
(16.8) (19.5) (24.7)
due to land 0.06
(2.0)
due to TFP 1.93 0.76 3.05
(63.1) (12.4) (43.9)
Pre liberalization 1980-1990
Growth in real value added 4.23 6.75 6.53
due to capital 0.65 4.52 1.70
(15.3) (67.0) (26.0)
due to labor 0.86 1.67 2.06
(20.4) (24.7) (31.5)
due to land 0.27
(6.5)
due to TFP 2.45 0.56 2.77
(57.9) (8.3) (42.5)
Post liberalization 1991-2003
Growth in real value added 2.02 5.52 7.44
due to capital 0.51 3.90 2.56
(25.2) (70.6) (34.4)
due to labor 0.18 0.90 1.41
(8.9) (16.3) (19.0)
due to land -0.17
(-8.5)
due to TFP 1.50 0.72 3.47
(74.4) (13.1) (46.6)
The number in parenthesis is the % contribution of the factor to real value added growth.
18
Table 2.3: Growth Accounting - Capital Share of One-third
Agriculture Industry Services
Factor share
capital 0.24 0.3 0.3
labor 0.56 0.7 0.7
land 0.2
Decomposition of average annual changes in real value added (%)
Entire period 1980-2003
Growth in real value added 3.06 6.14 6.95
due to capital 0.63 2.07 1.31
(20.7) (33.7) (18.8)
due to labor 0.70 2.19 2.39
(22.9) (35.7) (34.5)
due to land 0.03
(1.1)
due to TFP 1.69 1.88 3.25
(55.3) (30.6) (46.8)
Pre liberalization 1980-1990
Growth in real value added 4.23 6.75 6.53
due to capital 0.74 2.25 1.02
(17.5) (33.3) (15.6)
due to labor 1.18 3.00 2.88
(27.8) (44.4) (44.1)
due to land 0.14
(3.4)
due to TFP 2.17 1.50 2.63
(51.2) (22.2) (40.3)
Post liberalization 1991-2003
Growth in real value added 2.02 5.52 7.44
due to capital 0.58 1.93 1.55
(28.7) (35.0) (20.8)
due to labor 0.25 1.65 1.98
(12.2) (29.9) (26.6)
due to land -0.09
(-4.4)
due to TFP 1.28 1.94 3.92
(63.5) (35.2) (52.6)
The number in parenthesis is the % contribution of the factor to real value added growth.
19
Chapter 3
Steady State Analysis
The objective of this chapter is to build a three sector quantitative model which
can account for the rapid growth in the share of services’ in India’s GDP. Motivated by
examining the sectoral data, I introduce two exogenous changes in the model: growth in
TFP in each of the three sectors, and, trade in industrial as well as the service sectors.
The model is assessed on how closely can it replicate the sectoral composition of GDP
and the allocation of sectoral labor across two steady state years, 1980 and 1999. In
addition, the importance of each of the two exogenous changes is tested by conducting
a counter factual experiment in which one change is allowed to operate while the other
change is shut down.
3.1 The Model
There are three final goods, consisting of agricultural goods, industrial goods, and
services, three primary factors - capital, labor, and land (in agriculture); and trade con-
sisting of exports and imports of industrial goods and services. In addition, there is total
factor productivity (TFP) growth in each sector and it is assumed that this growth rate
is constant over the sample period. The production function for each final good displays
constant returns to scale and is assumed to be Cobb-Douglas.
3.1.1 Technology
The model is set up in terms of per capita quantities for simplicity. Agricultural
goods are produced using capital k
a
, land l
a
, and labor n
a
as inputs; industrial goods
20
and services are produced using capital and labor,k
i
,n
i
,k
s
andn
s
respectively. Time is
discrete and on a per capita basis the production functions are
y
at
= b
at
k
θ
at
l
γ
at
n
1−θ−γ
at
(3.1)
y
it
= b
it
k
α
it
n
1−α
it
(3.2)
y
st
= b
st
k
φ
st
n
1−φ
st
(3.3)
where b
jt
is the TFP level in sector j = {a,i,s}. The parameters θ,γ,α,φ ∈ (0,1) and
θ+γ ≤1. It is assumed that all firms behave competitively in all markets.
There are three market clearing conditions for produced goods:
c
at
= y
at
(3.4)
c
it
+k
t+1
−(1−δ)k
t
+x
it
= y
it
+i
it
(3.5)
c
st
+x
st
= y
st
+i
st
(3.6)
wherec
jt
istheconsumptionlevelinsectorj ={a,i,s}andi
jt
andx
jt
aretheimportsand
exports in sector j = {i,s} respectively. These conditions imply that while agricultural
consumption is met entirely from domestic production, the sum of domestic output and
importsofservicesequalsthesumofdomesticconsumptionandexportsofservicestothe
restoftheworld. Intheindustrialsector,domesticoutputandimportsofindustrialgoods
together equal the sum of domestic consumption, investment and exports of industrial
goods to the rest of the world.
There are also three market clearing conditions for primary inputs:
k
at
+k
it
+k
st
= k
t
n
at
+n
it
+n
st
= 1
l
at
= 1
where labor supply per capita is normalized at unity and where l
at
is the supply of land
per capita, also normalized to unity.
There is a role for foreign trade. When calibrating the model, industrial net imports
(ni
it
=i
it
−x
it
) are fixed at a level chosen to match the data, and net exports of services
(nx
st
=x
st
−i
st
) are assumed to adjust. I make a simplifying assumption of trade being
21
balanced in the steady state. Hence this implies that the net exports of services needed
to pay for net imports of industrial goods are
p
st
nx
st
=p
it
ni
it
(3.7)
where the price of the industrial good,p
i
is given and assumed it to be unity at all dates.
Thenlet{r
kt
,R
lt
,w
t
,p
at
,p
st
,}denotetherentalpricesforcapitalandland,thewagerate,
thepriceoftheagriculturalgood,andthepriceoftheservicegood,atdatet,respectively.
3.1.2 Preferences
There is an infinitely-lived representative household endowed with one unit of time
in each period. The lifetime utility function for the household is given by
∞
X
t=0
β
t
U(c
at
,c
it
,c
st
)
wherec
j
istheconsumptionofgoodj (j =a,i,s)inperiodtandβ isthediscountfactor.
The per period utility function is given by
U(c
at
,c
it
,c
st
)=ln(ω
a
c
ǫ
at
+ω
i
c
ǫ
it
+ω
s
c
ǫ
st
)
(1/ǫ)
with ǫ< 1 and
P
ω
j=a,i,s
= 1. Thus, the elasticity of substitution between c
a
, c
i
and c
s
is given by
1
1−ǫ
.
The parameter, ǫ, plays an important role in generating structural change in mod-
els with differential TFP growth across sectors. Specifically, if consumption goods are
complements, then, in the presence of differential TFP growth across sectors, resources
are transferred to the sector experiencing the lowest TFP growth. But if consumption
goods are substitutes, then resources are allocated to the sector witnessing highest TFP
growth. The underlying reasoning is that the sector witnessing highest TFP growth
also experiences the most rapid decline in the price of the good that it produces. If
the goods are substitutes, the household increases its share of consumption expenditure
on this relatively cheap good, and reduces the share of expenditure on the other goods.
The household then demands more of the cheap good and reduces the demand for the
relatively expensive good. As a result, when the two goods are substitutes, labor shifts
into the sector where TFP growth is the highest. The converse is true when goods are
complements. Since the growth accounting resultsrevealTFP growth to be largestin the
22
service sector for India, and the data show that the output and employment of this sector
have grown, I assume ǫ is <1 and therefore assume that the three goods are substitutes
in consumption.
Therepresentativehouseholdfacesthefollowingmaximizationproblemineachperiod
max
∞
X
t=0
β
t
U(c
at
,c
it
,c
st
)
subject to
p
at
c
at
+c
it
+p
st
c
st
+k
t+1
−(1−δ)k
t
=r
kt
k
t
+w
t
n
t
+R
lt
l
at
+p
st
nx
st
−ni
it
∀ t=0,1,..∞
given k
0
, all prices, the net import level of industrial goods ni
i
.
The above equations, together with assumptions that firms maximize profits and
markets are perfectly competitive, provide a complete description of the model. The
numerical appendix discusses how to solve the model in steady state.
3.2 Calibration
The two years considered as steady states are 1980 and 1999. In these two years,
the value of net imports of industrial goods (as a share of GDP) was approximately equal
to the value of net exports of services (as a share of GDP). In other words, trade balance
as a share of GDP was roughly small (-0.2 percent in 1980 and 0.7 percent in 1999) and
hence I assume balanced trade in these two years
1
.
To calibrate the model, I fix the level of net imports of industrial goods from the data
and solve for the level of net exports of services, by using the balanced trade condition.
Factor shares for each sector have been constructed as explained in chapter 1. The TFP
growthratesforeachsectorhavebeentakenfromthebaselinegrowthaccountingexercise.
The subjective discount factor, β, is calibrated to match the real interest rate in 1980
and the depreciation rate is set at 5 percent.
The remaining parameters - TFP levels in the initial period - b
a0
,b
i0
,b
s0
; the weight
on the agricultural and industrial good in the utility function -ω
a
,ω
i
,; and the parameter
dictating the elasticity of substitution between the three goods - ǫ, are calibrated to
minimize the sum of squared differences between the data and the model with respect
1
Agricultural trade as a share of GDP was relatively small for the sample period and hence I assume no trade
takes place in this sector.
23
to six targets in the initial steady state. These six targets are - the share of output in
agriculture, the share of output in services, the share of employment in agriculture, the
share of employment in services, the share of consumption expenditure on services and
the relative price of the service good, all in 1980. Specifically, if ˆ y
a0
,ˆ y
s0
,ˆ n
a0
,ˆ n
s0
,
ˆ
CS
0
,ˆ p
s0
are the model’s prediction for the six targets andy
a0
,y
s0
,n
a0
,n
s0
,CS
0
,p
s0
are the actual
observations in the data, then I solve the following problem:
{b
a0
,b
i0
,b
s0
,ω
a
,ω
i
,ǫ}=arg min
{x,y,z}
X
{(ˆ y
a0
−y
a0
)
2
+(ˆ y
s0
−y
s0
)
2
+(ˆ n
a0
−n
a0
)
2
+(ˆ n
s0
−n
s0
)
2
+(
ˆ
CS
0
−CS
0
)
2
+(ˆ p
s0
−p
s0
)
2
}
(3.8)
In order to calibrate the above parameters, I need data on private final consumption
expenditure as well as relative prices of service goods. CMIE reports disaggregated data
for private final consumption expenditures. To construct sectoral consumption expendi-
ture, I group the disaggregated final consumption expenditures under the three sectors,
following (Ech97)
2
. Since the industrial good is assumed to be the numeraire in the
model, relative prices for the service goods are got by dividing the GDP deflator series
for services with that of industry.
The parameter values are listed in the table below.
Table 3.1: Calibrated Parameters
Parameters Description Values
θ capital share in agriculture 0.22
γ land share in agriculture 0.2
α capital share in industry 0.55
φ capital share in services 0.44
ba0 initial TFP level in agriculture 3.9
bi0 initial TFP level in industry 1.1
bs0 initial TFP level in services 1.9
gat growth rate of TFP in agriculture 0.0173
git growth rate of TFP in industry 0.0098
gst growth rate of TFP in services 0.0311
β discount factor 0.98
δ depreciation rate 0.05
ωa weight on agricultural good 0.44
ωi weight on industrial good 0.19
ωs weight on service good 0.37
1/(1−ǫ) elasticity of substitution 4.3
nii (1980) share of net ind. imports in GDP 0.0019
nii (1999) share of net ind. imports in GDP -0.015
2
Details of the classification methodology are provided in the data appendix.
24
3.3 Results
Table 3.2 reports the results for the two steady state years, 1980 & 1999. The
model’s predictions for the composition of output are good. The model predicts that 38
percent of output is attributable to agriculture in 1980, which is equivalent to what is
observed in the data. In the industrial sector, the model slightly over predicts the output
share at 29 percent while in the data the corresponding share is 24 percent. The share of
output accounted by services is about 38 percent in the data and the model’s prediction
for this share is at about 33 percent for the initial steady state.
By 1999, agriculture’s share of output reduces to about 28 percent, measuring closely
to the share of 26 percent seen in the data. The model allocates about 29 percent of
output in the second steady state to industry, roughly equal to the 27 percent observed
in the data. With respect to services, the share of output accounted by this sector is
about 42 percent as predicted by the model, a little less than the 46 percent seen in the
data.
Withrespecttotheallocationoflaborin1980,themodelallocatesthelargestshareof
labortoagriculturesimilarto what isobservedinthe data, although themodelestimates
this share at 41 percent while in the data the share is larger at 65 percent. The model
allocatesabout 25 percent oflabor to industry, a littlemorethan the 15percent observed
inthedata. Withrespecttotheshareoflaborintheservicesector,themodelpredictsthe
sharetobe34percentin1980, higherthanthe21percentobservedinthedata. Although
the model cannot precisely capture the quantitative shares observed in the data, it does
pick up the qualitative pattern of labor allocation i.e. largest share of employment is
accounted by agriculture, followed by services and then industry.
By the final steady state, the share of labor accounted by the agricultural sector
decreases to about 30 percent; less than its data counterpart of 57 percent. In both
the industrial and service sectors, the share of labor increases to 25 and 45 percent
respectively, although these shares are higher than their data counterparts - 18 percent
in industry and 26 percent in the service sector.
With respect to foreign trade, the model predicts the level of services’ net exports
to be small and slightly positive at 0.2 percent of GDP in 1980; in the data this share
is small and negative at -0.03 percent of GDP. By the final steady state in 1999, the
model predicts net exports of services to be negative and large, at 2.3 percent of GDP;
the corresponding share observed in the data is about -0.7 percent of GDP.
25
Table 3.2: Results for the Two Steady States
1980 DATA 1980 MODEL 1999 DATA 1999 MODEL
Composition of GDP
Share in Agriculture 0.38 0.38 0.26 0.28
Share in Industry 0.24 0.29 0.27 0.29
Share in Services 0.38 0.33 0.46 0.42
Allocation of labor
Share in Agriculture 0.65 0.41 0.57 0.30
Share in Industry 0.15 0.25 0.18 0.25
Share in Services 0.21 0.34 0.26 0.45
Share of service net exports in GDP -0.0003 0.002 -0.007 -0.023
3.4 Counterfactual Experiment
The objective of this chapter is to analyze the relative importance of two factors
observed in the data with respect to service sector growth: The significance of trade in
services that occurred after the trade liberalization in 1991 versus the high growth in
TFP in the service sector. In this respect, I conduct two counter factual experiments.
The first experiment allows TFP growth to take place in all three sectors but does not
allow any trade to occur in industry and services. The second experiment allows trade to
take place in the industrial and service sector but shuts down the growth of TFP in each
of the sectors. In each experiment we examine the model’s prediction for the change in
the composition of GDP as well as the change in labor allocation in the second steady
state 1999.
The results are displayed in tables 3.3. For ease of comparison, the data values and
the original model’s predictions for 1999 are reproduced in the table. A comparison
revealsthat laborallocationas wellascomposition ofGDPdoesnot changeas onemoves
from an environment of trade to no trade. Thus the absence of trade does not affect the
sectoral composition of GDP as well as labor allocation in each sector.
The above inference does not apply in the case where TFP growth in each sector is
ceased while trade takes place in the economy. In 1999, the model predicts a much larger
share of output accruing in the agricultural sector at the expense of the service sector
i.e. in the absence of productivity growth, the share of output in agriculture is about
38 percent, higher than data value of 26 percent. With respect to services, the share
of output accounted by this sector is about 33 percent, falling short of the 47 percent
observed in the data. The absence of TFP growth affects the industrial sector relatively
less. Industrial share of output is about 29 percent as compared to the data value of
27 percent. The model allocates the largest share of labor to agriculture at 41 percent,
26
followed by services (34 percent) and then industry (25 percent). The model predicts the
same value of net exports of services (as a share of GDP) at -0.02 percent as that seen in
the original model; this is higher than the share of -0.007 percent observed in the data.
Table 3.3: Counterfactual Experiment
1999 DATA 1999 ORIGINAL NO TRADE NO TFP
MODEL GROWTH
Composition of GDP
Share in Agriculture 0.26 0.28 0.28 0.38
Share in Industry 0.27 0.30 0.30 0.29
Share in Services 0.47 0.42 0.43 0.33
Allocation of labor
Share in Agriculture 0.57 0.30 0.30 0.41
Share in Industry 0.18 0.25 0.25 0.25
Share in Services 0.26 0.45 0.45 0.34
Share of service net exports in GDP -0.007 -0.02 0.00 -0.02
3.5 Conclusion
Following the economic liberalization in India, the service sector has gained promi-
nence in the economy as it accounts for the largest share of GDP and, also that the
share of this sector in GDP has been growing very rapidly. Empirical data reveal two
significant trends in the service sector following liberalization in 1991: growth in service
sectorproductivityandgrowthinservices’trade. Theobjectiveofthischapteristobuild
a simple three sector quantitative model which can capture the increase in the share of
service sector in GDP after liberalization. In particular, the model is assessed on how
closely can it replicate the composition of output and allocation of labor for the three
principal sectors of the Indian economy namely, agriculture, industry and services across
twosteadystateyears. Withinthecontextofthemodel, therearetwoexogenouschanges
that occur across the two steady states: growth in sectoral TFP and trade in industrial
and service sectors. A steady state is defined as a year in which the trade balance as a
share of GDP is closest to zero. The two years which meet this criterion are 1980 and
1999.
The model developed here is successful in replicating the shares of sectoral output
and the change in this sectoral composition across the two steady states. It can correctly
capture the direction of structural change as the economy transforms from a situation
where the agricultural sector dominates the GDP to a situation where the service sector
gains primary importance. It cannot capture the shares of labor allocation in the two
27
years. Although the model predicts a much larger level of net exports of services, it does
estimate a negative level of net exports, similar to what is observed in the data.
The second focus of my analysis is to identify the relative importance of TFP growth
versus trade in industry and services. This is done by shutting down one source of
exogenous change and letting the other operate solely on its own. The results from the
counterfactual reveal that shutting down sectoral TFP affects the ability of the model
to capture the data trends whereas the absence of trade negligibly affects the results.
The mechanical method of modeling trade limits the model’s ability to address trading
opportunitiesasasourceofgrowthandisanavenuetobeexploredinthefuture. However,
the simple quantitative model does show that TFP growth can replicate the sectoral
composition of output and hence is a better candidate to examine the model’s dynamic
performance.
28
Chapter 4
Dynamic Analysis
In this chapter, I extend the two year analysis to a twenty three year period rang-
ing from 1980 to 2003 with the objective of evaluating the quantitative performance of
differential sectoral TFP growth in accounting for the structural transformation of In-
dia. I develop a three sector growth model in which the agents are assumed to view
consumption of agricultural, industrial and service sector goods as gross substitutes, but
their preferences over goods are homothetic. The model displays ‘unbalanced growth’ in
which the aggregate output, the aggregate consumption and the aggregate capital-labor
ratio, grow at different rates. The model is calibrated to Indian data and is examined on
how closely can it track the shares of sectoral output and employment over the period
1980-2003. In addition, an experiment is carried out to assess the importance of the
liberalization policies of 1991 for the Indian service sector.
4.1 Model
4.1.1 Technology
I develop a three sector, dynamic general equilibrium model in which an infinitely
lived representative household owns all land, labor and capital and is endowed with one
unit of productive time. Therefore, the model is set up in terms of per capita quantities.
Time is discrete and is indexed by t=0,1,...∞.
There are three sectors in the economy, agriculture, industry and services. In each
sector, the production function exhibits constant returns to scale and is assumed to be
Cobb-Douglas in form. The agricultural good is produced using capital k
a
, land l
a
, and
labor n
a
as inputs; the industrial good and the service good are produced using capital
and labor, (k
i
, n
i
), (k
s
, n
s
), respectively. θ and γ are the shares of capital and land in
29
agricultural output, α and φ are the capital shares in industrial and services’ output,
respectively. Labor shares are deduced as residuals.
The firms in each sector are assumed to behave competitively. They rent capital,
labor and land from the representative agent at rates, r
k
,w and R
l
, respectively. In the
agricultural sector, the firm solves
max
kat,nat,lat
{p
at
y
at
−r
kt
k
at
−w
t
n
at
−R
lt
l
at
}
subject to
y
at
=b
at
k
θ
at
l
γ
at
n
1−θ−γ
at
, θ+γ ∈(0,1) (4.1)
In the industrial sector, the firm solves
max
k
it
,n
it
{y
it
−r
kt
k
it
−w
t
n
it
}
subject to
y
it
=b
it
k
α
it
n
1−α
it
, α∈(0,1) (4.2)
In the services sector, the firm solves
max
kst,nst
{p
st
y
st
−r
kt
k
st
−w
t
n
st
}
subject to
y
st
=b
st
k
φ
st
n
1−φ
st
, φ∈(0,1) (4.3)
where b
jt
is the TFP level in sector j ={a,i,s}.
There are three market clearing conditions for produced goods:
c
at
=y
at
(4.4)
c
it
+k
t+1
−(1−δ)k
t
=y
it
(4.5)
c
st
=y
st
(4.6)
The market clearing conditions for agricultural and service goods imply that output
produced in these sectors is consumed. The industrial good can either be consumed or it
can be used for investment, where δ >0 is the constant rate of depreciation.
30
There are also three market clearing conditions for primary inputs:
k
at
+k
it
+k
st
= k
t
n
at
+n
it
+n
st
= 1
l
at
= 1
where labor supply per capita andl
at
, the supply of land per capita, are each normalized
to unity
1
.
The industrial good is assumed to be the numeraire and its price is normalized to
unity at all dates. Then let {r
kt
,R
lt
,w
t
,p
at
,p
st
,} denote the rental prices for capital and
land, the wage rate, the price of the agricultural good, and the price of the service good,
at date t, respectively.
4.1.2 Preferences
There is an infinitely-lived representative household endowed with one unit of time
in each period. The lifetime utility function for the household is given by
∞
X
t=0
β
t
U(c
at
,c
it
,c
st
)
wherec
j
is the consumption of goodj ={a,i,s} in periodt andβ is the discount factor.
The per period utility function is given by
U(c
at
,c
it
,c
st
)=ln(ω
a
c
ǫ
at
+ω
i
c
ǫ
it
+ω
s
c
ǫ
st
)
(1/ǫ)
with ǫ< 1 and
P
ω
j=a,i,s
= 1. Thus, the elasticity of substitution between c
a
, c
i
and c
s
is given by
1
1−ǫ
.
The parameter, ǫ, plays an important role in generating structural change in mod-
els with differential TFP growth across sectors. Specifically, if consumption goods are
complements, then, in the presence of differential TFP growth across sectors, resources
are transferred to the sector experiencing the lowest TFP growth. But if consumption
goods are substitutes, then resources are allocated to the sector witnessing highest TFP
growth. The underlying reasoning is that the sector witnessing highest TFP growth
1
In the data, stock of agricultural land is virtually fixed, and increases by less than 4 percent over the twenty-
three year time interval. In comparison, agricultural capital grows by 82 percent and labor grows by more than
100 percent.
31
also experiences the most rapid decline in the price of the good that it produces. If
the goods are substitutes, the household increases its share of consumption expenditure
on this relatively cheap good, and reduces the share of expenditure on the other goods.
The household then demands more of the cheap good and reduces the demand for the
relatively expensive good. As a result, when the two goods are substitutes, labor shifts
into the sector where TFP growth is the highest. The converse is true when goods are
complements. Since the growth accounting resultsrevealTFP growth to be largestin the
service sector for India, and the data show that the output and employment of this sector
have grown, I assume ǫ is <1 and therefore assume that the three goods are substitutes
in consumption.
Therepresentativehouseholdfacesthefollowingmaximizationproblemineachperiod
max
∞
X
t=0
β
t
U(c
at
,c
it
,c
st
)
subject to
p
at
c
at
+c
it
+p
st
c
st
+k
t+1
−(1−δ)k
t
=r
kt
k
t
+w
t
+R
lt
∀ t=0,1,..∞
with k
0
given.
4.1.3 Competitive Equilibrium
Givenk
0
, an equilibrium is defined as a sequence of prices{r
kt
,R
lt
,w
t
,p
at
,p
st
}
∞
t=0
and allocations {k
at+1
,k
it+1
,k
st+1
,n
at
,n
it
,n
st
,c
at
,c
it
,c
st
,l
at
}
∞
t=0
such that
1. Given prices, the sequence {c
at
,c
it
,c
st
,n
at
,n
it
,n
st
,k
t+1
}
∞
t=0
solves the household’s
maximization problem;
2. Given prices, the sequence {k
at
,k
it
,k
st
,n
at
,n
it
,n
st
,l
at
}
∞
t=0
solves the firms’ maxi-
mization problem;
3. The markets for primary inputs and final goods clear.
Numerical appendix describes in detail how an equilibrium is computed for this economy.
4.1.4 Model of Unbalanced Growth with Structural Change
The model is characterized by existence of an aggregate unbalanced growth path
consistent with structural change.
32
The process of structural change has been studied by previous authors using two
classes of models. The first class of models uses non-homotheticities in preferences and
neutraltechnologicalchangeacrosssectors. Theunderlyingpremiseisthatifincomeelas-
ticitiesofdemandarenotunitary,thenaseconomiesgrowricher,reallocationofresources
across sectors will take place. Examples of these models are (Ech97) and (KRX01). The
second class of models emphasizes that differential productivity growth across sectors
can generate structural transformation even with homothetic preferences. This is done
by assuming that the elasticity of substitution between goods is different from unity and
authors like (Bau67) and (NP07) use this class of models. Yet others, like (Rog07),
use a hybrid of both classes of models: uneven technological change across sectors cou-
pled with non-homothetic preferences. Rogerson states that while uneven technological
changecangeneratereallocationacrossindustryandservices,non-homotheticpreferences
are required to enable the reallocation of resources out of agriculture.
In this chapter, I use homothetic preferences with the assumption that the goods of
three sectors are substitutes in consumption, and assume differential productivity growth
across sectors, as measured by the growth rate of TFP derived in the growth accounting.
It is important to note that in India’s case, the high income elasticity of demand for
services is empirically implausible
2
.
The model characterizes the transitional dynamics of an economy. Here, unbalanced
growth exists because of the assumption of different values of factor shares in the sectors
and presence of a fixed factor (land) in one sector (agriculture). The equations of mo-
tion for the state variable (k) and the control variable (c) of the aggregate economy are
3
k
t+1
k
t
=
b
it
k
α−1
t
ˆ
λ
α

1−
x
at
c
t
X
t
y
t
1−θ−γ
1−α
Ω
1
−
x
st
c
t
X
t
y
t
Ω
2

−
c
t
X
t
k
t
+(1−δ) (4.7)
c
t+1
c
t
=β

1+αb
it+1
k
α−1
t+1
ˆ
λ
α−1
−δ

(4.8)
where
ˆ
λ =

θ
1−θ−γ
1−α
α
n
at
+n
it
+
φ
1−φ
1−α
α
n
st

Ω
1
=

1−α
1−θ−γ
n
at
+n
it
+
1−α
1−φ
n
st

Ω
2
=

1−φ
1−θ−γ
n
at
+
1−φ
1−α
n
it
+n
st

2
I elaborate on this point in section 9.
3
These are formally derived in the numerical appendix.
33
In this economy, unbalanced growth is characterized by aggregate output, aggregate
consumption and aggregate capital-labor ratio, growing at different rates. Notably, if the
values of factor shares were the same across the sectors, and no fixed factor (land) is
used in production in agriculture, then this model exhibits a balanced growth path. The
equationsofmotionforthestatevariable(k)andthecontrolvariable(c)oftheaggregate
economy will accordingly, be
k
t+1
k
t
=b
it
k
α−1
t
−
c
t
k
t
+(1−δ)
c
t+1
c
t
=β

1+αb
it+1
k
α−1
t+1
−δ

In this economy, then, the aggregate capital-labor ratio, the aggregate consumption and
the aggregate output, all will grow at the rate of labor augmenting technological progress
in the manufacturing sector. This is the case discussed in Ngai and Pissarides (2007)
where there exists a saddlepath equilibrium and stationary solutions for the aggregate
consumption and the aggregate capital-labor ratio.
4.2 Calibration
4.2.1 Methodology
Inowassesswhetherthismodelcanreplicatethesectoraltransformationwitnessed
by the Indian economy between 1980-2003. In particular, I evaluate the performance of
the model in matching the quantitative changes in sectoral output and sectoral employ-
mentsharesobservedinthedata. Ialsoreporttheaverageannualgrowthratesofsectoral
output and sectoral employment shares implied by the model and compare them with
their data counterparts.
Each period in the model is assumed to be one year. The computational experiment
conducted is as follows. For each sector, I use the calibrated factor income shares and
sectoral TFP growth rates from the baseline growth accounting exercise. The subjective
discountfactor,β, istakenas0.98andthedepreciationrateissetat5percentineachpe-
riod. Theremainingparameters-TFPlevelsintheinitialperiod-b
a0
,b
i0
,b
s0
; theweight
on the agricultural and industrial good in the utility function -ω
a
,ω
i
,; and the parameter
dictating the elasticity of substitution between the three goods - ǫ, are calibrated in the
following manner.
34
Assuming the initial period to be 1980 (t = 0), I compute the equilibrium trajectory
of the model in each period and calibrate these six parameters to minimize the sum
of squared differences between the data and the model with respect to six targets in
the initial period. These six targets are - the share of output in agriculture, the share
of output in services, the share of employment in agriculture, the share of employment
in services, the share of consumption expenditure on services and the relative price of
the service good, all in 1980. Specifically, if ˆ y
a0
,ˆ y
s0
,ˆ n
a0
,ˆ n
s0
,
ˆ
CS
0
,ˆ p
s0
are the model’s
prediction for the six targets andy
a0
,y
s0
,n
a0
,n
s0
,CS
0
,p
s0
are the actual observations in
the data, then I solve the following problem:
{b
a0
,b
i0
,b
s0
,ω
a
,ω
i
,ǫ}=arg min
{x,y,z}
X
{(ˆ y
a0
−y
a0
)
2
+(ˆ y
s0
−y
s0
)
2
+(ˆ n
a0
−n
a0
)
2
+(ˆ n
s0
−n
s0
)
2
+(
ˆ
CS
0
−CS
0
)
2
+(ˆ p
s0
−p
s0
)
2
}
(4.9)
The parameter values are presented in table 4.1. The numerical algorithm used to
compute the equilibrium trajectory is described in the numerical appendix.
4.2.2 Data for Calibration
Asmentionedabove,toassistinthecalibrationofsomeparameters,Ineeddataon
private final consumption expenditure as well as relative prices of service goods. CMIE
reports disaggregated data for private final consumption expenditures. To construct sec-
toralconsumptionexpenditure,Igroupthedisaggregatedfinalconsumptionexpenditures
under the three sectors, following (Ech97)
4
. Since the industrial good is assumed to be
the numeraire in the model, relative prices for the service goods are got by dividing the
GDP deflator series for services with that of industry.
4.3 Results
The trends in sectoral output shares implied by the model and those observed in
the data are presented in Figure 4.2. In the agricultural sector, the model tracks the data
closely and can capture the declining share of agricultural output in GDP over the whole
sample period. With respect to industry, the model captures the increasing trend of
output share in this sector, although it over predicts the output share by a small amount
in each time period. For the service sector, the model tracks the data very closely for the
4
Details of the classification methodology are provided in the data appendix.
35
Table 4.1: Calibrated Parameters
Parameters Description Values
θ capital share in agriculture 0.22
γ land share in agriculture 0.2
α capital share in industry 0.55
φ capital share in services 0.44
ba0 initial TFP level in agriculture 3.9
bi0 initial TFP level in industry 1.1
bs0 initial TFP level in services 1.9
gat growth rate of TFP in agriculture 0.0173
git growth rate of TFP in industry 0.0098
gst growth rate of TFP in services 0.0311
β discount factor 0.98
δ depreciation rate 0.05
ωa weight on agricultural good 0.44
ωi weight on industrial good 0.19
ωs weight on service good 0.37
1/(1−ǫ) elasticity of substitution 4.3
entire period, although it underpredicts the output share by a small amount in the last
few years.
SectoralemploymentsharetrendsaredisplayedinFigure4.3. Themodelhasdifficulty
in matching the levels of sectoral employment shares for the entire period. The model
impliesthatthesharesofsectoraloutputandsharesofsectoralemploymentaresimilarin
magnitude
5
,but,intheIndiancase,oneobservesahugedivergenceintheircorresponding
values. In the data, one observes a declining share of agricultural employment, although
at a very slow pace. The model can capture the declining share of employment in this
sector, but it implies a faster movement of labor out from this sector when compared to
the trend seen in the data. With respect to industrial and service sectors employment
shares, the model cannot match the magnitude of the employment shares over the entire
period, although it captures the trend of increasing employment shares in both these
sectors.
To gain further insight into the performance of the model, I also calculate the average
annual growth rates of the shares of output and employment in each of the three sectors
for the given time period. The growth rates implied by the model and those calculated
from the data are displayed in table 4.2. The model implies that the share of agricultural
output declined at an average annual rate of −2.1 percent. The calculated growth rate
from the data is about −2.4 percent; the model can account for about 86 percent of the
decline observed in the data. With respect to the share of industrial output, the model
implies a growth of 0.4 percent, which is very close to the growth of 0.5 percent seen in
5
See Numerical Appendix, section on ‘Solving the Model’.
36
the data and therefore, the model can account for about 80 percent of the growth in the
share of industrial output seen in the data. For the service sector, the model indicates
that the share of this sector in total output increases at an average annual rate of 1.2
percent. This share grows at an average annual rate of 1.3 percent in the data, and
therefore the model can account for about 88 percent of the growth in the service sector
output share in the data. With respect to employment shares, the model predicts the
share of agriculture declines at an average annual rate of 2.1 percent, whereas, in the
data, the movement of labor is much slower at 0.82 percent. For the industrial sector,
the data implies that the share of employment grows at a rate of 1 percent, while the
model suggests the growth rate to be 0.4 percent. With respect to the service sector, the
growth rate of the share of employment is about 1.3 percent in the data and the model
comes very close to matching this growth, implying a growth rate of 1.2 percent.
Table 4.2: Average Annual Growth Rates (%), 1980-2003
Variable Data Model
Share of output in agriculture -2.4 -2.1
Share of output in industry 0.5 0.4
Share of output in services 1.3 1.2
Share of employment in agriculture -0.8 -2.1
Share of employment in industry 1.0 0.4
Share of employment in services 1.3 1.2
4.4 Effect of Liberalization
The growth accounting results indicate that there was a rapid increase in service
sector TFP in the post liberalization period in India. Table 4.3 reproduces the pre
and post liberalization sectoral TFP growth rates obtained from the growth accounting
exercise. These results show that there was a rapid increase in services’ TFP from 2.73
percent in the pre liberalization period to 3.6 percent in the post liberalization period. In
the agricultural sector, TFP growth slowed from 2.19 percent during 1980-1990 to 1.33
percent in the 1991-2003 period. Industrial TFP growth increased by a small amount
from0.74percentinthepreliberalizationperiodto0.96percentinthepostliberalization
period.
In order to assess the importance of the changes in TFP growth rates that occur
following the economic liberalization in 1991, I ask the following: What would be the
level and the growth rate of each sector’s share in aggregate output in the 1991-2003
period if TFP growth rate had not changed after 1991? To start with, I simulate the
37
Table 4.3: Pre and Post Liberalization TFP Growth Rates
TFP growth rate (%) Agriculture Industry Services
Pre liberalization 1980-1990 2.19 0.74 2.73
Post liberalization 1991-2003 1.33 0.96 3.6
model by assuming that the average annual growth rate of sectoral TFPs for the 1980-
2003 period is equal to the pre liberalization (1980-1990) average annual growth rate.
Then, I compare this economy with the one in which, I take into account the higher TFP
growth rates observed in post liberalization era. Thus, I simulate the model by using the
average pre liberalization sectoral TFP growth rates for the 1980-1990 period, and the
average post liberalization sectoral TFP growth rates for the 1991-2003 period.
Figures 4.4 depict the time paths of sectoral output shares under the two scenarios
and compare these with the trends observed in the data. The results are also presented
in terms of average annual growth rates in table 4.4. In the simulation in which I use
the pre liberalization TFP growth rates, the share of output in agriculture declines at a
rate of -0.9 percent during 1991-2003. This rate is lower than -3.3 percent, which is the
growth rate implied by the model when I account for the post liberalization TFP growth
rate, and also the growth rate seen in the data.
Table 4.4: Average Annual Growth Rates during 1991-2003
Data Model Model
using pre liberalization TFPs using post liberalization TFPs
Share of output in agriculture -3.3 -0.9 -3.3
Share of output in industry 0.0 0.1 0.0
Share of output in services 2.0 0.6 2.1
Withregardtothetrendintheshareofindustrialoutput,thereisnotmuchdifference
between the time paths implied by the model, in both simulations. In fact, the model
implies a small positive growth in the share of industrial output of 0.1 percent in the
first simulation. This occurs at the expense of lower growth in the share of output in the
service sector.
IntheabsenceoftheTFPgrowthrateincreaseafterliberalization,theshareofservice
sector output increases at a rate of 0.6 percent. The corresponding growth in service
sector output share when I allow for TFP growth rate to increase following liberalization,
is about 2 percent in the model as well as in the data. Without the increase in TFP
following liberalization, the model can account for only 32 percent of the growth in the
share of service sector output. This low growth in the output share of services stems
from the fact that resource reallocation from agriculture to services is slower, compared
38
to the scenario in which TFP growth rises after liberalization. This is because, in the
absenceofthechangeinTFPgrowthratesfollowingliberalization,thedifferentialinTFP
growthrates,whichistheprincipalfactorguidingtherelativepricedifferencebetweenthe
agricultural good and the service good, becomes less potent in generating the movement
of resources from agriculture to services. Consequently, the differential between the price
of the service good and the price of the agricultural good becomes smaller and hence, the
household is less willing to substitute into consuming the service good.
4.5 Explaining the Rapid Growth of Share of Services in
Indian GDP
A number of explanations have tried to account for the rapid growth of the service
sector share in Indian GDP after liberalization. In this section, I discuss each of the ar-
guments and also present mine. I argue that the liberalization policies adopted by India
from 1991, and especially the deregulation and privatization of business and communica-
tions services, explain the improvement in service sector TFP, and hence the dominance
of service sector activity in India’s GDP growth.
Splintering
One ‘supply side explanation’ discusses the role of splintering. Splintering involves
switching to a more service-input intensive method of organizing production, which can
arise as a result of increasing specialization as the economy matures. (GG04) use input-
output coefficients for the 1989/1990-1993/94 period to measure the usage of services by
agriculture and industry in the early 1990s. They find that splintering could have added
only about one fourth of one percentage point to annual services’ value added growth
during the early 1990s. Following an identical methodology, (Sin06) uses input-output
coefficients from the 1998-1999 data and finds that splintering makes no contribution to
services’ value added growth during the entire 1990-2000 period.
Demand
The ‘demand side explanation’ argues that an increase in the share of services’
output in GDP is due to rapid growth of final demand for services, resulting from a high
income elasticity of demand for services. (GG04) find that this argument has little merit
in the Indian case. They argue that prior to the 1990s, final consumption of services
39
was growing at a lower rate than output of services and, after 1990, the two grew at
roughly equivalent rates. Hence, the income elasticity argument could only hold if there
was a behavioral change in the 1990s and there is no aprior reason to expect this to have
occurred. Moreover, they reason that, if the demand side explanation was true, the price
of services relative to the overall price level in the economy should have increased. The
Indiandatarevealthatthisratioactuallydecreasedafter1991. Additionally, recentwork
by(FG96)hastendedtorejecttheincome-elasticdemandforservicesoverallbutconfirm
a wide range of income elasticity estimates (above and below unity) across different types
of services.
TFP
The above explanations have little merit in explaining rapid service sector growth
in India. Moreover, the growth accounting results show that changes in TFP were crucial
for driving growth in the service sector, especially after liberalization. The question
is whether and which of the liberalization reforms of 1991 were the mechanisms which
resulted in productivity growth in the service sector? The economic liberalization of
1991 involved a myriad of policy changes. Some of the important policy reforms included
tariff reductions, reduction in export controls, removal of quotas, entry of foreign direct
investment (FDI) in some sectors and deregulation and privatization in the service and
industrialsectors. Whichofthesepolicychanges,ifany,canbestexplaintherapidgrowth
of service sector productivity and service sector output in India?
Trade liberalization
Major policy changes carried out within the ambit of trade liberalization involved
tariff reductions, reduction in export controls, repeal of quotas and removal of import
licensing. Prior to 1991, India had very high tariff rates, with the aim of turning quota
rents into tariff revenues. Pre liberalization, about 439 items were subject to export
controls, but this number was brought down to about 296 in 1992 ((Pan04)).
Although much progress was made in liberalizing the trade regime in India, India
remained a relatively closed economy during much of the 1990s. (RS04) use a gravity
model and conclude that India became a ‘normal’ trader only by 2000 (for the 1980-1999
period the coefficient of openness on India was negative and significant), as compared to
China, for which trade was significant during the entire 1980-2000 period. The (Wor04)
reports that the average tariff rate in India (inclusive of customs duties and other general
40
and selective protective levies) in 2002-03 was still high, at 35 percent. With respect to
exports of services, there is no refutation of the fact that, as a share of service sector
GDP, these exports grew following trade liberalization. However, by 2003, service sector
exports were about 8 percent of services’ GDP, and about 4 percent of aggregate GDP.
Given how small these numbers are, an export-led growth hypothesis of service sector
growth is difficult to support.
FDI in services
(GG04) and (Sin06) discuss the role of FDI in the service sector, particularly its
growth in the telecommunication sector after liberalization. The channel through which
FDI and foreign technology spills over to domestic firms deserves some merit as an ex-
planation of enhancing productivity growth in this sector. However, while it is true, that
services in general and telecommunications in particular have been attracting a large
share of FDI, the share of FDI inflow in service sector GDP has been very small. The
Handbook of Industrial Policy and Statistics 2003-05 reports the values of FDI inflows in
various sub sectors of the economy. Table 4.5 reproduces these values for the sub sectors
in services and also reports the values of service sector GDP for the 1991-2003 period.
During 1991-2002, the cumulative share of service sector FDI inflows in service sector
GDP is 0.3 percent and falls to 0.2 percent by 2003. The small share of FDI inflows
in the service sector seems unlikely to account for the magnitudes of productivity and
output growth in the Indian service sector.
Table 4.5: FDI and GDP in Indian Services
1991-2002 2003
FDI inflows: Rs. Million Rs. Million
cumulative
Telecommunications 98,994.43 7,272.59
Financial and Non Financial Services 65,938.62 13,903.59
Consultancy Services 4,354.96 2,480.26
Hotel & Tourism 6,276.92 2,594.21
Trading 11,982.54 831.46
Total FDI in Services 187,547.47 27,082.11
Total GDP in Services 66,368,910 11,434,480
Share of FDI/GDP in Services 0.3% 0.2%
Education
Since services are assumed to be relatively skill intensive, one could argue that
education has a big role to play in driving growth in this sector. (BCV07) conduct
41
growth accounting for each of the three sectors (agriculture, industry and services) for
theIndianeconomybetween1960and2004. Ineachsector,outputisproducedbycapital,
labor, and, human capital measured as education. For the 1980-2004 period, they report
that TFP in services grows at an average annual rate of 3 percent, similar to what I
find. Their results indicate that the average annual growth of education as a factor of
production in the service sector is small at 0.4 percent and accounts for 14 percent of
services’ output growth, suggesting that education alone could not be the driving force
of productivity and output gains in this sector.
Deregulation and Privatization
Prior to liberalization, the service sector had been subject to heavy government
intervention. There was a conspicuous dominance of the public sector in the key sectors
of insurance, banking and telecommunications.
Figure 4.1: Share of Output of Services Subsectors in Aggregate GDP
1980 1985 1990 1995 2000 2003
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Retail Trade
Hotels & Restaurants
Railways
Other Transport
Storage
Communication
Banking & Insurance
Real Estate & Business Services
Public Administration
Other Services
Pre liberalization
Post liberalization
42
Following liberalization, there was an active deregulation of some sectors, and en-
try of private firms was allowed in the service sector. Prior to 1991, insurance was a
state monopoly. In 1999, the Indian Parliament passed the Insurance Regulatory and
Development Authority (IRDA) Bill, which established an Insurance Regulatory and De-
velopment Authority and permitted private sector participation in the insurance sector.
Similarly, the banking sector was opened up to allow private banks to operate, following
the recommendations of the Narasimhan Committee in 1991-92. Another sector which
witnessed massive growth in its output was telecommunications. Until early 1990s, this
sector was a state monopoly but with the creation of the National Telecommunications
Policy in 1994, the doors were opened to provide for cellular, as well as basic and value-
added, telephone services by the private sector. The Handbook of Industrial Policy and
Statistics 2003-2005 lists that in 2003, the share of public sector investment in commod-
ity producing enterprises as 60.36 percent, while the corresponding share in enterprises
rendering services was much less, at 35.6 percent.
Figure 4.1 shows how the shares of the services sub sectors in aggregate GDP grew
for the 1980-2003 period. Clearly, the telecommunication, and the banking and insurance
sectors have witnessed rapid growth after liberalization
6
. The growth in the share of
the telecommunication sector is particularly notable. In this sector, massive deregulation
as well as technological progress occurred and may have promoted the rapid growth of
output in a short span of time. Information technology, as a sub sector of activity, is a
part of business services. Further disaggregated data are not available to see how this
sector grew but (SS04) report that the share of this sector in GDP was about 1 percent
in late 1990s. Even though this sector, in itself, may not account for a large share of
Indian GDP, its large spillover effects to the other sectors has enabled much growth in
the telecommunication, the banking and the insurance sectors.
I conclude that deregulation, privatization and quite possibly - technological progress
promoted the growth of output in the service sector during the 1991-2003 period.
4.6 Conclusion
In this chapter, I account for the rapid growth of the share of services’ in India’s
GDP for the 1980-2003 period. Using the sectoral TFP series obtained from growth ac-
counting, I develop a three sector growth model to evaluate the quantitative performance
6
The growthin bankingand insurance started duringlate 1980s as some deregulation took place then, although
major reform followed after liberalization in 1991.
43
of differential TFP growth across sectors in accounting for the structural transformation
of India. In the model, agents are assumed to view consumption of agricultural, indus-
trial and service sector goods as gross substitutes, but their preferences over goods are
homothetic. The model displays ‘unbalanced growth’ in which the aggregate output, the
aggregate consumption and the aggregate capital-labor ratio, grow at different rates. For
the period 1980-2003, the model is calibrated to Indian data in which average rates of
TFP growth by sector from India are the primary inputs. This model performs well in
accounting for the evolution of value added shares of the three major sectors of economic
activity over the period 1980-2003. It also accounts well for the growth rates of the GDP
shares of all three major sectors of economic activity over this period - for the structural
transformation of GDP. However, the model cannot match the evolution of employment
shares, primarily because of the large differences between the shares of sectoral value
added and the shares of sectoral employment observed in the Indian data.
In the latter part of this chapter, an experiment is conducted to highlight the im-
portance of the post 1991 increase in service sector TFP. I find that the performance
of the model improves significantly when the post 1991 increase in service sector TFP
growth is accounted for. I argue that the increase in service sector TFP is a result of the
liberalization policies adopted by India. The economic liberalization that India initiated
in 1991, involved a myriad of policy changes consisting of tariff reductions, reduction in
export controls, removal of quotas, entry of FDI in some sectors, and deregulation and
privatizationintheserviceandindustrialsectors. Amongallthesereforms,Ifindthatthe
deregulation and privatization of business and communications services and quite possi-
bly technological progress, explain the rapid growth in service sector TFP, and hence the
dominance of service sector activity in India’s GDP growth.
44
Figure 4.2: Shares of Sectoral Output, 1980-2003
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Share of Output in Agriculture
Year
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Share of Output in Industry
Year
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Share of Output in Services
Year
model
data
45
Figure 4.3: Shares of Sectoral Employment, 1980-2003
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
0.6
Share of Employment in Agriculture
Year
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
0.6
Share of Employment in Industry
Year
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
0.6
Share of Employment in Services
Year
model
data
46
Figure 4.4: Effect of Liberalization - Shares of Sectoral Output
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Share of Output in Agriculture
Year
data
liberalization
no liberalization
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Share of Output in Industry
Year
1980 1985 1990 1995 2000
0.1
0.2
0.3
0.4
0.5
Share of Output in Services
Year
47
Bibliography
[Ban06a] Abhijit Banerjee. The Paradox of Indian Growth: A Comment on Kochar et
al. Journal of Monetary Economics, 53:1021–1026, July 2006.
[Ban06b] Rashmi Banga. Critical Issues in India’s Services-led Growth. India Resident
Mission Policy Brief Number 2, 2006.
[Bau67] W.J. Baumol. Macroeconomics of Unbalanced Growth: The Anatomy of an
Urban Crisis. The American Economic Review, 57:415–426, 1967.
[BCV07] Barry Bosworth, Susan M. Collins, and Arvind Virmani. Sources of Growth in
the Indian Economy. NBER Working Paper 12901, February 2007.
[BS04] B.B. Bhattacharya and S. Sakthivel. Economic Reforms and Jobless Growth in
Indiainthe1990s. Institute of Economic Growth, Working Paper No. 245,2004.
[Cena] Center for Monitoring Indian Economy Private Ltd., Bombay. Business beacon
version 2.0.
[Cenb] Central Statistical Organisation. Ministry of Planning and Programme Imple-
mentation, New Delhi, Government of India. National Accounts Statistics.
[Col07] Wilbur J. Coleman. Accomodating Emerging Giants. Working Paper, 2007.
[DV06] RobertDekleandGuillaumeVandenbroucke. AQuantitativeAnalysisofChina’s
Structural Transformation. Working Paper, 2006.
[Ech97] CristinaEchevarria.ChangesinSectoralCompositionAssociatedwithEconomic
Growth. International Economic Review, 38(2), 1997.
[FG96] R. Falvey and N. Gemmell. Are Services Income-Elastic? Some New Evidence.
Review of Income and Wealth, 42(3), September 1996.
[GG04] Jim Gordon and Poonam Gupta. Understanding India’s Services Revolution.
IMF Working Paper, WP/04/171, 2004.
[Gol02] Douglas Gollin. Getting Income Shares Right. Journal of Political Economy,
110(2), 2002.
[Gup02] S. P. Gupta. Report of the Special Group on Targeting Ten Million Employ-
ment Opportunities per year over the Tenth Plan Period. Government of India,
Planning Commission, 2002.
48
[KP02] Timothy Kehoe and Edward C. Prescott. Great Depressions of the Twentieth
Century. Review of Economic Dynamics, 5, 2002.
[KRX01] P. Kongsamut, S. Rebelo, and D. Xie. Beyond Balanced Growth. Review of
Economic Studies, 68:869–882, 2001.
[MR04] Kaivan Munshi and Mark Rosenzweig. Why is Mobility in India so Low? Social
Insurance, Inequality, and Growth. BREAD Working Paper 092, 2004.
[NP07] RachelNgaiandChristopherA.Pissarides. StructuralChangeinaMulti-Sector
Model of Growth. The American Economic Review, 97(1), 2007.
[Pan04] Arvind Panagariya. India’s Trade Reform: Progress, Impact and Future Strat-
egy. 2004.
[Pan06] Arvind Panagariya. Transforming India. 2006.
[Roe] Terry Roe. http://www.apec.umn.edu/faculty/troe/class3sec-mor.xls.
[Rog07] RichardRogerson. StructuralTransformationandtheDeteriorationofEuropean
Labor Market Outcomes. NBER Working Paper 12889, 2007.
[RS04] Dani Rodrik and Arvind Subramanian. From Hindu Growth to Productivity
Surge: The Mystery of the Indian Growth Transition. NBER Working Paper
10376, March 2004.
[Sin06] Nirvikar Singh. Services-Led Industrialization in India: Assessments and
Lessons. MPRA Paper 1273, University Library of Munich, Germany, August
2006.
[Siv04] S Sivasubramonian. The Sources of Economic Growth in India 1950-1 to 1999-
2000. Oxford University Press, 2004.
[SS04] Nirvikar Singh and T.N. Srinivasan. Indian Federalism, Economic Reform and
Globalization. Public Economics, (0412007), September 2004.
[Sto01] Nancy L Stokey. A Quantitative Model of the British Industrial Revolution,
1780-1850. Carnegie-Rochester Conference Series on Public Policy, pages 55–
109, 2001.
[Syr88] Moshe Syrquin. Patterns of Structural Change, volume 1 of The Handbook of
Development Economics. Amsterdam: Elsevier Science Publishers B.V., 1988.
[Vir04] Arvind Virmani. Sources of India’s Economic Growth: Trends in Total Factor
Productivity. Indian Council for Research on International Economic Relations,
Working Paper no. 131, 2004.
[WG69] Thomas Weiss and Robert Gallman. The service industries in the nineteenth
century. Studies in Income and Wealth, NBER, 34, 1969.
[Wor04] World Bank Report. Trade Policies in South Asia: An Overview. Report No.
29949, 2004.
49
Appendix A
Data Appendix
1. Classificationaccordingtocurrentdailystatusapproach(cds): Theactivitypattern
of people particularly in the unorganized sector is such that a person might be
pursuing more than one activity during a week and sometimes even during a day.
In the current daily status, upto two activity statuses were assigned to a person on
each day of the reference week. The unit of classification was thus half day in the
cds. In assigning the activity status on a day, a person was considered working for
the entire day if he had worked 4 hours or more during the day. If he had worked
one hour or more but less than 4 hours, he was considered working (employed)
for half day and seeking/available for work (unemployed) or not available for work
(not in labor force) for the other half day depending on whether he was seeking
/available for work or not. On the other hand, if a person was not engaged in any
work even for one hour but was seeking or available for work for 4 hours or more,
he was considered unemployed for the entire day. If he was available for work for
less than 4 hours only, he was considered unemployed for half day and not in labor
force for the other half of the day. A person who neither had any work to do nor
was available for work even for half of the day was considered not in labor force
for the entire day and was assigned one or two non-economic activity status codes.
The aggregate of person days classified under the different activity categories for
all the seven days gave the distribution of person days by activity category during
an average week over the survey period of one year.
2. Expenditure on agriculture goods includes food, beverages, pan & intoxicant, to-
bacco & its products. Expenditure on industry includes clothing & footwear, gross
rent, fuel & power, furniture and household, personal transport equipment and op-
eration of personal transport equipment. Expenditure on services includes other
services in furniture etc., medical care & health services, equipment, recreation,
education & cultural services, miscellaneous goods & services, hotels & restaurants,
& transport & communication minus the sum of personal transport equipment and
operation of personal transport equipment.
50
Appendix B
Numerical Appendix
B.0.1 Solving the Model
Firms
At time t, using the firms’ first order conditions and the assumption that capital and
labor are completely mobile across sectors, one gets
1
r
kt
=p
at
θb
at
k
θ−1
at
n
−γ
at
=αb
it
k
α−1
it
=p
st
φb
st
k
φ−1
st
w
t
=p
at
(1−θ−γ)b
at
k
θ
at
n
−γ
at
=(1−α)b
it
k
α
it
=p
st
(1−φ)b
st
k
φ
st
R
lt
=p
at
γb
at
k
θ
at
n
1−γ
at
Since
wt
r
kt
is equal across the three sectors, one obtains the following relation
w
t
r
kt
=
1−θ−γ
θ
k
at
=
1−α
α
k
it
=
1−φ
φ
k
st
(B.1)
which implies
k
at
=
θ
1−θ−γ
1−α
α
k
it
(B.2)
k
st
=
φ
1−φ
1−α
α
k
it
(B.3)
The equality of the marginal product of capital across sectors implies
p
st
=
α
φ
b
it
b
st
k
α−1
it
k
φ−1
st
which can be further simplified, after substituting value of k
st
from above, into
p
st
=

α
φ

φ

1−φ
1−α

φ−1
b
it
b
st
k
α−φ
it
(B.4)
1
Here k
j
is the capital-labor ratio in sector j ={a,i,s}.
51
Also, from the equality of the marginal product of capital, one gets
p
at
=
α
θ
b
it
b
at
k
α−1
it
k
θ−1
at
n
γ
at
which can be further simplified, after substituting value of k
at
from above, into
p
at
=

α
θ

θ

1−θ−γ
1−α

θ−1
b
it
b
at
k
α−θ
it
n
γ
at
(B.5)
Household
At time t, the household’s intra-temporal optimization between c
at
,c
st
and c
it
imply
c
at
=

ω
i
p
at
ω
a

(
1
ǫ−1
)
c
it
(B.6)
c
st
=

ω
i
p
st
ω
s

(
1
ǫ−1
)
c
it
(B.7)
c
st
=

ω
a
p
st
ω
s
p
at

(
1
ǫ−1
)
c
at
(B.8)
And the inter-temporal Euler equation is

c
it+1
c
it

ǫ−1
C
t
C
t+1
=
1
β(1+r
kt+1
−δ)
where C
t
is defined for convenience as (ω
a
c
ǫ
at
+ω
i
c
ǫ
it
+ω
s
c
ǫ
st
).
At any time t, the inter-temporal Euler equation, the intra-temporal optimization
equation betweenc
st
andc
at
and the resource constraint of the industrial sector are used
tosolveforthreeendogenousvariablesk
it+1
,n
at
andn
it
. Oncethesearedetermined,n
st
is
determined asn
st
=1−n
at
−n
it
,k
at
,k
st
are determined from equations (B.2) and (B.3),
p
at
, p
st
are obtained from equations (B.5) and (B.4), y
at
,y
it
,y
st
are determined from
equations of the production functions, c
at
,c
st
are known from the resource constraints
andc
it
is known from the household’s intra-temporal optimization condition betweenc
at
and c
it
.
In this economy, output shares are functions of employment shares. Define A as
output share in agriculture, I as output share in industry and S as output share in
services. Then, using equations (4.1), (4.2), (4.3), (B.2), (B.3), (B.4), and (B.5), one
obtains
A=
p
a
y
a
p
a
y
a
+y
i
+p
s
y
s
=

1−α
1−θ−γ

n
a

1−α
1−θ−γ
−
1−α
1−φ

n
a
+

α−φ
1−φ

n
i
+

1−α
1−φ

I =
y
i
p
a
y
a
+y
i
+p
s
y
s
=
n
i

1−α
1−θ−γ
−
1−α
1−φ

n
a
+

α−φ
1−φ

n
i
+

1−α
1−φ

52
S =
p
s
y
s
p
a
y
a
+y
i
+p
s
y
s
=

1−α
1−φ

1−n
a
−n
i

1−α
1−θ−γ
−
1−α
1−φ

n
a
+

α−φ
1−φ

n
i
+

1−α
1−φ

B.0.2 Steady State
When the economy is in a steady state, the inter-temporal Euler equation implies
r
k
=
1
β
−(1−δ)
Since in equilibrium the marginal product of capital is equal to the interest rate, from
the above equation one can determine k
i
in steady state.
k
i
=

r
k
αb
i
1
α−1
Once k
i
is known, k
a
,k
s
,p
s
and w can be determined from the relations described in
the previous section. Using the resource constraint in the industrial sector and the intra-
temporaloptimizationequationbetweenc
st
andc
at
, onecansolveforthetwoendogenous
variablesN
a
andN
i
. Once these are determined, all the other variables can be solved for
as described in the previous section.
B.0.3 Model Properties: Unbalanced Growth and Structural Change
The aggregate capital-labor ratio can be expressed as
k
t
=k
at
n
at
+k
it
n
it
+k
st
n
st
Since k
at
=
θ
1−θ−γ
1−α
α
k
it
= λ
a
k
it
and k
st
=
φ
1−φ
1−α
α
k
it
= λ
s
k
it
, the aggregate capital-
labor ratio can be re-written as
k
t
=
ˆ
λk
it
∀ t=1....∞ (B.9)
where
ˆ
λ=
θ
1−θ−γ
1−α
α
n
at
+n
it
+
φ
1−φ
1−α
α
n
st
The resource constraint in the industrial sector can be expressed as
b
it
k
α
it
n
it
=c
it
+k
t+1
−(1−δ)k
t
Using equation (B.9), the above can be re-written as
k
t+1
k
t
=
b
it
n
it
k
α−1
t
ˆ
λ
α
−
c
it
k
t
+(1−δ) (B.10)
Now, consider the aggregate per capita consumption expenditure, c
c
t
=p
at
c
at
+c
it
+p
st
c
st
53
Dividing through by c
it
, this can be expressed as
c
t
=c
it
X
t
where X
t
=x
at
+x
st
+1, x
at
=
patcat
c
it
and x
st
=
pstcst
c
it
Similarly, the aggregate per capita output y
t
is
y
t
=p
at
y
at
+y
it
+p
st
y
st
(B.11)
Using the expressions forp
at
from (B.5),p
st
from (B.4), the equations for the production
functions, and equations (B.2), (B.3) imply after some algebraic simplification
y
t
=b
it
k
α
it

1−α
1−θ−γ
n
at
+n
it
+
1−α
1−φ
n
st

which can be expressed as
y
t
=b
it
k
α
it
Ω
1
(B.12)
where Ω
1
=
h
1−α
1−θ−γ
n
at
+n
it
+
1−α
1−φ
n
st
i
Now p
at
c
at
= x
at
c
it
=
xatct
Xt
. Using equation (B.5) and the resource constraint for the
agriculture good, one can derive the following expression for n
at
n
at
=
x
at
c
t
X
t
b
it
k
α
it
1−θ−γ
1−α
Using (B.12), the above can be written as
n
at
=
x
at
c
t
X
t
y
t
1−θ−γ
1−α
Ω
1
In a similar manner, one can derive the expression for n
st
as
n
st
=
x
st
c
t
X
t
y
t
Ω
2
where Ω
2
=
h
1−φ
1−θ−γ
n
at
+
1−φ
1−α
n
it
+n
st
i
Then n
it
=1−n
at
−n
st
can be expressed as
n
it
=1−
x
at
c
t
X
t
y
t
1−θ−γ
1−α
Ω
1
−
x
st
c
t
X
t
y
t
Ω
2
Therefore, equation (B.10) can be expressed as
k
t+1
k
t
=
b
it
k
α−1
t
ˆ
λ
α

1−
x
at
c
t
X
t
y
t
1−θ−γ
1−α
Ω
1
−
x
st
c
t
X
t
y
t
Ω
2

−
c
t
X
t
k
t
+(1−δ) (B.13)
54
Next, consider the utility function
U(c
at
,c
it
,c
st
)=ln(ω
a
c
ǫ
at
+ω
i
c
ǫ
it
+ω
s
c
ǫ
st
)
(1/ǫ)
Define ψ
t
(·)=(ω
a
c
ǫ
at
+ω
i
c
ǫ
it
+ω
s
c
ǫ
st
)
(1/ǫ)
The Euler equation implies
u
it
=u
it+1
β(1+r
kt+1
−δ) (B.14)
2
Now
u
it
=
1
ψ
t
(·)
ψ
it
ψ
it
=ω
i
(
ψ
t
c
it
)
(1−ǫ)
(B.15)
3
ψ
t
(·) is homogeneous of degree one. Then, using the Euler’s theorem one can express
ψ
t
(·) =ψ
at
c
at
+ψ
it
c
it
+ψ
st
c
st
Note that
u
at
u
it
=
ψ
at
ψ
it
=p
at
This impliesψ
at
=p
at
ψ
it
. Similarlyψ
st
=p
st
ψ
it
. Thereforeψ
t
=(p
at
c
at
+c
it
+p
st
c
st
)ψ
it
or ψ
t
=c
t
ψ
it
. Using this in equation (B.15), one getsψ
it
=ω
i
(
ctψ
it
c
it
)
(1−ǫ)
. But c
t
=c
it
X
t
which implies ψ
it
=ω
i
1/ǫ
X
1−ǫ
ǫ
t
.
Hence, equation (B.14) can be written as
1
c
t
=
1
c
t+1
β(1+r
kt+1
−δ)
Using r
kt+1
=αb
it+1
k
α−1
it+1
, one arrives at
c
t+1
c
t
=β

1+αb
it+1
k
α−1
t+1
ˆ
λ
α−1
−δ

(B.16)
Equations (B.13) and (B.16) are the equations of the state and control variables of the
aggregate economy. From these equations, one can infer that the aggregate capital-labor
ratio, the aggregate consumption and the aggregate output all grow at different rates.
Thisisbecauseoftheassumptionofdifferentsectoralfactorsharesandpresenceofafixed
input, land, in agriculture. If one relaxes these assumptions, i.e - assumes same values of
factorsharesandabsenceofland,thenbalancedgrowthexistsinthiseconomy. Thisisthe
case discussed in Ngai and Pissarides (2007) where there exists a saddlepath equilibrium
and stationary solutions for the aggregate consumption and the aggregate capital-labor
2
u
it
is the time t marginal utility of consumption with respect to the industrial good.
3
ψ
it
is the time t derivate of ψt(·) with respect to the industrial good.
55
ratio. In their framework, the aggregate capital-labor ratio, the aggregate consumption
and the aggregate output, all grow at the rate of labor augmenting technological progress
in the manufacturing sector.
56
B.1 Numerical Algorithm
This section describes the numerical algorithm used to simulate the model. The model is
simulated for T = 50 periods
4
. The solution technique involves solving three equations,
the inter-temporal Euler equation, the intra-temporal optimization equations betweenc
st
and c
at
and the resource constraint of the industrial sector, for t=1 to T-1 periods.

c
it+1
(b
at+1
,b
it+1
,k
it+1
,n
at+1
)
c
it
(b
at
,b
it
,k
it
,n
at
)

ǫ−1
C
t
(b
at
,b
it
,b
st
,k
it
,n
at
,n
it
)
C
t+1
(b
at+1
,b
it+1
,b
st+1
,k
it+1
,n
at+1
,n
it+1
)
=
1
β(1+r
kt+1
(k
it+1
)−δ)
c
st
(b
st
,k
it
,n
at
,n
it
)=

ω
a
ω
s

(
1
ǫ−1
)
c
at
(b
at
,k
it
,n
at
)

p
st
(b
it
,b
st
,k
it
)
p
at
(b
at
,b
it
,k
it
,n
at
)

(
1
ǫ−1
)
c
it
(b
at
,b
it
,k
it
,n
at
)+k
t+1
(k
it+1
,n
at+1
,n
it+1
)−(1−δ)k
t
(k
it
,n
at
,n
it
)=y
it
(b
it
,k
it
,n
it
)
The objective is to determine the time paths of k
i
5
, n
a
and n
i
, using the system of
equations specified above. To initialize the algorithm, I guess a path for k
i
, n
a
and n
i
,
i.e. {k
it
}
T
t=1
,{n
at
}
T
t=1
,{n
it
}
T
t=1
. At any time t, the endogenous variables are: k
it+1
,n
at
and n
it
, given k
t
and exogenous paths of sectoral TFPs. Then, I use the solutions of the
above equations to update the time t values of the vectors: {k
i
}, {n
a
} and {n
i
}. This
process is carried out for t = 1,...,T −1 and in the end I get the updated time paths.
I compare these vectors with those of the starting guesses and check if the difference
between the two is smaller than a defined threshold value. If the difference exceeds the
threshold, the guess is replaced by the new paths. This process is repeated until the error
becomes smaller than the threshold criterion.
4
I report results obtained by simulating the model for 50 periods, but these results are similar to those obtained
when the model is simulated for 100 periods.
5
k
i
is the capital-labor ratio in industry.
57

Abstract (if available)

Abstract The objective of this dissertation is to explain the rapid growth of value added in the service sector in India, and, to examine the factors driving this services' led growth in the economy. In the first chapter of this dissertation, I record the empirical data facts on output, employment and trade, for the three principal sectors of the economy, namely, agriculture, industry and services. This is followed by a sectoral growth accounting exercise which I conduct for the 1980-2003 period. Two empirical facts emerge from the analysis: changes in total factor productivity (TFP) were the largest source of service sector value added growth in India and a sharp acceleration in industrial and services ' trade occurred following liberalization in 1991. Motivated by these findings, in the second chapter of this dissertation, I build a simple three sector growth model with two main inputs: growth in sectoral TFP and trade in industry and services. The model is calibrated to Indian sectoral data across two steady state years during which trade is balanced. The results from this chapter indicate that productivity growth versus trade has a more important role in capturing the sectoral composition of GDP in India. In the second chapter of my dissertation, I extend the steady state analysis and develop a three sector growth model to evaluate the quantitative performance of differential sectoral TFP growth in accounting for the structural transformation of India during the 1980-2003 period.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Explaining the cross-sectional distribution of law of one price deviations

PDF

Essays on business cycle volatility and global trade

PDF

Structural transformation and globalization

PDF

Essays on firm investment, innovation and productivity

PDF

Effects of eliminating the unfunded social security system in an economy with entrepreneurs

PDF

Essays in macroeconomics

PDF

Inter-temporal allocation of human capital and economic performance

PDF

Essays in macroeconomics and macro-finance

PDF

Essays on macroeconomics and income distribution

PDF

Growth and development in Africa: macroeconomic and microeconomic perspectives

PDF

Empirical essays on trade liberalization and export diversification

PDF

Essays on political economy of privatization

PDF

Essays on interest rate determination in open economies

PDF

Cost structure and regulation in the telecommunications industry

PDF

Trade, training, employment, and wages: evidence from the U.S. manufacturing industry

PDF

The murky risk of trade protectionism in an interconnected and uncertain global economy: a state, industrial, and regional analysis

PDF

Financial crises and trade policy in developing countries

PDF

Three essays in international macroeconomics and finance

PDF

Essays on economic modeling: spatial-temporal extensions and verification

PDF

Essay on monetary policy, macroprudential policy, and financial integration

Asset Metadata

Creator Verma, Rubina (author)

Core Title Growth, trade and structural change in low income industrializing economies

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Economics

Publication Date 07/30/2008

Defense Date 04/10/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag economic liberalization,India,OAI-PMH Harvest,structural transformation,TFP

Place Name India (countries)

Language English

Advisor Betts, Caroline M. (committee chair), Kim, Yong (committee member), Quadrini, Vincenzo (committee member)

Creator Email rubinave@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1374

Unique identifier UC177933

Identifier etd-Verma-20080730 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-193309 (legacy record id),usctheses-m1374 (legacy record id)

Legacy Identifier etd-Verma-20080730.pdf

Dmrecord 193309

Document Type Dissertation

Rights Verma, Rubina

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu