Economic Growth and Convergence in Selected South Asian and East Asian Countries. A Data Envelopment Analysis

Transcription

1 Economic Growth and Convergence in Selected South Asian and East Asian Countries A Data Envelopment Analysis

2 Icfai Books An Introduction Icfai Books is the initiative of the Icfai University Press to publish a series of books in the areas of finance, management and allied areas with a special focus on emerging and frontier areas. These books seek to provide, at one place, a retrospective as well as prospective view of the contemporary developments in the environment, with emphasis on general and specialized branches of knowledge and applications. The articles are organized in a sequential and logical way that makes reading continuous and helps the reader acquire a holistic view of the subject. This helps in strengthening the understanding of the subject better and also enables the readers stretch their thoughts beyond the content of the book. The series is designed to meet the requirements of executives, research scholars, academicians and students of professional programs. The Icfai University Press has published over 950 books in this series. For full details, readers are invited to visit our website:

3 Economic Growth and Convergence in Selected South Asian and East Asian Countries A Data Envelopment Analysis Dr. Somesh Kumar Mathur Icfai Books The Icfai University Press

4 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS Author: Dr. Somesh Mathur 2007 The Icfai University Press. All rights reserved. Although every care has been taken to avoid errors and omissions, this publication is being sold on the condition and understanding that the information given in this book is merely for reference and must not be taken as having authority of or binding in any way on the author, editor, publishers or sellers. Neither this book nor any part of it may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording or by any information storage or retrieval system, without prior permission in writing from the copyright holders. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Only the publishers can export this book from India. Infringement of this condition of sale will lead to civil and criminal prosecution. First Edition: 2007 Printed in India Published by The Icfai University Press 52, Nagarjuna Hills, Punjagutta Hyderabad, India Phone: (+91) (040) , 369, 370, 372, 373, 374 Fax: (+91) (040) , info@icfaibooks.com, icfaibooks@icfai.org, ssd@icfai.org Icfai Editorial Team: Editorial Co-ordinator : Ved Prakash Sub-editor : K Prabhakar Visualizer : Ch Yugandhar Rao Designers : A G Bhaskar and K Malleswara Rao ISBN:

5 For my family, friends, colleagues and teachers

6

7 About the Author Dr. Somesh Kumar Mathur has nearly eleven years of teaching and research experience at the Department of Economics, Jamia Millia Islamia (Central University), Delhi. While teaching at the Jamia, he completed his M Phil and PhD degrees in economics from the Centre for International Trade and Development, JNU. He has joined RIS as Fellow in April, 2006 on deputation from the Department of Economics, JMI, for a term of two years. His areas of interest are in new trade and growth theories, TRIPS and other WTO issues. He has participated in various national and international conferences and has published articles etc., in refereed national and international journals. Dr. Mathur has taught papers like Pure Theory of International Trade, Quantitative Methods, International Finance and Banking, Microeconomics and Corporate Finance to the postgraduate students of the University.

8

9 Contents Preface Overview I III 1. Introduction 1 2. Review of Literature: Growth Theories, Previous Studies on Economic Growth and Convergence, Data Envelopment Analysis with its Uses and Criticism Objectives of the Study and Methodology Hypotheses and Data Source with Variable Description Convergence Analysis: Neoclassical Approach and Growth Model Providing Theoretical Justification for Per Capita Growth Equation Discussion of Results Time Series Approach to Convergence Industrial Sector Growth Accounting in Some Indian States and Union Territories: A Data Envelopment Analysis 172

10 9. Limitations of the Study Conclusions, Policy Implications and Directions for Future Research 201 Appendix (Tables) 209 Bibliography 271 Index 287

11 Preface The study arose out of my efforts to understand the relevance of different growth theories from neoclassical to endogenous in analyzing the economic growth experiences of some South Asian and East Asian economies since 1960s. Also, we try to address the core inadequacies in the system like persistent inequality of incomes across nations and within the Indian state despite many years of development. We further work out different determinants of economic growth of the selected countries along with their technical efficiencies using the method of the Data Envelopment Analysis from 1960 onwards. In the end, we do growth accounting with a twist, without assuming any production function or market structure not only for the sampled countries in our study but also for some states and union territories of the Indian Union. This is my second endeavor in bringing out a book with Icfai University Press. I am privileged to have an opportunity to work in close quarters with the Icfai University Press, Hyderabad. This publication will cater to the large number of students and practitioners. We sincerely thank the Icfai University Press for such an initiative and

12 II look forward to work closely with the organization to bring out a few more books on various aspects of economy. There are many people who have shown their helping hand at Icfai during the execution of this project. I would like to make a special mention of a few people to whom we are really indebted. I sincerely thank Prof. N Rajshekar, Dean, the Icfai University Press, Mr. Koshy Verghese, Director, Icfai Research Center, Mr. Ved Prakash, Associate Consultant, Icfai Books and the whole of the editorial team for their kind help and support in bringing out this volume. My sincere thanks to Icfai! Somesh Kumar Mathur.

13 III OVERVIEW Chapter 1 introduces one of the stylized facts of economic growth, presence of growth and differential level of standard of living in terms of per capita income till date across some South Asian and East Asian nations. The chapter gives some data to explain the point, while growth rates have differed across nations; the growth rates of the South Asian nations are benign, robust and upward looking, while those of East Asian nations which grew at a much faster rate than any other region from (till they faced the banking and currency crises), growth rates are stable and healthier. The EU nations, though they had and still have relatively higher per capita income levels across the nations, have shown declining growth rates since 1960s and have had to face the difficult time of aligning themselves with ten new entrants (countries) in the EU region since March Also, the chapter uses kernel densities to explain that, from

14 IV , the distribution of labour productivity was transformed from a trimodal distribution in 1966 into a bimodal distribution in 2000, depicting that the world has been divided into categories now: the rich and the poor. One of the chapters will test the hypothesis whether higher economic growth is leading to a process, where the laggards of 1960s can catch up with their rich counterparts and, if so, in how many years. Chapter 2 reviews the literature on previous studies of economic growth and convergence hypotheses. The chapter also discusses the method of data envelopment analysis (along with its criticism), which works out the technical efficiency of the decision making units (DMUs). Technical efficiency is defined as to how judicious are DMUs in transforming inputs into outputs. The chapter outlines the basic tenets of the growth theories, right from classical growth theories put forward by the classical economists like Adam Smith, David Ricardo, Thomas Malthus, Frank Ramsey, Frank Knight, Joseph Schumpeter, Karl Marx, among others to neoclassical growth theory of Solow and Swan and the recent endogenous growth literature by Lucas junior, Romer, Srinivasan and others. Chapter 3 gives the main objectives of the study and discusses the methodology of testing absolute and conditional convergence, the data envelopment analysis and growth accounting with a new twist without assuming any production function and market structure. Index numbers are used to theoretically decompose labour productivity growth into its Tripartite and Quadripartite decomposition into capital accumulation, technical change, efficiency change and human capital accumulation. The chapter also outlines a model of the data envelopment analysis using two inputs and one output.

15 V Chapter 4 outlines the hypotheses and data source with variable description. The hypotheses pertains to absolute and conditional convergence and its speed, time series convergence using unit root literature, tripartite and quadripartite decomposition of labour productivity across nations and states of India and on the technical efficiency of the sampled countries and Indian states, over time. Chapter 5 gives an extended neoclassical model of Solow with additional assumptions on endogenizing technical progress. The same model is used to theoretically justify and derive the per capita growth equation and all its right-hand side variables. The same equation is used to test for convergence hypothesis and its speed using linear and non-linear least squares method. Chapter 6 discusses the regression and growth accounting results at length and on the same basis some policy conclusions are suggested so that some important insights are developed on growth, efficiency and convergence process across some South Asian, East Asian, EU and some CIS countries over time. Chapter 7 is on time series approach to convergence using time series analysis. Chapter 8 does industrial sector growth accounting and works out technical efficiency in some Indian States and Union Territories, using the data envelopment analysis. Some suggestions are given at the end on how some laggard industrial states can catch up with their richer counter parts. Chapter 9 gives limitations of the study and Chapter 10 provides conclusions, policy implications and directions for future research.

16 Introduction 1 1 Introduction There has been considerable research inquiry into the causes and nature of differences in growth rates across countries and regions over time. Even small differences in these growth rates, if cumulated over a long period of time, may have substantial impact on the standards of living of people. Despite considerable research on the subject, cross-country and cross-regional income disparities are on the rise over time. Understanding the causes behind such inequalities is essential to formulate appropriate policies and bring about the required institutional changes in order to spread the benefits of growth processes across regions. Convergence refers to the process by which relatively poorer regions or countries grow faster than their rich counterparts. Few subjects in applied economic research have been studied as extensively as the convergence hypothesis advanced by Solow (1956) and documented by Baumol (1986) and Barro and Xavier Sala-i-Martin (1995). In simple terms, the hypothesis states that poor countries or regions tend to grow faster than rich ones. In its strongest version (known as absolute convergence), an implication of this hypothesis is that, in the long run, countries or regions should not only grow

17 2 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS at the same rate, but also reach the same income per capita. The present study tests the convergence of GDP per capita within and across four regions South and East Asia, European Union (15+United Kingdom) from and for some CIS countries since 1966 till 2001 and for other Commonwealth of Independent States (CIS states) from mid-1980s 1 to Convergence can be conditional (conditional beta convergence) or unconditional (absolute beta convergence). Conditional convergence implies that a country or a region is converging to its own steady state, while the unconditional convergence implies that all countries or regions are converging to a common steady state. These hypotheses have been tested by many researchers using different methodologies and data sets and appear to have been strongly rejected by some data sets and accepted by others. In view of these results, several modifications of the absolute and conditional convergence hypothesis have been advanced and tested. However, they usually lack both theoretical foundations and econometric rigor and discipline. Our study gives the theoretical foundation to the growth equation. Using Solovian model and its assumptions on factor accumulation, we derive the steady state level of per capita income. Linearizing the model around the steady state, we derive growth rate of income per-effective labor as a function of speed of convergence (the rate at which country/region reaches the common steady state position of per capita income levels in a year) and the gap between steady state (potential level) and actual level of per capita income. This differential equation is solved further and technological progress is endogenized to derive the estimating growth equation. This equation relates growth rate of income per capita as a function of steady state level of income (which, in turn, is a function of savings rate, technological progress, rate of growth 1 Please see Appendix Table I for the list of countries included in the four regions included in our study.

18 Introduction 3 of population, and depreciation rate) and initial level of GDP per capita, share of industry in GDP, human capital and trade openness. Speed of absolute convergence is estimated for all regions included in our study. One of the stylized facts of economic growth today is that the levels of GDP per capita and growth rates have differed across countries and regions of the world. This is indeed the case for the regions included in our study (Table I). While the EU region (industrialized economies) has relatively the highest GDP per capita income levels in 2001, East Asian region has shown relatively higher growth rates in all periods from 1961 to Decadal growth rates, however, do show that the performance of the EU and East Asian region shows a declining trend in economic growth rates, while that of the South Asian region shows an upward trend. As evident from factual data, economic growth varies tremendously across different regions. With the new era of free market philosophy, countries are competing with one another for resources. How are the countries doing relative to one another? Have they been diverging away from one another? These are critical questions for three reasons: (1) Central planning in Soviet Union (Now Russian Federation and CIS countries), China and India has explicitly sought to reduce regional disparities. Also, with 10 new East European and Baltic states joining the EU on May, , 2 2 Poland, Hungary, Lithuania, Estonia, Latvia, Slovakia, Slovenia, Cyprus, Malta and the Czech Republic are new entrants to EU. In practice, according to Eurostat figures, average per capita GDP in the 10 new entrant East European countries now stands at around 40 percent of the EU average. Indeed, the gap between the average EU income level and that of the new entrant countries has widened considerably since The EU s real GDP grew by 30 percent between 1989 and 2002, whereas for the 10 East European accession countries the increase amounted to only 8 percent during the same period. The widening of the gap is mainly attributable to the sharp fall in GDP in most of the region immediately after the shift from central planning. Lowering the gap has not diminished appreciably even since the beginning of growth, generally in the mid-1990s, in part because of macroeconomic mismanagement in some countries; slowing structural change in others; and the impact of external shocks, such as the 1998 Russian financial crisis.

19 4 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS Table I: Per Capita Income Levels and Growth Rates Across Regions ( ) Average GDP Per Capita (In Current US$) in Average Annual Per Capita GDP Growth Rate South Asia (5) East Asia(8) 3, European 21, Union (16) CIS (15) 1, (Latvia, Russia and Georgia) ( ) (Estonia & ( ) Moldova) 10 CIS countries excluding Latvia, Russia, Georgia, Estonia and Moldova Note: For calculating growth rates, figures for GDP per capita (in constant 1995 US$) are used from World Bank World Development Indicators on CD ROM, The first column shows weighted average with population as weights. Source: Author s calculations by using GDP per capita (constant 1995 US$) data and GDP (Current US$) and population figures from World Bank Development Indicators on CD ROM. reducing regional inequalities within the EU would be the explicit goal of the EU enlargement policies, (2) Rising regional disparities cause regional tensions, and (3) Poor regions should not remain poor for generations to come. 3 3 Has the globalization process had any impact on the poor? Mathur (2004) finds that there is unambiguous empirical evidence from economies around the globe and for some of the Asian economies included in the sample that trade openness promotes economic growth. Raising economic growth in a sustained manner reduces poverty. Further, for cross-section of fourteen Asian economies included in the study, no significant relationship could be found between changes in Contd...

20 Introduction 5 Some pertinent questions as mentioned above are whether relatively backward regions could catch up with their richer counterparts (absolute convergence) and whether countries which are further away from their steady state level of GDP per capita can grow at a faster rate (conditional convergence). Such analysis is attempted in the present study by testing the convergence hypotheses for the four regions using cross-sectional (on countries and regions) and time series data of GDP per capita growth rates and initial level of incomes from 1960 onwards. The other reason for studying convergence is to test the relevance of the neoclassical model by Solow in analyzing the economic growth experiences of the regions included in our study. We would also examine why even after so many years of development, there exist gaps in terms of labor productivities between developed and developing world. We use kernel densities and Data Envelopment framework to examine the latter question. Very much in the spirit of Quah s (1993, 1996b, 1997) suggested approach (also adopted by Galor [1996] and Jones [1997]), we analyze the evolution of the entire distribution of the four growth factors: technological change, technological catch-up, capital accumulation and human capital accumulation. 4 Quah has argued compellingly that analyses based on standard regression inequality and poverty, and inequality of incomes with economic growth rates and trade openness. However, most of the poor in the developing economies are in the agricultural sector and therefore raising growth in the agricultural sector is essential ingredient for making the reform process successful. The paper closely examines the benefits of the globalization process as manifested by the unilateral trade liberalization efforts of some South Asian countries included in our study. Some guidelines are given for making trade work for the poor after identifying how the trade regime works. These provide a benchmark against which to judge the prevailing trade regime and provide guidance for the direction of reforms for the poor in future. The objective of this study is, however, different. 4 This approach to Growth Accounting is not dependent on particular assumptions about the technology, market structure, technological change and other aspects of the growth process.

21 6 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS methods focusing on first moments of the distribution cannot adequately address the convergence issue. These arguments are buttressed by the empirical analyses of Quah and others posing a robust stylized fact about the international growth pattern that begs for explanation. A plot of the distribution of output per worker across 29 countries consisting of 5 South Asian, 8 East Asian and 16 EU countries (country names are given in Appendix Table I at the end) in 2000 and 1966 appears in Figures I and II, respectively. (The data and the kernel-based method of smoothing the distribution are described in the chapter on objectives and methodology). Over this 34-year period, the distribution of labour productivity was transformed from a tri-modal distribution in 1966 into a bimodal distribution in 2000 with a higher mean 5 (data on output per worker is available in Appendix Tables XIII and XIV below column II and column III). This transformation, in turn, means that while in 1966 there were countries in the lower, middle income and upper income groups, in 2000 the world had become divided, as a stylized fact, into two categories: the rich and the poor. It seems that almost all of the East Asian economies have joined the elite rich group. Quah (1996a, b, 1997) refers to this phenomenon as two-club, or twin-peak convergence, a phenomenon that renders suspect analyses based on the first moment (or even higher moments) of this distribution. South Asian countries have remained at the lower end of the distribution. Our 5 Sample Kolmogorov- Smirnov Test (non parametric test) is used to test whether two sets of observations could reasonably have come from the same distribution. This test assumes that the samples are random samples, the two samples are mutually independent, and the data are measured on at least an ordinal scale. In addition, the test gives exact results only if the underlying distributions are continuous. data: x: output per worker in 1966 (Continuous), and y: output per worker in 2000(Continous) ks = , p-value = alternative hypothesis: cdf (cumulative density function) of x: output per worker in 1966 does not equal the cdf of y: output per worker in 2000 for at least one sample point. We conclude from the test that two sample probability distributions of output per worker in 1966 and 2000 are indeed different statistically.

22 Introduction 7 Figure I: Distribution of Output Per Worker, 2000 (Bimodal) 1.5 *10 ^-5 Density of x t 10^ -5 5*1 0^ ,000 40,000 60,000 80,000 1,00,000 1,20,000 x values Figure II: Distribution of Output Per Worker, 1966 (Trimodal) Density of x 1.5*10^-5 2.5*10^-5 3.5*10^ ,000 20,000 30,000 40,000 x values

23 8 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS analysis is aimed at explaining this bipolarization of the distribution of output per worker, as well as its growth pattern, in terms of the tripartite and quadripartite decomposition described below. As such, it builds upon Quah s insights about the need to examine the dynamics of the entire cross-section distribution (Quah, 1997, p.29). In this study, we will further identify policies which may reduce differential levels of per capita income levels and growth rates of regions and work out the reasons for the existence of bimodal distribution of per capita income across countries. Also, related to the concept of labour productivity is the concept of efficiency, i.e., amount by which outputs can be increased without requiring extra inputs. We will also work out the efficiency levels of countries included in our sample and prefectures of the Indian federation by using linear programming method of data envelopment analysis. In a nutshell, in this study, we will provide theoretical justification for the per capita growth equation, test for absolute and conditional convergence across and within regions, work out speed of absolute and conditional convergence and identify policies which may reduce differential levels of per capita income levels and growth rates of regions, calculate technical efficiency index across countries and Indian States and do growth accounting of labour productivity changes from and give some reasons for the now existing bimodal distribution of per capita income across countries. Also, to understand the regional economic performance, we do industrial sector growth accounting for the Indian States. Neoclassical growth models (Cass 1965; Koopmans 1965; Solow 1956, Swan 1957) have been used as a framework to study convergence across regions within countries. The main variable in use will be GDP calculated at constant 1995 US$, capital (calculated at constant 1995 US$) labour, life expectancy in years (proxy for human capital), GDP per capita (GDP divided by population)

24 Introduction 9 and labour productivity (GDP divided by labour force) prevailing in different countries/regions included in our study. The study is organized as follows: Chapter 2 presents a review of the literature of neoclassical and endogenous growth theories, provides empirics of growth and convergence, gives an account of the data envelope analysis for measuring technical efficiency; Chapter 3 states the objectives and the methodology; Chapter 4 is on hypotheses and data sources; Chapter 5 provides the definition of absolute and conditional convergence within the neoclassical framework and gives the theoretical derivation of the per capita growth equation; Chapter 6 discusses the regression results of absolute and conditional convergence analyses along with the growth accounting results; Chapter 7 discusses the time series approach to convergence; Chapter 8 is on the industrial sector growth accounting of the Indian States; Chapter 9 is on limitations of the study; and Chapter 10 provides conclusions, policy implications and directions for future research.

25 10 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS 2 Review of Literature: Growth Theories, Previous Studies on Economic Growth and Convergence, Data Envelopment Analysis with its Uses and Criticism II.I From Classical and Neoclassical Growth Theories to Endogenous Growth Theories In our study, we combine features of exogenous and endogenous growth models to provide theoretical justification for all variables included in our per capita growth equation (see Chapter V). We describe below the chronological description of the development of all growth theories from eighteenth century till date. Basic ideas of modern growth theories are found in the celebrated works of classical economists such as Adam Smith (1776), David Ricardo (1817), Thomas Malthus (1798), Frank Ramsey (1928), Allyn Young (1928), Frank Knight (1944), and Joseph Schumpeter

26 Review of Literature: Growth Theories, Previous Studies (1934), among others. These ideas include the basic approaches of competitive behavior and equilibrium dynamics, the role of diminishing returns and its relation to the accumulation of physical and human capital, the interplay between per capita income and the growth rate of population, the effects of technological progress in the form of increased specialization of labor and discoveries of new goods and methods of production, and the role of monopoly powers as an incentive for technological advance. From a chronological viewpoint, the starting point was the classic articles by Ramsey (1928) and Fisher (1930). They introduced optimality conditions which are used in consumption theory, asset pricing and theories of business cycles. Ramsey intertemporally separable utility function is used widely in applied economics. Harrod (1939) and Domar (1946) provided the dynamic extension of the Keynesian model by integrating Keynesian analysis with elements of economic growth. They raised the issue of stability of growth path that would be consistent with maintaining the savinginvestment equilibrium and the natural growth rate as determined by the growth of the labor force and technical change. In these models, unless the economy s behavioral and technical parameters keep it on the knife-edge of equality between warranted and natural growth rates, there would be either growing underutilization of capacity if warranted rate exceeds the natural rate or growing unemployment if the natural rate exceeds the warranted rate. This problem resulted from the assumption that capital and labor are used in fixed proportions. Between Harrod and Domar, Von Neuman (1945) provided theory of balanced growth at maximal rate. In their model, production technology is characterized by a finite set of constant returns to scale activities with inputs being committed at the beginning of each discrete production period and outputs emerging at the end. There are no non-produced factors

27 12 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS of production such as labor or exhaustible resources. In the primal version, Von Neuman characterized the vector of activity levels that permitted the maximal rate of balanced growth (i.e., growth in which outputs of all commodities grew at same rate) given that the outputs of each period were to be ploughed back as inputs in the next period. In the dual version, a vector of commodity prices and an interest rate were derived which had properties that the value of output of each activity was no higher than the value of inputs inclusive of interest, and the interest rate was the lowest possible. Under certain assumptions about the technology, Von Neuman showed, first, that the maximal growth rate of output of the primal version was equal to the (minimal) interest rate of the dual and, second, that the complementary slackness relations were observed between the vector of activity levels, prices, growth and interest rates. The next and more important contributions were those of Solow (1956). The key aspect of the Solow-Swan model is the neoclassical form of production function, a specification that assumes constant returns to scale, diminishing returns to each input, and some positive and smooth elasticity of substitution between the inputs. In these models, if there were no technological progress, then the effects of diminishing returns would eventually cause economic growth to cease. Therefore, the model predicts that in order to sustain a positive growth rate of output per capita in the long run, there must be continual advances in technological knowledge. Technology being a public good, those models assume identical rate of technological progress across countries, thereby affecting the same long run steady state growth rates among various countries. Policy changes have level effect on steady state growth rates. Another central features of neoclassical growth models are the absolute convergence and the conditional convergence property. Absolute convergence takes place when poorer areas grow faster then rich

28 Review of Literature: Growth Theories, Previous Studies ones assuming similar structural characteristics across countries. Conditional convergence occurs when a country (or a region) grows faster the further it is below it own steady state. If there are diminishing returns to capital, the level of income per capital should converge toward its steady state value, with the speed of convergence increasing in the distance to the steady state. In other words, lower initial values of income per capita generate higher transitional growth rates, once the determinants of the steady state are controlled. The obvious shortcoming of the Solow-Swan model is that the long run per capita growth rate is determined entirely by an element the rate of technological progress that is exogenous to the model. Endogenous growth theories as discussed below provide the missing link by providing theory of technological progress. Cass (1965) and Koopmans (1965) brought Ramsey analysis of consumer optimization back into the neoclassical growth model and thereby, provided for an endogenous determination of the saving rate. This extension allows for richer transitional dynamics but tends to preserve the hypothesis of convergence of income per capita and growth rates. The endogeneity of saving did not eliminate the dependence of the long run per capita growth rate on technological progress. The contribution of Phelps (1961) focused only on the steady state level of consumption per worker rather than on transitional dynamic path to steady state and solved the problem for that saving rate which maximized the steady state level of consumption per worker. Turning to the exceptions, Solow (1956) himself drew attention to the possibility that steady state need not even exist and that even if one existed, it need not be unique. Output per worker could grow indefinitely even in the absence of labor-augmenting technical progress, if the marginal product of capital were bounded below by a sufficient high positive number (Srinivasan, 1993; Helpman, 1992).

29 14 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS There were also exceptions to the exogeneity of technical progress and of the rate of growth of output along a steady state. In the one-sector models of Harrod and Domar and the two-sector models of Feldman (1928, as described in Domar, 1957) and Mahalanobis (1955), marginal capital-output ratios were assumed to be constant, so that, by definition, the marginal product of capital did not decline. The growth rate was endogenous and depended on the rate of saving (investment) in such one-sector models and on the aggregate rate of investment and its allocation between sectors producing capital and consumer goods in the two-sector model. Kaldor (1957) proposed abandoning altogether the notion of an aggregate production function and the distinction between increases in productivity due to capital and those due to technological progress. Instead, he introduced a technical progress function relating the rate of output growth to the rate of investment, the underlying rate of new ideas and society s adaptability to these new ideas. However, the steady state rate of growth was independent of savings behavior and was determined entirely by the exogenous properties of the postulated technical progress function. Kaldor and Mirlees (1962) endogenized technical progress (and hence the rate of growth of output) by relating productivity of workers operating newly produced equipment to the rate of growth of investment per worker. Arrow (1962) and Sheshinski (1967) constructed models of learning by doing, in which factor productivity was an increasing function of cumulated output or investment. Learning by doing was assumed to be purely external to the firms who did the producing and to the firms that acquired the new capital goods. Thus, capital and labour could continue to receive their marginal products, because in competitive equilibrium no additional compensation is paid to factor productivity. Nevertheless, the growth of factor productivity became endogenous, in the sense

30 Review of Literature: Growth Theories, Previous Studies that an increased saving propensity would affect total time path. The Arrow model, however, was fully worked out only in case of a fixed capital/labor ratio and fixed labor requirements. This implied that in the long run the growth of output was limited by growth in labor, and hence was independent of savings behavior, as in the Solow-Swan model. Uzawa (1965) also endogenized growth of technological progress by postulating that it is a concave function of the ratio of labor employed in the education sector to total employment. The education sector is assumed to use labor as the only input. The work of Cass (1965) and Koopmans (1965) completed the basic neoclassical growth model. Thereafter, growth theory became excessively technical and steadily lost contact with empirical applications. Due to lack of empirical evidence, growth theory effectively died as an active research field by the early 1970s particularly on the eve of rational expectations revolution and oil shocks. For about 15 years, macro economic research focussed on short-term fluctuations. Major contributions included the incorporation of rational expectations into business cycle models, improved approaches to policy evaluation and the applications of general equilibrium methods to real business cycle theory. The recent revival of growth theory started with the influential papers of Lucas (1988) and Romer (1986). Lucas motivated his research by arguing that neoclassical growth theory cannot account for observed differences in technological progress and growth rates of per capita income across countries. He draws on the theory of human capital to provide an explanation to technological progress. He defines human capital as the time spent by an individual on accumulating skills. The acquisition of skills by a worker not only increases his productivity by increasing the average level of skills in the economy as a whole, but has a spillover effect on the productivity

31 16 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS of all workers by increasing the average level of skills in the economy as a whole. There is a dynamic spillover or externality from the growth-generating activity to the rest of the economy and this spillover allows economies to escape the strait-jacket of diminishing returns, which otherwise inevitably brings growth back to the exogenous rate in neoclassical growth models. Lucas builds his model on the assumption of increasing returns to scale and positive asymptotic marginal product of aggregate capital. In the Lucas model, a policy that leads to a permanent increase in the time individuals spend for obtaining skills generates a permanent increase in the growth of output per worker. Building on the work of Arrow (1962) and Sheshinski (1967), Romer (1986) gives an equilibrium model of endogenous technical change in which long-run growth is driven primarily by the accumulation of knowledge by forward-looking profit-maximizing agents. While the production of new knowledge is through a technology that exhibits diminishing returns, the creation of new knowledge by one firm is assumed to have a positive external effect on the production possibilities of other firms so that production of consumption goods as a function of stock of knowledge exhibits increasing returns. To put it more precisely, knowledge may have an increasing marginal product. It should be noted that the spillover effect of the average stock of human capital per worker in the Lucas model and of knowledge in the Romer model are externalities that are unperceived (and hence not internalized) by individual agents. However, for the economy as a whole, they generate increasing scale economies even though the perceived production function of each agent exhibits constant returns to scale. Thus by introducing non-convexities through the device of marshalling externality, Lucas and Romer were able to work with an inter-temporal competitive (albeit socially non-optimal) equilibrium.

32 Review of Literature: Growth Theories, Previous Studies Thus both avoided facing the problem. 6 Research and development (R&D) efforts that lead to technical progress are naturally associated with imperfect competitive markets as Schumpeter (1942) had forcefully argued (Stiglitz, 1990,25). Later works by Grossman and Helpman (1991, 1992), Romer (1987, 1990) and Aghion and Howitt (1992) formulate models in which firms operating in imperfect competitive markets introduce R&D Barro and Xavier Sala-i-Martin (1995) that provide expositions and extensions of these models. In these settings, technological advances result from purposive R&D activity, and this activity is rewarded, along the lines of Schumpeter (1934) by some form of ex-post monopoly. If there is no tendency to run out of ideas, then growth rates can remain positive in the long run. The various linkages between trade and growth in endogenous growth models of knowledge accumulation via R&D have been thoroughly catalogued and analyzed in Grossman and Helpman (1992). International trade may affect the underlying conditions for growth by expanding the potential market size, allowing firms to spread the costs of R&D over greater volumes. This will stimulate innovative activities, but it will not lead to sustained growth unless the R&D sector generates spillovers or additions to the general stock of knowledge from which subsequent innovations can benefit. Essentially, the cost of each innovation must fall because of the larger stock of knowledge available to draw from. This is a testable hypothesis that seems to be partially refuted by data, which, instead, suggests that more resources need to be devoted to R&D over time to keep up the momentum in particular innovative activities. Endogenous growth models explained above can be divided into two broad categories. The first one maintains the assumption of perfect competition and assumes externalities. The second category 6 However, in Romer (1990) innovation is driven by profit-maximizing entrepreneurs.

33 18 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS models are those that drop the assumption of perfect competition and model technological progress as a function of R&D activity. There is a third strand of endogenous growth model called the AK model associated with Romer (1987) and Rebelo (1987, 1990). The AK model can be seen as a reduced form of the class of endogenous growth models with constant returns to broad capital. Broad capital includes not only privately held machines but also other assumable factors: human capital, public infrastructure, and possibly knowledge. These models are referred to as AK models, because they result in a production function of the form Y=AK, with A constant. The key result of the AK model is that the growth rate of the economy is an increasing function of the investment rate. Therefore, the government policies that increase the investment rate of the economy permanently will increase the growth rate of the economy permanently. The fourth strand of endogenous growth models relies on the existence of specialized intermediate inputs. Growth depends on the range or quality of such inputs that increase over time. One shortcoming of the early versions of endogenous growth theories is that they do not predict conditional convergence. Since this behavior is a result of empirical regularities in the data for countries and regions, it was important to extend the theories to restore the convergence property. One such extension involves the property. Another one involves the diffusion of technology (Barro and Sala-I-Martin, 1997). They construct a model that combines elements of endogenous growth with the convergence of the neoclassical growth model. In the long run, the world growth rate is driven by inventions in the technologically leading countries. Followers converge toward the leaders because copying is cheaper than innovation over some range. A tendency for copying costs to

34 Review of Literature: Growth Theories, Previous Studies increase reduces followers growth rate and thereby generates a pattern of conditional convergence. They discuss how countries are selected to be technological leaders and assess welfare implications. Poorly defined intellectual property rights imply that leaders have insufficient incentive to invent and followers have excessive incentive to copy. Endogenous growth theories that include the discovery of new ideas and methods of production are important for providing possible explanations for long-term growth. Yet the recent cross-country empirical work on growth has received more inspiration from the older, neoclassical model, as it extended to include government policies, human capital, and diffusion of technology. Theories of basic technological change seem more important for understanding why the world as a whole can continue to grow indefinitely in per capita terms. But these theories have less to do with the determination of relative rates of growth across countries, the key element studied in cross-country statistical analyses. With the introduction of endogenous growth models, there has been a renewed interest in the neoclassical growth models. Renewed interest in empirically testing some of the implications of neoclassical growth theories and estimating the contributions to growth of various factors has also been stimulated by the introduction of endogenous growth theory and new growth theory (see section on growth empirics). A neat description of the recent advances in neoclassical growth model has been discussed in Barro and Sala-I-Martin (1994), Barro, Mankiw and Sala-I-Martin (1995), Jones (1998), and Raut and Srinivasan (1991), among others.

35 20 ECONOMIC GROWTH AND CONVERGENCE IN SELECTED SOUTH ASIAN AND EAST ASIAN COUNTRIES: A DATA ENVELOPMENT ANALYSIS Barro and Xavier Sala-i-Martin (1994) discusses that technological progress takes the form of improvements in quality of an array of intermediate inputs to production. In an equilibrium that is standard in the literature, all research is carried by outsiders, and success means that the outsider replaces the incumbent as the industry leader. The equilibrium research intensity involves three considerations: leading-edge goods are priced above the competitive level; innovators value the extraction of monopoly rents from predecessors; and innovators regard their successes as temporary. They show that if industry leaders have lower costs of research, then the leaders will do all the research in equilibrium. However, if the cost advantage is not too large, then the equilibrium research intensity and growth rate depend on the existence of the competitive fringe and take on the same values as in the standard solution. They discuss the departures from Pareto optimality and analyze the determination of the economy s rate of return and growth rate. Barro, Mankiw and Xavier Sala-i-Martin (1995) paper discusses that the neoclassical growth model accords with empirical evidence on convergence if capital is viewed broadly to include human investments, so that diminishing returns to capital set in slowly, and if differences in government policies or other variables create substantial differences in steady-state positions. However, openeconomy versions of the theory predict higher rates of convergence than those observed empirically. The authors show that the open-economy model conforms to the evidence that if an economy can borrow to finance only a portion of its capital, for example, human capital, it must be financed by domestic savings. Jones (1998) gives an extended Solow model where an increase in the share of the labor force engaged in R&D generates level effects on steady state per capita income rather than long run growth effects. For example, a government subsidy that increases the share

36 Review of Literature: Growth Theories, Previous Studies of labor in research will typically increase the growth rate of the economy, but only temporarily, as the economy transits to a higher level of income. Jones (1998) also provides extended neoclassical model which assumes technological transfers and openness to trade. In these models, technology transfer occurs because individuals in an economy learn to use more advanced capital goods. This model provides one justification for the key assumption of neoclassical growth model that all countries share the same long-term growth rate, given the rate at which world technological frontier expands. Other extensions include models which answer the following questions: Why is it that some countries invest more than others, and why do individuals in some countries spend more time learning to use new technologies? These models suggest that conductive economic environment is key to attract investment from abroad in the form of transfer of better technologies and encourage accumulation of skills by individuals. Conducive economic environment is built on factors like efficiency of the government bureaucracy, the efficacy of the rule of law, stable government policies and better infrastructure of an economy. Raut and Srinivasan (1991) model adopts a different approach to endogenize technical progress and growth by assuming fertility and savings to be endogenous and the size of the total population to have external effect (of a Hicks-neutral type) either through the negative influence of congestion or the positive stimulation of faster innovation. The model generates a rich set of growth paths for per capita income and consumption, some of which do not converge to a steady state and are even chaotic. A higher rate of population growth lowers the steady state level of capital and output per worker and tends thereby to reduce the per capita growth rate for a given initial level of per capita output. The standard model does not, however, consider the effects of per capita income and wage rates on population growth the kinds of effects stressed by Malthus and