1 CROSS-COUNTRY DIFFERENTIALS IN ECONOMIC GROWTH AND INEQUALITY IN SOUTH ASIA ------A PANEL DATA APPROACH. Ratan Kumar Ghosal Professor of Economics, Department of Commerce, University of Calcutta, Kolkata, West Bengal, India. Address for correspondence: Dr. Ratan Kumar Ghosal, 178, B.K.street, Uttarpara, Hooghly West Bengal, PIN-712258 (INDIA) Phone: +91 33 2 663 5827 E-mail: [email protected]Acknowledgement: The author records his deep indebtedness to his PhD students Mr. Surajit Sengupta and Mr. Saikat Bhattacharya for their immense help in the collection of data required for this paper .The usual caveat applies.
Transcript
1
CROSS-COUNTRY DIFFERENTIALS IN ECONOMIC GROWTH AND INEQUALITY IN SOUTH ASIA ------A PANEL DATA APPROACH.
Ratan Kumar Ghosal Professor of Economics, Department of Commerce,
University of Calcutta, Kolkata, West Bengal, India.
Address for correspondence: Dr. Ratan Kumar Ghosal,
178, B.K.street, Uttarpara, Hooghly West Bengal, PIN-712258 (INDIA)
The author records his deep indebtedness to his PhD students Mr. Surajit Sengupta and Mr. Saikat Bhattacharya for their immense help in the collection of data required for this paper .The usual caveat applies.
CROSS-COUNTRY DIFFERENTIALS IN ECONOMIC GROWTH AND INEQUALITY IN SOUTH ASIA ---A PANEL DATA APPROACH.
Ratan Kumar Ghosal
Professor of Economics, Department of Commerce, University of Calcutta, Kolkata,
West Bengal, India.
Abstract
This paper intends to investigate the proximate explanatory factors behind the cross-country differentials in the growth rates and the various dimensions of inequalities in the south Asian (S.A.) region since 1965 by using both the cross-country regression technique and the panel data approach . We find that in spite of the inheritance of some structural constraints as an outcome of the imperialistic exploitation and also the presence of some country specific constraining factors, most of the countries in the S.A.region have been able to achieve high rates of growth of both the real GDP and PCI especially in the 90s with slightly declining tendency in the cross-country inequalities but increasing tendency of intra-country inequalities. The cross-country and cross-time variations in the country specific factors and the cross-time variations in the effect of the investment, health and education capital are more important for the differentials in the level and the inter-temporal growth rates of per-capita income in this region. Since in most of the cases of our panel regression results it is found that the effect of the human capital more than offsets that of physical capital and further since the growth accounting exercise reveals that the contribution of Solow residual to the productivity growth has declined and that of education capital has increased in the 90s, one can plausibly conclude that the individual nation states of this region must play a crucial role through massive public investment in boosting the social sector development even under the on-going process of globalization. JEL Classification No: O 41, O 47, O 57. Key Words: Economic Growth; cross-country differentials; convergence: Inequalities: South Asia; Panel Data.
Address for correspondence: Dr. Ratan Kumar Ghosal,
178, B.K.street, Uttarpara, Hooghly West Bengal, PIN-712258 (INDIA)
CROSS-COUNTRY DIFFERENTIALS IN ECONOMIC GROWTH AND INEQUALITY IN SOUTH ASIA---A PANEL DATA APPROACH . I. Introduction. The countries in the globe are in inequality traps which are reflected in its various facets and dimensions of divides viz; the economic (income), the social, political, ethnical, religion, castes, and geographical divides. The contemporary dimensions of inequalities persisting across the countries in the world have some historical inheritance also. Actually the initial inequality in the distribution of resources, income and wealth either across the countries or at the intra-country level leads to the inequality in opportunities in its various forms which in turn, lead to the inequalities in the capabilities vis-a-vis entitlements of the people across the countries thereby leading to further economic, socio-cultural and political inequalities and thus completing the vicious circle of inequality i.e. the inequality trap. In fact, the interaction between the economic, socio-cultural and the political inequities shapes the institutions and rules in all societies. The ways the institutions function affect the access of the people to various opportunities and their abilities to invest and prosper. The unequal economic opportunities yields inegalitarian outcomes and reinforce inequity in political power which in turn, shapes the institutions and policies that tend to foster the persistence of the initial conditions. So in any study of the cross-country inequality one has to take note of the base level inequity in income and wealth which has a close bearing on the future pattern of economic growth and inequality. Interestingly, as a fall out of the on-going process of globalization of technology, finance and socio-culture which is basically based on market fundamentalism, there has been a rapid transformation of the economies in the world from the state of bureaucratic control over trade, investment and finance to market through the policy of deregulation. This process of policy liberalization has indeed produced immense impact on economic growth and the distribution of its fruits both across the countries and also across the people at the intra-country level. The countries in the South Asian (S.A.) regions are not exception to this and also not outside the purview of the on-going process of globalization and its impact. There has indeed been substantial policy reform in S.A, albeit the pace of reform has been slow but steady. But one should recognize the fact that all the countries in S.A region have some colonial history of British imperialism which dominated the economies in this region almost up to the middle of the 20th century. As a fall out ,the countries in this region have got to face some structural constraints also. Actually the British imperialist deliberately tried to keep the economic structure of the societies of this region backward so as to perpetuate the exploitation during their regime. For instance, it is also well recognized that the British rulers developed the infrastructures in those regions in India viz in Calcutta, Madras, Bombay which would help smoothening the process of movement of goods and resources from hinterland to port (Frankel, 1978; Dutta, 1975). This has indeed caused regionalization of the process of development thereby causing inequality in development. So the inequality across the countries and also at the intra-country level in the SA region has some historical inheritance (Ludden, 2005).In fact, with the failure of
4
the imperial governance in managing the disastrous effect of depression, Bengal famine and partition, the countries in S.A switched over to the formulation of the strategy of nationally planned economic development with great emphasis on the acceleration of economic growth and the reduction of poverty ,inequality and the ignorance in the early 50s.As Ludden(2005) observes “The 25 years between 1950 and 1975 were the heyday of nationally planned development in S.A.Uniquely in Bangladesh, however, independence arrived only on 1971.--------..Notably in Bangladesh, but to some degree for all post-colonial regimes, national planning faced serious constraints: financial, infrastructural, administrative, political and intellectual.------During the heyday of planning, systematic inequalities in wealth and power among social groups and regions remained starkly visible in development thinking……”.But the neo-liberal free market orthodoxy with development strategies emphasizing on the private sector leadership in market-driven economic growth started dominating the national development planning strategies in S.A since1980s.So while studying the cross-country differentials in the economic growth and inequality in this region one has to take into consideration of the structurally constraining factors. But the inadequacy of the comparable set of data prevents us to do so. Further, it is also recognized that the binding constraints on economic growth of the low income countries are present amongst the countries in S.A. and these vary from country to country such that the most important of which are weak governance in case of Bangladesh; civil conflict in Nepal and Sri-Lanka, fiscal deficit in India and a fixed exchange rate in Bhutan etc. (Deverajan,2005).Astonishingly, in spite of the presence of these factors coupled with the persistence of the historically inherited structural constraints, the countries in S.A are growing at a faster pace such that both the growth rates of GDP and the per-capita GDP have revealed remarkable increase since 80s and especially since the 90s in varying degrees. So one has to undertake an in-depth study on the trajectories behind the growth process of the countries in S.A region. Interestingly, the World Bank in its recent World Development Report (WDR) 2006 has concentrated on the analysis of the global equity and development pattern. According to this report, the evaluative judgment about whether the income inequality has been increasing or decreasing actually depends inter alia on which concept of inequality is under the microscope. The World Bank has used three concepts of inequality viz; the inter-country inequality in the distribution of unweighted country means; the inter-national inequality in the distribution of country means weighted by their respective population sizes; and the global inequality in the distribution of individual incomes. It is found that the inter-country inequality (unweighted) has been increasing steadily since 1980. But the international inequality (weighted) has been declining steadily because of tremendous economic growth in the populous countries like China and India. However, the inter-country and international inequality without China and India track each other quite closely from 1980 onward coinciding with the period of rapid growth in these two countries, the slower average growth rates in other developing countries, and the declines in measured output in Eastern Europe and former soviet Union countries(WDR,2006).In fact, it is found by the World Bank that while the World became reacher,the income inequality –relative and absolute, international and global increased tremendously during the long period of time i.e.1820-1992.In fact, it is found that while the international inequality is falling , the intra-country inequality in income is rising and so the global
5
inequality depends on the relative strength of these two effects. But since china and S.A are gradually catching up to world average, their equalizing effect will eventually be diminishing thereby causing international inequality to rise in future. But what is surprising is that the rising intra-country inequality has been accompanied by falling trend in the international inequality in the educational attainment and health. So how can one reconcile? Surprisingly as per UNDP assessment of countries across the regions of the world, South Asia remains one of the world poorest region with highest proportion of people (42%) living below the international poverty line 1$ a day (ppp), 42 % of undernourished people during 1998-2000; 35% of primary age children not enrolled; 41% of primary age girl not enrolled; 35% of under five death; 38% of people living without adequate sanitation etc.Moreover as per UNDP classification all the countries in S.A belong to the medium human development range of countries with six of them belonging to low-income and 3 to middle income group pf countries . Further, if we look at the cross-country growth and in equalities in its different forms across the countries in south Asia then the data tell us a varied picture. So far as the growth rate of per-capita income is concerned it is found that it ranges from -0.3% p.a for Iran Islamic republic to 4.05 p.a for Bhutan during 1975 -2003 with a very high degree of cross country differentials( coefficient of variation (C.V) being 58.6%). But if we consider the period 1990-2003 then the growth rates of some countries like India, Nepal,, Bangladesh, Iran ,Islamic rep.,Maldives are found to improve while that of others reveal declining tendency and the cross country differentials are found to decline to some extent ,C.V. being 38.47.However the cross-country inequality in growth remains very high. So an intensive probe in to the nitty gritty on the various dimensions of intra-country and cross-country inequalities in S.A so as to find out the proximate explanatory factors behind the trajectories of growth with inequality and to derive some policy implications is quint essential. Hence is our study. It is well-known that the empirical and the theoretical literature on economic growth since the advent of neo-classical growth model of Robert Solow (1956) have been concentrated on the exploration of the proximate explanatory factors responsible for the tremendous inequities in the level and growth of per-capita income and also the levels of living of the people across the countries in the world. The two broad approaches in this respect have been (i) .the behavioristic approach and (ii) the optimizing approach based on micro foundation. In fact the neo-classical Solow-model has helped engendering a new branch of growth theory known as the endogenous growth theory and a vast literature concentrating on the convergence controversy i.e. whether the poor countries are catching up the rich countries. Some of the studies by using large sample countries concluded that the convergence hypothesis holds. But some others are of the view that the diverging tendencies and sometimes the conditional convergence across the countries to hold(Solow (1956, 2000), Cass (1961), Koopman (1965) Baumol (1986), Barro (1991), Lucas (1988) and Rebelo (1991), Romer (1986), Mankiw et al 1992), Bils et al (2000), Caballe and Santos (1993);Benhabib and Spiegel,(1994); Islam (1995, 1998,2003);( Barro and Martin,1994);Ghosal,(2002,2007); Lee et al., 1997; Rebelo (1991), Lucus (1988), Caballe & Santos (1993), Barro (1994)etc.All these studies have used large no of sample countries and some have classified the countries according to UNDP classifications. However, there is hardly any intensive study concentrating on the cross-country differentials in the growth and inequalities in the south Asian region.
6
Moreover most of these studies have used inadequate proxies for human capital especially for education. Some have not considered health capital as an in important ingredient of human capital (Mankiw et al, 1992).Further one has to take note of the serious limitation of the above studies by considering the fact that human capital is also a stock variable. So the use of gross enrolment ratio or the weighted average of the enrolment ratios across countries as proxies of education capital as is done by us also (Ghosal, 2002; 2007) is rather misleading. To remove this limitation we, in our present study have considered the number of people enrolled in the tertiary education in all disciplines across countries in S.A as proxy of education capital on the basis of the assumption that all the students enrolled in tertiary education acquire the degree and the drop-out rate in this sector is almost nonexistent. Under this backdrop we, in our paper intend to investigate the proximate explanatory factors behind the cross-country differentials in the growth rates and the various forms of inequalities in the south Asian region since 1965 by using both the cross-country regression technique and the panel data approach and the various conventional measures of inequality. This paper is structured as follows. Section II will discuss data and methodology and the model; section III presents a descriptive analysis of the various dimensions of inequality and growth; section IV makes econometric analysis of the cross-country inequality in growth; finally section V gives concluding observation and the policy implication of our study. II.Data and Methodology. In our study on the cross-country and intra-country with cross time differentials in the economic growth and inequality in its various dimensions in S.A region, we have considered eight out of nine countries as the most the data on the relevant variables are not available for Afghanistan. So we have excluded Afghanistan from our study. The period that we covered in our study ranges from 1965 to 2004.The choice of the period is absolutely determined on the basis of the availability of data. Our empirical analysis is entirely based on the secondary data available from various sources like World Development Reports and World Development Indicators (various issues) of World Bank and the various issues of Human Development Reports of UNDP, the Pen World Tables 6.2 version and the reports of UNESCO. It is well recognized that the comparable set of time series data on the human resources like education and health capital and also on some of the macro economic fundamentals are not available for all the countries in S.A region. So in the cases of minor data gap for few intermediate periods we have got to use average figures in order to keep the series comparable and suitable for our quantitative analysis. However this has been done for very limited cases. In our study, we have tried to examine the various dimensions of inequalities viz; the inequality in income and its distribution, the inequality in the attainment of education and health across the countries in S.A and also within the country across time. To judge the cross-country variability in growth and inequality in its various dimensions we have used very conventional tool i.e. the coefficient of variation (C.V) on the basis of which we have tried to have some insight about the tendency of convergence and divergence. In fact since the no of countries as observation is very small, we have very small degrees of freedom and so the application of the most contemporary econometric technique for making convergence analysis becomes difficult.
7
Now to investigate into the proximate factors responsible for the cross-country differentials in the economic growth and inequality over the period under consideration we have used the cross –country regression method by using per-capita real GDP(y) (PCRGDP) as dependent variable and investment-gdp ratios, human capital in the form of health capital (H1) and education capital (H2) and the effective rate of depreciation of all capital (n+g+δ) as explanatory variables, where n is the rate of growth of population, g stands for the growth rate of knowledge and δ for the depreciation of the stock physical and human capital. Like Mankiw et al (1992) we have also assumed that g and δ remain constant across countries and also across time, the value of g+δ being 0.05. We have used life expectancy at birth as a proxy of health capital and the number of persons enrolled in the tertiary education in its various disciplines as surrogate of the stock of educational capital. However it is worth mentioning that the empirical specification of the regression model actually descends from the model that we have developed by augmenting the neoclassical Solow model through the incorporation of the human capital in its two forms in to the model which will be presented latter (Ghosal, 2002; 2007). On the other hand, to examine the cross-country and cross-time inequality in growth simultaneously in S.A region we have made panel data regression analysis by following the model developed by us, as the panel data estimate can take the cross-time and cross-country heterogeneity more explicitly into account. However, because of the non-availability of time series data, we have used five yearly panel data on all the relevant variables for the period 1965 to 2003 by choosing 8 countries out of 9 countries of S.A. The choice of the period and the sample countries is made on the basis of the availability data on all the variables. Here also we assume that the growth rates of knowledge (g) and value of depreciation of stock of capital (δ) remain constant across countries such that g+δ=0.05. Now since we are focusing on the specific set of countries in the region, our inference is restricted to the behavior of this set of countries only. In our panel data estimation we assume that the country specific effect is fixed. Now since the number of countries is very small and we have got 5yearly panel of time series data for the period from 1965 to 2004, it seems that the fixed effect model is an appropriate specification and so we use the fixed effect model such that the intercepts may vary across countries but it becomes time invariant. Moreover due to the lack full time series data on all the variables considered in our model causing smaller degrees of freedom we could not estimate the structural break and also the impact of the structural as well as the policy variables on the cross-country inequality in this region. For empirical estimation we have used the software Eviews 0.3 version. The Model:
We actually augment the Solow model by incorporating human capital in the form of two additional variables viz., (a) health capital (H1) and (b) education capital (H2) into a Cobb-Doglous production function. Unlike Mankiw et al model and other models here the stock of education capital consists of the no of persons enrolled in the tertiary education in its various disciplines. Further, the life expectancy at birth is used as a proxy of health capital and the physical capital (k) is measured in terms of the ratio of gross domestic investment to the GDP(i) of the respective countries.
We write the production function as:
8
Yit = KitH1it
H2it(AitLit)
................................. (1): , > 0 and < 1. Where Y = aggregate output; K, L, H1, H2 are stocks of capital, labour, health and education capital; A=level of technology. And the subscripts denote country (i) and time (t). Here L(t)=L0ent and A(t)=A0egt. Here, we consider one-sector production technology with labor-augmenting technological progress and assume constant return to scale to operate in physical and human capital. In fact, we assume that (i) the availability of natural resources is not a constraint on growth and (ii) although the gain from specialization for large countries gets exhausted there is further scope of specialization in small countries. So one may say that if the new inputs are used in the same way like that of the existing inputs, then the doubling of K & H is likely to double the output. Lucas (1988) has also assumed that the returns to all reproducible capital (Human plus Physical) are constant. Now, since we consider a one-sector production technology and further since there is no standard estimate of depreciation of physical and human capital across the countries we assume that the physical and human capital depreciates at same constant rate (). We also assume that constant fractions of real GDP (i.e. sh and se) are spent for health and educations respectively. The transitional dynamics gives rise to the following empirical specification of the augmented Solow model in logarithmic form:
In(Y/L) it = InA0it + gt + / (1--1-2) In (sk) it + 1/(1--1-2) In(sh) it + 2/ (1--1-2) In(se) it - ( + 1 + 2)/(1--1-2) In(n+g+)it ... (2)
It indicates that per-capita income (PCRGDP(y=Y/L)) depends on growth of
population, technical progress and accumulation of physical capital and education and health capital. All variables are in logarithmic form. We use this for both the cross-country regression analysis and our dynamic panel data analysis. In fact the cross country regression models assume the common initial state of technology and constant rate of technological progress. But one can cast doubt against the empirical results drawn on the basis of this assumption. Robert Solow (2000) has also cast doubt about the constancy of the parameters of the production function across the countries. So the dynamic panel data approach seems to be appropriate and we applied the same. Islam (1995) has also estimated the augmented Solow model by using panel data investigation with fixed effect model. III.Various Dimensions of Cross-country Inequalities in S.A: Income Inequality. In this section we proceed to analyse various dimensions of cross-country and intra-country inequalities inflicting the countries in the S.A.region.It is well known that countries in this region inherit some colonial history with the tremendous domination of mainly the British imperialist on their economic and social structure which has led to the persistence of some structural constraining factors causing regionalization of
9
development vis-a-vis the inequality. We have already mentioned that because of the non availability of suitable data set on the structural constraint we could not highlight these factors in our quantitative analysis. Now the various dimensions of inequalities are concerned so far we first consider the economic inequality (i.e. the income inequality) and see what the data given in table-1 tell us. It is evident from the table that all the countries in the region excepting Iran Islam Republic achieved very high annual average growth rates of GDP during the 80s ranging from 4.0% for Srilanka to 6.3% for Pakistan. On the other hand, all the countries excepting Pakistan have experienced the increase in their growth rates of GDP during 90s.Interestingly for Srilanka, India and Iran the rates of increase are found to be very high but for Nepal and Bangladesh it is marginal. Although the cross-country inequality in the growth of GDP was found to be very high during 80s as is revealed by the value of coefficient of variation (C.V) reading the figure of 36.31%, it has declined substantially during the 90s the value of C.V being 16.11%.So what we find is the converging tendency of growth rates across the countries in the S.A region. Table-1: Cross-country Income Inequality in South Asian Region during 1975-2003 Country Auuual
growth rate of GDP%
Per-capita GDPat const.prices
Annual Growth rate (%) of PCGDP
Share of Richest 10% to Poorest 10%
Share of Richest 20% to Poorest 20%
Gini Index
!980-90 !990-8
1975 2003 1975-03 1990-03
Bangladesh 4.3 4.8 1331.22 2154.6 1.9 3.1 6.89(2000) 4.6(2000) 31.8(2000) Bhutan - - 266.56 933.46 4.0 3.6 - - - India 5.8 6.1 1178.53 2989.77 3.3 4.0 7.3(1999) 4.9(1999) 32.5(199900) Iran 1.7 4.0 7066.9 6398.13 -0.3 2.1 17.2(1998) 9.7(1998) 43.0(1998) Maldives 917.4 5153.9 - 4.7 - - - Nepal 4.6 4.8 900.47 1440.94 2.1 2.2 9.3(1995) 5.9(1995) 36.7(1995) Pakistan 6.3 4.1 1336.83 2591.96 2.5 1.1 7.6(1998) 4.8(1998) 33.0(1998) Srilanka 4.0 5.3 1506.41 4274.17 3.4 3.3 8.1(1999) 5.1(1999) 33.2(1999) C.V. 36.31 16.11 119 58.2 58.6 38.47 41.81 33.34 12.15 The table-1 clearly brings out the fact that all the countries in this region have been able to bring about remarkable increase in their levels of real per-capita GDP in varying degrees over the period between 1975 and 2003. Now so far as the growth rate of per-capita income is concerned it is found that it ranges from -0.3% p.a for Iran Islamic republic to 4.05 p.a for Bhutan during 1975 -2003 with a very high degree of cross country differentials( coefficient of variation (C.V) being 58.6%). But if we consider the period 1990-2003 then the growth rates of real per-capita income of some of the countries like India, Nepal,, Bangladesh, Iran Islamic rep.,Maldives are found to improve remarkably while that of others reveal declining tendency and the cross country differentials in the same which was very high (C.V.=58.2%) during 1975-03 are again found to decline to some extent during 1990-03, the C.V. being 38.47%.Now if we consider the income inequality within the countries in the region by using the ratio of relative shares of the richest 10% to poorest 10% of people in their respective GDP then
10
the figures are found to be very high during 90s, ranging from 7.3 for India to 17.2 for Iran Islamic rep .Again if we use the ratio of shares of richest to poorest 20% then also the figures are found to be high ranging from 4.8 for Pakistan to 9.7 for Iran Islamic Rep.This obviously reveals high degree of intra-country inequalities in the distribution of income to persist during the period of liberalization also. The conventional measure of intra-country in equality in the distribution of income i.e. the Gini Ratios also reveal high degree of inequality to persists in almost all the countries in S.A region, the value of which ranges from 43.0 for Iran to 36.7 for Nepal. The possible explanations behind this remarkable achievement in the higher growth rates of GDP and also the level and growth of per-capita income despite presence of structural binding constraints seem to be the operation of some country specific factors, the realization of the benefit of market through liberalization and the structural shift in the composition of GDP of the respective countries. So far as the country specific factors are concerned mention may be made of the effective and massive participation of NGOS in the development process especially the development of social sectors in Bangladesh; the tremendous increase in the remittance to Nepal due to the high rate of migration to India and West Asia; the rapid development of tourism industry in Maldives and finally, the established democracies in India, Bangladesh and Sri Lanka leading to effective policy formulation. Further, Sri Lanka and India have been able to reap the benefit of trade liberalization. Apart from these factors, the other crucial reasons behind the conspicuous achievement in the growth rates of GDP and the PCGDPacross the countries in South Asia seem to be the tremendous structural shift in the composition of GDP in favor of industry and especially to the services sector. For instance, presently in India 52% of GDP accrues out of the services sector and in the new millennium the growth rates of services sector highly dominates the growth rate of overall GDP.The Table 2 represents an overview of the structural changes in the composition of GDP and its growth rates across the countries of S.A. Table-2: Changes in the Sectoral Composition of GDP (%) in South Asia. Country Year Agriculture Percent
point change
Industry Percent point change
Service Percent point change
Bangladesh 1980 2004
50.0 21.0
-158 16 27
68.75 34 53
55.88
India 1980 2004
38.0 22.0
-42.11 26 36
38.46 36 52
44.44
Iran 1980 2004
18 11
-38.8 32 41
28.13 50 48
4.0
Nepal 1980 2004
62 40
-35.48 12 23
91.66 26 37
42.31
Pakistan 1980 2004
30 23
-23.33 25 24
-4.0 46 54
17.39
Sri Lanka 1980 2004
28 17
-39.29 30 25
16.67 43 58
34.88
The table-2 clearly brings out the fact that there has been a tremendous transformation in the sectoral composition of gdp of the countries in S.A.region from agriculture to industry and service sectors during 80s and 90s.In Pakistan, however the fall in the share
11
of agriculture and industry during 90s has been complemented by the gain in the same in service sector. Surprisingly, in Bangladesh, India , Nepal and SriLanka there has been a remarkable increase in the shares of service sector in their respective GDP during 90s by 55.88;44.44,42.31and 34.88 percentage points respectively. Moreover it is astonishing to note that Bangladesh and Nepal have experienced tremendous acceleration in their shares of the industrial sector in GDP during 90s, the figures being 68.75 and 91.66 percentage points respectively. This seems to be due to the rapid expansion of garment industry in Bangladesh and the expansion of the hydroelectric sector and tourism industry in Nepal. On the other hand, the persistence of the high degree of intra-country inequality in the distribution of income might be the result of the failure of the countries in the region in especially in India to implement the strategy of participatory growth process through the state action of programme. In fact, it is argued by Amartya Sen that in China, the inequality in the distribution of income has fallen not because of the introduction of any redistributive measures but because of making the growth process participatory, which India has failed to do (Dreze and Sen, 2002). On the other hand, a more concrete estimate of growth accounting i.e. the determination of the relative contribution of the growth of physical capital, human capital in the form of education per effective worker and the total factor productivity i.e.Solow residual on the growth of output per worker has been made by Barry Bosworth and Susan Collins(2003) for a large no of countries in the world during 1960to 2000.We reproduce the result for S.A in table-3 so that we may have the insight about the factors which are responsible for the high growth rate in productivity in the region. Table-3: Growth Accounting in South Asia during 1960-2000(Average annual growth rate %) Year Growth of
output per worker
Contribution of physical capital per worker
Contribution of education per worker
Contribution of total factor productivity
1970s 0.68 0.56 0.34 -0.23 1980s 3.67 1.02 0.40 2.25 1990s 2.78 1.19 0.42 1.17 Source: Susan, M.Collins and Barry Bosworth, the Empirics of Growth-An Update” Brooking Papers on Economic Activity, 2(2003), 113-79. It follows that the annual growth rate of productivity per worker in S.A reached a climax in 80s (3.67%) which is by a decline to the figure of 2.78% in the 90s.The contribution of physical capital to this growth rate has increased from 0.56% in the 70s to1.02% in 80s and further to1.195 in the 90s and that of the growth rate of education to the total growth has also increased from 0.345in 70s to 0.425 in the 90s.However, the contribution of the average growth rate of total factor productivity to the overall growth rate which was negative (-023%) in the 70s has increased tremendously to 2.25% in 80s followed by a decline to 1.17% in 90s. Surprisingly, in this study only education has been treated as human capital barring other crucial component of human capital viz; health. So this study suffers from specification error and we, in our model have tried to take into consideration of this limitation by considering the health capital as an ingredient of human capital also.
12
Astonishingly, this growth accounting approach clearly reveals that the contribution of the growth of Solow residual has fallen and that of physical capital and education have increased marginally in the era of liberalization which is mainly characterized by the competition of technology. Inequality in education and health: Inequality in social sector development across the countries in S.A.region has been conspicuous nevertheless most of the countries in the region have been able to experience rapid increase in the growth rates of both GDP and the per-capita GDP during the last two decades. What is more surprising is that despite the tendency of convergence of the growth rates of per-capita income and that of real GDP, there is a strong tendency of divergence in respect of two fundamental aspects of health viz; the infant mortality rate and the under five mortality rates which are the good indicator of the status of health across the countries. However, it is worth mentioning that all the countries in this region have been able to experience the substantial increase in the life expectancies at birth of their people. The table-4 gives a clear picture about the status of education and health. Table-4: Cross country inequality in Education and Health in South Asia. Country Adult
Bangladesh 41.1 45.2 62.6 48 145 46 239 69 Bhutan - 41.5 62.7 19 156 70 267 85 India 61 50.3 63.1 47 127 63 202 87 Iran 77 55.2 70.2 11 122 33 191 39 Maldives 97.2 51.4 66.3 30 157 55 255 72 Nepal 48.6 44.0 61.4 48 165 61 250 82 Pakistan 48.7 51.9 62.9 38 120 81 181 103 Sri Lanka 90.4 63.1 73.9 29 65 13 100 15 C.V. 33.37 13.73 6.8 41.45 24.3 41.07 26.05 41.46 It follows that the adult literacy rate is highest (97.25) in Maldives which is followed by Sri Lanka (90.4%), Iran (77%) and India (61%) respectively. Nepal and Pakistan are at the same levels. However the cross-country differentials in this respect is not very low (C.V =33.37) and further it surprising to note that in spite of the adoption of the planned development strategies by all the countries of S.A. in the early 50s Maldives ,Sri Lanka and Iran have been able to scale further while India ,Pakistan Bangladesh and Nepal trail behind. It seems that the structural adjustment programme as a part of economic reform seems to have led to a serious set backs to the social sectors especially in India. For instance, the fiscal austerity as part of structural adjustment programme has made
13
social expenditure as a soft target for the financial axe (Dreze and Sen, 2002). Moreover the neglect of social opportunities in India since the inception of reform in 1991 is also a continuation of earlier distortion of public priorities. It really surprising to note that presently only 3.5% of GDP is expended on education and less than 2% on health in India, nevertheless the Kothari Commission (1966) recommended a 6% of GDP to be expended on education. In fact it seems that the reason behind the setback of social sectors lies in the political economy of economic reform. In Bangladesh, however, many aspects of education and health are left to the NGOs. Now if we accept the life expectancy at birth as a surrogate of the status of health of the respective countries then the table clearly indicates that the all the countries in the S.A region have been able to increase the same in varying degrees since 1970 and there is a tendency of cross-country convergence in this respect. However, if we look at other aspects of health status then it really surprising to note that the proportions of under weight of below 5 years are still very high in some of the countries of this region and the cross-country variability in this respect is also substantial(C.V=41.47%).Moreover ,although all the countries in this region have been able to reduce the infant mortality rate and under five mortality rates in varying degrees since 1970, the cross-country variability in respect of this achievement has been increased remarkably i.e. there has been a diverging tendency in this respect. So in spite of the increase in life expectancy of these countries one can plausibly argue that the status of health still remain in a fragile state. IV. Econometric Analysis of Cross-country Inequality in Growth. This section presents the results of our dynamic panel data regression analysis for the cross-country differentials in the level and growth of real per-capita income for the period 1965-2004. We have estimated various forms of panel regressions in terms of fixed effect model for the set of five yearly panel data. The logic behind the choice of this model is given in section II above. Initially we have tried to estimate the dynamics of change as per our empirical specification of the regression equation ( Equation 2) which follows from our model and then expressed each explanatory variable i.e. physical, health and education capital net of effective rate of depreciation (i.e. n+g+δ) in order to find out the net effect of each explanatory factor on the cross country as well as cross-time variations of in the level of per-capita income(PCI).Finally, we have estimated the effect of inter-temporal growth rates of each explanatory variable on the inter-temporal growth of per-capita income both across time and across countries over the period of our study so that we may have the growth effect on the cross-country differentials of growth of per-capita income. Now while estimating the panel regression model and also the pooled regression model by applying GLS method, we make some assumptions about the intercept measuring the effect of country specific factors, the slope coefficients and the error term. First, we assume that the slope coefficients remain constant across countries but not across time.However, the intercepts vary both across country and time. So in this framework the variation is explained in terms of variation in intercept. The results are given in the table no I; II; III; IV.
14
Second, we assume that both the slope coefficients and the intercepts vary both across time and countries. The results are given in table no: V&VI. Third, we assume that the intercepts are time invariant but not the slopes. In other words, we tried to find out the cross-country variations in the panel framework. The results are given in table VII. Now so far as the cross-country and cross-time variations in the inter-temporal growth rates of per-capita income(gy) in S.A region is concerned we have used pooled regression method i.e. cross country regression in panel framework where the intercepts change across countries, not across time but the slope coefficients remain time invariant. The results of this analysis are given in table no VIII; IX and X. The results given in table I clearly brings out the fact that 99% of the cross-country and cross-time variation in the level of per-capita income is explained by the variations in the investment, health and education capital such that all the explanatory factors excepting the population growth are statistically significant with investment and health are very highly significant. So the results are consistent with the reality of the poorest developing region in the world. Although the low Durbin Watson statistics indicate some specification error, the high value of log likelihood indicates that the results are highly realistic. The signs of the coefficients are at the desired level excepting for investment .It seems that the contribution of human capital more than offsets that of the investment variable. Table: I Results of Panel Regression (Dep. Variable lnyit)
So one can plausibly conclude that the cross-country and cross-time variations in the country specific factors and the cross-time variations in the effect of the explanatory factors are more important for the differentials in the level of per-capita income in this region.
15
Table-II: Results of Panel Regression Analysis (Dep. Variable lnyit).
Bhutan 0.416 Adjusted R-squared 0.999 India 0.570 S.E. of regression 0.228 Iran 2.386 Log likelihood 35.610
Nepal 0.805 Durbin-Watson stat 0.911 Pakistan 1.421 Sri Lanka 0.094
The table –II presents the effect of the explanatory factors net of effective rate of depreciation on the cross-country and cross time differentials in the levels of PCI. In this case also we have the same results with a mild improvement in the explanatory power of the model such that the explanatory factors net of the effective rate of depreciation explain almost entire variability of the cross country and cross-time variation of PCI..However if we consider only the investment and population growth variables we find investment variable is highly significant and the population growth be insignificant Moreover if we consider investment variable net of effective rate depreciation it becomes highly significant with the explanatory power remaining almost same in both of the cases (See table III&IV). Table-III: Results of Panel regression (Dep. Variable lnyit)
Table-IV: Results of Panel regression (Dep. Variable lnyit)
Variable Coefficient t-Statistic Prob.
lni-ln(n+g+) 0.250 3.561 0.0006
Fixed Effects Weighted Statistics
Bangladesh 7.100 R-squared 0.996 Bhutan 5.737 Adjusted R-squared 0.996 India 7.130 Log likelihood -27.669 Iran 8.078 Durbin-Watson stat 0.355 Maldives 7.405 Nepal 6.605 Pakistan 7.207 Sri Lanka 7.300
Now if we allow the slope coefficients and intercept to vary both across time and across countries then all the explanatory factors together explain about 89% of the cross-country and cross time variations in the levels of PCI of this region with health, population growth being highly significant and education capital also be significant without specification error (Durbin Watson stat=2.65) (see table-V).Further if we express the Table-V: Cross Country regression results. (Dep. Variable lnyit)
Variables Coefficient t-Statistic Prob.
Intercept -22.491 -3.812 0.062 lni 0.069 0.183 0.872 lnH1 6.737 4.226 0.052 lnH2 0.099 1.615 0.248 ln(n+g+) 1.634 2.296 0.149 R-squared 0.965 Mean dependent var 7.480 Adjusted R-squared 0.894 S.D. dependent var 0.722 S.E. of regression 0.235 Akaike info criterion 0.118 Sum squared resid 0.110 Schwarz criterion 0.079 Log likelihood 4.588 F-statistic 13.649 Durbin-Watson stat 2.650 Prob(F-statistic) 0.069
17
Table-VI: Cross Country regression results. (Dep. Variable lnyit)
Coefficients of Countries Fixed Effects lni-ln(n+g+) lnH1-ln(n+g+) lnH2-ln(n+g+)
Note: Figures in parentheses are t values the explanatory factors net of effective rate of depreciation then about 97% of the cross-country and cross-time differentials of PCI is explained by investment ,education and health capital( see table-VI).The Durbin –Watson statistics (1.474) and the log likelihood(74.617) indicate the absence of specification error and the result is highly realistic. The negative coefficient if investment variable for some of the countries seem to be due to the fact that the positive contribution of health and education variable more than offset that of the investment variable. This has happened in cases of Sri Lanka, India, Bhutan, and Nepal. For Sri Lanka, India and Bhutan the coefficients of health and education capital are also found to be highly significant. The empirical information also supports this result. For some countries like Bhutan, India and Sri Lanka the country specific factors are found to negative impact on the cross-time variations in the levels of PCI.However, if we keep the intercept time invariant then the investment variable net of effective rate of depreciation explain about 74% of the variation of PCI across time and countries (see table-VII).In such case However the coefficients if the variable are
18
Table-VII: Regression Results of cross-country variation in Panel framework
significant for all the countries excepting Iran. Further the country specific factors produce positive impact on the variations in the levels of PCI across the countries in S.A region with a colonial background. Now so far as the results of pooled regression analysis on the cross-country and cross-time variations in the inter-temporal growth rates of per-capita income(gi) in S.A region is concerned ,we find that since we have used five yearly panel data ,the growth effect of all the variables taken together is not very remarkable as the values of adjusted R squared are very poor (See Tables VIII,IX and X below).However, the values of log likelihood and Durbin-Watson indicate very poor specification errors with results being realistic. In these results since the intercepts change across countries but not across time and the slope coefficients remain unchanged countries, the effects of country specific factors are also captured and found to produce positive impact on the cross country variation on the inter-temporal growth rates of PCI over the period of our study.
19
Table VIII:Results of Pooled Regression(Dep.Var(gy) Variable Coefficient t-Statistic Prob. gi -0.088 -2.510 0.015 gh1 0.390 2.235 0.029 gh2 0.029 1.853 0.068 (n+g+) 0.006 0.000 1 Fixed Effects Weighted Statistics Bangladesh 0.031 R-squared 0.307 Bhutan 0.111 Adjusted R-squared 0.202 India 0.085 Log likelihood 91.226 Iran 0.019 Durbin-Watson stat 1.503 Nepal 0.036 Pakistan 0.064 Sri Lanka 0.113
Table IX: Results of Pooled regression (Dep.Var(gy) Variable Coefficient t-Statistic Prob. gi – (n+g+) -0.089 -1.806 0.075 gh1 – (n+g+) 0.067 0.272 0.787 gh2 – (n+g+) 0.030 1.246 0.217 Fixed Effects Weighted Statistics Bangladesh 0.055 R-squared 0.184 Bhutan 0.137 Adjusted R-squared 0.075 India 0.105 Log likelihood 73.598 Iran 0.044 Durbin-Watson stat 1.362 Nepal 0.061 Pakistan 0.088 Sri Lanka 0.121 Table X: Results of Pooled regression (Dep.Var(gy) ) Variable Coefficient t-Statistic Prob. gi – (n+g+) -0.100 -1.825 0.072 Fixed Effects Weighted Statistics Bangladesh -0.034 R-squared 0.196 Bhutan 0.049 Adjusted R-squared 0.114 India 0.033 Log likelihood 71.452 Iran -0.057 Durbin-Watson stat 1.229 Maldevs 0.081 Nepal -0.023 Pakistan -0.021 Sri Lanka 0.094
20
V. Concluding observations. This paper intends to investigate the proximate explanatory factors behind the cross-country differentials in the growth rates and the various dimensions of inequalities in the south Asian region since 1965 by using both the cross-country regression technique and the panel data approach and also the various conventional measures of inequality on the basis of the secondary data available from various sources like World Development Reports and World Development Indicators (various issues) of World Bank and the various issues of Human Development Reports of UNDP, the Pen World Tables 6.2 version and the reports of UNESCO. It is well known that the process of policy liberalization as an outcome of globalization has indeed produced immense impact on economic growth and the distribution of its fruits both across the countries and also across the people at the intra-country level. The countries in the South Asian (S.A.) regions are not exception to this and also not outside the purview of the on-going process of globalization and its impact. There has indeed been substantial policy reform in S.A since 1980, albeit the pace of reform has been slow but steady. Further, one should recognize the fact that all the countries in S.A region have some colonial history of British imperialism which dominated the economies in this region almost up to the middle of the 20th century. As a fall out the countries in this region have got to face some structural constraints also.So, an intensive probe in to the nitty gritty of the various dimensions of intra-country and cross-country inequalities and differentials in economic growth in S.A is quint essential so as to capture the proximate explanatory factors behind the trajectories of growth with inequality and to derive some policy implications. Hence is our study. It is found that all the countries in this region have been able to bring about remarkable increase in their levels of real per-capita GDP in varying degrees over the period between 1975 and 2003. Now so far as the growth rate of per-capita income is concerned it is found that it ranges from -0.3% p.a for Iran Islamic republic to 4.05 p.a for Bhutan during 1975 -2003 with a very high degree of cross country differentials( coefficient of variation (C.V) being 58.6%). But if we consider the period 1990-2003 then the growth rates of real per-capita income of some of the countries like India, Nepal,, Bangladesh, Iran Islamic rep.,Maldives are found to improve remarkably while that of others reveal declining tendency and the cross country differentials in the same which was very high (C.V.=58.2%) during 1975-03 are again found to decline to some extent during 1990-03, the C.V. being 38.47%.Now if we consider the income inequality within the countries in the region by using the ratio of relative shares of the richest 10% to poorest 10% of people in their respective GDP then the figures are found to be very high during 90s, ranging from 7.3 for India to 17.2 for Iran Islamic rep .Again if we use the ratio of shares of richest to poorest 20% then also the figures are found to be high ranging from 4.8 for Pakistan to 9.7 for Iran Islamic Rep.This obviously reveals high degree of intra-country inequalities in the distribution of income to persist during the period of liberalization also. The conventional measure of intra-country in equality in the distribution of income i.e. the Gini Ratios also reveal high degree of inequality to persists in almost all the countries in S.A region, the value of which ranges from 43.0 for Iran to 36.7 for Nepal.
21
The possible explanations behind this remarkable achievement in the higher growth rates of GDP and also the level and growth of per-capita income despite presence of structural binding constraints seem to be the operation of some country specific factors, the realization of the benefit of market through liberalization and the structural shift in the composition of GDP of the respective countries. In fact there has been a tremendous structural shift in the composition of GDP in favor of industry and especially to the services sector in this region. The growth accounting of S.A clearly reveals that the productivity per worker has increased tremendously in the 80s followed by a declining tendency and it is accompanied by remarkable increase in the contribution of physical capital and Solow residual. Astonishingly, this growth accounting approach also reveals that the contribution of the growth of Solow residual has fallen and that of physical capital and education have increased marginally in the era of liberalization which is mainly characterized by the competition of technology. Further the inequality in social sector development across the countries in S.A.region has been conspicuous; nevertheless most of the countries in the region have been able to experience rapid increase in the growth rates of both GDP and the per-capita GDP during the last two decades. What is more surprising is that despite the tendency of convergence of the growth rates of per-capita income and that of real GDP, there is a strong tendency of divergence in respect of two fundamental aspects of health viz; the infant mortality rate and the under five mortality rates which are the good indicator of the status of health across the countries. However, it is worth mentioning that all the countries in this region have been able to experience the substantial increase in the life expectancies at birth of their people. The results of our dynamic panel data regression analysis for the cross-country differentials in the level and growth of real per-capita income for the period 1965-2004 reveal that about 97% to99% of the cross-country and cross-time variations in the level of per-capita income is explained by the variations in the investment, health and education capital such that all the explanatory factors excepting the population growth are statistically significant with investment and health are very highly significant. So the results are consistent with the reality of this region. Although the low Durbin Watson statistics indicate some specification error, the high value of log likelihood indicates that the results are highly realistic. Moreover from the results of the various forms of panel regression one can plausibly conclude that the cross-country and cross-time variations in the country specific factors and the cross-time variations in the effect of the explanatory factors are more important for the differentials in the level and the inter-temporal growth rates of per-capita income in this region. However if we consider only the investment and population growth variables we find investment variable is highly significant and the population growth be insignificant On the whole we find that in spite of the inheritance of some structural constraints as an outcome of the imperialistic exploitation and also the presence of some country specific constraining factors, most of the countries in the S.A.region have been able to achieve high rates of growth of both the real GDP and PCI especially in the 90s with slightly declining tendency in the cross-country inequalities but increasing tendency of intra-country inequalities. The cross-country and cross-time variations in the country specific factors and the cross-time variations in the effect of the investment, health and education
22
capital are more important for the differentials in the level and the inter-temporal growth rates of per-capita income in this region. Since in most of the cases of our panel regression results it is found that the effect of the human capital more than offsets that of physical capital and further since the growth accounting exercise reveals that the contribution of Solow residual to the productivity growth has declined and that of education capital has increased in the 90s, one can plausibly conclude that the individual nation states of this region must play a crucial role through massive public investment in boosting the social sector development even under the on-going process of globalization.
23
REFERENCES:
Barro, Robert J. (1991),“Economic Growth in a Cross Section of Countries”, Quarterly Journal of Economics, 106 (May), 407-443.
Barro, Robert J. (1997), Determinants of Economic Growth, Cambridge, MIT Press
Barro, Robert J. (1999a), "Notes on Growth Accounting," Journal of Economic Growth, Vol.4 No. 2 (June), pp. 119-37
Barro, Robert J. (1999b), "Determinants of Democracy," Journal of Political Economy, Vol. 107, No. 6 (Part 2, December), pp. S158-83
Barro, Robert J. (2000), "Inequality and Growth in a Panel of Countries," Journal of Economic Growth, Vol. 5, No. 1 (March), pp. 5-32
Barro, Robert J–and Xavier Sala-i-Martin (1995), Economic Growth, Mc- Growhill, Inc.
Baumol, William (1986),“Productivity Growth, Convergence and Welfare”. American Economic Review, 76 (December), 1072-1085. Benhabib and Spiegel (1994), “The Role of Human Capital in Economic Development: Evidence from Aggregate Cross-country data”, Journal of Monetary Economics, 34, 143-173 Bils, Mark, Klenow, J. Peter (2000), “Does schooling cause growth?” The American Economic Review, 90, 5, 1160-1183 Baltagi.H Badi (1995), Econometric Analysis of Panel Data, John Wiley &Sons, New York. Caballe, Jordi & Santos, S. Manuel (1993), “On Endogenous Growth with Physical and Human Capital”. Journal of Political Economy, 101, No.6, 1042-1067. Cass, David (1965), “Optimum Growth in an Aggregative Model of Capital Accumulation”, Review of Economic Studies, 32 (July), 233-240. Chatterjee and Price (1997), Regression analysis by Example, John Wiley & Sons, New York. De Long, J. Bradford (1988), “Productivity Growth Convergence and Welfare Comment”, American Economic Review, 78, 5 (December), 1138-1154. Deverajan, S (2005), “South Asian Surprise”, Economic and Political weekly, Vol.XL, No.37, pp.4013-4015. Dreze and sen (2002), India Development and Participation, Oxford University Press, New Delhi.
24
Dutt, Romesh (1975),The Economic History of India,Director,Publication Division, Ministry of Information and Broadcasting,Govt.of India.
Ghosal, Ratan Kumar(2002),“Impact of Social Expenditure on International Differential in the Level and Growth and Income – An Empirical Analysis Based on Solow Model and its Augmented Form for the periods (i) 1960-85 and (ii) 1980-98” – Presented at the XIIIth World Congress of the International Economic Association held on September 9-13, 2002 at Lisbon, Portugal and published in CD-ROM version.
------------- (2007), “Augmented Solow Model of Growth and Its Empirical Relevance”, Journal of Quantitative Economics, vol.5, no. 1, January.
-------------- (2005), “Globalization and Inequality”, Indian Economic Journal, vol. 53, no 2, pp 71-86.
--------------- (2006),“Inter-state Disparities in Economic Growth and Human Development in India---What Happens During Liberalisation?”Presented at the 89th conference of IEA held at Kurukshetra University, India in Dec.2006 and published in the conference volume.
Islam, N (1995), “Growth Empirics: a Panel data approach”, Quarterly Journal of Economics, 110, 1127-1170
Islam, Nazrul (1999), "International Comparison of Total Factor Productivity: A Review," Review of Income and Wealth, Series 45, No. 4 (December), pp. 493-518
Islam, Nazrul (2003a), "What Have We Learned from the Convergence Debate? A Review of the Convergence Literature," Journal of Economic Surveys, Vol. June, pp. 309-62
Koopman, Tjalling C. (1965), “On the Concept of Optimal Economic Growth”, in the Econometric Approach to Development Planning, Amsterdam, North Holland, 1965.
Lee, K., Pesaran, M. H., Smith, R. (1997), “Growth and Convergence in a multi-country empirical stochastic Solow model”, Journal of Applied Econometrics, 12, 357-392.
Lucas, Robert E. Jr. (1988),“On the Mechanics of Development Planning”, Journal of Monetary Economics, 22, 1 (July), 3-42.
Ludden, David (2005),”Development Regimes In South Asia”, Economic and Political weekly, Vol.XL, No.37, pp.4042-51. Mankiw, N. Gregory, David Romer and David N. Weil (1992), “A Contribution to the Empirics of Economic Growth”, Quarterly Journal of Economics, 107, 2 (May), 407-437. Romer, Paul M. (1986), “Increasing Returns and Long-run Growth”, Journal of Political Economy, 94, 5 (October), 1002-1037. Rebelo, Sergio (1991), “Long-Run Policy Analysis and Long-Run Growth”, Journal of Political Economy, 99, 3 (June), 500-521.
25
Ramsey, Frank (1928), “A Mathematical Theory of Saving”, Economic Journal, 38 (December), 543-559.
Solow, Robert M. (1956), “A Contribution to the Theory of Economic Growth”, Quarterly Journal of Economics, 70, 1 (February), 65-94. –––––––– (2000), Growth Theory: An Exposition, Oxford University Press. The World Bank: The World Development Report 2006, Equity and Development. Summers, R. & Heston, N. (1988); “ A New Set of International Comparison of Real Product and Price Levels Estimates for 130 Countries, 1950-85”, Review of Income & Wealth, Vol. 34, 1988, 01-25.
Susan M,Collins and Boseworth,Barry,(2003),”The Empirics of Growth: An Update”Brookings papers on Economic Activity,vol.2,pp113-79.
