Anda di halaman 1dari 8

A Review of Journal Article : Jih Y. Chang and Rati Ram.

Level of Development, Rate of Economic Growth, and Income Inequality. Economic Development and Cultural Change, Vol. 48, No. 4
(July 2000), pp. 787-799

By : La Ode Sabaruddin
Introduction

The study of the relationship between income inequality and the level of economic development was heavily influenced by the Kuznets inverseU hypothesis. According to Kuznets (1955) there should exist an inverted-U correlation between income inequality and the level of economic development. The inverted-U correlation suggests that for low levels of economic development there will exist low levels of inequality in per capita GDP, and inequality will increase as income increases. However, as incomes reach a particular level, i.e., as the economy develops past a certain point, the relationship reverses and income inequality decreases. In other words, income inequality would change as economic growth changes or more precisely, income inequality would rise first and then decline with economic growth. As a result, structural transformation experienced by low-income countries would ignite a sequence of increasing and then decreasing inequality. Kuznets empirically documented this relationship for nations transitioning from agricultural to industrial economies. Numerous studies have tried to assess the pattern of evolution of income inequality during the course of modern economic growth whether the data are consistent with the inverted-U pattern suggested by S. Kuznets. However, the overall empirical evidence does not show a clear picture. Several researchers, including S. Anand and S. Kanbur (1993), and K. Deininger and L. Squire (1996), have concluded that multicountry data do not support Kuznetss hypothesis. Saith (1983) is skeptical about the validity of the paradigm and argues that the U-Hypothesis is more of a hindrance than an aid to understanding the relationship between economic development and income inequality. On other hand, some other studies including those by S. Randolph and W. Lott (1996), R. Ram and S. Jha(1996), Fielding and Torres (2005), have reported evidence that is

the results conformed to Kuznets hypothesis (supportive of the inverted-U pattern). Fishlow (1996) has also suggested that a complete dismissal of the Kuznets proposition might be in error. Nevertheless, very few studies have investigated how the rate of growth is related with inequality at various levels of development (or income), except perhaps incidentally and through an additive growth term in Kuznets-type models of inequality. Motivation of the Paper The principal objective of Chang and Ram paper is to extend the earlier research in order to estimate the income parameters for the high growth and the low-growth countries and to compare the evolution of inequality in the two groups at various income levels in terms of role of growth rate in the determination of inequality. However, Chang and Ram paper does not necessarily presume that there is an inverted-U pattern. Presence or absence of the inverted-U in either group is determined by the data and become secondary purpose for their study. Chang and Ram study is mainly enhanced by recent interest in the relationship between inequality and growth, which has been studied by A. Alesina and D. Rodrik; T. Persson and G. Tabellini; R. Benabou; J. Furman and J. Stiglitz; and H. Li and H. Zou. Through their study, Chang and Ram would explore more completely the role of growth rate in the determination of inequality. Argument of the Paper In their paper, Chang and Ram evaluate the Kuznets hypothesis using a common functional form that includes an additive growth term. They estimate the following regression : Ineqi = a1 + b1(LYi) +c1 (LYi) + d(GYi) + u1i
2

(1)

where Ineqi is the Gini coefficient for country i, LYi is the natural log of income per capita, GYi is the rate of growth of income per capita, and u1i is a disturbance term that is assumed to be normal and identically distributed (NID).

The data that Chang and Ram use come from a variety of sources. The Gini coefficient measuring income inequality are taken from the the high quality observations of the Deininger and Squire (1996) data set for or around the year 1985. The growth variable comes from the United Nations Human Development Report (1993) and is averaged over the years 1980-90. On the other hand, the per capita GDP variable comes from the Penn World Tables (PWT), Mark 5.6a of Summers and Heston. The usable observations is 65. To investigate difference between the high-growth and the low-growth countries, Chang and Ram stratify their multi-country data by growth rate : high-growth countries (2% and above) of 17 observations, low-growth countries (zero or negative growth) of 23 observations, and excluded 17 observations of medium (between zero and 2%) in order to sharpen the comparison. Their final regression estimates the following : Ineqi = a2 + b2(LYi) + c2(LYi) + a22D + b22(D LYi) + c22(D (LYi) ) + u2i
2 * * 2

(2)

where D is a dummy variable takes the value of 1 for high growth countries and 0 for low growth countries. The coefficient of dummy variable indicates the difference in the constant term between the two groups, with the low-growth group being the base. While D(LY) and D(LY)2 indicates the interaction terms, and their coefficients express the difference between the two groups with regard to the parameters of the income and income-squared terms. The Main Claims of the Paper The main claims of Chang and Ram in their paper found that high economic growth is associated with low income inequality at all levels of income. The estimates and simulations show a statistically significant and quantitatively substantial structural difference between the two groups, and the high-growth scenario is characterized by lower inequality at all income levels. High growth countries have lower levels of income inequality over all levels of income than do low growth countries. Moreover, the high-growth advantage seems particularly large at low income levels. The results are consistent with the view that there need
3

not be a trade-off between equity and growth and that the two could be mutually reinforcing. Assesment of extent which the Claims are Justified Chang and Ram estimates their model using Ordinary Least Square Method (OLS). The Results are provided in the following table : Table 1 Estimated Parameters of the Basic Kuznets Quadratic for High-Growth and Low-Growth Regimes

Coefficient Constant LY LY2 D D X LY D X LY2 R2 N

Parameter t-statistic

-212.774 -1.62

65.336* -4.106* -193.792 43.875 1.92 -1.87 -0.89 0.82

-2.529 0.40 -0.77 a

48

* P 0.10 a The standar error of the regression is 7.086 and the F-statistic is 3.42 for the null hypothesis that the coefficient of D, D X LY, and D X LY2 are all zero is 2.66 while the P value is 0.06 Notes : - High-growth countries using critea an annual growth rate of at least 2% in real GNP per Capita during 1980-1990 - Low-growth countries using critea an annual growth rate of zero or negative in real GNP per Capita during 1980-1990 - The low-growth countries is the base, and D takes the value of 1 for highgrowth countries and 0 for low growth countries. The D X LY and D X LY2 are interaction terms, and their coefficients express the difference between the two groups with regard to the parameters of the income and income-squared terms. Table 1 reports estimates for equation (2) and resulting support of the Kuznets paradigm in both groups, and the parameter estimates for the income terms (for the base group) are statistically significant at least at the 10% level. While the individual parameter estimates for the high-growth dummy and the interaction terms have low statistical significance due to collinearity, the three terms are jointly significant at least at the 10% level. To reveal more clearly the

implications of the parametric structure for the growth-inequality nexus in the two groups, Chang and Ram generates from the estimates in table 1 to have simulated (predicted) values of the Gini coefficient for various income levels in each group and compared the simulated inequality profiles. The simulated results are : Table 2 Simulated (Predicted) Values of Gini Coefficient in High-Growth and Low-Growth Countries at Selected Levels of Real GDP per Capita Predicted Coefficient in the cases of Real GDP per Capita* 750 1,000 1,500 2,000 3,000 4,000 5,000 7,000 9,000 11,000 13,000 15.000 17.000 High Growth 25.84 31.45 37.48 40.44 42.75 43.06 42.55 40.53 38.04 35.44 32.88 30.39 27.99 Low Growth 40.00 42.83 45.66 46.85 47.36 46.92 46.10 44.09 41.99 39.94 37.98 36.11 34.35

* Real GDP per capita is in international dollars at 1985 prices Table 2 shows the value of predicted Gini for each country in the lowgrowth and high-growth groups on the basis of the estimates in Table 1. Simulated values for each group at selected income levels between $750 and $17,000 show that the simulated (predicted) values of inequality are substantially lower in the high-growth countries than in the low-growth group, at all income levels, and the difference is particularly large at lower income levels. Graphically, the following figure shows plots of the simulated values for the Gini coefficient in each group.

Figure 1 Plots of the simulated values of Gini coefficient (GINI) in high-growth and low-growth countries, based on the estimates in table 2.

Figure 1 shows clearly that the high-growth countries are marked by considerably smaller inequality at all income levels than the low growth countries and that the difference is particularly large for high-growth countries than low growth countries at low income levels. The results are consistent with the view that there need not be a trade-off between equity and growth and that the two could be mutually reinforcing.

Identify the Weaknesses of the Paper

The data on the Chang and Ram study was Multi-country cross-sectional which comes from different regions, possibly leading to regional fixed effect. Theoritically, it might expect that for political and socio-economic reasons,

inequality in incomes may spill over from one nation to another within a particular geographical region. Unfortunately, Chang and Ram doesnt provide assessment to determine whether significant regional fixed effects exist or not in their specifying model. It is very critical aspect because ommiting relevant regional fixed effect may take biased and inconsistent results. If regional fixed effect exists, it causes the assumption of identical distribution of errors in Ordinary Least Squares regression to break down resulting in biased and inconsistent estimators. Hence, no inference can be drawn accurately. Thus, knowing of regional fixed effect is very useful to drawing accurate statistical inference from the estimated parameters. Relevant critics on this issue was reported by Jeff Edwards and Anya McGuirk (2004). Chang and Ram test Kuznetss hypothesis and examine

simulated/predicted gini coefficient profiles for high-growth and low-growth countries. The quality reliability and properties of the inferences drawn from both these activities depends on the validity of the statistical assumptions underlying the model estimated for the data they are using. Like many researchers, Chang and Ram choose to estimate the simplest, most widely used model in this field of work. This simple model selection criterion is very common in applied work. Unfortunately the popularity and simplicity of a regression model does not guarantee that the model assumptions are reasonable for the data being analyzed and, thus, that the analysis carried out using the model will lead to reliable inferences. It is in this sense that the model estimated by Chang and Ram is ad hoc. They simply choose the model, subjected the model to the data, and drew conclusions regarding the Kuznets hypothesis using the reported t-statistics, and regarding the differences between high-growth and low-growth countries, using predictions from the model. While Without attempting to assess whether the statistical assumptions underlying this model are valid for the data choosen. It is also unclear why this type of quadratic structure was chosen among many others that produce the same shape, whereas Randolph and Lott (1993) address seven possible forms, with supporting results, for the inverted-U pattern of the Kuznets hypothesis.

References : Albert Fishlow. 1996. Inequality, Poverty, and Growth: Where Do We Stand? in Annual World Bank Conference on Development Economics, 1995, ed. Michael Bruno and Boris Pleskovic (Washington, D.C.: World Bank) pp. 25 39.
Chang, Jih Y., and Rati Ram. 2000. Level of Development, Rate of Economic

Growth, and Income Inequality. Economic Development and Cultural Change 48(4): 787-799 Jeff Edwards and Anya McGuirk. 2004. Kuznets Curveball : Missing the Regional Strike Zone, Econ Journal Watch, Volume 1, Number 2 pp 222-234. Klaus Deininger and Lyn Squire. 1996. A New Data Set Measuring Income Inequality. World Bank Economic Review 10(3): 565-591. Kuznets, Simon. 1955. Economic Growth and Income Inequality. American Economic Review 45: 1-28 Randolph, Susan, and William F. Lott. 1993. Can the Kuznets Effect Be Relied on to Induce Equalizing Growth? World Development 21(5): 829-840. Saith, A. 1983. Development and Distribution: A Critique of the Cross-Country U-Hypothesis. Journal of Development Economic 13. Sudhir Anand and S. M. R. Kanbur.1993. Inequality and Development: A Critique. Journal of Development Economics 41: 1943

Anda mungkin juga menyukai