نابرابری درآمد :دوره پس از آزادسازی بازار سهام در بازارهای در حال ظهور

Early research has documented that the large-scale equity market liberalizations of the last decade led the subsequent rise in aggregate equity indices, investment booms, capital flows and economic growth. An important and unaddressed issue is the normative question of whether and how these reforms shifted the distribution of incomes in the aftermath of equity market liberalization. In careful empirical analysis, we find a pattern indicating that income share growth accrued almost wholly to the top quintile of the income distribution at the expense of a “middle class” that we define as the three middle quintiles of the income distribution. A surprising finding is that the lowest income share remained effectively unchanged in the event of liberalization. These patterns are robust to the inclusion of a wide variety of controls for global shocks, country-specific factors, and contemporaneously implemented privatization and stabilization policies.

In the latter half of the 1980s and the early years of the 1990s, the governments of over two dozen sovereign nations began to implement a wave of major economic reforms, which included capital account liberalization, privatization and/or a host of stabilization policies. This large-scale experiment has fuelled an active academic and popular debate on the causes and consequences of these reforms, in part because the aftermath of such reforms is important to nations which, to date, are considering liberalizing reforms. Empirical evidence on the consequences of these reforms will therefore be an important tool in assessing how best to implement similar liberalization policies in reform-minded countries. In this context, a nascent but important body of empirical research has begun to emerge, analyzing important questions such as the relation of capital account liberalization to (i) emerging market equity prices (e.g., Bekaert and Harvey, 2000a, Bekaert and Harvey, 2000b, Kim and Singal, 2000, Henry, 2000a and Froot et al., 2001), (ii) liquidity (Levine and Zervos, 1998), (iii) private investment Levine and Zervos, 1998, Henry, 2000b, Bekaert and Harvey, 2000a and Bekaert and Harvey, 2000b, (iv) equity flows (Bekaert et al., 2002) (v) and economic growth (Bekaert et al., 2001). This literature has found strong evidence that suggests capital account liberalization is associated with higher equity prices, lower cost of capital, investment booms, greater capital flows and higher growth. To date, an important but unaddressed question in this literature is the issue of whether and how these reforms have shifted the distribution of incomes in the reforming countries. This is an important area of research for several reasons. For example, the finding in Bekaert et al. (2001) that average economic growth increased after liberalization raises normative issues about the allocation of the generated wealth. One would presumably evaluate the success of liberalizing reforms differently when average growth uniformly raised incomes for all quantiles of the distribution, from a finding that the average growth post-liberalization was only influenced by gains to the upper tails of the income distribution. For many countries that are still considering capital account liberalization, it is important to evaluate the benefits of market liberalization such as investment booms relative to the potential downsides of such policy reforms, so that future reforms may be tailored to alleviate any negative fallout from undertaking such economic reforms. This paper takes a step in this direction by presenting the first set of results on the association of capital account liberalization with income inequality. In this paper, we heuristically and empirically describe the dynamics of the shifts in income distributions in a sample of 11 countries that undertook extensive economic reforms between 1986 and 19951. Prior to our discussion of the empirical work, we consider various mechanisms that link capital account liberalization to income inequality. We do not intend these mechanisms to be causal in either direction, but to provide a framework in which the finding of an empirical link between the two variables should not necessarily be discarded as spurious. We analyze income distribution changes by comparing the size of three income shares before and after liberalization, conditional on a set of country-specific factors, and contemporaneous global shocks. We study the share of GDP held by the top quintile, the lowest quintile, and a group we will henceforth refer to as the middle class, which represents the sun of the three middle quintiles. Because we track the changes in income shares which plausibly respond slowly to economic reforms (relative to equity prices or the dividend yield), we analyze movements in the distribution over the “short run” which we take as the first 2–4 years beyond the year of the first liberalizing reform. This aspect of the analysis is discussed in more detail when we present our methodological framework and the construction of event windows. We focus mainly on capital account liberalization, but also control for other reforms such as privatization and stabilization policies that were contemporaneously implemented. We will use the terms financial market liberalization, stock market liberalization and capital account liberalization synonymously for our purposes. Regression results in this study are given both for simple linear models as well as event-study models, which are somewhat incongruous in this context but useful nevertheless. The principal findings of this research indicate systematic shifts in the income distribution in the first 4 years after a country's first liberalization reform. First, the raw data indicate that in the 11 emerging markets we consider, nine experienced a growth in the top quintile's share of income, and the mean share held by the top quintile rose by 1.3 percentage points. Second, one mildly surprising finding is that there is no discernible change in the mean income share of the lowest quintile. In the sample of 11 countries, four experience declines in the lowest quintile's income share, seven experience growth, and the mean share held by this group rises by 0.1%, which is statistically indiscernible from zero in regression analysis. This finding supports a recent finding in Dollar and Kraay (2001) that income growth of the poor is one-for-one with mean income growth of the population, which implies a constant income share for the poor. Based on these two strands of evidence and a basic adding-up theorem, there must be a reduction in the aggregate income share held by the middle class in the aftermath of capital account liberalization, and we find this to be the case. Specifically, in 9 of the 11 emerging markets in the sample, the aggregate income share of the middle class falls after liberalization. The raw data indicate a mean reduction of 1.45 percentage points in their share of income. We test this pattern in careful regression analysis, to control for variations in domestic fundamentals, contemporaneous domestic and global business cycle effects, and country-specific effects. The regression results indicate that the income share of the middle class is strongly negatively associated with liberalization, and this relation persists in the presence of a wide variety of controls. In general, this finding is strongest for the 3rd quintile but holds statistically for the 2nd and 4th quintiles as well. In contrast, there is no statistically significant relation between liberalization and the income share of the lowest quintile in the presence of a rich set of controls. The empirical evidence for the upper quintile points to a positive and statistically significant relation between its income share and the event of liberalization, in accordance with the raw mean increases in their income shares. However, for certain regression specifications this relation attenuates. Because our regression estimates are sensitive to the choice of the regressor vector, we report a variety of results. We do not lend any causality interpretation to our results because of the well-known problems of identification. Instead, all of our results are presented as conditional correlations, and are open to varying interpretations, some of which we will discuss in the context of the regression results. It is important to stress that the described patterns hold for income shares. In the sample of emerging markets we consider, average income of the top quintile rose in all 11 nations, average income of the middle class as well as the lowest quintile rose in 10 of 11 nations.2 The key results obtained in this research complement, and should be viewed in the context of, some important research that precedes this, e.g., Levine and Zervos (1998), Henry, 2000a and Henry, 2000b, Bekaert and Harvey, 2000a and Bekaert and Harvey, 2000b and Bekaert et al. (2001). For instance, our finding that average middle class income share decreased while the average income share of the upper quintile simultaneously increased, coupled with the finding of Bekaert et al. (2000) that mean growth increased post-liberalization, indicates that the “pie grew” upon liberalization, but that the generated wealth was disproportionately allocated to the upper tail, suggesting a mean-increasing distributional shift. Without the result of Bekaert et. al., our finding could not reject the hypothesis that liberalization leads greater inequality with a mean-reducing distributional shift. The entire body of results jointly raises normative issues about capital account liberalization and the subsequent welfare of the population, and questions of the trade-off in mean income growth and the concomitant growth in inequality. Although the latter topics are of academic interest, they are not explicitly discussed here, but will be addressed in future research. To better comprehend the dynamics of these shifts in the distribution post-liberalization, we complement the regression analysis with nonparametric tests of stochastic dominance of the pre-liberalization and post-liberalization income distributions. Logically, combining our results with that of Bekaert et al. (2001) yields a very precise set of null hypotheses. First, patterns of income levels described in the current research indicate that the average post-liberalization distribution of incomes should exhibit First Order stochastic dominance (FOSD) over the pre-liberalization distribution. Similarly, SOSD should be obtained. Second, the finding that liberalization on average raises mean income growth in Bekaert et al. (2000), coupled with our finding that this growth appears to be driven entirely by the upper quintile, indicates that we should not reject Lorenz dominance.3 We provide formal tests of each of these hypotheses to provide a larger context for the results of our regression analysis. An important issue that often arises in studying emerging markets phenomena is that of dating a country's first liberalizing reform. In this respect, our research was aided greatly by Bekaert and Harvey, 2000a and Bekaert and Harvey, 2000b, and Henry (1999, 2000a), whose work has compiled official liberalization dates on overlapping sets of the sample of 11 liberalizing countries we consider.4 Also useful to this paper is the work of Deninger and Squire (1996), which has assembled panel data on socioeconomic metrics for a large cross section of nations. Although the focus of this research is on the association of financial reforms with income inequality, it is directly related to the vast literature in growth that has studied amongst other issues, growth and income inequality (some examples include Barro, 1990, Barro and Salai-Martin, 1995, Perotti, 1996, Banerjee and Duflo, 2000 and Dollar and Kraay, 2001). This paper integrates some aspects of the growth research into studying emerging markets phenomena, by exploring income inequality in the aftermath of stock market liberalization. We will focus on the share of income owned by the jth quintile as the dependent variable for the greater part of our empirical analysis (where j=l represents the lowest, j=h represents the highest quintile and, aggregating the three middle quintiles yields a “middle class” with j=m). One might consider studying the variation in the Gini coefficient of inequality to analyze income distributions pre- and post-liberalization; one drawback of this measure, however, is its inability to distinguish between distributions that are unequal in very different ways, e.g., thick-tailed at the upper end, versus thick-tailed at the lower. Furthermore, income shares are the relevant variable to analyze in studying relative gains, and in inferring the process of resource allocation in the liberalizing nation. Normative analyses about the welfare of the middle class whose income share falls in the event of capital account liberalization must therefore be tempered to allow for the possibility that the level of income in the middle class rises after liberalization. As indicated earlier, middle class income increased contemporaneously with the decrease in their share of national income in 10 of the 11 nations in our sample of liberalizing nations. The remainder of the paper is organized as follows. In Section 2 we discuss the sample of emerging market nations in our study, provide some preliminary findings, and describe various mechanisms that link capital account liberalization with the observed changes in income shares. Section 3 describes our empirical methodology; results from both standard regression analysis and event study models are reported here. In Section 4, we discuss tests of stochastic dominance and Lorenz dominance, then describe the findings of our tests. Section 5 considers whether it is possible to give a causality interpretation to our findings. Conclusions follow. Appendix A details the data construction of this study, and the construction of some of our test statistics.

This paper presents evidence of a strong statistical association between the event of liberalization and income shares. The data strongly support a positive coefficient between liberalization and the highest income quintile's share of mean income, and a negative coefficient between liberalization and the middle class income share. For our study, the middle class represents the aggregate sum of the three middle quintiles. We find no evidence of any statistical association between liberalization and the lowest income quintile. Although the middle class “suffers” in the wake of a liberalizing reform while the upper quintile gains, this statement is true for income shares. We find that income levels in liberalizing nations almost universally rise after liberalization. We provide tests of stochastic dominance of the pre- and post-liberalization income distributions to complement our regression analysis. The patterns we describe persist in the presence of a wide variety of controls for domestic fundamentals, world business cycle movements and country-specific factors. They are also robust to the addition of controls for contemporaneous domestic reforms. We find that equity revaluations affect income shares differentially before and after liberalizations. They also affect income shares differentially across liberalizing and non-liberalizing nations. Because liberalizing and other major domestic reforms are rarely implemented in a vacuum, we do not press on causality. It is possible that endogenous policy decisions will attenuate some of the aforementioned correlations that we document. However, it is important to note that there are mechanisms which should relate capital market liberalizations to income shares under a wide variety of hypotheses that are true in emerging markets (e.g., differential access to credit markets, limited stock market participation, and the tight links between upper quintiles and policy makers). In any event, the patterns we describe should prove useful in the debate on emerging markets phenomena, and add to the research that analyzes the aftermath of capital account liberalization in emerging markets.