A Meta-Analytic Reassessment of the Effects of Inequality on Growth

Neves, P.C., Ó. Afonso, S. Silva (2016), "A Meta-Analytic Reassessment of the Effects of Inequality on Growth", World Development, 78, 386–400.

This paper develops a meta-analysis of the empirical literature that estimates the effect of inequality on growth. It covers studies published in scientific journals during 1994–2014 that examine the impact on growth of inequality in income, land, and human capital distribution. We find traces of publication bias in this literature, as authors and journals are more willing to report and publish statistically significant findings, and the results tend to follow a predictable time pattern over time according to which negative and positive effects are cyclically reported. After correcting for these two forms of publication bias, we conclude that the high degree of heterogeneity of the reported effect sizes is explained by study conditions, namely the structure of the data, the type of countries included in the sample, the inclusion of regional dummies, the concept of inequality and the definition of income. In particular, our meta-regression analysis suggests that: cross-section studies systematically report a stronger negative impact than panel data studies; the effect of inequality on growth is negative and more pronounced in less developed countries than in rich countries; the inclusion of regional dummies in the growth regression of the primary studies considerably weakens such effect; expenditure and gross income inequality tend to lead to different estimates of the effect size; land and human inequality are more pernicious to subsequent growth than income inequality is. We also find that the estimation technique, the quality of data on income distribution, and the specification of the growth regression do not significantly influence the estimation of the effect sizes. These results provide new insights into the nature of the inequality–growth relationship and offer important guidelines for policy makers.