Re-estimating the relationship between inequality and growth

In this paper, we revisit the inequality-growth relationship using an enhanced panel data set with improved inequality data and special attention to the role of transition countries. We base our analysis on the specification of Forbes (2000), but also address the functional form concerns raised by Banerjee and Duflo (2003). We arrive at three main findings: First, similar to Forbes we find a significant positive association between inequality and subsequent economic growth in the full sample, but this is entirely driven by transition (post-Soviet) countries. Second, this positive relationship in transition countries is not robust to the inclusion of separate time effects. Lastly, it therefore appears that this association is not causal but rather driven by the particular dynamics of the transition. Our finding is consistent with the claim that the relationship between inequality and growth emerges due to the particular timing of inequality and growth dynamics in transition countries. In particular, the rise in inequality in the 1990s coincided with a sharp output collapse, leading us to find an association between the large increase in inequality in the early 1990 and a growth recovery in the late 1990s. In sum, once the transition country dynamics are accounted for, we find no robust, systematic relationship between inequality and subsequent growth, neither for levels nor for changes in inequality. These results hold for different lag structures as well as in the medium- rather than the short term, and the empirical patterns observed are robust to the use of different data sets on inequality.