This paper extends the ongoing discussion on optimal exchange rate regimes to the issue of pro-poor growth. To analyze empirically the poverty effects of exchange rate regimes, we estimate the distribution effects of different exchange rate arrangements on the poorest 20 and 20 to 40 percent. In addition, we test the total effect, i.e. the distribution and growth effect, to capture potential trade-offs between poverty effects through overall economic growth and distribution. To analyze this question, we collect an irregular and unbalanced panel of time-series cross-country data on the first and second quintile share from 76 countries and use two recently proposed de facto exchange rate regime classifications, Levy-Yeyati/Sturzenegger (2002) and Reinhart/Rogoff (2003). To cover econometric issues, cross-country variation and dynamic aspects of within-country changes of the income of the poor, we apply two econometric specifications, a growth equation and a system GMM estimation. We estimate the poverty effects of different exchange rate regimes for all countries and, separately, developing and industrial countries due to considerable differences in economic structure, access to international capital markets and soundness of domestic financial systems. Empirical findings vary considerably with respect to three aspects. First, findings for the Levy-Yeyati/Sturzenegger (2002) and Reinhart/Rogoff (2003) classification differ significantly with respect to similar exchange rate categories. Thus the classification process of exchange rate regimes affects critically the policy conclusions. Second, statistically significant exchange rate regimes in the Reinhart/Rogoff (2003) classification impact positively on the poor in developing countries, but negatively on the poor in industrial countries. Thus exchange rate regimes affect very differently the poor in developing and industrial countries in the Reinhart/Rogoff (2003) classification. Third, statistical significance of exchange rate regimes in the system GMM approach differs considerably for adjusted and unadjusted income inequality measures. Due to these varying and only weakly robust empirical findings, a concise policy recommendation with respect to poverty-reducing exchange rate regimes is difficult. Nevertheless, positive effects of intermediate regimes of the Reinhart/Rogoff (2003) classification in developing countries should be emphasized, showing at least a tendency to not negative and possible positive effects of intermediate regimes on the poorest 40 percent in developing countries.
