Devaluations in Latin America (1993-2023): A study of their effect on GDP using heterogeneous dynamic panel estimators

Abstract

This study examines the effect of real and nominal exchange rates on output in Latin America using a panel of 20 countries with quarterly data from 1993 to 2023. The pooled mean group (PMG) and mean group (MG) approaches were applied, and the Pesaran-Yamagata test revealed heterogeneity in the individual slope coefficients, indicating MG as the preferred estimator. Given the presence of cross-sectional dependence, MG and PMG were employed while accounting for it. Using MG with control for cross-sectional dependence—deemed the most suitable approach—we find that, for the region, the short-term effect of the real exchange rate is positive, although heterogeneous effects are observed within the sample. Argentina and Venezuela exhibit a significant negative effect, while Brazil, Colombia, Costa Rica, Jamaica, Mexico, Paraguay, and Uruguay show a negative but statistically insignificant effect. In the long term, a generally negative impact is observed. Regarding the nominal exchange rate, the short-term effect is negative but not significant at the regional level. When analyzed by country, it is significantly negative in Argentina and Venezuela, persisting in the former even when controlling for government consumption. In most countries, the effect remains negative but not significant. Similarly to the real exchange rate, the long-term impact is generally negative. These findings suggest that devaluations should not be indiscriminately implemented as an exchange rate policy across the region, as multiple transmission mechanisms with opposing effects are at play. Instead, they should be considered only in countries where a net expansionary effect is evident.