Elections, exchange rates and reform in Latin America

In this paper, we study the link between real exchange rate (RER) depreciation and elections in Latin America. Our contribution is threefold. First, we employ a statistical model that takes into account the pervasive conditional heteroskedasticity found in financial data and includes a wide range of macroeconomic variables as regressors. Second, we test whether the wave of central bank reform that swept the region has had any effect on the existence or strength of the electoral cycle in exchange rates. Third, we test an additional hypothesis, namely, that financial liberalization may also be an important variable explaining changes in electoral effects on the real exchange rate. In a panel of 9 Latin American countries with available macroeconomic data and a history of exogenous election dates, we confirm the previous findings that real depreciation intensifies after elections even when modeling the significant conditional heteroskedasticity in these data. We also show, for the first time in the literature, that post-election exchange rates are significantly less predictable. We go on to test whether central bank reform has influenced the way in which elections affect the RER in Latin America. If reform has been effective at reducing political manipulation of the exchange rate, then any relationship we see between elections and the RER before central bank reform should be mitigated in the post-reform era. We find that the relationship disappears after reform and that post-reform real exchange rates are also significantly less volatile. Finally, we show that financial liberalization seems to have a stronger effect on the conditional variance of the RER than does central bank reform, but reform has a stronger impact on the conditional mean. (C) 2009 Elsevier B.V. All rights reserved.