Description

Given the theoretical disagreement on the e ff ect of a devaluation on output, empirical evidence plays an fundamental role. Empirical fi ndings have favored the view that devaluations are contractionary. In this paper the author argues that what stands behind these results is selection bias. In theory, the same variables that determine the probability of a devaluation determine the rate of growth of output. The author controls for selection bias using matching estimators and extensive dataset of 155 countries for the period 1970-1999 that includes as many as 264 devaluation episodes. Not controlling for selection bias, devaluations appear to be contractionary, in line with previous fi ndings. Nevertheless, when the author controls for this bias, the contractionary e ff ect vanishes. Extensive sensitivity analysis is provided.