Abstract
In recent years, governments around the world have been opening up their banking systems to foreign competition with the expectation of making them more efficient, stable, and resilient to shocks. Academics and policymakers have therefore been exploring the effects of foreign bank entry from both a theoretical and an empirical point of view. To explore the impact of foreign entry, this paper builds on the methods employed by Martinez Peria and Mody to study interest rate spreads in foreign banks in Latin America. The results suggest that researchers interested in the impact of foreign bank entry should be cautious about drawing inferences from either pooled ordinary least squares or random-effects regressions. The analysis of the Mexican data suggests that foreign acquisition is not randomly assigned and that the characteristics of the banks offered for sale to foreigners may correlate with the outcomes of interest. Unless researchers expunge such unit heterogeneity, they may draw spurious conclusions or infer causality erroneously.