Bank Competition, Cost of Credit and Economic Activity: evidence from Brazil

We estimate the effect of bank competition on financial and real variables. To do so, we use regional heterogeneous exposure to mergers and acquisitions (M&A) of large banks as an instrument for changes in competition of local banking markets in Brazil. We use detailed administrative data on loans and firms and a difference-in-differences empirical strategy that compares changes in outcomes for markets affected or not by the M&A episodes. We find that, following M&A episodes, spreads increase and lending decreases persistently in exposed markets relative to controls. We show that this is larger for more concentrated markets at the time of the M&A episode, and is unlikely to be driven by other factors, such as branch closures. We find that non-agricultural employment falls .2% for an increase of 1% in spreads. We develop a model of heterogeneous firms and concentration in the banking sector that is consistent with the micro evidence. In our model, the semi-elasticity of credit to lending rates is a sufficient statistic for the effect of concentration on output, which we estimate to be -3.17. Among other counterfactuals, we show that if Brazilian spreads fell to world levels, output would increase by approximately 5%.