The Bank Branch Exit Game

We study the process of capacity reduction by multi-plant firms competing in many markets, following a permanent, negative aggregate demand shock. The resulting insight on strategic plant closure is relevant to competition policy. We focus on the measurement of strategic delay in local markets, using data on the closures of bank branches in Spain during the Great Recession. We geolocate each branch and identify its competitors: those that lie within 150 meters. We also cluster the local markets in the same census tract, enabling the use of fixed effects in our regressions. We find that branches with competitors are less likely to close in any given year than branches without, indicative of strategic behavior. Further -only when controlling for possible correlation between demand size and vulnerability to the shock, through fixed effects - the probability of exit is decreasing in the number of competitors. This is the opposite slope of what the literature -unable to use fixed effects - tends to find. We argue that this sign reversal is also rationalizable by a war of attrition. Finally, we confirm that branches are more likely to close if their parent bank has multiple branches in the same local market.