ABSTRACT The average FDIC loss from selling a failed bank is 28% of assets. We document that failed banks are predominantly sold to bidders within the same county, with similar assets business lines, when these bidders are well capitalized. Otherwise, they are acquired by less similar banks located further away. We interpret these facts within a model of auctions with budget constraints, in which poor capitalization of some potential acquirers drives a wedge between their willingness and ability to pay for failed banks. We document that this wedge drives misallocation, and partially explains the FDIC losses from failed bank sales.

We develop a structural empirical model of the US banking sector. Insured depositors and run-prone uninsured depositors choose between differentiated banks. Banks compete for deposits and endogenously default. The estimated demand for uninsured deposits declines with banks' financial distress, which is not the case for insured deposits. We calibrate the supply side of the model. The calibrated model possesses multiple equilibria with bank-run features, suggesting that banks can be very fragile. We use our model to analyze proposed bank regulations. For example, our results suggest that a capital requirement below 18 percent can lead to significant instability in the banking system. (JEL E44, G01, G21, G28, G32)