An entropy approach to size and variance heterogeneity in U.S. commercial banks

In this paper, we investigate the effect of bank size differences on cost efficiency heterogeneity using a heteroskedastic stochastic frontier model. This model is implemented by using an information theoretic maximum entropy approach. We explicitly model both bank size and variance heterogeneity simultaneously. We find that non-performing loans, federal insurance premium, legal expenses and director fees drive bank inefficiency as the bank becomes larger. Moral hazard, bank management and a “too big to fail” doctrine are likely explanations for the results from this study.