The Relative Cost Efficiency of Stock Versus Mutual Thrifts: A Bayesian Approach

Abstract

The relative cost efficiency of the mutual versus stock forms of ownership for thrifts has been a relevant issue in an era of deregulation and competition in the financial services industry. In this study Bayesian-based Markov chain Monte Carlo (MCMC) re-sampling methods are used to solve a stochastic cost frontier model and effectively determine cost efficiencies for the stock and mutual thrift groups. We find a statistically significant difference between both the cost frontiers and cost efficiencies of the two groups, with the stock group operating at the lower-cost point. Agency problems explain a significant portion of the cost efficiency difference. Capital structure differences, though not helping to explain differences in cost efficiency, do help to explain differences in cost structure and managerial attitudes toward risk.

Keywords: thrifts, cost efficiency, stochastic cost frontier, Gibbs sampler, data augmentation

JEL Classification: G21, C11, C15