The Efficiency of Bank Branches

Most of the early studies of bank efficiency focused on scale and product mix efficiency, or how close banks are to producing at the scale or product mix that minimizes costs per unit of output. Despite these past efforts, something rather important is missing from most of the analysis according to the authors - evaluation of bank branch efficiency. There has been very little empirical research on bank branch efficiency because branching data generally are confidential and not required by regulators. This paper tries to add to the limited information available about bank branch efficiency. It specifies the Fourier-Flexible nonparametric form for the cost function to characterize the efficient frontier for bank branches, the first application of the form in a frontier efficiency context. This form is applied to over 760 branches of a large U.S. commercial bank for three years of 1989, 1990, and 1991. The authors specify and compare models that employ two different concepts of bank costs and output - the intermediation approach and the production approach. The data set includes information on several types of transactions, allowing for more accurate efficiency estimation. The authors also have information on interbranch transactions, which allows them to account for the services that one branch provides for the customers of another branch. The study is a first of its kind from many perspectives in its methodological approaches, its conceptual approaches that compare and contrast the "intermediation" and "production" approaches to defining branch output, linking the relationship between branch efficiencies and bank efficiencies, the problems associated with bank mergers, and the effects of branch efficiency on price or branch offices to be sold. The empirical results are mutually consistent between the intermediation and production approaches and robust to other variations in specifications. The data suggest that most branches are considerably smaller than efficient scale - there are roughly twice as many branches as are needed to minimize bank cost. However, the average cost curves are relatively flat. The cost of "overbranching" is at most about 13.9% of operating costs and about 3.3% of total branching costs. Some cost scale inefficiency may be optimal from a profitability standpoint, according to the authors, since additional offices provide convenience for the bank's customers that may be recaptured by the bank on the revenue side. The branch level X-inefficiencies dominate the scale effects, similar to the findings in bank-level research. The branching data reveal that some of the bank's branches do not perform to the level of its own best practice branch, which is a necessary condition for full efficiency. The bank may still be fully efficient relative to a conventional bank frontiers which allows for banks to be measured as efficient even if they have branches that are inefficient. The dispersion of measure X-efficiency also suggests that the quality of the local managers also is important in determining performance. The authors believe branch inefficiencies may help explain the often-measured scale diseconomies at the bank level for large branching banks, as the additional offices attract more total customers for the bank as a whole. They also may help explain the large cost X-inefficiencies typically found at the bank level, since the total cost of producing the bank's same total output level could be reduced by having fewer branches, each producing more output. The empirical results also suggest that it may be very difficult for banks to achieve large branch cost saving through mergers. The ability to achieve savings by closing branches that are below cost-efficient scale or are very X-inefficient depends upon the proximity of other branches that are both relatively X-efficient and can absorb the additional output without significant scale diseconomies. The potential for improvements from superior bank management also appears to be limited by the importance of the quality of local branch mangers. The empirical results also suggest there are substantial differences in the value of bank branches that depend on efficiency. An efficient branch may be worth more than twice as much as an inefficient branch with the same deposit base. Premium paid data confirm this implication. The authors suggest that banks may be able to improve the efficiency of their branching networks by using efficiency measures along with their own performance measures. Relative efficiencies may be used as an incentive or monitoring device. Observation of the most efficient and least efficient offices also could help discover efficient and inefficient practices, respectively, that may be used to improve management policies and procedures.