MEASURING THE EFFICIENCY OF AGRICULTURAL BANKS

The first paper in this session outlined the competitive and structura l changes that have occurred among commercial banks in the last decade and the need to explicitly address the efficiency of agricultural ban ks. Frontier efficiency analyses may be used to evaluate the effects o f these changes on agricultural banks. Although many studies have eval uated the efficiency of banks and other types of financial institution s, few have specifically addressed the efficiency of agricultural bank s. The objective of this paper is to examine and discuss several of th e issues involved in estimating the efficiency of agricultural banks. In addition, the results of two agricultural bank efficiency analyses also are presented and discussed. The most common approach used to exa mine bank efficiency is to estimate individual bank cost inefficiencie s utilizing a frontier cost function. Studies that have employed cost frontiers include Ferrier and Lovell, Berger and Humphrey, and Ellinge r and Neff. Individual approaches differ by data availability and form , bank input and output definition, the functional form specified, the time period examined and the empirical inefficiency estimation method employed. Most of these issues are discussed separately below. Estima tes of individual bank inefficiency are used to examine relationships between inefficiency and structural or environmental bank characterist ics. A common approach is to present average bank inefficiency measure s within alternative bank size categories. Also, inefficiency is commo nly regressed on bank profitability measures, bank size measures. and measures of local market share or concentration. These regressions, wh ile typically low in explanatory power, provide a basis for making pol icy conclusions based on results from different samples, geographic re gions, etc. In a recent paper, Berger, Hancock, and Humphrey derive es timates of bank inefficiency using a variable profit function frontier . The profit frontier approach has an advantage over cost frontier app roaches in that bank output inefficiencies also are incorporated into the analysis. Furthermore, because the profit function treats inputs a nd outputs nearly symmetrically, bank input and output definitional de cisions become less important. For example, bank deposits have both in put and output characteristics. In a cost function, their inclusion or exclusion as bank outputs has greater estimation implications. Modeli ng deposits as a netput in a profit function, however, may decrease th is possible misspecification bias (Berger, Hancock, and Humphrey).