Thrift Scale Economies: An Alternative Approach

Article excerpt


The trend toward consolidation in the thrift industry received a substantial boost from the passage of the Financial Institutions Reform, Recovery and Enforcement Act (FIRREA) in August 1989. FIRREA came after ten years of increased financial pressure on the industry as a result of earlier moves toward deregulation. FIRREA imposed higher capital requirements and higher deposit insurance premiums on thrifts while it allowed less liberal asset powers. The question of how this consolidation will affect operating efficiency is a subject of considerable interest.

The two principal recent studies of thrift scale economies are LeCompte and Smith (1990) and Mester (1987). While the findings differ somewhat depending on model specification and sample period, the basic conclusion from both studies is that, for the early 1980s, returns to scale are approximately constant.(1) Both studies employ a multiproduct approach, which has been used increasingly in the scale economies literature over the past several years. This approach can result in severe multicollinearity, however, when applied to a small sample. The Mester (1987) study produces a negative estimate of marginal cost, which, according to Berger, Hanweck, and Humphrey (1987, p. 509), "theoretically invalidates measures derived from these cost functions." In addition, in a major review of bank cost studies, Gilbert (1984) has written that because of small, geographically restricted samples, existing studies do not provide evidence that is relevant for many policy issues. This article argues that there are advantages to a large sample, single product study as a check on the conclusions of smaller sample, multiproduct studies of scale economies. We apply such an approach to the thrift industry using data for the period 1986 to 1989.(2) Another purpose of this paper is to provide estimates of scale economies for several specific size classes, something that is not always done in other studies.(3) We briefly summarize recent research on scale economies at commercial banks and thrifts and then discuss single product and multiproduct cost studies. Our data and estimation method and empirical results follow.


Evidence on scale economies in the financial services industry is mixed. Early banking studies summarized in Benston (1972) conclude that scale economies are present throughout a wide range of size classes. Some later studies of commercial banking cost structures (i.e., Benston, Hanweck, and Humphrey, 1982; Benston, Berger, Hanweck, and Humphrey, 1983; and Gilligan, Smirlock, and Marshall, 1984) conclude that scale economies are exhausted quickly beyond a relatively small asset size and that cost curves are either flat or upward sloping at higher output levels. This literature is summarized in Gilbert (1984). In contrast, Murray and White (1983) find evidence of scale economies in credit unions.

Studies of savings institutions similarly have yielded conflicting results. Goldstein, McNulty, and Verbrugge (1987) and McNulty (1982, 1983) report evidence of scale economies, while LeCompte and Smith (1990) and Mester (1987) suggest approximately constant returns to scale in most model specifications. The Mester (1987) study uses a sample of California institutions in which over one-half of the firms are state-chartered stock firms that have engaged in diversified activities for some time. The study by LeCompte and Smith (1990) uses a sample of 431 institutions from the southwestern portion of the United States for two years, 1978 and 1983. In contrast, the study by Goldstein, McNulty, and Verbrugge (1987) is based on four years of pre-1980 data for all FSLIC-insured institutions. Thus, the latter results reflect the effects of more specialized activities (primarily residential lending) in a highly regulated era. In addition, this study excludes interest cost from the definition of total cost. …