A canonical correlation analysis

The banking structure-performance relationship has been the subject of many studies (Heggestad, 1979). This paper addresses two problems associated with previous research through analysis of the structure-performance relationship in the savings and loan association industry. One problem is that most studies estimate the structure-performance relationship with multiple regression analysis. The problem with using this technique is that only a single measure of performance may be studied at one time, rather than a set of measures. Such a set might include hours of operation, loan rates, deposit rates, service charges and profitability. A more general study ofperformance should include more than one measure. A second problem is that it is difficult to determine the relative contributions of all factors which affect performance, such as demand, product market, costs and market structure factors. It would prove useful to determine the relative importance of factors which affect performance when considering changes in regulatory policy. This paper avoids these problems through the use of canonical correlation analysis. This technique allows one to test for the effects of explanatory factors upon multidimensional measures of performance. While this technique is not widely used in the structure-performance literature, it has been used by Fraser et al. (1974) in their study ofcommercial banks in Texas for the years 1969 and 1970. Fraser et ai. (1974) find that market structure variables are of the second order of importance in explaining measures of performance, with cost, deposit and loan composition, and demographic factors being relatively more important. 1 The tests presented here are on savings and loan associations operating within 153 Standard Metropolitan Statistical Areas (SMSAs) in 1979. The conclusion that is reached in this paper is that market structure is very important in determining performance. Reasons to explain the difference between the two studies are presented as well as implications for regulatory policy.