A Canonical Correlation Analysis of Commercial Bank Asset/Liability Structures

Abstract

Commercial banks have been the subjects of a large body of empirical re? search employing regression and econometric models and discriminant analysis. The purpose of this paper is to empirically identify and describe relationships, including hedging behavior, between the asset side and the liability/capital side of the balance sheets of a cross-section of large U.S. banks. Canonical correla? tion analysis is the statistical technique that is employed. Unlike regression anal? ysis which explains the behavior of a single dependent variable as a function of a set of independent variables, canonical correlation analysis relates two sets of variables. In the present case, one set of variables is the composition of the lefthand side of the balance sheet and the other set is the right-hand side. The vari? ables used in this study are asset and liability/capital categories expressed as a proportion of total bank assets (i.e., a percentage breakdown of the balance sheet or a common size statement). These proportions are used in lieu of the more usual financial ratios and no information exogenous to the bank is employed. A recent article by Stowe, Watson, and Robertson [22] examines the rela? tionships between the two sides of the balance sheets of a cross-section of 510 large, nonfinancial corporations. They used canonical correlation analysis to ex? plore relationships between the structure of the left- and right-hand sides of the balance sheet.1 The empirical results suggested several relationships: (1) the firms seemed to use hedging?matching the maturity structure of assets and lia? bilities; (2) some assets such as accounts receivable and real estate could be used as collateral for short-term bank or factor loans and mortgages, respectively; (3) commodity-producing firms should have higher levels of both inventories and accounts payable than service-providing firms; and (4) high-risk businesses may