Corporations - Finance; Bank investments; Financial leverage

Neoclassical models of the banking firm (for example, Michael Klein, 1971) treat all deposit liabilities as fully variable factors of production. Banks then maximize profits (minimize liability costs) by equating the marginal costs of all liability types during each period. The empirical relevance of these models is difficult to establish because binding deposit rate ceilings (Regulation Q) force bank competition for many retail deposits into implicit interest channels that are not readily measured. It is therefore noteworthy that during two recent periods when deposit rate ceilings were not binding, banks paid retail deposit rates considerably in excess of the rate at which they could borrow via large, unregulated certificates of deposit. Such behavior seems inconsistent with the cost minimization prescribed by neoclassical bank models. However, if retail deposit accounts are interpreted as quasi fixed (Gary Becker, 1962; Walter Oi, 1962; Donald Parsons, 1972; Sherwin Rosen, 1968) inputs to the banking firm, these important historical observations can be reconciled with bank profit maximization. This paper first describes two historical episodes during which the bank retail deposit rate exceeded the negotiable certificate of deposit rate for substantial periods of time. While no profit-maximizing (cost-minimizing) bank would pay such a rate differential if retail deposit quantities are costlessly variable, interpreting retail deposit accounts as quasi-fixed inputs to the bank explains the peculiar rate structures. A simple two-period model of bank liability selection formalizes the analysis. I. Two Puzzling Historical Episodes