Banks' Buffer Accounts, Uncertainty, and Institutional Variables

Abstract

variables in the shortrun variables [1]. Most types of loans and deposits fall in this category. Because balance sheets do balance even in the shortrun a commercial bank's balance sheet also includes a set of accounts, accounts, which individually can be controlled in the shortrun and vary in response to changes in the banking variables. Excess reserves and securities are examples of buffer accounts. A bank's average holdings of buffer accounts should depend on the anticipated distribution of banking variables, cost and returns of the various accounts, and preferences, say, with respect to risk. In Section I we assume a highly aggregated balance sheet and construct a single-period inventory model with uncertainty. There are two buffer accounts, cash and borrowings. The main objective of the section is to show th t deposit variability increases desired cash holdings and that the expected growth of deposits reduces desired balances. However, xpected borrowings are independent of the distribution of deposits. We derive the general form of the equations to be estimated using cross-sectional data for the 350 member banks in the Richmond Federal Reserve District. Section II describes the data and discusses measurement problems and the introduction in the buffer-account equations of institutional variables, which may be viewed as proxies for bank costs and returns but also for bank preferences. The crosssectional results for buffer assets and borrowings are presented in Section III. Some of the institutional variables, particularly the logarithm of deposits, and the variability of depo its in the cash equations are significant. However, actual cash holdings are indepen ent of deposit growth, a factor which suggests that growth is anticipated. Banks reduce their desired cash position in anticipation of large deposit increases, so when their expectations are realized, they are holding no more cash than banks with unchanged deposits. The empirical results confirm that bank borrowing is independent of both deposit growth and variability, contradicting a need theory of borrowing and * The Federal Reserve Bank of Richmond kindly provided the data and computer time. The authors alone are responsible for the conclusions. 1 Of course, the shortrun distribution of these variables is affected by the longer run policy decisions.