A GROWING BODY of monetary theorists has become convinced that our monetary authority should concentrate on the size and rate of change in size of some monetary aggregate, such as the supply of money or credit, as a target and indicator of monetary policy. The monetary authority can manipulate monetary aggregates by altering the size of the monetary base-an operation it can perform quickly and with precision. Altering the monetary base is aimed at exerting precise control over the target monetary aggregate. To have such control the monetary authority must know the relationship between the monetary base and the target monetary aggregate. That is, precise estimates of the size of the relevant credit expansion multiplier are essential if the monetary authority is to be able, ex ante, to make the size, or rate of change in size, of the monetary aggregate hit some specific target. There are two popular approaches to evaluating the size of credit expansion multipliers. One simple approach is to look at past observations of the ratio of the monetary aggregate to the base and use the observed ratios as a forecast of the expected future ratio. Many monetary theorists would argue that the world is not so cooperative, that the value of a credit expansion multiplier is likely to be different at different stages of the business cycle or even for different sizes of policy actions so that this simple approach will introduce serious errors into monetary policy actions. There is an alternative method which may yield better forecasts of the size of the multipliers. This alternative approach requires: (1) that we build credit expansion models which specify the relevant behavioral relationships (for the commercial banking system's portfolio selection behavior as well as that of the nonbank public) affecting the size of the multiplier; and (2) that we estimate the parameters in the model to allow us to use it for forecasting purposes. While considerable research has been conducted to cast light on some of the behavioral relationships necessary to forecast multipliers, other relationships have received relatively little attention. This study deals with one of the essential behavioral patterns which has been neglected: the currency-holding behavior of the nonbank public. Traditional credit expansion models rely on highly simplified assumptions as to the currency-holding behavior of the nonbank public during credit expansion or contraction. On the other hand, any modern portfolio analysis would require that currency holdings be influenced by an array of variables which in turn are affected by monetary policy actions; those variables include a portfolio constraint and yields on alternative assets. If systematic currency demand responses to changes in variables affected by monetary policy actually occur, the strength of monetary controls, (the impact on intermediate or ultimate target variables of a given monetary policy action) may be considerably different from what traditional credit expansion models imply. For example, if a tight money policy raises the interest rate on marketable securities inducing a shift out of currency and into marketable securities, slippage will be introduced into monetary policy in the sense that the endogenous shift from