Optimal Choice of Monetary Policy Instruments in a Macroeconometric Model


Ray C. Fair

Document Type

Discussion Paper

Publication Date


CFDP Number


CFDP Pages



It has been nearly twenty years since Poole (1970) wrote his classic article on the optimal choice of monetary policy instruments in a stochastic IS-LM model. Poole assumed that the monetary authority (henceforth called the Fed) can control the interest rate or the money supply exactly. These are the two “instruments” of monetary policy. If the aim is to minimize the squared deviation of real output from its target value, Poole showed that the choice of the optimal instrument depends on the variance of the error term in the IS function, the variance of the error term in the LM function, the covariance of the two error terms, and the size of the parameters in the two functions. Most people would probably agree that between about October 1979 and October 1982 the Fed tried to use the money supply as its primary instrument. This attempt does not appear to have been successful in the sense that since about October 1982 the Fed seems to have gone back to using the interest rate as its primary instrument. If the interest rate has won out, it is interesting to ask if this decision can be justified on the basis of the Poole analysis. Is the economy one in which the relevant variances, covariance, and parameters are such as to lead a la the Poole analysis, to the optimal instrument being the interest rate? The purpose of this paper is to examine this question using my United States econometric model. Are the variances, covariances, and parameters in the model such as to favor one instrument over the other, in particular the interest rate over the money supply? This question can be examined in an econometric model by the use of stochastic simulation. Interestingly enough, Poole’s analysis has never been tried on an actual econometric model. The closest study in this respect is that of Tinsley and von zur Muehlen (1983), although they did not analyze the same question that Poole did. Other studies that have extended Poole’s work, such as those of Turnovsky (1975) and Yoshikawa (1981), have been primarily theoretical.

This document is currently not available here.