This paper seeks to evaluate monetary policy rules which generalize the rule proposed by Taylor (1993). In particular, we consider rules in which the Fed sets the Federal funds rate as a function of the history of inflation, output and the Federal funds rate itself. Even though this is not part of Taylor’s original formulation, we introduce the possibility that the Federal funds rate depends on the history of the funds rate itself in order to allow for interest-rate smoothing of the kind that appears to be an important feature of current Fed policy. We also consider the character of optimal policy, i.e., the policy that maximizes the utility of the representative agent, assuming unlimited information about the exogenous disturbances to the economy. We then compare optimal policy in this unrestricted sense with the best rule of the generalized-Taylor family.

We evaluate these rules under the assumption that interest rate, inflation and output determination in the U.S. economy can be compactly represented by the small structural model whose parameters we estimate in Rotemberg and Woodford (1997). This is a rational expectations model derived from explicit intertemporal optimization, in which firms are unable to change their prices every period, and in which purchases are determined somewhat in advance of when they actually take place. In evaluating different monetary rules we use two approaches.

The first approach is simply to compute the welfare of the representative household according to our model of the U.S. economy. Because this places great strain on the assumptions that the model contains accurate descriptions of the preferences of American residents, and that we have correctly identified the nature of the real disturbances to which monetary stabilization policy must respond, we also study separately the variability of output, inflation, and interest rates induced by different policy rules. This latter way of characterizing economic performance under alternative rules is less dependent upon the “deep structural” interpretation of the residuals in our structural equations, although it is, of course, still dependent upon the specification of those structural equations and upon the statistical properties of their disturbance terms.

We proceed as follows. In section 1, we describe the structure of the model, which is discussed more thoroughly in Rotemberg and Woodford (1997, 1998). Section 2 is devoted to the analysis of simple policy rules that represent variations upon the rule proposed by Taylor (1993), while section 3 considers optimal policy. Section 4 concludes. cash advance payday loans