This project is concerned with the analysis of alternative rules for monetary policy. The rules to be considered consist of alternative specifications of feedback rules for setting the Federal funds rate, the instrument of monetary policy in the U.S. While the numerical results of the study will address U.S. policy choice in particular, the more general conclusions are likely to be of application to other economies as well. The class of rules that will receive primary attention makes the funds rate a function of the current and recent past evolution of inflation and real economic activity, and of its own recent past values. Such a specification incorporates the two types of variables that most clearly affect current U.S. policy, and allows consideration of a range of popular policy proposals as well. Under at least certain simple assumptions, a policy within that class can also be shown to be optimal. Issues to be addressed include the degree to which it is desirable for policy to respond to these two variables, the consequences of responding to such data only with a delay, and the consequences of adjusting policy only gradually over time rather than in one sudden move. Consideration will also be given to the possible advantages of the use of additional indicators, especially when direct measures of inflation and output are not available that are sufficiently timely or accurate. Among such other indicators, the proper use of inflation forecasts will receive particular attention. The project will analyze alternative policies in the context of a small structural model in which the model equations are derived from explicit intertemporal optimization on the part of the suppliers and demanders of goods, and as a result incorporate forward-looking behavior; the analysis will thus take account of the `Lucas critique` of more traditional methods of econometric policy evaluation. At the same time, the model parameters are estimated on the basis of U.S. aggregate time series, and the model is able to mimic the properties of U.S. data on output, inflation and interest rate variations in considerable detail. A benchmark version of such a model has been introduced in recent work with Julio Rotemberg of Harvard; but the model will be further elaborated under this project. Comparability of the results obtained with this model with those of other small structural models will also be considered, in attempt to reach conclusions that are relatively robust to alternative assumptions in areas where model uncertainty is especially significant. Alternative policies will be evaluated from a variety of points of view. Particular emphasis will be given to the consequences of policy for the variability of inflation, and for the average rate of inflation (which turns out to be closely related to a policy's consequences for the variability of short term interest rates). The use of a model with optimizing foundations means that it is also possible to consider macroeconomic performance under alternative policies in terms of the level of private welfare attained, according to the preferences reflected in the structural relations of the model. Thus the project will also compute measures of the `deadweight loss` resulting from price level instability, in the spirit of standard public finance exercises, and seek to characterize the policy rules that are optimal in the sense of minimizing this loss, under various assumptions about the information available to the Fed, and the degree of complexity that the rule may involve. Theoretical deadweight loss will be shown to be increasing in both the average rate of inflation and its variability, owing to the relative price distortions caused by inflation when prices are not all changed simultaneously; hence policies that are desirable in terms of their consequences for these intuitive criteria are closely related to those that are optimal in welfare terms.
