Nonstandard Indicators for Monetary Policy: Can Their Usefulness Be Judged from Forecasting Regressions?

Abstract

From the chapter: "In recent years there has been a great deal of interest in proposals to use nonstandard indicator variables as a guide to the conduct of monetary policy. By nonstandard indicators I mean indicators other than the various measures of the money supply that the Federal Reserve System computes, and to which academic economists in the monetarist tradition have long directed their attention, and other than the variables that can be more or less directly controlled by the Fed, such as borrowed and nonborrowed reserves or the federal funds rate, and measures of the Fed’s success at achieving its ultimate objectives, such as measures of inflation and economic activity in the recent past. Some of the new indicators that have been discussed include commodity price indexes, nominal exchange rates, and spreads between the interest yields on longer- and shorter-maturity Treasury securities. Interest in the new indicators seems mainly to have been a response to the perceived instability in the 1980s of the relations between traditional monetary aggregates and nominal aggregate demand. The various new proposals just mentioned all represent variations upon the idea that a desirable monetary policy, that would be able to respond to, and counteract, incipient inflationary pressures before much inflation had developed, could be conducted by monitoring various indicators that are known to be valuable as forecasts of future inflation, rather than by giving one’s sole attention to the evolution of variables such as the money supply that are thought to be proximate causes of inflation. I do not attempt here to evaluate the likely consequences of any of these specific proposals. Instead, the present note addresses a general issue raised by proposals of this general type, that of how to determine which variables, of all the many types of data available to the Fed, are reasonable indicators to use as the basis for a feedback rule for one or another instrument of monetary policy. The discussion of the new indicators has tended to assume that one should simply look for any or all available variables that have proved to be useful in forecasting inflation over some relevant horizon. The advocates of the new variables within the Federal Reserve System have stressed their usefulness as leading indicators of inflation, and much of the academic commentary has addressed itself to formal econometric evaluation of their forecasting ability, typically within a completely atheoretical vector autoregression (VAR) framework. The question I consider here is whether analysis of this sort is a sufficient ground for choosing variables to be used in making monetary policy. I will argue that such investigations are no substitute for an analysis of the consequences of making policy on the basis of one indicator or another in the context of a specific structural model of the economy, which models both the determination of the indicator in question and the effects of possible monetary policy interventions. Before developing this point further, however, I first review from a purely statistical point of view the literature that has addressed the issue."

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 152 citations exist upstream.