How Forward-Looking is Optimal Monetary Policy?

We calculate optimal monetary policy rules for several variants of a simple optimizing model of the monetary transmission mechanism with sticky prices and/or wages. We show that robustly optimal rules can be represented by interest-rate feedback rules that generalize the celebrated proposal of Taylor (1993). Optimal rules, however, require that the current interest-rate operating target depend positively on the recent past level of the operating target, and its recent rate of increase, in a way that is characteristic of estimated central-bank reaction functions, but not of Taylor’s proposal. We furthermore find that a robustly optimal policy rule is almost inevitably an implicit rule, that requires the central bank to use a structural model to project the economy’s evolution under the contemplated policy action. However, calibrated examples suggest that optimal rules do not place nearly as much weight on projections of inflation or output many quarters in the future as in rules often discussed in the literaure on inflation targeting, or in the current practice of inflation-forecast targeting central banks. ∗This paper is excerpted from a longer working paper, circulated under the title “Optimal Interest Rate Rules: II. Applications.” We thank Ed Nelson, Julio Rotemberg and Lars Svensson for helpful discussions, and the National Science Foundation, through a grant to the NBER, for research support. Both positive and normative accounts of monetary policy are often expressed in terms of systematic rules for determining the central bank’s operating target for a short-term nominal interest rate in the light of current macroeconomic conditions, especially following the widely discussed proposal of Taylor (1993). Empirically estimated central-bank reaction functions are typically similar in form to the Taylor rule, but incorporate additional dynamics. For example, estimated reaction functions, such as those of Judd and Rudebusch (1998), Clarida et al. (2000), or Nelson (2001), frequently imply that one or more recent past levels of the interest-rate target are important determinants of its current value, along with macroeconomic indicators such as an inflation rate or a measure of the output gap. This implies that the bank’s interest-rate target at any given time depends on past conditions as well as those measured at that time. Furthermore, some estimated reaction functions, such as those of Clarida et al. and Nelson just mentioned, imply that the central bank responds to forecasts of future inflation and/or output rather than to the current values of those variables, as proposed by Taylor. Indeed, the representations of policy incorporated into the econometric models used by some central banks for policy simulations imply that the bank implements a forward-looking Taylor rule, perhaps with a forecast horizon as far as two years in the future. Furthermore, the official explanations that inflation-forecast targeting central banks offer for their decisions, such as in the Inflation Reports of the Bank of England, typically emphasize inflation forecasts with a horizon of that length. This paper considers the optimal choice of an interest-rate rule as a basis for the conduct of monetary policy. We shall be particularly concerned with the question of the extent to which it is desirable for the policy rule to be forward-looking (as has been emphasized in particular by the literature on inflation targeting), as opposed to seeking to establish a purely contemporaneous relation, as in Taylor’s proposal, or even a backward-looking relation of the kind indicated by many econometrically estimated reaction functions. The question has been extensively discussed in recent years. However, most of the recent See, e.g., xxxxx. See, e.g., the papers in Taylor (1999).