Stok206.tex Comments on Nancy Stokey, 'rules versus Discretion' after Twenty-five Years 1

Nancy Stokey's (2002) interesting and thought-provoking paper has two parts. Part I, " Ob-servability, " discusses the choice of monetary-policy instruments by relying on Atkeson and Kehoe (2001). This discussion is in terms of a tradeoff between observability and " tightness " (the correlation with the monetary-policy goal). Part II, " Robustness, " discusses the choice between discretion and commitment to a simple rule. This discussion is in terms of a tradeoff between flexibility and myopia on one hand and rigidity and farsightedness on the other. I believe part I is better described as concerned with the choice of an intermediate target for monetary policy rather than a monetary-policy instrument. The setting of the monetary-policy instrument (the Fed funds rate in the U.S.) is usually directly observable, whereas the relation between the instrument setting and the monetary-policy goals is complex, making it difficult to infer the central bank's intentions from its instrument setting. Thus, I interpret part I as a discussion of the pros and cons of either an exchange-rate target or a money-growth target as intermediate targets, when the final target (the goal) is inflation. The choice of an intermediate target is a classic problem in the design of monetary policy. An ideal intermediate target is (1) highly correlated with the goal, (2) easier to control than the goal, and (3) easier to observe than the goal. The idea is that, if such an ideal intermediate target can be found, it may be better to aim for the intermediate target rather than to aim directly for the goal, and this way indirectly achieve the goal. In current real-world monetary policy, the idea of intermediate targeting has largely been abandoned (except in a specific sense mentioned below). Instead, central banks nowadays aim directly for their goals, typically low inflation and (to some extent) stable output gaps, as in (flexible) inflation targeting. The main problem with inflation targeting is that the control of inflation (and the output gap) is very imperfect, due to the lags in, and different strengths of, the various channels in the transmission process from instrument adjustments to actual inflation and output. This makes it difficult to judge whether current policy settings are, and past policy 1 These comments were prepared for the NBER Macroeconomics Annual 2002, Cambridge, MA, April 2002. I thank Annika Andreasson for secretarial and editorial assistance.