Economics serving society

Can We Identify the Fed’s Preferences?

Short link to this article:

Kirsten Ralf and Jean-Bernard Chatelain

In 1973, two engineers, Marwan Simaan and Jose Cruz, showed that an actor in a dominant position (a central bank, for example) that influences choices made by other actors in tracking positions (e.g. the private sector) have an initial optimal policy that changes with each period. Their finding is very specific and a little surprising: the optimal policy choice changes with each date (this is called dynamic inconsistency) even though all future information about private sector behaviour has been anticipated correctly (i.e. there are no surprises) and the bank’s aims have not changed. In practice, it may be optimal for the central bank to have a rate of inflation higher than expected at the earlier date, in order to lower it more rapidly at later dates than was planned at the preceding date, and so on at each future date. This result arises from the fact that price decisions can be modified quickly by private sector agents depending on central bank announcements, which anchor the nominal values at each date. In 1977, Finn Kydland and Edward Prescott interpreted this dynamic inconsistency finding as a logical impossibility for optimal monetary policy to stabilise inflation. In 2004, the Swedish central bank awarded them the Nobel prize in economics for two contributions to macroeconomic dynamics: the dynamic inconsistency of policies designed to stabilise economic cycles, and the forces at the source of economic cycles, assuming that central bank stabilisation policies have no effect.

In this article, Ralf and Chatelain address two theoretical responses that followed Kydland and Prescott’s view of the logical impossibility of an optimal stabilisation policy. The first was proposed in 1989 by William Roberds, who believed that the change in optimal policy at each date is not certain. There is a non-zero probability that the central bank will maintain the optimal policy of the preceding date. If an increase in the real interest rate leads to a decrease in investment demand, which contributes to the reduction of inflation in the private sector, and if the expected inflation reaches 3% (overtaking by 1% a desired inflation target of 2%), then the central bank should change its key nominal interest of 2% to a little more than 3%, in order to increase the real interest rate (the net nominal rate of inflation). The alternative theory, proposed by Gilles Oudiz and Jeffrey Sachs in 1985, views the change in optimal policy at each date as certain. Otherwise, the central bank would lose credibility and its dominant position. It would re-optimise at each date and yet that would not change the inflation trajectory. To obtain this result, two optimal responses of the central bank and the private sector were chosen. If a rise in the real interest rate leads to a drop in investment demand, which contributes to the reduction of inflation in the private sector, and if the expected inflation reaches 3% (passing by 1% a desired inflation target of 2%), then the central bank’s optimal response would be to make public the fact that it could raise its key nominal interest rate from 2% to a very high, even infinitely high level, or even reduce it instead of raising it. In 2007, Stephanie Schmitt-Grohé and Martin Uribe found that the best monetary policy would be an interest rate that reacts at 332 times the inflation differential from its target. If this differential were 1%, then the relationship would entail an interest rate of 332%. Inflation would be immediately stabilised at 2%, rather than allowing the massive deflation implied in the other theory. In recent years, Schmitt-Grohé and Uribe have proposed that key interest rates be stiffly increased (up to 6%), in order to create inflation in Europe, Japan and the United States. The optimal response of the private sector would be to avoid an infinite hyper-inflationary or -deflationary trend so that we could never see such a strong interest rate reaction to inflation differentials from its target. Relationships between inflation expectations and non-anticipated (and sometimes non-observable) variables in the private sector would be assumed to be known exactly by the central bank and the private sector. They would choose a single trajectory that would stabilise inflation very quickly.

In the past 20 years or so, this second theory has been prevalent in hundreds of macroeconomic models (1), simulated and quantified in the research departments of central banks and published in academic journals. The authors test the two theories for the case of the US Federal Reserve Bank, using data related to the key interest rate, inflation and production during various periods between 1960 and 2006. The idea of the test is as follows: To avoid infinite instable trajectories created by the threat of a colossal over-reaction to interest rates, the second theory places important restrictions on the data compared with the first theory, which allows dependence on the past interest rate. For example, while the first theory predicts that the interest rate is correlated to two variables, current inflation and inflation in the preceding trimester (or the key rate in the preceding period), the second theory insists that the interest rate can only be correlated with one variable, regardless of current inflation, inflation in the previous trimester, or the key rate in the preceding period. The results of this new test lead to a rejection of the second theory.

(1) And, more precisely, models of dynamic stochastic general equilibrium models.

Original title of the article : “Can We Identify the Fed’s Preferences?” Kirsten Ralf, Jean-Bernard Chatelain

Published in : PSE Working Papers n°2017-25. 2017

Available at :

© Sergey Nivens -