In lieu of an abstract, here is a brief excerpt of the content:

  • Desirability of Nominal GDP Targeting under Adaptive Learning
  • Kaushik Mitra (bio)

Nominal GDP targeting has been advocated by a number of authors since it produces relative stability of inflation and output. However, all of the papers assume rational expectations on the part of private agents. In this paper I provide an analysis of this assumption. I use stability under recursive learning as a criterion for evaluating nominal GDP targeting in the context of amodel with explicit micro-foundations which is currently the workhorse for the analysis of monetary policy.

Monetary policy rules that utilize as their principal target variable the level or growth rate of some aggregate measure of nominal spending, such as nominal GDP, have had considerable academic support since the early 1980s. More recently, arguments in favor of nominal GDP targeting have also been made by, among others, Hall and Mankiw (1994), McCallum (1997c), and McCallum and Nelson (1999a) (see Hall and Mankiw, 1994, and McCallum and Nelson, 1999a, for a more extensive discussion of this literature). Nominal output targeting has two desirable features as a strategy for monetary policy. First, it automatically takes into account movements in both prices and real output, which in practice are the two variables central banks care about most. Second, nominal GDP can serve as a long-run nominal anchor for monetary policy, given the common belief that monetary policy cannot affect the real economy in the long-run.

Rudebusch (2002) pointed out that two distinct developments have also boosted an interest in nominal GDP targeting in recent years. The formation of the European Central Bank (ECB) in Europe has encouraged a lively debate about the appropriate strategy for European monetary policy. The announced ECB strategy contains an element of monetary targeting which is closely related to nominal output targeting if there are no large shifts in monetary velocity. In fact, the ECB (1999) has explicitly [End Page 197] derived its 4.5% reference value for M3 growth from a desired growth rate for nominal output. The ECB's announced monetary strategy, therefore, provides support for the consideration of nominal output targeting. The second development that has increased interest in nominal GDP targeting has been the behavior of the U.S. economy in recent years. A number of macroeconomic forecasters have been making forecasting errors - both overpredicting inflation and underpredicting output growth (see, for instance, Figure 1 of Brayton, Roberts, and Williams 1999). In light of this uncertainty about the level of potential output and the dynamics of the U.S. economy, several authors like McCallum (1999) and Orphanides (2000) suggest that monetary policy should focus on nominal GDP growth.

In this paper I study the desirability of nominal GDP targeting in the context of a standard forward-looking model that is currently the workhorse in the analysis of monetary policy. I use the New Phillips curve model, which has explicit microeconomic foundations and has been derived in a number of papers and reviewed, for instance, in Clarida, Gali, and Gertler (1999) and Woodford (1999). I first show that a monetary policy of targeting the growth rate of nominal income every period yields a locally unique stationary equilibrium when the private sector has rational expectations (RE) of inflation and output. Such a question is of utmost importance, because it is well known since the paper of Sargent and Wallace (1975) that uniqueness of equilibrium cannot be taken for granted in such models. This question takes added importance in the context of nominal GDP targeting since Ball (1999) has argued that this policy would be destabilizing for output and inflation.

The novel contribution of this paper, however, comes from a different angle. The entire literature of targeting nominal GDP has, to the best of my knowledge, assumed RE on the part of economic agents. By now it is well known that this assumption need neither be innocuous nor realistic. Agents are somehow assumed to be able to coordinate on a particular rational expectations equilibrium (REE). However, it is not obvious whether or how such coordination may arise. In order to complete such an argument, one needs to show the potential for agents to learn the equilibrium of the model being analyzed. An...