The New Economy
United States economic performance during the latter portion of the 1 990s far exceeded even optimistic forecasts. From 1996 through 2000, nonfarm business sector productivity grew by about 3.0 percent per year, on average. In the ten years previous to this period, from 1986 through 1995, it had increased at an average rate of only 1.4 percent per year. The late 1990s coincided with a spell of accelerated progress in computer technology and a widening adoption of the Internet by businesses and consumers. U.S. real output increased about 4.3 percent per year, on average, from 1996 through 2000, while, at the same time, inflation pressures remained rather subdued, with the personal consumption expenditures price index increasing at an average rate of only about 1.9 percent per year.
Economists in the United States have been cognizant of these changing trends. Many commentators have argued that technological change may be increasing American productivity, making it possible for the economy to grow at a faster rate without creating inflation. And, in fact, Federal Reserve officials have made many such arguments in recent years. Consider, for example, the May 6, 1999, Congressional testimony by Federal Reserve Chairman Alan Greenspan: "...the evidence appears to be mounting that, even if productivity does not continue to accelerate, the pickup already observed does seem to explain much of the extraordinary containment of inflation despite the ever-tightening labor markets of recent years." The next day the Washington Post reported: "Greenspan said the unexpected jump in productivity is the major reason that for the past three years so many forecasters, including those at the Fed, have underestimated economic growth while overestimating inflation."
This set of events is sometimes collectively called "the new economy," and we will use this meaning of the term for the purposes of this paper.
Optimal Monetary Policy Rules in the New Economy
The U.S. monetary policy debate has been importantly influenced by Taylor (1993), who argued that simple, nominal interest rate-based monetary policy rules might produce good stabilization performance.  Taylor's (1993) ideas were based on a given, constant inflation target for the monetary authorities and, especially important for this paper, a given, constant long-run level of productivity. Nearly all rules in this literature are then specified relative to these fundamental objects. In addition, Taylor's (1993) analysis was not of an optimal policy rule, but of an ad hoc rule that Taylor reasoned would perform well based on historical experience. Svensson (1997) showed how a version of the Taylor rule could be viewed as the optimal monetary policy rule in a simple dynamic macroeconomic model, again for a given inflation target and a given underlying level of productivity. In addition, the papers in the Taylor (1999) volume generally favor the idea that something close to optimal stabilization performance could be obtained by adhering to a Taylor rule, across a wide variety of macroeconomic models.
However, one of the key "new economy" events is the shift in productivity. It seems natural that a fully optimal monetary policy rule would take account of the changing nature of the supply side. Our main goal in this paper is to derive an optimal monetary policy rule in an environment with unobserved shifting productivity, so that the policy authorities must infer the underlying regime from observed data. We wish to accomplish this in the simplest possible framework, but one in which we are sure that a Taylor-type rule would be optimal were it not for the productivity changes. Accordingly, we adopt Svensson's (1997) model as a baseline, and we augment the model with two-state regime switching in the level of long-run productivity. We wish to understand how a Taylor-type rule would have to be altered to allow for the possibility that underlying productivity shifts may occur. Although we use this model to keep our exposition relatively simple, we also think that it is reasonably clear that the basic findings here would hold in far more elaborate models.
Our main finding is that the optimal policy rule in an environment with unobserved productivity shifts involves important lagged terms in inflation and the output gap. The role of these lagged terms is to help the policy authority react with appropriate interest rate adjustments when unobserved shifts in underlying productivity occur. In certain special cases, our optimal policy rule collapses to Svensson's (1997) rule for the same model, which involves only contemporaneous data. These special cases occur when (i) the probability of remaining in each regime is exactly one-half, so that productivity regimes are not persistent and can be interpreted simply as noise, or (ii) when the levels of productivity in the two regimes approach one another, so that there is effectively no difference between the two regimes. Intuitively, we think our main finding is an important one that would extend to a wide variety of models: In the face of possible unobserved changes in regime, the policy authority must optimally consi der recent trends in the data to infer whether the regime shift has occurred.
While our main results are analytical, we also consider a calibration of the model in order to fix ideas and provide illustrations of our findings. Adhering to a Taylor-type rule as derived by Svensson (1997) when there are, in fact, unobserved switches in productivity regimes implies significantly worse macroeconomic performance, relative to the optimal rule that we derive. Policymakers using a Svensson-Taylor rule would typically observe inflation that is persistently above or below target. This would appear to them to be due to unobserved special factors. But, with the optimal rule, inflation remains near target at all times and output fluctuates in response to the changes in productivity regimes and normal macroeconomic shocks. Thus the shift from low to high productivity in conjunction with a policy authority adhering to a Svensson-Taylor rule produces the events associated with the "new economy" as described in the opening paragraph: output persistently higher than expected, inflation persistently lowe r than …