Stanford Professor John Taylor has suggested that monetary policy could be summarized in terms of a simple rule, lowering interest rates when output is too low and raising them when inflation is too high. A number of academic papers have investigated this rule from the perspective of describing what the Federal Reserve has historically done. In a new paper co-authored with Federal Reserve economist Seth Pruitt and Office of Immigration Statistics economist Scott Borger, I take a look at what monetary policy rule the market perceived the Fed to be following over different historical periods.
Our basic idea is that, back in the days before we ran into the zero lower bound on interest rates, if there was a major surprise in a macroeconomic news release, you would see a big change in fed funds futures prices. For example, when the BLS announced on March 8, 1996 that nonfarm payrolls increased by 705,000 workers in February (wouldn’t we love to see another report like that one about now!), the interest rate associated with the August fed funds futures contract jumped from 5.01% to 5.25%. That response reflected a market belief that the development would cause the Fed to choose a higher interest rate than it otherwise would have.
We then built simple forecasting models for how news like this would alter a rational forecast of future inflation and output growth and ask, if market participants were revising their expectations in a way consistent with those forecasting relations, and if they thought that the Fed would respond to those future fundamentals according to a Taylor Rule, what parameters for the Taylor Rule are most consistent with the observed response of fed funds futures prices?
Our estimates are similar to those derived from more conventional econometrics, and suggest that, since 1994, the market believed that monetary policy was consistent with the Taylor Principle, raising the nominal interest rate more than 1-for-1 with an increase in inflation. Output targeting, particularly over the 2000-2007 period, appears to have been assigned secondary importance by the market.
One of the advantages of our approach is that we are also able to come up with estimates for the degree of inertia in perceived monetary policy. For example, while the August 1996 contract experienced a 24-basis-point jump on March 8, the May contract only moved 10 basis points. Since we can describe how the March 8 news should have altered a forecast for both May and August inflation, we can come up with direct measures for how long the market expected it would take the Fed to adjust interest rates fully in response to the news.
We document two important changes in the perceived policy rule over time. After 2000, the market believed that the Fed would eventually have a stronger response to inflation than it had prior to 2000, but also that the Fed would take longer to implement those changes, responding to news more sluggishly than it had before 2000.
We study the consequences of these changes using a simple new-Keynesian model. We find that the first change (a stronger long-run response to inflation) would be something that would have made output less variable, whereas the second change (a smaller immediate response) would have made output more variable. According to these simulations, increased Fed inertia undid some of the benefits it could have otherwise obtained with its anti-inflation policies.
Our conclusion is that the measured pace at which Greenspan increased interest rates over 2004-2005 may have been counterproductive, and that economic performance might have been improved if the Fed instead had raised interest rates more quickly to the higher warranted levels.