Bernanke’s speech at the AEA meetin suggested that monetary policy was not to blame for the housing boom based on his version of Taylor rule. Today, John Taylor responds to Bernanke on the WSJ opinion page:
In his speech, Mr. Bernanke’s main response to this critique was to propose alternatives to the standard Taylor rule—and then to use the alternatives to rationalize the Fed’s policy in 2002-2005.
In one alternative, which addresses what he describes as his “most significant concern regarding the use of the standard Taylor rule,” he put the Fed’s forecasts of future inflation into the Taylor rule rather than actual measured inflation. Because the Fed’s inflation forecasts were lower than current inflation during this period, this alternative obviously gives a lower target interest rate and seems to justify the Fed’s decisions at the time.
There are several problems with this procedure. First, the Fed’s forecasts of inflation were too low. Inflation increased rather than decreased in 2002-2005. Second, as shown by economists Athanasios Orphanides and Volker Wieland, who previously served on the Federal Reserve Board staff, if one uses the average of private sector inflation forecasts rather than the Fed’s forecasts, the interest rate would still have been judged as too low for too long.
Third, Mr. Bernanke cites no empirical evidence that his alternative to the Taylor rule improves central-bank performance. He mentions that forecasts avoid overreacting to temporary movements in inflation—but so does the simple averaging of broad price indices as in the Taylor rule. Indeed, his alternative is not well defined because one does not know whose forecasts to use. Moreover, the appropriate response to an increase in actual inflation would be different from the appropriate response to an increase in forecast inflation.
The entire piece is a must-read, but I would like to focus attention on Bernanke’s use of the Taylor rule. What is troubling about the recent debate and framing it in terms of the Taylor rule is that it seems that everyone has their own definition. Over time, many economists have statistically fit the parameters of the Taylor rule in order to estimate the Fed’s reaction function. However, we have to be careful about what these estimates actually mean. These types of estimates are certainly useful for policy comparisons and other positive analyses. However, they are not useful for drawing normative conclusions because the fitted parameters incorporate policy mistakes in the estimation period.
As Taylor notes, Bernanke commits a related error by plotting the interest rate implied by the Taylor rule using the Fed’s forecast of inflation. Taylor’s rule is not based on the inflation forecast, but rather on the actual inflation rate.
Why is this important?
In a working paper, I make the case that the Federal Reserve’s policy during the Great Inflation was the result of an incorrect doctrine. Specifically, the Federal Reserve was convinced, in Arthur Burns’ words, that the rules of economics had changed and that inflation was driven by cost-push forces. I argue that this caused the Fed to misinterpret positive aggregate demand shocks for negative aggregate supply shocks. What’s more, this view implies that inflation forecasts based on the Phillips curve would result in systematically lower predictions than the actual value. This is, in fact, what the data show about the Fed’s forecasts. As a result, this misplaced view ultimately led the Fed to have a much strong response to forecasts of inflation than to the actual values observed ex post.
Bernanke’s analysis similarly misguided. What good is it to use the forecast of inflation in plotting the Taylor rule if that forecast is systematically lower than the actual rate observed ex post?