The Brookings Institution held an interesting conference yesterday organized by David Wessel on “Should the Fed Stick with the 2 Percent Inflation Target or Rethink It?” Olivier Blanchard and Larry Summers argued, as they have elsewhere, that the Fed should increase its inflation target—say from 2% to 4%. Others—such as John Williams—argued that the Fed should change the target in some other way such as by focusing on the price level. Sarah Binder, Peter Hooper and Kristen Forbes were on a panel to answer questions about political, market, and international issues, respectively. I was on that panel to answer questions about monetary policy rules, and the first question posed by David Wessel was about the inflation target in the Taylor rule. Here’s a summary of my answer and later remarks during the course of the panel:
For the policy rule that came to be called the Taylor rule, first presented in 1992, I used a 2% inflation target (π*). This was long before the official adoption of a 2% target by the Fed, the BOJ or the ECB. The central banks of New Zealand, Chile and Canada were moving toward inflation targeting about that time, but not with the single number of 2% as a target. John Murray presented the Canadian history at the conference.
I chose 2% rather than zero back then because of the upward bias in measuring inflation, which was widely discussed at that time, and because of the zero bound problem for the interest rate. It was not an arbitrary choice. I also chose an equilibrium real interest rate (r*) of 2%. That was not arbitrary either with the real GDP growth rate trending a little over 2%. The actual rule for the interest rate (i) was i=π+.5y+.5(π-π*)+r* with π*=2 and r*=2. This meant that the equilibrium nominal rate was 4%. In equilibrium the output gap (y) equal zero and the inflation rate (π) equals π*.
Regardless of whether or not the Fed changes its inflation target π* going forward, it is important that its monetary framework be based on policy rules. The good economic performance during the Great Moderation was due largely to policy becoming more consistent with a rules-based framework, and the devastation of the Great Recession was due in part to deviating from rules-based policy.
It is also important for the new research on π* to be based on policy rules. In fact, virtually all economic research on the matter has been conducted using policy rules, including the important recent work by Fed economists Michael Kiley and John Roberts for the Brookings Papers on Economic Activity—a paper which was widely cited at the conference. The whole new section on policy rules in the Fed’s recent Monetary Policy Report and speeches last year by Fed Chair Janet Yellen also use this approach.
All the alternative proposals considered at this conference can and should be evaluated using policy rules, including price level targeting, nominal GDP targeting, and different inflation targets. If you want to evaluate a higher inflation target, you just stick in a higher value for π*. If you choose an inflation target of 4% with r* still 2%, then the average nominal rate will be 6%. If r* was 0% rather than 2%, an inflation target of 4% would mean an equilibrium nominal interest rate of 4%, exactly as in the original Taylor rule. In each case one can evaluate performance of the economy over a range of models as in Volker Wieland’s model data base.
Such policy rules or strategies would fit into the legislative language in recent bills in Congress, including the “Monetary Policy Transparency and Accountability Act,” which simply require the Fed to describe its policy strategy and compare it with policy rules of its own choosing. The rules would also help clarify the Fed’s actions to the markets and to policy makers in other countries.
The main motivation for the newer inflation targeting proposals is concern about the zero lower bound (ZLB), or the effective lower bound, on the interest rate. But the lower bound is not a new thing in economic research. Policy rule research took that into account long ago. In my 1993 book, for example, I noted that the policy rule “must be truncated below some nonnegative value.” We used 1% then: whenever the policy rule “calls for a nominal interest rate below 1 percent, the nominal interest rate is set to 1 percent.”
Another alternative is to move to a money growth regime. For example, in 1996 I noted that the interest rate rule needed “to be supplemented by money supply rules in cases of either extended deflation or hyperinflation.” Recently, Peter Ireland and Michael T. Belongia have suggested a return to money growth rules in the case of the ZLB.
Other proposals for dealing with the zero bound have been made over the years. In 1999 David Reifschneider and John Williams proposed that the interest rate be kept extra low following an ZLB period. For example, the interest rate would be kept at zero until the absolute value of the cumulative sum of negative deviations of the actual interest rate from the ZLB equals what occurred during the period of the ZLB.
Many at the conference thought that the ZLB is more of a problem now than in the past because estimates of r* have fallen. But those estimates are uncertain and may reverse soon. Volker Wieland and I demonstrated this uncertainty, especially in current circumstances, in considering the influential research of Thomas Laubach and John Williams. The low estimates of r* may be due to “fog” cause by unusually low policy interest rates and unconventional monetary policies at many central banks. Permanently changing the target inflation rate may not be the best response.
There are also international considerations. As we all know the original 2% inflation target is becoming universal for central banks around the world, and there is also a clamoring for a more rules-based international monetary system. One reason for the clamoring is research showing the increased exchange rate and capital flow volatility of recent years has been due in part to a deviation from a rules-based system. Now is an opportune time to move in the direction of a rules-based international system by simply reporting on the policy strategy in each country. Changing the inflation target in these strategies unilaterally will make this more difficult.
For all these reasons, I would be hesitant to change the inflation target introduced 25 years ago. But as research on policy rules at the Fed and elsewhere continues, I hope two related concerns are addressed.
First, there is a danger in the way that the numerical inflation target has come to be used in practice. It seems that even if the actual inflation rate is only a bit below the 2% inflation target—say 1.5% or 1.63%—there is a tendency for people to call for the central bank to press the accelerator all the way to the floor. This is not good monetary policy; it is not consistent with any policy rule I know, and it could create excesses or even bubbles in financial markets. This problem could be remedied as the Fed continues to clarify its strategy.
Second, the greater attention to a numerical inflation target may have reduced attention to other aspects of the policy rules, including the idea that we need a policy rule at all. In other words, trying to give more precision to π* may have led to less precision about other parameters, including the sizes of the responses. Recall that the Fed and other central banks moved toward rules-based policy well before they adopted formal numerical inflation targets. Most of the move to rules-based policy occurred during the period when Paul Volcker and Alan Greenspan simply said that inflation should be low enough that it did not interfere with decision- making. Again, I think this problem can be remedied as the Fed continues to clarify its strategy.