访谈联邦储备银行主席伯南克(第一部分)

Interview with Ben S. Bernanke*

INFLATION TARGETING

Rolnick: For several years now, you've argued that inflation targeting will improve monetary policymaking by anchoring the public's inflation expectations and by improving Fed accountability. Some of us would argue that we're already practicing something like that de facto, with our public commitment to price stability. In what sense is your proposal a substantive change in policy? If we did move to an explicit target, what would be the benefits?

Bernanke: It's true that the Federal Reserve is already practicing something close to de facto inflation targeting, and I think we've seen many benefits from that. My main suggestion is to take the natural next step and to give an explicit objective, that is, to provide the public with a working definition of price stability in the form of a number or a numerical range for inflation. I believe that that step, though incremental, would have significant marginal benefits relative to current practice.

First and very importantly, such a step would increase the coherence of policy. Currently, the FOMC [Federal Open Market Committee] makes its decisions without an agreed-upon definition of price stability or of the inflation objective, and one wonders how oarsmen pulling in different directions can get the boat to go in a straight line. I think the FOMC's decision-making process would be improved if members shared a collective view of where we want the inflation rate to be once the economy is on a steady expansion path.

Second, there's a great deal of evidence now that tightly anchored public expectations of inflation are very beneficial, not only for stabilizing inflation but also in reducing the volatility of output and giving the Federal Reserve more ability in the short run to respond flexibly to shocks that may hit the economy.

Inflation expectations in the United States are better anchored than they used to be but are still too volatile for optimum performance of the economy. Announcing an actual number or range would serve to anchor public expectations of inflation more firmly and avoid the risk of “inflation scares” that might unnecessarily raise nominal bond yields.

Third, from a communications viewpoint, financial markets would be well served by knowing the medium- to long-term inflation objective of the Fed. An explicit inflation objective would help market participants accurately price long-term assets, both by anchoring long-term inflation expectations and by giving the market better information about the likely path of short-term policy as the Fed moves toward its long-term target. And fourth and finally, I think an inflation target does introduce an additional measure of accountability for the Federal Reserve, although I would put that as least important of the things I've mentioned.

Rolnick: Much of what you mention we've achieved over the last 20 years without an explicit target. According to current market expectations we appear to have credibility. Inflation expectations, as best we can measure, are now pretty low. So even though there may be different oars in the water, we seem to be going in the same direction.

However, there is a rationale in the economic policy literature to justify inflation targets, that is, the problem of time inconsistency. Are you saying that an explicit inflation target is more of a commitment, harder to overturn and hence better equipped to deal with time inconsistency?

Bernanke: No, I think we have achieved a substantial degree of credibility based on the record of low inflation and that reduces the inflation bias problem already in itself. However, I think that announcing a target would strengthen our commitment to the price-stability objective and also give more emphasis to long-run considerations in policymaking. Our policy meetings are very often focused on near-term developments. Having a medium- to long-term inflation objective would force us to keep in view where we want the economy to be in the longer run.

I question, though, whether we have achieved as much as you suggest in terms of anchoring inflation expectations. In fact we see significant fluctuations in expected inflation. There are a number of research papers providing evidence that inflation expectations respond significantly to changes in actual inflation, for example, suggesting that long-term expectations are partially adaptive and not as well tied down as we would like. Bond yields are easier to understand under the hypothesis that long-term inflation expectations change significantly at times. And inflation compensation in indexed securities is higher (suggesting a large inflation risk premium) and more volatile than we would like to see it. So while we have indeed achieved a considerable improvement in the degree to which inflation expectations are anchored, more could be done, and there would be very little cost to taking this extra step to strengthen that commitment.

Rolnick: Is there any danger that once you make your inflation target explicit, the market will give it too much importance? Most misses from the target will not matter that much to the overall economy. Hence, the advantage of an implicit target.

Bernanke: Well, I don't advocate a strict inflation-targeting regime that attempts to maintain inflation at the target level all the time, by any means. Rather, I think it's worth specifying our objective for the average steady-state inflation rate over a longer period of time, understanding that the actual inflation rate would deviate around that level. Everyone understands that the Fed cannot maintain inflation at a fixed level even if it wanted to, given the shocks that buffet the economy; nor is doing so even desirable, because of short-run stabilization objectives. But I still think it would be useful additional information to know what the Fed's objective is at a medium-term horizon.

FED COMMUNICATION

Rolnick: That's certainly consistent with your views on the importance of Fed communication and transparency. Presumably, though, there is some value in maintaining a level of control over information related to FOMC deliberations: limiting miscommunication, giving the Fed room to change its mind, so to speak, recognizing the high-frequency noise. Should we think of it as a trade-off?

Or if transparency in and of itself is the right principle, shouldn't we release more information about our policy process; for example, shouldn't we release the Greenbook [which contains the Fed staff's national model and economic forecasts] after each FOMC meeting? Wouldn't releasing the Greenbook be consistent with transparency and improve the way the market responds to our policies? Or is there some point at which providing this much information is counterproductive?

Bernanke: I think that the kind of information you want to release is information that helps the market and the public achieve more accurate expectations of future policy and the future state of the economy. Not all information is beneficial. One would not want to release information that would compromise the decision-making process; for example, televising the FOMC meeting would not be productive. For the same reason, I'm not sure that releasing the Greenbook after each meeting would be productive because it would put too much emphasis on the staff's forecasts. The staff's forecasts are for the information and the assistance of the FOMC decision-making process. The FOMC, not the general public, is the client of the staff.

I am in favor of making greater use of the FOMC's central tendency forecasts for communication. In one of my earlier speeches I suggested that the FOMC release more information (for example, for longer forecast horizons) about our view of the state of the economy, with the purpose of trying to help the public and the markets better understand what our perspective is and what policy is likely to be doing in the future. That's the kind of transparency I think would be most useful. I have also advocated earlier release of the minutes of FOMC meetings, again in order to provide timely information about the views of the Committee and its appraisal of the economy.

DEFLATION

Rolnick: You participated in the Minneapolis Fed's November 2003 workshop on deflation. [See the December 2003 Region.] At that workshop it was clear that there is still a serious debate over the link between deflation and depressions. Some would argue that low levels of deflation, in fact, may actually be good. China is a recent example of a country that experienced deflation and economic growth. What's your current thinking on the downside risks of deflation? Are there good deflations, bad deflations, or is deflation just a red flag that should make us more cautious?

Bernanke: The critical issue in thinking about deflation is whether or not the zero bound on nominal interest rates is binding. The lower the rate of inflation, the lower are nominal interest rates, on average. If a deflation is large and protracted, the chances are much greater that you'll be at the zero bound, the point at which nominal interest rates can no longer fall further. The combination of interest rates at the zero bound and deficient aggregate demand, a situation where the economy is in recession and producing below its productive capacity, is potentially quite dangerous, for several reasons.

First, monetary policy is unable to respond in a conventional manner by lowering interest rates to affect aggregate demand. Although, as I emphasized in my November [21] 2002 speech, there are alternative means by which monetary policy can expand the economy, we have limited experience with them and they are less well understood than the conventional interest-rate tool.

And secondly, at the zero bound there is some possibility of a downward spiral of worsening deflation. With the nominal interest rate stuck at zero and deflation intensifying, the real interest rate becomes higher. That weakens aggregate demand still further and conceivably could generate a vicious circle.

In short, my principal reason for avoiding deflation is that you need to keep short-term nominal interest rates sufficiently above zero to have some room to respond in the case that aggregate demand falls short. There are other reasons as well; for example, deflation can be very hard on the financial system by increasing the real burden of debts. You mention China as a counterexample. Generally speaking, a deflation that accompanies a productivity boom would not be serious because you would have a high real interest rate and the nominal rate would therefore be well above zero. However, the United States is a very interesting example currently of a country experiencing a productivity boom, but for some reason real interest rates don't seem to fully reflect what would appear to be a higher marginal product of capital. The nature of the productivity boom—labor-saving rather than capital-enhancing—may help explain that puzzle.

So for those reasons I think that—going back to the inflation targeting issue—if I were picking an inflation target, I would pick one that was above zero, even above zero accurately measured. The reason would be to have at least a bit of buffer so that the average nominal interest rate is sufficiently above zero that the chances of running into the lower bound are minimized.

Rolnick: In a broad class of general equilibrium models, a Friedman rule—a zero nominal interest rate and a low level of deflation—is optimal. These models seem to be fairly robust to changes in assumptions. Do you think these models are flawed? Are there better alternatives?

Bernanke: I would say that those models are deficient in at least two respects. One problem is that they focus almost entirely on so-called “shoe-leather costs,” that is, the cost of minimizing cash holdings (by more frequent trips to the bank, for example) to avoid losses to inflation. But shoe-leather costs are not the only relevant costs of inflation. Even in a steady state there are many other inflation costs, such as costs of price changes and the distortions of tax and accounting rules expressed in nominal terms, which these models don't incorporate. So putting aside cyclical issues entirely and just asking what is the growth-maximizing rate of inflation, it's not evident to me at all that it would be negative.

The second issue is that these models abstract from sticky wages and prices and hence from the possibility of short-run deviations of output from full employment. They don't have any substantial role for stabilization policies. I believe that we live in a world where stabilization policy—stabilization of inflation as well as output—is sometimes needed. For reasons I have just given, the occasional need to use monetary policy for stabilization argues against having negative inflation and zero interest rates on average.

DEBT, DEFICITS AND FISCAL CONSTRAINT

Rolnick: The federal deficit has once again become a major economic concern. By some estimates, the present value of future debt is over $40 trillion. Can you envision constraints on fiscal policy that are similar to the kind of constraints you're advocating with inflation targeting? And what is the public role of the central bank on this issue? As a central bank, how should we respond to this kind of scenario?

Bernanke: Well, the central bank has the responsibility to be a nonpartisan adviser on general matters of macroeconomic and financial stability. So to the extent that deficits and debt are threatening macroeconomic and financial stability, the central bank is one actor that can provide advice and counsel to the fiscal policymakers.

Ultimately, though, the determination of the fiscal debt and deficits is the responsibility of the president and the Congress. To the extent that rules or caps can be useful in forcing Congress to look at the whole budget, as opposed to each individual component and then adding them all up, I think that's certainly worth looking at. But again, that's a decision for the president and Congress to make.

Rolnick: Caps might be an analogy to inflation targeting, providing some constraint?

Bernanke: Some constraint, yes. Certainly there's a constraint that we want to have a balanced budget over a certain period of time. There's the formal constraint that the Europeans have, the Stability and Growth Pact, although experience with that pact shows that rules are not necessarily going to constrain legislators who do not wish to be constrained. Although fiscal policy is different in many ways from monetary policy, it seems to me that, nevertheless, some overall structure for making budgetary policy, some rules about linking spending and taxing decisions, might well be helpful in allowing 535 people to collectively make rational decisions about the state of the budget.