Remarks prepared for the NABE session on “Recent and Prospective Developments in Monetary Policy Transparency and Communications,” ASSA Meeting, New Orleans, January 5, 2008
Not having had advance access to the remarks of the central bankers on the panel, I shall have, with apologies, to offer remarks addressed more to the topic proposed to them than to their actual remarks. I can perhaps offer an external perspective on some of the year’s developments in central-bank communication policy. I thought it might also be useful for me to contrast the policies of the major central banks, represented on this panel, with those of some of the smaller central banks that happen to have been leading the way with regard to transparency.
Probably the most interesting development of the past year in communications policy, however, has been at the Federal Reserve. Along with the minutes of its October 30-31 meeting, the FOMC released a summary of the members’ forecasts for several economic variables, the first example of a type of information that is now to be released four times a year, following the 1st, 3d, 4th and 7th FOMC meetings of each year. The change in communication policy, announced in a press release by the FOMC on November 14, and explained in further detail in a speech by Chairman Bernanke to the Cato Institute’s Annual Monetary Conference the same day,1 represented the outcome of a year-long process of deliberation within the FOMC about ways in which the transparency of Fed policymaking might be improved.
An obvious question about the new policy is the extent to which it represents an attempt to implement for the U.S. the kind of “forecast targeting” that has been used by a number of central banks with official inflation targets, such as the Bank of England and Sweden’s Riksbank, for the past 15 years. Both Chairman Bernanke and Governor Frederic Mishkin were well-known as leading American fans of inflation targeting before taking their current positions at the Fed,2 and many have wondered whether the Bernanke Fed might adopt some form of inflation targeting. Regular publication of quantitative projections of the future evolution of inflation and other variables --- in the Inflation Reports of the Bank of England, that appear 4 times per year, or the Monetary Policy Reports of the Riksbank, published 3 times per year --- is a characteristic feature of those banks’ approach to monetary policy. To what extent does the Fed’s new “enhanced communication strategy” achieve a similar end? Should it be viewed as inflation targeting in everything but the name?
It is useful first to recall the point of the publication of forecasts by the other central banks. “Forecast targeting” is a particular kind of decision procedure for monetary policy, under which the policy committee seeks, each time that it meets, to determine the level of its policy rate consistent with a projected evolution of the economy that would satisfy a quantitative target criterion. For example, the Bank of England often explains its policy as aiming to ensure that the projected rate of CPI inflation at a horizon 8 quarters in the future should equal 2.0 percent. Given a criterion of this kind, that refers to the future conditions that policy is intended to bring about, the bank’s internal forecasts play a crucial role in policy deliberations.
The regular publication of the bank’s forecasts also plays a central role in these banks’ communication policies. The point of publishing the forecasts is to make the nature of the banks’ policy commitments evident to the public, and to increase the credibility of these commitments, by allowing the public to observe how policy decisions are actually shaped by the banks’ efforts to ensure the fulfillment of the target criterion. The Inflation Reports or Monetary Policy Reports include not only forecasts and discussion of the reasoning behind them, but also, crucially, an explanation of how the bank’s most recent policy decisions are justified by the projections presented in the report. The argument presented, essentially, is that (i) projections are displayed conditional on a particular assumption about policy; (ii) the projections can be seen to have the desired properties; hence (iii) it may be verified that the assumed policy is an appropriate one.
The commitment to justify policy to the public in this way increases public understanding of the policy, and so should improve the public’s ability to correctly anticipate the future conduct of policy, increasing the effectiveness of policy.3 It also serves the goal of making the central bank more accountable, which provides democratic legitimacy for the central bank’s grant of operational independence.4 And finally, it can improve policy itself, by providing a check on possible temptations to base policy purely on short-run considerations, losing sight of whether policy remains consistent with the bank’s medium-run objectives.5
Does the Fed’s commitment to release additional projections, and additional discussion of those projections, serve a similar function? Both Chairman Bernanke’s speech, and a speech soon after by Governor Mishkin,6 make it clear that at least for these two members of the FOMC, the point of the new policy is to better communicate the nature of the FOMC’s policy commitments --- essentially, revealing that the Fed conducts policy in the way advocated by proponents of “flexible inflation targeting,” albeit without any official announcement of targets for policy.
Governor Mishkin’s speech is clearest about the way in which the “Summary of Economic Projections” in the minutes is intended to reveal the goals of Fed policy. First of all, the new projections, unlike those previously summarized in the semi-annual Monetary Policy Report to Congress, extend three years into the future, rather than only two. In addition to providing projections more similar to those of the forecast-targeting central banks, this means that the projections extend far enough into the future that one might usually expect the economy to converge over that horizon to something like the long-run (or “steady state”) values of inflation and unemployment associated with the contemplated approach to policy. Hence, it is suggested, the public should be able to tell from the projections for three years in the future the place that policy is trying to reach, relatively independently of where it may currently be.
To be specific, according to Mishkin, the three-year projections for inflation “provide information about each FOMC participant’s assessment of the inflation rate that best promotes [the Fed’s dual stabilization] objectives --- a rate that I will refer to as the ‘mandate-consistent inflation rate’.” This is essentially the inflation target in a quadratic objective function for the individual FOMC member, though Mishkin is careful to describe the target as an assessment --- a matter of fact, rather than a value judgment --- of the inflation rate required to achieved the goals assigned to the Fed by Congress, rather than a goal chosen by the FOMC itself. Here the inclusion of a projection for headline PCE inflation, and not just the core PCE inflation projection previously emphasized in the Monetary Policy Report, also seems to be intended to help communicate the FOMC’s inflation target (or rather, the members’ collection of individual targets). While the FOMC has in the past paid greater attention to projections of core PCE inflation in making short-run policy decisions, as it is believed to be a more reliable indicator of risks to price stability over short periods of time, they would clearly prefer to express their medium-run inflation target in terms of a broader index, which better conforms to the public’s concerns.
Mishkin similarly argues that the three-year unemployment projection indicates the participants’ views of the “sustainable unemployment rate” or natural rate of unemployment. He emphasizes even more strongly in this case that the long-run forecast represents a judgment about an economic fact rather than a goal chosen by the members of the FOMC, as “the Federal Reserve most emphatically cannot choose the level of maximum sustainable economic activity.” Nonetheless, it is clear that in his view, this “sustainable unemployment rate” plays the role of an unemployment target in the FOMC member’s loss function, in the sense that policy should aim (other things being equal) to minimize departures of the actual unemployment rate from the sustainable rate. Thus target values for two variables, representing the twin stabilization objectives of the Fed’s “dual mandate,” can both be discerned from the projections. Mishkin suggests that the relative weight placed on the two goals can also be discerned, not from the three-year projections alone, but from the relative rates of convergence of the inflation and unemployment projections to their eventual values.
There are certainly some advantages to communicating the Fed’s stabilization goals in this way, rather than through an explicit announcement of policy targets. One is that it allows the Fed to maintain a degree of symmetry in the way that it treats the two stabilization objectives --- projections are presented for both (in fact, there are projections for two inflation measures and for two measures of real activity), and one can make similar inferences about medium-run objectives for both variables --- unlike the inflation-targeting central banks, which have an official target for the inflation rate but no equally explicit target for any measures of real activity.7 This circumvents one of the most important objections raised to proposals that the Fed adopt an inflation target, which is that an apparent focus on inflation alone might be found to violate the Fed’s legislative mandate. At the same time, the fact that the members of the FOMC present only their forecasts, and not targets or objective functions, avoids the delicate problem of having to justify an unemployment “target” higher than what some politicians might feel that society should set itself as a goal.
Nonetheless, while the approach taken treads a careful path through a political minefield, it is far from providing a full substitute for an explicit announcement of policy targets. The fact that members of the FOMC all currently forecast that PCE inflation should eventually decline to a rate between 1.5 and 2.0 percent per year does not convey the same thing as a commitment to aim to return it to that level. On the hypothesis that each member of the FOMC is a covert inflation-targeter, then their forecasts (of how inflation would involve under an “appropriate” path for policy) for a long enough horizon should reveal their inflation targets; but the fact that their current forecasts happen to converge to some inflation rate hardly proves that their objective function involves an inflation target at all. The important thing that is achieved by an official inflation target is that medium-run inflation expectations should remain firmly anchored even in the face of disturbances that cause inflation to temporarily depart from the target rate. But the knowledge that members of the FOMC currently expect PCE inflation to fall below two percent gives one no reason to suppose that they will continue to expect that (or aim at it) in the event of a disturbance that makes disinflation more difficult than had previously been expected.
Over time, of course, one can expect to learn whether the three-year forecasts of inflation remain constant or not. But even if they do, and this eventually creates confidence in the existence of fixed inflation targets, requiring people to learn in this way is a slower and more uncertain process than would be possible with a public commitment from the start, with likely consequences for the stability of inflation expectations in the meantime. Even after several years of relatively constant three-year forecasts, there might remain considerable uncertainty about the FOMC’s intentions regarding inflation in the event of a relatively unusual disturbance that had not been experienced in the previous five years. Moreover, the absence of an explicit statement of the targets makes it less likely that even those members who privately think in terms of inflation targets will feel a need to stick to the same target when circumstances change. An assessment of the mandate-consistent inflation rate, treated as a judgment of fact about the economy, can reasonably change as the perceived facts change; hence no member need be embarrassed if his or her assessment is observed to shift over time. If instead one believes that stability of inflation expectations is more important than the particularinflation rate that is expected, then a commitment to a particular target, that one should have little ground to subsequently revise, has much to recommend it.
More generally, the Fed’s “Summary of Projections” is far from a complete substitute for the Monetary Policy Report of a forecast-targeting central bank, because the projections are not used to justify a policy decision. In the absence of an explanation of how the projections are supposed to relate to the decision that has been taken, publication of the additional information, while arguably an increase in transparency of one sort, does not do much to clarify the nature of FOMC decisionmaking, and hence to make future policy more predictable.
The summary of projections is presented separately from the summary of the discussion leading to the interest-rate decision, and contains no comments on any implications of the projections for policy. Nor do the minutes of the discussion at the meeting contain many references to the projections supplied by the FOMC members. There is considerably more discussion (at least in the minutes of the October 2007 meeting) of the forecast prepared by the staff of the Board of Governors for the meeting; but no information about this forecast is currently released except with a considerable delay.
Moreover, even if the members’ projections are the relevant ones for understanding the policy deliberations, the information provided in the “Summary of Projections” lacks two key elements required for a justification of policy under a forecast-targeting procedure. On the one hand, banks like the Bank of England and the Riksbank explain the assumptions about future policy reflected in their projections; and on the other hand, they explain what acceptable projections must look like. It is then possible to judge that the projections under a particular policy should be acceptable, and to know what policy it is that has thereby been shown to be appropriate. In the case of the Fed, instead, there is no explicit target criterion, apart from what may be inferred from looking at the projections themselves; and while each member’s projections are supposed to be based on his or her “assessment of appropriate monetary policy,” no information is supplied about what policy is being assumed by any of the members. One therefore has no way of saying whether the decision actually taken at the meeting has any similarity to the policies assumed in the projections or not.
So the Fed has not yet taken too great a step toward implementation of inflation targeting in the U.S. (This is likely to come both as a relief to some and as a disappointment to others.) Does the new policy matter at all?
I think that it will make a difference, but not primarily to the degree to which outsiders are better able to understand or predict Fed policy, in the first instance. Rather, the most important consequence of the new strategy, in the short run, will be for the Fed’s own deliberations. Requiring the members of the FOMC to consciously consider the way in which the economy is likely to evolve over the next several years, and the nature of appropriate policy not just in the short run, but also over the next several years, is likely to increase the extent to which policy decisions are considered as part of a coherent strategy rather than as a sequence of unrelated decisions. Encouraging them to discuss with one another the extent to which their forecasts differ and why will facilitate the development of a shared understanding of what a sensible strategy would be like. And pressing them to consider the reasons for the changes in their forecasts from one release to the next should ultimately favor the maintenance of consistency over time in the way that policy is approached.
Changes of these sorts in the way that policy decisions are approached are highly desirable, given the extent to which the anticipation of policy, and hence its intelligibility, matters to its effectiveness. Once policy decisions are made to a greater extent on the basis of a consistent strategy, it should be possible for the Fed to explain more about that strategy to the public. In this respect, the new strategy does represent a step toward an eventual improvement in public understanding of Fed policy.
Another dimension along which central-bank communication policy continues to evolve, at many central banks, is the degree to which, and the means by which, central banks offer “forward guidance” about the likely future path of policy rates. All three of the central banks represented on this panel have experimented over recent years with more explicit forward guidance through their official communications, generally through the use of “code words,” such as removal of policy accommodation at a “measured pace,” or the exercise of “strong vigilance” toward inflation risks. At the Fed, at least, there has been a step back from this practice over the past year, with recent FOMC post-meeting statements no longer containing the explicit signals about future policy that had become routine --- and the subject of much scrutiny by market commentators --- in the period between 2003 and 2006. I suspect that other central banks are becoming more cautious as well about the use of code words that are taken to directly indicate future interest-rate decisions, under the current rapidly changeable conditions in financial markets.
Is it in fact appropriate for central banks to try to communicate about future interest rates, or is this a point on which they are better advised to say as little as possible? I think that talking about the likely future course of rates can be quite valuable, at least at certain times. The economic effects of monetary policy depend almost entirely on the anticipated future path of the policy rate, rather than on the current level itself of a rate such as the federal funds rate; announced changes, or non-changes, in the funds rate operating target matter only to the extent that they also often imply changes in the expected path of the funds rate months or even years into the future. Hence it is certainly relevant to a central bank’s stabilization objectives what the private sector understands about the likely forward path of the policy rate; and when the central bank perceives that the private sector has not reached a correct understanding on its own, there is reason for it to seek to clarify matters.
A good example is the situation faced by the Fed in the summer of 2003.8 The funds rate operating target had been reduced to only one percent, and could surely be reduced little further, if at all; market speculation consequently focused on how soon and how sharply rates would be raised again. Indeed, speculation that the Fed would raise rates substantially within a matter of months was already causing long-term interest rates to rise, creating a situation that, it was feared, could actually tip the US economy into deflation. The FOMC could not, or at any rate was certainly reluctant to counter these expectations through any further cut in the current operating target; but instead it was able to calm market fears of an early tightening of rates by explicitly committing to maintain low rates “for a considerable period,” beginning with the statement following its August meeting. Somewhat similar considerations had led the Bank of Japan more than two years earlier to commit itself explicitly to maintain its zero-interest-rate policy until deflation had clearly ended, and also in that case the policy signaling facilitated policy objectives by helping to keep long-term interest rates low.9
While explicit discussion of the level of interest rates that is expected to be chosen at future meetings can be useful on occasions like those just mentioned, it is clearly not possible under all circumstances to expect that interest-rate decisions will invariably be signaled many months in advance. The events of the past fall, in which central banks have had to keep abreast of rapidly changing market conditions --- and in which they have needed to be free to announce changes in policy, precisely because of their concern to respond to perceived changes in market expectations --- have for obvious reasons made central banks reluctant to announce in advance what they might or might not wish to do even a few weeks later.
Does this mean that there is no useful way for central banks to communicate about future policy? Here again I think that banks like the Fed have much to learn from the communications policies of the forecast-targeting central banks. Banks like the Reserve Bank of New Zealand (for the past decade), joined more recently by the central banks of Norway and Sweden as well, include quantitative projections for the future path of the policy rate along with the projections for inflation and real activity that are discussed in their Monetary Policy Reports.10These quantitative projections have a number of advantages over the use of “code words” as practiced by banks like the Fed, the ECB, or the BOJ. One is the simple fact that they are much less ambiguous. But another is that they clearly represent forecasts, conditional on what is known at the time of the forecast, rather than advance commitments of policy --- whereas the code words have frequently been understood as announcements of policy intentions, and have had to be kept ambiguous precisely so as to reduce the extent to which future policy decisions can be regarded as having already been announced. An interest-rate fan chart, of the kind published by the Norges Bank or the Riksbank, does not suggest anything of the kind: first, because it is published alongside similar fan charts for other variables, which clearly represent forecasts rather than promised outcomes; and second, because the presence of the widening probability distributions, the farther into the future one looks, rather than a single path, makes it clear that no specific outcome is being promised in advance.
Could a similar approach be adopted by the Federal Reserve? The FOMC is already publishing a summary of the FOMC members’ forecasts for several other variables. And these forecasts are supposed to be made under each member’s assessment of “appropriate monetary policy”; conceptually, at least, it would be a relatively small step to ask each member to describe his or her projected path for the federal funds rate under the assumption of “appropriate monetary policy” as well. If, for example, such a procedure had already been in place in the summer of 2003, the publication of projections indicating a substantial degree of unanimity among FOMC members about the judgment that the funds rate would remain at or near one percent through the following summer would have achieved the same end as the commitment to “a considerable period” of accommodation --- and it would likely have been even more effective, as a consequence of its greater precision.
Would this run the risk of requiring FOMC members to commit themselves far in advance to particular interest-rate decisions? I think it could be done in a way that would create little expectation of that kind. Stressing the conditionality of the projections on data available at the time would be important; and over time, as the discussion of reasons for each meeting’s projections to differ from those made at the previous meeting become a routine part of each “Summary of Economic Projections,” this should become well understood by readers of the Summary. And allowing for changes in policy in response to developments that could not previously be forecasted is as much flexibility as any committee member should seek; in fact, advance commitment of policy, to the extent that things work out in the way that had previously been anticipated, should lead to superior policy decisions, as a decision in advance of this kind leads the policymaker to internalize the consequences of policy anticipations.
The risks of misunderstanding would also be reduced if policymakers announced confidence intervals for their projections, rather than point forecasts. This is a weakness of the format currently used for the FOMC’s “Survey of Economic Projections,” which emphasizes the distribution of views across committee members, but not the degree of uncertainty associated with each of these forecasts. The current questionnaire does, however, solicit rough assessments of uncertainty from each member, and this could be extended and used to present the forecasts in a way that further emphasized their uncertainty.
Even more to the point would be to clarify the contingency of the projections by providing more information about the kind of possible future developments that should be expected to affect future policy. To some extent, this can be achieved by explaining the general framework within which policy decisions are made --- something that forecast-targeting central banks say a good deal about, though the Fed’s “enhanced” communication strategy still seeks to avoid any explicit statements about this.
I believe that it would also be useful to illustrate the general policy strategy by talking through particular alternative scenarios and how the FOMC would expect to respond should they occur. The questionnaire used to solicit the forecasts of FOMC members could be expanded to ask them to comment on a small number of alternative scenarios, and not simply on their view of the most likely evolution given current knowledge. In fact, since the Board staff currently prepares its own projections under several alternative scenarios, and circulates these for discussion prior to each meeting, it would only be necessary to solicit the FOMC members’ views of the scenarios that they are already asked to consider. A summary of views of appropriate policy under each of the scenarios could then be included in the “Summary of Economic Projections” in the published minutes of the FOMC meeting.
Extension of the “enhanced projections” in this way would improve the private sector’s ability to predict Fed policy, for at least two reasons. First, it should make FOMC members more comfortable talking about the interest-rate paths associated with their projections, by making it clear that multiple possible scenarios are possible. And second, discussion of multiple scenarios should also help people in the private sector understand how the outlook for policy should be affected by at least some of the kinds of news that may arrive between publications of the projections. But at the same time, discussion of the appropriate response to alternative scenarios should also help to clarify the nature of the FOMC’s longer-run policy objectives. In this way, it should also serve the purpose of stabilizing medium-to-longer run expectations. One hopes that an evolution of the policy along lines of this sort will be considered by the FOMC, as it gains further experience with its new communications policy.
1 Ben S. Bernanke, “Federal Reserve Communications,” a speech at the Cato Institute 25th Annual Monetary Conference, Washington, D.C., November 14, 2007.
2 See, for example, Ben S. Bernanke, Thomas Laubach, Frederic S. Mishkin, and Adam S. Posen, Inflation Targeting: Lessons from the International Experience, Princeton: Princeton University Press, 1999, which includes a chapter arguing that inflation targeting would be desirable for the U.S.
3 For evidence that the Inflation Reports of forecast-targeting central banks do actually increase the predictability of policy, see Andrea Fracasso, Hans Genberg, and Charles Wyplosz, How Do Central Banks Write? An Evaluation of Inflation Targeting Central Banks, Geneva Reports on the World Economy Special Report 2, London: Centre for Economic Policy Research, 2003.
4 It is no accident that the Bank of England was finally granted the authority to set interest rates independently in 1997, after several years of experience with a forecast-targeting regime.
5 For further discussion, see Michael Woodford, “The Case for Forecast Targeting as a Monetary Policy Strategy,” Journal of Economic Perspectives, Fall 2007, pp. 3-24.
6 Frederic S. Mishkin, “The Federal Reserve’s Enhanced Communication Strategy and the Science of Monetary Policy,” a speech to the Undergraduate Economics Association, M.I.T., Cambridge, Mass., November 29, 2007.
7 A partial exception is Norway, where the official target criterion requires that the projections maintain “a reasonable balance” between the gap between actual and target inflation on the one hand and the rate of “capacity utilization” on the other. However, even in this case, greater prominence is given to the requirement that the inflation rate be projected to converge to the target rate. See the box “Criteria for an appropriate interest-rate path,” Norges Bank Monetary Policy Report 2007/3, p. 9.
8 For further discussion, see Michael Woodford, “Central-Bank Communication and Policy Effectiveness,” in The Greenspan Era: Lessons for the Future, Kansas City: Federal Reserve Bank of Kansas City, 2005.
9 For empirical evidence of the effects of the BOJ’s communications policy, see Nobuyuki Oda and Kazuo Ueda, “The Effects of the Bank of Japan’s Zero Interest Rate Commitment and Quantitative Monetary Easing on the Yield Curve: A Macro-Finance Approach,” Bank of Japan working paper no. 05-E-6, April 2005.
10 For further discussion of this aspect of current practice at forecast-targeting central banks, and its desirability, see Woodford, “The Case for Forecast Targeting,” op. cit.