Thank you, Mr. Chairman. I’d also like to lend my voice to the vote of thanks to the staff on what I thought was an excellent set of memos. I think the staff did an outstanding job in summarizing a lot of the nuances and the details that are involved.
Perhaps it comes as no surprise that I’m in favor of defining the Committee’s price objective numerically and announcing that goal to the public. I believe that specifying our long- run price stability objective numerically would focus our policy discussions and help anchor expectations. By reducing the public’s uncertainty about our goal, long-run expectations would become less responsive to changes in short-run inflation. This should help enhance monetary policy’s flexibility to respond to economic shocks as we may deem appropriate. Thus, rather than being unduly constraining, I believe it would actually add to our flexibility. In this way, it’s consistent with the other goals of monetary policy and increases monetary policy effectiveness. It would also increase social welfare to the extent that it helps us avoid time-inconsistent policy, since we know welfare under commitment generally exceeds welfare under discretion when agents are forward looking. Thus, I believe a numerical goal would be a long-run anchor to our monetary policy and help coordinate our own discussions of what appropriate policy might be.
Now, as Vince laid out the questions, we have to make a number of decisions in making that numerical goal operational. I feel more strongly about some aspects of the design than about others, and many of the choices are interrelated. For example, a long horizon should mean a tighter control range. A long horizon makes the choice between headline and core less important. A long horizon makes the choice of core in fact less compelling. I think that any proposals will have their pros and cons. As I said, I’m not necessarily wedded to all of the particulars, but I will make a proposal. But let me emphatically stress that what is more important than the specifics is that we agree on a numerical objective and a horizon for its achievement.
Regarding which price index, I prefer the headline CPI even though it’s likely to be harder to control and forecast in the short run than the core. Many foreign central banks with experience have tended to move toward the headline CPI number, indicating that it can in fact work. The headline CPI is a measure that’s more understood by the public, so the communication arguments are important here. Unlike the PCE, as has been mentioned a couple of times, or the GDP deflator, the CPI is not revised, which helps us in assessing our accountability in reaching our goal. I also prefer headline to core for our goal because I don’t want to convey the idea that we are insensitive to the wider array of prices that influence behavior—particularly because, as Governor Mishkin indicated, the stochastic processes or properties of individual elements of the CPI may change over time. If we started defining a subset of prices, we could find ourselves in trouble. I think we would also have the opportunity to use the core in our communications when explaining why we might or might not react with policy to a temporary blip in headline inflation. I view focusing on headline and using the core for other purposes as a means of enhancing our communication efforts. Again, this practice is similar to the practices of other central banks that announce their inflation goals. Of course, I think it’s going to be terribly important for us to consider how we will respond to and communicate about the inevitable misses from our target.
Regarding a point goal or range, I strongly favor announcing a point goal. In reality, there is a range around this point that reflects the precision with which we policymakers think we can control inflation, and this control range will differ depending on the inflation measure used and the time horizon selected. However, I’m reluctant to announce a range as part of our inflation goal because I think it would be very difficult to ensure that the public would not interpret it as a range over which we are indifferent. So for the headline CPI, I would specify and announce a target of 1 percent. That’s consistent with our goal of price stability and the estimated measurement bias of the CPI of being something slightly less than 1 percent. I picked 1 percent because I take seriously our mandate for price stability. Since I do not believe that there is any long-run tradeoff between inflation and employment, we have no reason not to seek and meet that goal over a reasonable period. I recognize that some may feel that a 1 percent target is too low as the risk of deflation or zero bound restrictions on nominal interest rates might call for a greater cushion. I understand those arguments, and they are certainly plausible. But for various reasons, some articulated by people around this table, I’m less concerned about our ability or the economy’s ability to deal with those issues, both of deflation and zero bounds. But I accept that some people may have more concern about it than I do.
Regarding time horizons, I feel strongly that we should specify a time horizon by which we think we can achieve our target so that we can be held accountable for meeting or missing a goal. Since I favor using the headline CPI, which is a little more difficult to control, more volatile, than a core measure, I think a two-year horizon would be appropriate and, indeed, achievable given the typical shocks that hit the economy and the volatility of the CPI. For example, for the past ten years, the standard deviation of the monthly twenty-four-month CPI headline inflation rate has been about 0.5 percent. I could also make the case that initially we may want to consider a slightly longer horizon, especially if we choose a number like 1 percent and given that we are currently well above 2 percent. Specifying a longer horizon at first may provide markets with more opportunity to adjust to the new regime and mitigate some of the transition costs. As we converge to our target, we might be able to shorten the horizon. I much prefer that idea to adjusting the goal. I think the goal ought to be the goal, and we use the horizon to give us some flexibility, depending not only on initial conditions but also perhaps in future discussions on the nature of the shocks that may cause us to do that. I don’t like the idea of not picking an optimal target simply because we’re not there yet. I think that’s not the right strategy.
There remains the issue of how we treat deviations from our goal—that is, whether we let bygones be bygones or whether, for example, when inflation has been above our target for a year, we must get it below our target for a year. Our chosen inflation goal will imply a price- level path with that goal. We need to decide whether or not we will permit permanent deviations from that price-level path. Deviations from the price-level path occur when inflation deviates from our goal. If inflation increases above our goal for a time, then we would need to bring inflation below our goal to return to the price-level path that was consistent with our initially announced inflation goal. Similarly, if inflation moved below our goal, we would need to have a period of above-target inflation to return to the price path. Alternatively, the Committee might decide to accept permanent deviations from the price path and choose to implement policy only to return to the inflation level. I would prefer the former as a price-level path because it prevents base drift, and I suspect that we will, on average, more likely be above the target than below, and the ensuing gradual erosion of purchasing power will be higher than we might have anticipated. In either case, I think it’s important, regardless of which way we decide, that we make that choice consciously and weigh the costs and benefits of it and decide which regime we want to be in.
I do think it’s critical that the FOMC members reach a consensus, or at least a decision, on the goals and the definition of price stability and essentially not dispute those in public. I don’t think members need to agree on the model of the economy or the channel through which monetary policy affects the economy. Indeed, given the state of economic science, the differences in the models and the channels can actually aid in policy formation. However, the point of announcing a numerical definition of price stability is to anchor expectations and improve our accountability. Without agreement on that definition, the benefits of such an announcement would be critically diminished. The Committee may want to periodically review the definition of particulars such as the horizon as it gains experience operating under this structure, but I think it’s very important that we reach a consensus on our announced goal. While I have offered my own choices on the particulars, I believe that they really are of secondary importance to our public commitment to an objective.
As far as the trial run is concerned, I strongly favor having a trial run for producing a forecast narrative in May, and I agree with the discussion last time in that I like the idea of doing it four times a year. I also like the idea of incorporating a forecast narrative into the minutes, which gives participants the opportunity to comment on the draft narrative. This is not the only way to proceed, but I think that it is a good first step and that refinements can follow. As I mentioned earlier, our discussion this morning highlighted the real need for us to have a way of communicating our policy views more effectively and outside the narrow confines of the policy statements as they are currently constructed.
Regarding the narrative itself, I have two comments. First, I suggest adding the assumed policy path as a variable in the forecast. Participants are asked to assume appropriate monetary policy, and conditioning on policy paths that can differ across participants embeds differences in participants’ preferences over outcomes as well as differences across their models. If we are thinking about the forecast as a communication device that enhances the transparency of our policymaking process, then we want to convey something about the reaction function that is likely to arise out of our Committee’s decisionmaking process. Aggregated information, such as the range and central tendency of the fed funds rate in the fourth quarter of each of the three years of the forecast, as we do with the unemployment rate, might be useful without holding the Committee to any particular path. I think the idea of conveying information regarding the uncertainty of the forecast is also important so that the public does not place too much emphasis on point forecasts or narrow ranges. Moreover, uncertainty is clearly larger than the range of the point forecasts of the Committee, and yet I’m not sure, as has been mentioned already, that using the forecast standard errors out of the FRB/US model is the best solution since they are not consistent with the forecast that was generated by the Committee. We might, for example, even consider asking Committee members when they submit their forecasts to submit their own range of uncertainty at the four-quarter horizons and then use those estimates to create something like a fan chart. In any case, I’m looking forward to a trial run in May. Thank you, Mr. Chairman.