Changes to Fed Communications #3

Post 3 of 3 coauthored with my former Federal Reserve colleague John Roberts

You can find John here.

Could the SEP fill the void?

The Fed already releases forecasts four times each year in the SEP. As we noted in our previous posts, the SEP scores high in terms of relaying policymaker views, including about the future path of monetary policy. However, it does not currently provide a narrative, nor does it provide alternative scenarios. In this post, we consider ways that the SEP could, with modest changes, move closer to a “best practice” projection with these features.

The SEP is released in conjunction with FOMC meetings in March, June, September, and December. After seeing the staff forecast, policymakers submit their own forecast for the most likely outcome (that is, a modal forecast) for real GDP growth, the unemployment rate, inflation, and the target federal funds for the current year and several subsequent years, as well as the values to which these variables are expected to converge over the longer run. Policymaker forecasts are conditioned on individual judgments about appropriate monetary policy, defined as “the future path of policy that each participant deems most likely to foster outcomes for economic activity and inflation that best satisfy his or her individual interpretation of the Federal Reserve’s objectives of maximum employment and stable prices.”[1] Thus, policymakers do not write down what they think the FOMC will do with monetary policy but rather what they would do if they were making the decision on their own.

By linking each policymaker’s forecasts and policy rate path, the SEP could, in principle, provide the information needed for assessing each individual’s monetary policy reaction function. But under current procedures, the public SEP release only provides forecasts variable-by-variable and does not link the forecasts that FOMC participants make for one variable with the forecasts they make for other variables. The Fed’s presentation of the SEP focuses on the median of each forecasted variable in isolation. Because there is no linkage across forecasts, the median forecast for the variables reflects forecasts submitted by different individuals. Thus, the median SEP projections don’t necessarily provide information about the reaction function of, for example, a typical FOMC participant.

As many observers have noted, including the G-30 report, this shortcoming is easily remedied. Linking the forecasts in the SEP has long been debated inside the halls of the Eccles building.

Another key shortcoming of the SEP is that it does not provide a narrative for the projections. However, the current SEP questionnaire provides many opportunities for narrative response and, taken together, the collective narrative is substantial. For example, in March 2019, narrative responses were solicited for the following prompts:

If you anticipate that the convergence process will take SHORTER OR LONGER than about five or six years, please indicate below your best estimate of the duration of the convergence process. You may also include below any other explanatory comments that you think would be helpful.

If you have any explanatory comments regarding your judgment of the uncertainty attached to your projections relative to levels of uncertainty over the past 20 years, you may enter them below.

If you have any explanatory comments regarding your judgment of the risk weighting around your projections, you may enter them below.

Please describe the key factors informing your judgments regarding the appropriate path of the federal funds rate. If, in your projections for any year in the projection period, the unemployment rate for that year is close to or below your projection for its longer-run normal level and inflation for that year is close to or above 2 percent, and your assessment of the appropriate level of the federal funds rate for that year is still significantly below your assessment of its longer-run normal value, please describe the factor or factors that you anticipate will make the lower-than-normal funds rate appropriate. If you have revised your estimate of the longer-run normal value of the federal funds rate since the previous SEP, please indicate the factor or factors accounting for the change. You may include any other comments on appropriate monetary policy as well.

Please describe the key factors, potentially including your assumptions about changes to government policies, shaping your central economic outlook and the uncertainty and risks around that outlook.

Please describe the key factors, potentially including revisions to your assumptions about changes to government policies, causing your forecasts to change since the previous SEP.

Please describe any important differences, potentially including those related to your assumptions about changes to government policies, between your current economic forecast and the Tealbook.

Policymakers do not necessarily pen answers to all of the prompts or provide similar depth or reasoning in their answers, but even so, the SEP as currently structured offers ample raw material for generating a forecast story and for describing uncertainties and risks to the forecast narrative.[2] If these replies were to be used as the basis for a published narrative, the Fed could consider how best to structure the prompts so as to focus answers and elicit a full set of responses from all participants.

Assuming the FOMC decides to link the forecasts across individual policymaker submissions, a key question arises: How would one select a single “headline” forecast and narrative from the linked projections? We suggest one way the Fed could go about addressing this issue: Begin with a statistical procedure to identify the most “typical” SEP submission. We won’t hazard a specific suggestion about how to do this; the Fed staff is chock-a-block with skilled technicians who could propose ideas. Potentially, some principle of weighted least-squares, or minimum distance, could be invoked. The narrative associated with that submission could then provide a starting point for the SEP-associated narrative. Other narratives associated with projections that are statistically close could be drawn on for additional detail and color.

One critique of the way the current SEP is used by the public is the disproportionate attention directed to the central tendency, currently represented by the median forecasts. Providing a narrative for the central forecast selected by the statistical procedure would only exacerbate this concern. To address this, we would again turn to the SEP itself. We would suggest that the information in the SEP be used to choose extreme forecasts for the target federal funds rate (aka the dot plot) that could be highlighted, together with the accompanying forecasts for the macro variables and narratives, to construct alternatives that a handful of policymakers (at the tails) see as most likely.

Providing multiple narratives would help address some of the chief concerns that arise from providing only a single forecast and narrative—in particular, helping avoid excessive focus on the central projection. We recognize that it would differ from true alternative scenarios of the sort that Bernanke advocates—it would, after all, represent only an alternative modal forecast. But we nonetheless think that exploring the diversity of views of the Committee itself is valuable in and of itself. And it would leverage information that is already collected in the SEP: the projections and their narrative.[3]

There are a number of governance and process issues raised by this proposal. In the past, the SEP and modifications to it have been the purview of the FOMC’s informal subcommittee on communications headed by the Board’s Vice Chair. We see nothing wrong with delegating the process and governance questions to that subcommittee. If not the subcommittee, then an appropriate policymaker, or small group of policymakers, should be ultimately responsible. As always, the Fed’s capable staff can provide support in helping policymakers assess options.

Wrapping Up

To sum up our three posts, there are clear advantages to releasing an official central bank forecast, especially in terms of narrative—which enhances communication beyond raw numbers—and alternative scenarios—which appropriately diffuse attention away from the central outlook. If done with sufficient policymaker input, the forecast and alternative scenarios can reveal key information about the central bank’s reaction function.

As we have emphasized, to be most useful, a published forecast must have substantial policymaker input, so that it can provide information to the public about the policymaker reaction function. We note that while the current SEP falls short on the dimensions of narrative and alternative scenarios, it does provide information about the reaction function.

We pointed out key hurdles to the adoption of a policymaker-informed staff forecast, emphasizing issues around the determination of key assumptions and the centralization of power relative to the current SEP approach.

Although past efforts to adopt a policymaker “consensus” forecast failed, we believe that the advantages are sufficiently great that it is worth making another attempt via modifications to the SEP.

Our message to the Fed is: Don’t abandon the SEP prematurely!

---

[1] SEP minutes, p. 1, March 2014 FOMC meeting, https://www.federalreserve.gov/monetarypolicy/files/fomcminutes20140319.pdf.

[2] The narratives along with a complete set of linked forecasts for each policymaker are released after five years. The most recent available historical material is 2019, which was released in early 2025: https://www.federalreserve.gov/monetarypolicy/fomchistorical2019.htm.

[3] This new approach would provide an incentive for policymakers to provide more-complete narratives.