US: FRB-US Update and FOMC Reaction Function

Preamble. We have updated our near term bottom up forecasts, which are now available on our website. We currently expect core CPI inflation to stand at 3.4-3.5% in December 2026. This forecast does not yet incorporate the forthcoming BEA revisions. We continue to expect core PCE inflation to be revised downward by around 20 basis points. All other details, in both CPI and PCE space, are available on our website. For convenience, we provide the two relevant links here, one for CPI and one for PCE.

The Exercise and New Assumptions

This is a somewhat unusual model update, because our objective was to construct a baseline that matched the September SEP as closely as possible in order to understand what assumptions could generate the policy path implied by the new projections.

This was not straightforward. Ultimately, we had to assume an r star of 1.2%, particularly strong Q3 real GDP growth of 3.7% annualized, and core PCE inflation of 3.4% on a Q4 over Q4 basis in 2026. Under these assumptions, the model comes very close to reproducing the September SEP.

Given the difficulty of obtaining this result, we also tested several alternative specifications. We first estimated a less inertial Taylor rule, reducing the inertia coefficient from 0.85 to 0.75, but this did not improve the fit. We then replaced the output gap in the Taylor rule with the unemployment gap, again without success. Finally, we imposed stronger GDP growth in 2027 in an attempt to keep the federal funds rate higher that year, but this specification also failed to reproduce the SEP path. The broader point is that, at present, it is not straightforward to model the Fed’s reaction function precisely. 

The New Baseline

Overall, the baseline comes reasonably close to the conclusions of the September SEP. The main differences are that growth in 2027 and 2028 is somewhat weaker than assumed in the SEP and, more importantly, that there remains a significant disconnect in the inflation dynamics. While the SEP projects inflation returning broadly to target, the model suggests that the disinflation process will take considerably longer.

As a result, the model continues to imply that, to bring inflation back to target within a reasonable time horizon, the federal funds rate would need to follow a substantially more restrictive path, rising well above 5%.

An important caveat concerns the core PCE projection. As discussed above, we currently assume core PCE inflation of 3.4% on a Q4 over Q4 basis in 2026. This is a relatively generous assumption, as it does not incorporate the forthcoming BEA revisions, which we expect to lower the measured inflation path. If the core PCE path is indeed revised downward, as discussed in our previous communications, it would become very difficult to reconcile the federal funds rate path signaled in the September SEP with the model. Under a lower core PCE trajectory, the model would imply a correspondingly lower path for the federal funds rate.

All charts and the “build your scenario” tool have been updated on our website (underlyinginflation.com). Simply log-in. We will update FRB-US baseline again next week after PCE data.

Figure 1. SEP forecast (blue lines and bands) and FRB-US forecast (red lines)

Note: Real GDP growth and core inflation are expressed on a year-over-year basis. Core inflation refers to core PCE price inflation. The blue lines and bands represent the latest SEP projections, while the red lines depict the “inconsistent FRB-US” forecast—i.e., the model-based baseline without the Fed staff’s add-factors used to align with the latest SEP. This forecast has been updated as outlined in the text.

Did the FOMC reaction function change overtime? If so, how?

Given the difficulty of modeling the FOMC reaction function, we conducted a complementary empirical exercise to assess whether it has changed over time. The model is estimated using 44 SEP vintages from 2012 through 2022, with observations for the one year ahead and two year ahead horizons, for a total of 88 observations. The dependent variable is the federal funds rate projected in the SEP. The explanatory variables include the core PCE inflation gap relative to 2%, the unemployment gap relative to the equilibrium rate reported in the same SEP, the neutral rate, and the post meeting federal funds rate as a control for the initial policy stance. We then keep the coefficients estimated through 2022 fixed and use each SEP from 2023 onward to calculate the policy path implied by the historical reaction function. Similar results are obtained when the estimation sample ends in 2019.

The result is striking – see Figure 2 below. The post 2022 federal funds rate path was substantially below the path implied by the FOMC’s own historical reaction function. The divergence was particularly large for the two year ahead rate, which at its peak in 2024 was roughly 150 basis points below the historically implied level. Our interpretation is that, once inflation had declined from its post pandemic peak, the FOMC placed less weight on completing the normalization of inflation and more weight on labor market risks, while underestimating the persistence of inflation. The gap has since narrowed substantially. The current two year ahead policy rate is much closer to the level implied by the historical reaction function, although it remains around 20 to 30 basis points below it. The latest SEP can therefore be interpreted as a partial normalization of the Fed’s reaction function toward its historical behavior.

One important qualification is that the exercise is conditional on the SEP forecasts themselves. If we instead assume a core PCE path that does not converge to 2%, the historically implied policy rate becomes materially higher. This is consistent with our FRB US simulations, which suggest that a stronger commitment to returning inflation to target would require a federal funds rate of at least 5%, and potentially around 5.5%. Overall, the exercise points to a substantial dovish shift in the FOMC reaction function during 2024 to 2026, which now appears to be reversing (although not yet completed).

Figure 2. SEP reaction function

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.