TL;DR: Recent data have not altered the Fed outlook. The baseline FRB/US model continues to imply that the Fed remains on hold unless core PCE rises above 3.5% on a spot basis. Alternative Taylor-rule specifications point to a marginally more hawkish Fed, but even under these assumptions they imply only one hike in 2026. Stochastic simulations deliver the same message: one hike in 2026 cannot be ruled out, but anything beyond that looks more like a bet than a conclusion supported by robust evidence.
FRB-US BASELINE VS SEP
Following last week’s labor market report, we received several questions about whether the outlook for the Federal Reserve had changed in light of the latest developments, particularly the evolution of the unemployment rate. The short answer is essentially no.
In this note, we present a new set of results from the FRB model and discuss some of the features of the simulation that have generated the most frequent questions. The first point to emphasize is that we have updated the conditioning assumptions of the model. At the margin, the U.S. economy now looks somewhat weaker than the model had previously expected, especially in the second quarter of 2026, for which the current nowcast is relatively low. At the same time, we have also updated the labor market variables, while also revising the labor force participation rate and, most importantly, the inflation variables.
The results are shown in Figure 1, which reports the model’s current forecast and its baseline scenario, represented by the red lines, for the four variables included in the SEP, together with the most recent SEP median, central tendency, and range. As already noted after the June meeting, the path for the federal funds rate implied by the model remains very close to the SEP median, only marginally higher, with the difference between the two being extremely small. In terms of real GDP growth, the model is slightly less optimistic than the SEP median, but it still lies well within the central tendency and therefore in the middle of the distribution. For the unemployment rate, the model forecast is essentially identical to the SEP median. The only meaningful difference concerns core PCE inflation, where the model remains more pessimistic and continues to estimate a higher degree of inflation persistence than the SEP, in line with the signal coming from our other models.
Against this background, the key question is whether the Federal Reserve can still raise interest rates. We remain skeptical about this possibility for a simple reason: in the June SEP, core PCE inflation for 2026 was set at 3.3%, and all our models are currently clustered around this level. In other words, the Fed’s forecast now looks realistic, at least for 2026, and it will not necessarily have to be revised upward in September.
This has important implications for the near-term policy outlook. When we think about the September meeting, our starting point is that there will be no rate hike in July and that, in September, the Fed may also decide to leave interest rates unchanged. The only way to see a rate hike as early as September would be for the inflation forecast to move above 3.5%, a scenario that we have already tested in previous simulations and that remains possible, but would require significant upside surprises in the inflation data over the next two months.
Figure 1. FRB-US forecast (red lines) and most recent SEP (blue lines).
SOME Q&As
How is the path of the FF rate constructed in FRB? Is it fitted on history?
The path of the federal funds rate in the FRB model takes the historical data as given and then projects the policy rate forward according to a Taylor rule. This Taylor rule responds to deviations of core inflation from target and to the output gap. More specifically, it is an inertial Taylor rule, with an inertia coefficient equal to 0.85.
Would the current level of the federal funds rate differ if the model had been allowed to set the Taylor rule in history?
To address this question, we take the model and allow the Taylor rule to determine the path of the federal funds rate directly from the post-Covid period onward. Figure 2 shows the model baseline, again represented by the red line, together with this simulated counterfactual scenario (the green line).
As the figure shows, the model would have started raising interest rates earlier than the Federal Reserve actually did. However, the resulting terminal rate would not have been materially different, and neither would the timing of the first rate cuts. Most importantly, the current level of the federal funds rate would remain only marginally above its actual level, by roughly 10 to 20 basis points.
The answer to the question is therefore that, even if the model had been free to set the path of the federal funds rate endogenously through the Taylor rule from the post-Covid period onward, the current level of the policy rate would not be meaningfully different from where it stands today.
Figure 2. Current FRB baseline (red) and alternative Taylor rules (green and gray)
Would the implied level and trajectory of the federal funds rate change if the Taylor rule were specified in terms of the unemployment-rate gap rather than the output gap?
This is, of course, an adverse scenario, and we simulate it in Figure 2 using the grey lines. As the figure shows, the level of the federal funds rate would be higher than in the current baseline. Moreover, the model would indeed imply a rate hike toward the end of 2026. In any case, even in this scenario there would be only 1 hike this year.
Ok but can you give us the implied probabilities of the model?
Sure. Using stochastic simulations, here are the current implied probabilities for the level of the FF rate in 2026:Q4.
< 350: 0.74%
350-375: 17.7%
375-400: 72.7%
400-425: 7.8%
425-450: 0.7%
450: 0.3%
(Note: Please, note that the current level of the FF rate in the model right now is 3.8%)
CONCLUSION
We have little reason to depart from the approach we have consistently followed, namely using the FRB model in its baseline specification, as this framework has proved useful in interpreting the Federal Reserve’s reaction function. On this basis, we continue to be cautious about expecting a particularly aggressive Fed, unless core PCE inflation were to move decisively above 3.5%, a possibility that cannot be ruled out at this stage but that is not the model’s baseline scenario.
With respect to alternative specifications, it is true that the implied level of the federal funds rate would be higher than in the current baseline. At the same time, even when we push the model beyond its baseline assumptions, it remains difficult to generate more than one rate hike this year. We therefore continue to think that, if markets want to price some additional Fed tightening, the most defensible outcome would be a single rate hike.
Our own view remains that the Fed is more likely to wait. However, if one wanted to assign some probability to a more hawkish outcome, then one additional hike could be justified. Pricing more than one hike, at this stage, looks more like a bet than a conclusion supported by the evidence from the models.