The Fed staff (and SEP) forecast of core PCE price inflation has been consistently below model-based forecasts since the third quarter of last year.
In this note we explain why, in our view and experience, this has happened, what are the risks around the Fed staff (and SEP) forecast and what to expect going forward.
To keep in mind
The Fed staff (and SEP) forecast of core PCE price inflation has been well below model-based forecasts due to the inertia of the staff judgmental forecast. In our view and experience, the Fed staff (and SEP) is at risk of repeating the same mistake it made last year: underestimate the risks, especially at the end of the medium-term. The consequence is that interest rates might remain too low and inflation might end up higher than the staff (and the FOMC) are expecting next year and the following.
The Fed staff forecast
In order to understand the mechanics of the Fed staff judgmental forecast, we need to go back to the end of 2021. According to the December 2021 FOMC minutes, the Fed staff expected inflation to return quickly towards target: “PCE price inflation was therefore expected to step down to 2.1 percent in 2022 and to remain there in 2023 and 2024”. A reverse engineering of the Fed staff forecast at the December 2021 meeting resulted in the contributions shown in Figure 1.
Figure 1. Reconstruction of the Fed staff core PCE price inflation decomposition in December 2021
Note: the chart shows the evolution of core PCE price inflation and its contribution at the time of the December 2021 Tealbook. The chart shows our reverse engineering of the Fed staff forecast, as inferred from the FOMC minutes. The bars show the contributions, while the black line shows the Q4/Q4 (% change) of core PCE price inflation.
However, the estimated “main” Phillips curve model (based on Detmeister et al. (2014)) was delivering a significantly higher path of core inflation (Figure 2). According to the model, core PCE price inflation was expected at 3.1% in 2022 (Q4/Q4), 2.7% in 2023, and 2.5% in 2024. Not only, but the model estimated the probability that core inflation would reach 2% at the end of the forecast horizon to be quite low (around 30%).
Figure 2. Forecast of core PCE price inflation of the “main” Phillips curve model in December 2021.
Note: the chart shows the “main” Phillips curve model forecast of core PCE price inflation (YoY, %) at the time of the December 2021 Tealbook.
Why was the Fed staff judgmental forecast lower that the estimated model?
There are two reasons. First, in our experience, the Fed staff judgmental forecast is based on a calibrated version of the estimated “main” model. In other words, the Fed staff *sets* judgmentally the contributions. (Needless to say, the calibration is, indeed, based on the estimated parameters of the model but at any point in time there can be a significant gap between the two, inertia being a big factor driving the wedge) Second, while the forecast of the current year is a mix of incoming data and model-based predictions, the forecast of the years after the current one are based only on the calibrated fundamentals. Put it differently, the “other” factors (the yellow bars in the decomposition, which are the quasi-residuals of the model) are set to zero for the years after the current one. This explains why the Fed staff revised up its 2022 forecast only in March of this year: the staff needed some data before deviating from the calibrated 2022 fundamentals.
Both factors (the calibration and the zero residuals) are reasonable assumptions in a low inflation environment because the gap between the calibrated and the estimated models is tiny. But in a high inflation environment, setting the yellow bars to zero in the forecast is very risky. Put it differently (and more directly): you cannot correctly forecast inflation if you *assume* that it has no persistency, despite the evidence signaled by the estimated model. Unfortunately, this is what the Fed staff has done in 2021 and what seems to keep doing right now.
Does the current Fed staff (and SEP) forecast suffer from the same issue?
The current situation is very similar to what happened last year. According to the July 2022 FOMC minutes, the Fed staff expected the following: “core inflation was expected to be 4.0 percent [in 2022]. Core PCE price inflation was expected to step down to 2.6 percent in 2023 and to 2.0 percent in 2024”. Figure 3 below shows our reverse engineer of the Fed staff forecast and its contributions.
Figure 3. Reconstruction of the Fed staff core PCE price inflation decomposition in July 2022
Note: the chart shows the evolution of core PCE price inflation and its contribution at the time of the July 2022 Tealbook. The chart shows our reverse engineering of the Fed staff forecast, as inferred from the FOMC minutes. The bars show the contributions, while the black line shows the Q4/Q4 (% change) of core PCE price inflation.
Compared to the forecast made in December 2021, the Fed staff had to revise up its 2022 forecast because the inflation process proved to be persistent. Unsurprisingly, the latest decomposition (Figure 3) shows a large yellow bar not only in 2021 but also in 2022.
Crucially, the Fed staff continues to think that (conditional on the FF rate reaching about 4% and the unemployment rate gradually converging to the natural rate) inflation will quickly return to target. At the same time, the “main” model continues to flag significant upside risks (Figure 4), especially at the end of the medium-term (for the record, the “main” model forecast is 2.9% in 2025).
Figure 4. Forecast of core PCE price inflation of the “main” Phillips curve model in September 2022.
Note: the chart shows the “main” Phillips curve model forecast of core PCE price inflation (YoY, %) at the time of the September 2022 Tealbook.
In other words, the Fed staff (and the SEP) forecast seems to be in a similar spot compared to where it was at the end of last year because the risks continue to be skewed to the upside. At this point, the baseline scenario is that the Fed staff might revise up its medium-term forecast going forward. But if it follows its normal procedures, the Fed staff will wait until the beginning of 2023 to revise its 2023 forecast and by doing so it might end up being late again.
Can the Fed staff be right? (Alternatively: what can go wrong?)
As usual, the answer is “yes” but the probability is quite low. In a nutshell, the Fed staff can be right but only if either (i) supply comes back vigorously (which might or might not happen at this point of the story) or (ii) demand slows down significantly (a.k.a. “recession” which is in any case not the baseline of the Fed staff).
Durable goods prices will probably slowdown/drop in September/October (see Manheim index). But given the distribution of price changes, the effect would probably be short-lived. The baseline is that, conditional on the FF rate peaking at 4%, inflation is expected to remain well above target in the medium-term (please note that some investment banks are now predicting core CPI at or above 3% beyond 2024). The Fed can change the future. But it should start by admitting the reality of the data: inflation is way more persistent than the Fed staff (and FOMC) anticipated and moderating it will probably require a more aggressive monetary policy.