We have updated our core PCE price inflation forecast using our “main” model (as a reminder, the “main” model is the Fed Board staff model as described in Detmeister et al. (2014)).
As usual, our forecast includes a point forecast, as well as two types of confidence bands: A. confidence bands calculated using 100,000 draws from estimated parameters’ distributions, and B. confidence bands calculated using quasi-out-of-sample methods.
The model includes 2021:Q3 in the sample and the forecast extends to 2024:Q4. Compared to the last update, the starting point of the forecast (2021:Q3) is slightly higher and we have revised marginally the path of core import prices (downward), and the path of the unemployment rate (downward). In the next update we will start including 2021:Q4 in the sample using our forecast of it.
Results: The inclusion of 2021:Q3 in the sample has triggered significant upward revisions to the forecast of core PCE price inflation, as the model is now taking a lot of signal from the recent strength of the data. The model is currently forecasting elevated inflation in 2022 before dropping towards target at the end of the medium-term.
Implications for the Fed Board staff: The Fed Board staff forecast is significantly lower in 2022 (and beyond) than the “main” model. As previously discussed, the staff is working under the assumption that its own information set is larger than the model. Under the staff view, most of the strength of 2021:Q2 and 2021:Q3 will prove transitory, as the level of durable goods will stop increasing and possibly drop in 2022. In any case, the staff has admitted (see latest FOMC minutes) that the risks around its forecast are skewed to the upside as we have been signaling in the last 6 months or so. Going forward, we re-iterate our call that the Fed Board staff will be reluctant to make significant changes to its 2022 core PCE price forecast, unless it will receive high incoming data at the beginning of 2022.
We will update our “thick modelling” set of models in the next few days. As we previously flagged, given the current environment, the distribution of point forecasts from a large set of Phillips curve models is essential to complement the forecast of the “main” model.
Inflation forecast (YoY of core PCE price inflation, %) from "main" model
A. Confidence intervals calculated using 100,000 draws from estimated parameters' distributions
B. Confidence intervals calculated using quasi-out-of-sample methods
Note: the confidence bands are estimated using quasi-out-of-sample methods (that is, estimate the model over a sub-sample, forecast from that point onwards, and calculate the associated root mean squared forecast errors).