The data will be soft. The Fed will not pivot (for now).
The October CPI report showed that the distribution of price changes is shifting. In this note we use our models to predict (if and) when the Fed staff might start sending dovish signals to the FOMC.
Keep in mind
There is now a “tension” between the monthly incoming data that suggest softness and the medium-term models that suggest persistency of inflation.
If the question is “will the monthly pace of core (CPI and) PCE price inflation moderate in the near term?” then the answer seems to be “yes”. But if the question is “can the Fed staff (and FOMC) pivot soon?” then the answer seems to be “not yet” because monetary policy is set by the medium-term forecast which remains against the Fed.
Assumptions
Sample: This is the first time we include 2022:Q4 in the sample using our initial nowcast (4.4% a.r.). The forecast starts in 2023:Q1. Our Q4 nowcast for core PCE prices assumes a MoM path of 0.25, 0.34 and 0.36 in October, November and December, respectively.
Exogenous: the unemployment rate is assumed to gradually trend higher over the forecast horizon and reach 4.5% at the end of it. We assume that U* is currently at 4.5%. Given the behavior of the broad dollar, we assume that core import prices will contract 5% (a.r.) this quarter, that they will be flat next year before growing 1.3% per year in 2024 and 2025. Finally, as usual, we follow the futures contracts for energy prices.
Results
According to our “main” model, core PCE prices will expand 3.8% in 2023, 3.8% in 2024, and 3.5% in 2025 (see Figure 1 below). Importantly, the model continues to signal a (very) low probability of returning back to target by the end of the medium-term.
Comparison with previous forecasts
Compared to forecast at the time of the November 2022 FOMC meeting, the current forecast is stronger (about 1/2 percentage point) in each year.
Why the model revised up its forecast despite the fact that the CPI was quite soft?
There are two reasons:
- The October CPI report was softer than expected by market participants (2 tenths to rounding) and in our view a bit lower than the Fed staff expected (less than 1 tenth). However, our Q4 nowcast is still higher than the forecast of the model stopping the estimation in Q3. Therefore, the new quarter in the sample is a net upward surprise for the model and the forecast is revised up accordingly.
- When a new quarter is added in the sample, the model re-estimate the parameters, including the autoregressive coefficients which determines the persistency of the process. Because all recent quarters are now very strong, the model estimates a higher persistency and projects the recent strength forward.
How much signal should we take from the “main” model?
We monitor the evolution of the “main” model in order to understand the risks around the Fed staff (and FOMC) forecast.
In the last year the “main” model has proven to be ahead of the curve and anticipated well the SEP dots and the Fed staff forecast (a comparison of the evolution of the forecasts is available upon request). However, we might be now at a turning point because the near-term models (CI-C, etc..) and the monthly distributions are sending strong signals. In other words, in our experience using the “main” model, we now see a risk that the model might overstating future inflation.
What would it take for the “main” model to revise down its forecast going forward and when might this happen?
The short answer is: a large downward surprise in incoming data (or a large revision in the assumed path of some exogenous, such as a much higher path of the unemployment rate).
The model is running on the assumption that core PCE prices will grow 4.4% (a.r.) in 2022:Q4. Therefore, we doubt that the model will revise significantly its forecast this quarter. Rather, the current forecast of the model for 2023:Q1 is 3.7% (a.r.). Accordingly, there are some potential downside risks next quarter, should the monthly prints continue as suggested by the monthly distributions and the CI-C model.
All told, the model is unlikely to send dovish signals this quarter. Instead, it might start sending some small dovish signals in 2023:Q1, should the data cooperate. Having said so, at the moment in our view and estimates the baseline is that the real dovish signal will come in 2023:Q2.
Figure 1. Main model forecast (YoY of core PCE, %)
A. C.I. calculated from parameters distributions
Note: the confidence intervals (C.I.) are estimated using 100,000 draws from the estimated parameters distributions to simulate the path of core PCE price inflation going forward.
First quarter of forecast: 2023:Q1.
B. C.I. calculated using quasi-out-of-sample methods
Note: the confidence intervals (C.I.) are estimated using quasi-out-of-sample methods (estimate the model over a sub-sample, forecast, and calculate the root mean squared forecast errors).
First quarter of forecast: 2023:Q1.
What about other specifications of the “main” model?
The “main” model is constructed under several assumptions, including a specific lag structure, a specific measure of “slack” (the unemployment gap), etc…
Because some of these assumptions have progressively become more and more important in determining the final forecast, we checked the “main” model result against a large set of alternatives (around 100,000 alternative specifications). Each alternative differs from the other in one dimension, such as the number of lags of the endogenous variable or the measure of slack (V/U instead of the unemployment gap).
(For the record, so far we have not stressed this exercise, even if we regularly reported in our “Pre FOMC Meeting Package” because the forecast of the main model was close to the one from this exercise)
The distribution of point forecasts in each year of the medium-term is presented in Figure 2 (the left panel shows the results of all models using Ugap as a measure of slack, while the right panel of all specifications).
There are two main takeaways from this exercise: (i) the “main” model is delivering a higher forecast compared to other specifications mainly because its lag structure, and (ii) in any case the alternative specifications continue to see a low probability (around 25%) of core PCE price inflation to hit target in 2025. Therefore, in our view and estimates it is still early for the Fed to pivot.
Figure 2. Distribution of point forecasts of alternative Phillips curve models
All anchored models using Ugap as slack
Note: the figure shows the distribution of point forecasts of a large set of Phillips curve models with “anchored” expectations and using the unemployment gap as a measure of slack. This set of models restricts the coefficient of the long-term expectations to sum up to one with the lags of the endogenous variable (so that movements in long-term expectations eventually translate into equivalent movements in actual inflation). Whiskers show the 25th, 75th, 10th, and 90th percentile of the distribution. The red line and the light blue dot in each whisker show the median and the mean of the distribution, respectively.
All anchored models
Note: the figure shows the distribution of point forecasts of a large set of Phillips curve models with “anchored” expectations. Total number of models is about 100,000. This set of models restricts the coefficient of the long-term expectations to sum up to one with the lags of the endogenous variable (so that movements in long-term expectations eventually translate into equivalent movements in actual inflation). Whiskers show the 25th, 75th, 10th, and 90th percentile of the distribution. The red line and the light blue dot in each whisker show the median and the mean of the distribution, respectively.
Implications for the Fed Board staff (and the FOMC)
As explained in previous communication, in our experience the Fed Board staff sets the medium-term forecast using a calibrated version of the “main” model, which can be therefore used as a way to assess the risks around the forecast.
Right now, the monthly incoming data are likely to remain soft, given the shift of the distribution. At the same time, the “main” model is likely to send hawkish signals for another few months.
Therefore, if the question is “will the monthly pace of core PCE price inflation moderate in the near term?” then the answer seems to be “yes”. But if the question is “can the Fed staff (and FOMC) pivot soon?” then the answer seems to be “not yet” because ultimately monetary policy is set by the medium-term forecast which remains against the Fed.
This is why we continue to think that the most likely scenario is that the Fed will raise the FF rate to 5%-ish and keep it there for a few quarters. Can the Fed be “late” and “overdoing” it? Sure. But in this situation, we are confident that both the Fed staff and the FOMC will want to see “the white eyes of disinflation” before start cutting.