What follows is our standard analysis of the November 2025 CPI report. This month, however, readers should interpret both our model outputs and our broader analysis with particular caution, given the current circumstances.
Specifically, the results presented here are affected by missing data points and rely on a set of assumptions, including imputations, linear interpolations, and other technical adjustments. As such, we strongly recommend that these findings be read with an appropriate degree of prudence and take little signal from the models right now.
Very Soft, No Signal. The November CPI release came in significantly below expectations. Yesterday, the Bureau of Labor Statistics (BLS) provided additional details regarding the methodology used to impute missing observations—information that was not available at the time we prepared our preview and, in retrospect, pointed to downside risks. That said, we are not drawing strong conclusions from this surprise at this stage as some of the details are puzzling (i.e. flat rent/OER in October?). We will take the necessary time to analyze today’s report in greater depth before updating our assessment.
CPI/PCE Translation: Based on the CPI and PPI data, we estimate that core PCE will print at 0.15% MoM in November. As for headline, we expect 0.15% MoM translation. See here for details. Please note that this month’s translation is necessarily approximate, as it is based on a number of assumptions, interpolations, and imputations of missing data.
Reaction to the incoming data. We have revised our near-term inflation forecast upward. While we are still finalizing the updated projections, it appears that due to imputation effects, the “true” level of many underlying series is likely higher than reported in the latest release. As a result, we anticipate stronger inflation readings in the near term. Specifically, we have increased our forecast for core PCE by approximately 35 basis points over the next three months. We now expect the December core CPI to register at 0.37% month-over-month (seasonally adjusted), and the January core CPI at 0.45%.
Models Insights: limited signal. Both our Common Inflation (CI) model and the associated distributions are currently heavily affected by missing data. While we provide a model run, we emphasize that it relies on a number of assumptions. As such, we strongly recommend drawing no meaningful signal from these outputs at this time. Regarding the medium term, our main model has been revised slightly downward. However, as explained below, this adjustment should also be interpreted with caution given the current data limitations. Importantly, the model continues to project inflation above the Federal Reserve’s target over the forecast horizon. We will continue working with the models in the coming days and will provide updates as we refine our assumptions and incorporate additional information. Further revisions to the forecast are likely.
A PDF containing all relevant CPI charts has been posted. You can download it here.
Evidence from the distributions
The distribution: large movement, NO signal. This month, the distribution is very dispersed, similar to last month and remains very different than pre-Covid (see ridge plot). The median (Figure 2) ticked up in November and remains volatile. As shown in Figure 1, the distribution shows large movements in the last 3 months (Sep-Oct-Nov) but we take no signal for obvious reasons.
Figure 1. Kernel of CPI excluding food and energy items changes (MoM %, a.r.)
Note: the Figure shows the fitted Kernel (Epanechnikov) distribution of MoM percent changes at annual rate of CPI prices excluding food and energy items.
Figure 2. Median (core) CPI metrics
Note: the Figure shows the median (MoM %, a.r.) of the distribution of CPI prices changes excluding food and energy items (left panel) and the YoY (right panel).
Evidence from our CI-C model
Our CI model estimates that, excluding idiosyncratic shocks, the common component is weak. Figure 3 illustrates the decomposition of the MoM core CPI into “common” and “idiosyncratic” components. This month, the model estimates that the common component increased by 14bps, below the average of previous months, while the idiosyncratic shock was negative (-7bps). The 3m/3m of the “common” component (Figure 4) stands at 3.3%. We take no signal from today’s reading, as we expect the data to offset today’s weakness in coming months.
An Excel file containing the results shown in Figure 3 and 4 can be downloaded here.
Figure 3. Contributions to MoM changes of CPI excluding food and energy items (CI model)
Note: the Figure shows the decomposition of the MoM percent changes of CPI prices excluding food and energy items. The contributions are estimated using our CI model.
Figure 4. Estimated “Common” component: YoY, 3m/3m a.r. and 6m/6m a.r.
Note: the Figure shows the 3m/3m at annual rate (green line), the 6m/6m at annual rate (red line), and the YoY (blue line) of the “common component” estimated using our CI model.
Implications for the medium-term forecast of core PCE price inflation
The medium-term is unrevised. Relative to the run conducted at the time of the preview, the medium-term forecast from our main model has been revised downward. Based on the latest data, our Q4 nowcast has been revised to 2.1% QoQ SAAR. However, as noted above, we have offset today’s downside surprise by raising our Q1 forecast to 3.9% QoQ SAAR. In addition, we are now running the model with Q1 placed in-sample. While this represents a significant methodological adjustment, we believe it is warranted. Without this step, the model would place excessive weight on today’s data point, given its limited information set relative to what is currently available to analysts and forecasters. The model’s latest Q4/Q4 forecasts are as follows: 2.7% in 2025, 2.6% in 2026, 2.4% in 2027, and 2.4% in 2028. The forecast remains above target going forward.
Figure 5. “Main” Phillips curve model forecast, core PCE price inflation (YoY, %).
Note: the figure shows the latest run of our “main” Phillips curve model. 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).