Following several questions along the lines of “What is your estimate of trend inflation in the U.S. once idiosyncratic shocks are stripped out?” and “Are we sure that the recent inflation dynamics are not simply driven by idiosyncratic movements in goods prices?”, we present the results of some models that we do not typically circulate.
The TL;DR is straightforward: the cross-sectional distribution of inflation across items remains inconsistent with a 2 percent target; estimated trend inflation is running at around 3 percent in both CPI and PCE space; and there is evidence that the strength in goods inflation is not merely idiosyncratic, but instead reflects a broader common trend.
The evidence
Evidence from DFM model
We run a multivariate common-trend model (in the spirit of the New York Fed framework) on CPI and PCE sectors. Documentation for the model is available here. The objective of the model is to separate sector-specific idiosyncratic shocks from the common inflation trend shared across sectors. The results are presented in Figure 1.
The bottom line is straightforward: the model estimates underlying core CPI trend inflation to be above 3 percent.
The strongest evidence comes from the sectors currently under the greatest pressure—namely goods. In these sectors, the rise in inflation is being driven both by idiosyncratic shocks and by a common component. Figures 3 and 4 show the estimated trend for CPI Apparel and PCE Durable Goods. The trend in these sectors (the blue lines) has increased substantially and is now approaching levels last seen in 2021–2022. The key point, however, is that while the idiosyncratic component (the red lines) has also risen, it does not come close to explaining the full increase in the overall trend. Put differently, the gap between the blue and red lines—that is, the contribution of shocks common to all sectors—is currently substantial. This stands in marked contrast to the prevailing narrative in many policy discussions and in financial markets. Translated in simple terms: inflation is here to stay and likely persist more than expected because it is more broad-based that “it’s mainly goods inflation”.
Figure 1. Core CPI mom (saar) and estimated trend

Figure 2. Core PCE mom and estimated trend

Figure 3. CPI Apparel – Total trend and sector-specific trend (contribution)

Figure 4. PCE Prices Durable Goods – Total trend and sector-specific trend (contribution)

Evidence from the trend models
We run our full suite of trend-inflation models. We have also added several new specifications, bringing the total number of models to fifteen. Documentation for the trend models is here. Results for core PCE are reported in Figure 5.
The models have revised recent quarters modestly upward and now point to trend inflation of approximately 2.8 percent in core PCE space. Because we typically report only the simple average across models, Figure 6 also presents alternative methods for extracting signals from the model set, including the median, the trimmed mean, RMSE-weighted estimates, and other aggregation approaches. The key conclusion is that these alternative methodologies often produce estimates of trend inflation that are even higher than the simple average.
For comparison, Table 1 reports all metrics for core and headline CPI as well as core and headline PCE. Overall, the evidence suggests that trend inflation is currently running at roughly 3 percent in core PCE, and around 2¾ percent in core CPI, headline CPI, and headline PCE. These estimates have moved broadly sideways relative to their 2025 averages and remain well above the levels observed during the 2016–2019 period.
Figure 5. Core PCE – Estimated trend

Figure 6. Core PCE – Alternative ways of taking signals from estimated trends

Table 1. Inflation Trend – Summary

Evidence from the distributions
For several years now, we have argued that the distribution of price changes is not consistent with the inflation target (documentation of the distribution charts is here). Figure 7 shows the cumulative density function of cross-sectional month-on-month price changes in CPI space. The difference between recent months and the 2016–2019 period (when core CPI inflation was running close to 2 percent) is readily apparent.
To facilitate the analysis, we have also prepared a table reporting a variety of distributional statistics, together with several measures designed to quantify the distance between the current distribution and the 2016–2019 benchmark distribution.
The conclusion is straightforward. Whether one looks at the most recent months or at the past year as a whole, virtually none of the indicators have returned to their pre-Covid levels. Particular attention is drawn to measures such as percentile ranges and the third and fourth moments. (An alternative approach is to look at the evolution of the percentiles over time – see here – but this also brings the same conclusion).
Figure 7. Cumulative density function of cross-section MoM (saar) core CPI items.

Table 2. Distributions – Summary of Indicators

Conclusion
The conclusion is straightforward: we are not at target, and the situation is deteriorating rather than improving. This is not a story about “just a few items.” Unfortunately, the underlying picture is considerably worse than that.