US: How “crazy” was the CPI report?

Given the number of questions we have received regarding the December distribution of core CPI items, this note is intended to provide a brief explanation of what occurred in the data and to outline the implications for monetary policy and for our forecast.

Dispersion

The starting point is that, along several dimensions, the December report appears to be genuinely anomalous, if not outright unique. We already highlighted this point yesterday when looking at the full distribution. To be more precise, however, Figure 1 shows that the interquartile range (IQR, the difference between the 75th and the 25th percentile). In December 2025, the IQR is the highest observed over the past 35 years and, in fact, the second highest on record when extending the sample back to the 1970s. In this sense, what occurred in December was truly unprecedented.

At the same time, as some of you have correctly noted, the variance of the cross section distribution was not particularly elevated. As we discuss in more detail below, however, this does not imply that the distribution was less dispersed. On the contrary, the much higher IQR indicates that the distribution was significantly more dispersed than usual. Inflation has therefore become more volatile and less predictable, which in turn makes the process of assessing and managing its path back toward target more complex going forward.

Figure 1. Interquintile range (75th-25th pct)

Note: the Figure shows the difference between the 75th and the 25th percentile of the cross section distribution of core CPI items (MoM saar)

Variance vs IQR

From a theoretical standpoint—and this is essentially what happened in December, in a way that is not just unusual but genuinely rare in inflation space—it is entirely possible for two distributions to share the same variance while having very different interquartile ranges. To illustrate this point, we constructed a simple simulation in MATLAB, shown below. The blue and orange distributions plotted there are built to have exactly the same variance by construction. However, the orange distribution has a materially larger interquartile range (the difference between the 75th and the 25th percentiles) than the blue one.

If the variance is unchanged while the interquartile range increases, the distribution becomes more “boxy” in shape—more squared-off—relative to a distribution that looks closer to a Normal. In this case, the mass in the tails remains broadly unchanged, while a much larger share of the probability mass shifts into the shoulders of the distribution. As a result, the distribution is more dispersed overall than the blue one, but not because of heavier tails—rather because of greater dispersion in the central part of the distribution (the “shoulders”). Indeed, the variance captures movements in the tails, with the IQR caputres movements in the shoulders of the distribution.

Varianve vs IQR: which one is more problematic for a central bank?

An increase in the interquintile range (IQR) of price-change distributions is in theory more problematic for a central bank than a rise in overall variance because it signals broader-based pressures rather than volatility driven by the tails. The IQR captures the dispersion of price changes across the middle of the distribution, where the bulk of consumption lies. When it widens, it indicates that a large share of prices are moving at increasingly different rates, making inflation less “sector-specific” and more volatile. This undermines the usefulness of relative-price adjustment narratives and suggests that inflation is becoming harder to isolate and contain through targeted or temporary explanations.

Conclusion

As noted yesterday, the December CPI print came in lower than expected, but in a sense by chance, given the specific weighting of individual components within the CPI basket. Indeed, when the same report is translated into PCE space, it implies a materially higher print—roughly 40 basis points. For this reason, we do not take particular comfort from the December CPI release at this stage, while we await today’s Supreme Court decision.

More broadly, the key question we have been grappling with for some time is no longer whether disinflationary pressures are present—that has been evident for a while—but rather whether inflation can realistically return to target, and over what horizon. If distributions continue to shift so markedly from one release to the next, and remain as dispersed as they currently are, the task becomes particularly challenging—not only for forecasters, but also for central banks. It is inherently difficult to reach stable conclusions when the underlying distributions themselves are unstable, implying a need for either more time or substantially stronger evidence.

At present, the data suggest that we are operating in an inflation environment that is significantly more volatile than in the pre-COVID period. As such, it remains an open question whether inflation can return to target, or whether instead it will settle moderately above target—potentially in a regime where core CPI averages between 2.5 (3?) percent going forward.

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.