US: September 2026 CPI

The CPI report came in broadly in line with our expectations for headline inflation, but stronger than expected for core inflation. This is the second consecutive month in which this has happened, and at this point we need to take the signal seriously. We have not yet revised our forecasts, but we will do so between today and tomorrow and circulate an update. (Please, note that our ML forecast model has been delivering very accurate forecasts for the past 6-12 months – see dashboard online)

At this stage, we are increasingly reluctant to focus on individual CPI components. Looking at the distribution, it is clear that inflation remains inconsistent with the Fed’s 2% target, with price changes still showing considerable dispersion. One month we can discuss a particular item that unexpectedly jumps, and the following month it will be another. We have been going through this exercise for more than five years, and by now we have seen virtually every CPI component discussed by the sell side. The underlying problem remains the same. Inflation persistence continues to be a significant issue.

Our translation into PCE space is particularly strong this month. We will reassess it tomorrow following the PPI release and subsequently incorporate the import price data.

Regarding the Fed, today’s CPI report does not invalidate the conclusions of our previous FRB US simulations. It simply shifts the implied policy path marginally higher, while leaving the broader picture essentially unchanged. According to the model, if the Fed wanted to ensure that inflation returned to target by 2029, the federal funds rate would need to reach approximately 4.85%. If economic growth remained stronger or inflation proved somewhat more persistent, the implication would primarily be higher for longer, rather than a higher terminal rate. In other words, the Fed could maintain a restrictive policy stance for longer, with the terminal rate remaining broadly unchanged or potentially even slightly lower. The picture would change dramatically, however, if the Fed signaled that it intended to accelerate the disinflation process and bring inflation back to target by the end of 2027. Under that scenario, our simulations suggest that the federal funds rate would need to rise well above 6%.

In any case, if our translation is correct and core PCE inflation is indeed running at around 30bp this month, we would be very surprised if the Fed did not raise interest rates in October. Failing to do so would risk damaging its credibility and bringing the debasement trade back into focus.

NOTE: All models and charts are on our website: underlyinginflation.com. Simply log-in.

REACTION TO THE INCOMING DATA

We have not reacted to the incoming data, we will soon. We currently expect core CPI to print 0.15% MoM in October (YoY 2.3%), and headline CPI to print 0.47% MoM (YoY 3.5%). These numbers are likely too low and we will revise them higher soon (likely, the YoY of both headline and core CPI will be 1-2 tenths higher in December after revision).

For the record, our Q4/Q4 forecast for core PCE is now at 3.2%. Likely, it will increase to 3.3%-3.4% after folding-in our reaction.

CPI/PCE Translation:

Please, find below our translation of today’s data in PCE space.

EVIDENCE FROM THE MODELS

The models continue to point to a common inflation component above target, a distribution of price changes that remains inconsistent with inflation at target, and a medium-term forecast that takes time to converge to the Federal Reserve’s objective.

DISTRIBUTIONS

Last month, the distribution showed some improvement, but that progress has reversed entirely this month. Overall, the shape of the distribution remains inconsistent with the 2% target. There are NO signs of improvement in recent months.

Figure 1a. Cumulative density function (and table stats) of core CPI cross-section price changes.

Figure 1b. Kernel of core CPI cross-section price changes.

Figure 1c. Median of core CPI cross-section price changes.

CI MODEL

Our Common Inflation (CI) model estimates that, once idiosyncratic shocks are stripped out, the underlying common component remains firm. This month, the model estimates that the common component increased by 26 bps—above the average of recent months—while the idiosyncratic component contributed a 7 bps. The signal we take from today’s release is the following: the common component of inflation remains strong (with trend inflation a bit below 3% – see our models on our website for more details).

Figure 2a. Contributions to MoM changes of CPI excluding food and energy items (CI model)

Figure 2b. Estimated “Common” component: YoY, 3m/3m a.r. and 6m/6m a.r.

IMPLICATIONS FOR MEDIUM-TERM MODEL-BASED PCE FORECAST

The medium-term outlook is a bit stronger than the previous run.

Figure 3a. “Main” Phillips curve model forecast, headline PCE price inflation (YoY, %).

Figure 3b. “Main” Phillips curve model forecast, core PCE price inflation (YoY, %).

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.