(Important) Update: Michigan Inflation Expectations Jumped—What Now?

This is an update to our previous note: The Times They Are a-Changin’. In the final reading of the Michigan consumer survey, long-term inflation expectations jumped to 3.5%—the highest level since 1995. Following this update, the Fed’s Common Inflation Expectations (CIE) index is trending upward and has reached one of its highest levels in the series (excluding the COVID period).

Bottom line: At this point, it’s no longer an exaggeration to ask whether inflation expectations remain anchored. With core CPI running at 5% MoM annualized (January) and expectations climbing, the Fed now faces two choices: acknowledge the growing risk and take action, or downplay the issue and risk pouring gasoline on the fire. Given the current political climate, we’re not sure the first option is on the table.

For Q&As on the CIE, please see the end of this note.

The Updated Fed CIE Nowcast

Figure 1 presents our latest nowcast for the Fed CIE for Q1 2025. Today’s update is notable, as the model has revised both the level (up by 4 basis points) and the trajectory of recent readings. The CIE is now trending higher in our nowcast, with the Q1 reading ranking among the highest in the series’ history—excluding the COVID period.

Figure 1. Fed CIE and Underlying Inflation nowcast for 2025:Q1

Some Q&As replying to questions we generally get on the CIE

Does the Fed CIE get revised?

Yes, the Fed CIE is derived from a dynamic factor model (DFM), which is estimated using the Kalman filter. This statistical method reprocesses the entire historical series each time a new data point is added. As a result, our nowcast—built on the same DFM—also revises past estimates with every update.

Why is the published index different in history compared to Underlying Inflation nowcast?

There are two reasons for this. First, as mentioned earlier, our model functions as both a nowcast and a backcast, since the DFM revises historical data along with new observations. This means our model also serves as a preview of the index the Fed will eventually publish. Second, while the list of series included in the CIE is publicly available (see here) , certain details—such as the sample used—are not. As a result, our estimates can come very close to the final index but will never align perfectly.

The Fed CIE has a very low variance. How come?

This is intentional by design. The DFM estimates a common trend across the series, which is then “projected” onto a selected reference series. In this case, the Fed has anchored the CIE to have the same mean and variance as the SPF 10-year inflation expectations—a particularly stable series. Had a different reference series been chosen, the CIE would exhibit greater volatility. In other words, the mean and variance of the CIE are not inherently meaningful (a common characteristic of any principal component analysis).

If the CIE is at 2%, does this mean that expectations are consistent with a 2% reading in actual inflation?

No, not at all. As mentioned earlier, the mean of the CIE is meaningless. In fact, the CIE could have a mean of 200% instead of 2%, and it would still convey the exact same information.

If so, how should one read the CIE?

The CIE is designed to convey two key insights. First, its trajectory indicates whether inflation expectations remain anchored. Simply put, the fact that the CIE is now trending higher is concerning, even if the increase appears “small” (as discussed earlier regarding the series’ variance). Second, while the absolute level of the CIE should be interpreted in the context of historical trends, it becomes particularly meaningful when compared to past inflation dynamics. The fact that the index is now at a historical peak (excluding COVID) and that some components are aligning with early-to-mid 1990s levels—when inflation was well above 2% and expectations were unanchored—is a significant concern, especially given the prevailing upside risks.

Is the Fed CIE important for the Fed staff?

(Riccardo here) The CIE was created during my time at the Board in response to the frequent question: Are inflation expectations anchored? To address this, we developed a synthetic index to track expectations more systematically. I have always found—and still find—the index useful for summarizing movements across multiple series.

For the record, there is no clear consensus in the literature on which expectations are most relevant for actual inflation. However, expectations directly factor into the estimation of pi* (see Rudd (2020)). If this trend continues, it is unlikely that π* will remain at 2.4% in PCE terms. We will update our π* estimate early next week.

Conclusion

After today’s reading, it should be clear why we believe the FOMC should be concerned about both inflation expectations and actual inflation. The real question is whether, given the current political landscape, the Fed has the ability to take aggressive action once again.

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.