Rent Slowdown – A Discussion of What Research and Our Models Say

In March both CPI rent of primary residence and the owner equivalent rent (OER) slowed down on a MoM basis. Because rents are crucial for the inflation outlook, we re-assess the forecast and we review the recent literature. Overall, despite the positive news in the March CPI, our models continue to suggest that disinflation will be gradual and that risks around the Fed staff and SEP forecast remain to the upside.

(At the end of this note, we review a new paper by Michael Kiley released yesterday by the Fed Board)

Recent research

A new paper by Boston Fed economists Christopher D. Cotton and John O’Shea constructs an error-correction model using data at the metropolitan statistical area level to forecast how CPI shelter will evolve, based on the evolution of market-based rents. We have not reviewed the paper in details because, unfortunately, the sample ends in September 2022. Nevertheless, the paper is interesting because it is a careful attempt to tackle the issue at the micro level. The paper has some limitations (for instance, the dataset covers only 16 MSAs and the estimated coefficients are not stable across specifications) but the result is convincing: CPI rents are persistent even when market rents fluctuate sharply. Indeed, the authors argue that CPI rents are set to disinflate but the process is expected to take time and CPI rents can remain above their pre-Covid mean even at the end of the medium-term. In the paragraphs below we show that our research comes to similar conclusions, although with a macro approach.

Our own research

Research suggests CPI Rents will remain elevated for quite some time. The YoY of CPI rents and CPI OER is expected to decelerate, but can remain above its pre-Covid mean this year and the next. This is the third time we write about CPI rents. In our first note (here) in February 2022 we argued that the Fed (staff) was putting its reputation at risk because we flagged huge upside risks around its forecast due to accelerating rents. The second time, in December 2022 (here), we argued that some moderation was to be expected and that the YoY of CPI rents could peak in the second quarter of this year. Admittedly, CPI rents (and OER) in the first 3 months of 2023  -on net- came in a bit above our expectations at the time of the second note but overall, we think that we got the big picture right. Now that we are around the peak, where are we going from here? How fast will rents disinflate?

Figure 1 and 2 show the updated forecast of our models for CPI rents and CPI OER. Specifically, Figure 1 shows history and the evolution of the QoQ (ar) forecast for CPI rents (left panel) and CPI OER (right panel). Figure 2 shows the history and the latest YoY forecast together with the forecast errors (the yellow bands). According to our models, CPI rents and CPI OER are expected to growth 7.6% (ar) and 7.3% (ar) in the second quarter of the year, before decelerating and ending the year at around 5-6% (ar). The quarterly figures in Q2 are consistent with an average MoM of about 55bps for both, rents and OER. As for the evolution of the forecast in recent quarters, the model for CPI rents (left panel in Figure 1) has performed very well in the last 2/3 quarters, while the model for OER has been upwardly surprised in 2023:Q1.

Figure 1. History and model-based forecast of CPI rents and CPI OER.

CPI rents (QoQ, ar).

CPI OER (QoQ, ar).

Note: the figures show history and the evolution of the model-based forecast. Figures are QoQ growth at annual rate. “Two quarters ago” refers to forecast of the models stopping the sample in 2022:Q3. “One quarter ago” refers to the forecast of the models stopping the sample in 2022:Q4. Finally, “current” refers to the latest forecast (with 2023:Q1 in-sample).

Figure 2. History and current model-based forecast of CPI rents (YoY) and CPI OER (YoY).

CPI rents (YoY).

CPI OER (YoY).

Note: the figures show history and the evolution of the model-based forecast. Figures are YoY growth rates. The yellow bands are quasi out-of-sample forecast errors (that is, how much on average the model has missed at a given horizon in the past).

Some Q&A

Can the models be downwardly surprised going forward?

Yes, as usual a model will miss something going forward, as it did in the past. If the model is correctly specified, it should miss both ways, with the errors begin zero-mean and non-serially correlated. Given that the OER model missed it on the upside in Q1, the risks around the forecast are probably skewed to the downside in the next few quarters. Having said so, the reader should notice that -on average- the models performed remarkably well out of sample, as indicated by the narrow yellow bands in Figure 2.

Why CPI rents and OER are not expected to decelerate sharply?

First of all, rents and OER are expected to decelerate quite notably from here. However, the dynamics of CPI rents/OER depend on N factors, including its own persistency and the BLS sampling methods. A simple way of putting it is the following: it took 7-8 quarters for CPI rents/OER to accelerate from about 2% (ar) to about 9% (ar). Therefore, when asked about future developments, a model tends to think that it will take (at least) about the same time to disinflate. A forecaster can judgmentally adjust the forecast and assume a faster disinflation but given that some markets rents are still running above their pre-Covid levels ultimately, we agree with the models in this moment. In other words, given what happened, we would like to see the models being downwardly surprised for 1 (or 2) quarter before re-setting the medium-term forecast.

Conclusion and implications for the Fed staff

In our view, monetary policy can turn dovish (or less hawkish) only when core CPI will print 0.2% MoM. If our analysis is correct, this is unlikely in the next 4-6 months. A necessary condition for core CPI to print 0.2% MoM is that CPI rents and OER slowdown to 0.3% MoM (the pre-Covid average QoQ ar was about 3.5% for both rents and OER). It is a necessary condition because in a high inflation environment it is unrealistic to assume large deflationary pressure on goods, as some have been wrongly assuming for two years. If we put everything together, we are left with a core CPI MoM forecast that can remain above 0.2% for another several months. For this reason, unless the US will experience a sharp recession, it is very hard to see the Fed cutting rates in 2023. Indeed, the baseline continues to be the one we have been working since 2021: the risks around the Fed staff and SEP forecasts are skewed to the upside.

Post-Note

The Fed Board has released a new paper by Senior Adviser Michael Kiley. The paper (link) is a statistical exercise that estimates, using an unobserved component model, the role of nominal wages in trend inflation. Without entering into the details, the paper shows that wages do help explaining (and forecasting) inflation, although only during certain periods: the ‘70s, the early ‘80s… and right now. (for the record, Kiley (2023) estimates trend PCE inflation at 4.5% at the end of 2022). While Kiley (2023) model cannot identify causality, his results represent a possibly large upside risk around the Fed staff and FOMC medium-term forecast.

Comment and (possible) implication for the Fed staff. Chicago Fed President Austan Goolsbee recently pushed back on Twitter (yes, on Twitter) against the idea that wages are a useful indicator for future inflation.  Goolsbee’s tweet is based on a recent Chicago Fed Letter (here) by Gadi Barlevy and Luojia Hu in which the authors do not find a strong relation between current unit labor costs and future inflation. Kiley (2023) is interesting because it shows that, once the model allows the contribution to vary over time (that is, contrary to Barlevy and Hu (2023)), wages can have a role in inflation dynamics. In our experience, before the pandemic the Fed staff considered the passthrough from nominal wages to consumer inflation to be very small, if any (see Peneva and Rudd (2017)). Nevertheless, in the May 2022 FOMC minutes (here) the Fed staff  raised the forecast based on higher wage growth, admitting that the passthrough is probably not zero right now. All told, we suspect there is an ongoing discussion at the Board about the magnitude and timing of the passthrough. Given the situation, dismissing the upside risks from wage growth would certainly not be wise.

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.