美国经济：CPI篮子中的苹果与橙子：为何市场租金指标在住房方面具有误导性

Apples and oranges in the CPIbasket: Why market rent gauges SIGNATURE Jonathan Millar+1 212 526 4876jonathan.millar@barclays.comBCI, US CPI rents have been gradually slowing for some time now,and pricing anticipates more of the same given market rentindicators. While many stress downside risks, we think theupside is underappreciated, given indications that rents for Pooja Sriram+1 212 526 0713pooja.sriram@barclays.com Marc Giannoni+1 212 526 9373marc.giannoni@barclays.com •Inflation markets are pricing the return of CPI inflation to target-consistent levels overthe course of 2026. A big part of the rationale is an expectation that the gradualdeceleration of CPI rents over the past few years will continue.FOMC dove Stephen Miranis an extreme example of this view, projecting that CPI shelter will decelerate from its current Colin Johanson+1 212 526 8536colin.johanson@barclays.comBCI, US •This optimism mostly reflects the market rent indicators, which, as a whole, havedecelerated below their pre-pandemic run rates following a breathtaking surge duringthe pandemic.Miran's argument implies that it is simply a matter of time until the CPI shelter We think the logic that CPI rents will slow to a pace resembling these measures is deeply Level versus growth rate:This logic is seemingly based on the expectation that thegrowth °rateof CPI rents will eventually converge toward market measures, entirely overlooking thediscrepancy inlevelterms. We show that market rent measures surged more than the CPImeasure during the pandemic, opening up a large, positive gap that has been closing only °New leases versus renewals:More importantly, the logic is based on an apples-to-oranges comparison, as market measures primarily capture new leases, which account for only asmall share (~10%) of the rental market. Instead, the CPI sample, which is meant to berepresentative of the entire housing stock, is largely informed by dynamics of renewalleases, which tend to be stickier, and are currently running higher than new-lease rent Although market rent measures can be a useful benchmark for CPI rents when thehousing market is on a stable equilibrium trajectory, they can be misleading whenmarkets are hit by large shocks.In our view, this was the case during the pandemic, when asharp – and, partly temporary – strengthening in housing demand caused rents for such units While CPI rents are expected to decelerate further, the inherent stickiness of continuingrenters would significantly limit disinflationary pressure.We show that error-correctionmodels in which CPI shelter inflation adjusts to close the gaps between CPI rents and market Our baseline forecast expects CPI rent and OER inflation to slow to 2.6-2.8% y/y inend-2026 (from an estimated 3.2%-3.5% y/y in December 2025), and further to2.3%-2.5% y/y in end-2027.While this trajectory aligns with the base-case scenario resultsfrom our error-correction model, we view risks skewed to the upside, mainly because we’reskeptical market rent inflation will continue to run at its current low pace throughout the Afterthree years, the storm is still closing in More than three years ago, in late 2022, Chair Powell gave the speech that directed markets tofocus on so-called “supercore” measures of core CPI that exclude goods and housing rents, onthe grounds that CPI shelter inflation would “begin falling sometime in 2023" given the sharpdeceleration signaled by various measures of market rents, "...the market rate on new leases is atimelier indicator of where overall housing inflation will go over the next year or so." In other Rates markets generally bought into Powell's logic. In retrospect, his argument seems overly optimistic. Although CPI shelter inflation has clearlydecelerated from 2023 peaks, growth rates still well exceed those for market rents more thanthree yearsafterPowell's prediction. Primary rents, which peaked at about 9% y/y in 2022, wereabout 3.4% y/y in September 2025, prior to distortions related to the absence of data collectionduring the government shutdown. OER shows a similar pattern, having gradually decelerated the absence of October data.1By comparison, indicators of market rents — such as the Zillow (2.6% y/y in Q3 2025) and REIS (1.2% y/y) measures shown in Figure 2 — have continuedto run at very slow rates for some time now. It remains to be seen whether CPI rent inflation approaches these market benchmarks. To besure, sequential rates for primary rents (0.2% m/m) and OER (0.13% m/m) plunged inSeptember 2025 to rates not seen since the onset of the pandemic. Taken at face value, theaverage monthly pace of increase in these two components from September-November 2025—0.06% and 0.13%, respectively—seems similarly slow if one does not adjust for the BLS's But is this reasonable, or are markets misreading the situation? Given the ongoing divergencebetween CPI rents and market indicators, we think it is reasonable to question whether the lagsat play are m