Over the 12 months ending May-23, the Consumer Price Index (CPI) increased 4.0%. The energy index decreased 11.7% and commodities rose by only 2.0%. Core CPI rose 5.3% from a year ago (Table 1) and is still above the Federal Reserve’s 2% target.
Other sections of the economy, however, are not slowing as fast as energy and commodities. The U.S. rental market is still running too hot. The BLS is trying to measure monthly aggregate personal expenditures on housing. The BLS measures housing costs using its “cost of shelter”.
The CPI cost of shelter is essentially the sum of two components: The first, is a measure of the rents paid by apartment tenants in multi-unit structures for their primary residences. This measure is called CPI rent (or tenants’ rent). The second is an estimate of the rent that owner-occupied housing could command called Owners’ Equivalent Rent (OER). These measures tend to move together as the OER of a specific owner-occupied unit is estimated in part by observed actual rents on similar types of properties. Owner equivalent rent, tenant’s rent and combined shelter represent 29.9 percent, 9.6 percent and a total 42 percent of core CPI, respectively. In Table 2, we see that shelter increased by 8.05% in May-23.
Going forward, this high growth rate (above the 2% desired by the Federal Reserve) is likely to continue. Charts 1 through 3 show the problem.
The solid line in Chart 1 shows data for OER_YOY. The far-right observation is the 8.05% for shelter for May-23 that we see in Table 2. The dotted line is the year-over-year percentage change in rents on three-bedroom properties (lnRRA3bd_YOY) for the 20 CBSAs tracked by the CHTR lagged 24 months. The data -- the average of rents on 3-bedroom properties for twenty CBSAs -- is a closer apples-to-apples match to OER than rent data on apartment units from Zillow.com. We notice that with a lag of 24 months the two series appear highly correlated (correlation=0.9) from the time of Mar-20 to May-23.
The survey process used by the BLS results in OER being reported with a 24-month lag vis-à-vis Altisource.com which collects the average of rents in each city each month. This is true because the BLS takes a massive, nationwide, rolling sample of housing units, splits them into panels, and then surveys each panel once every six months. The data then must be cleaned, checked and matched to the same property twelve months earlier. They then take an average rent and a one-month year-over-year change of that average. The surveying process essentially delays reporting changes in market conditions by about 12 to 24 months. If we recognize the roughly 24-month lead between lnRRA3bd and OER, we can use a projection of lnRRA3bd (p_lnRRA3bd), to see the problem facing the Federal Reserve in Chart 2.
OER is falling from its high of early 2023, but the dotted line (actual data already reported by Altisource.com of rents paid on three-bedroom properties pushed forward) shows that the declines only go so far. The chart pushes those observations of lnRRA3bd forward 24 months. If we would use only this variable (p_lnRRA3bd), OER to be reported in 2025-05 will likely range near 5.00%. This is higher than the Federal Reserve’s target of 2.0% (dotted red line). We, however, also need to consider rents of apartments in multi-unit structures.
Chart 3 shows the relationship between rent appreciation for apartments in multi-unit structures tracked by Zillow.com and the BLS’s tenants’ rent in CPI.
The lag between the two is much shorter (about 9 months), but the variation is much larger than the lag between OER and RRA3bd. The correlation between the two variables is 0.87.
This larger variation during the pandemic occurred because of the way that Zillow.com constructs its rent index. Zillow constructs repeat-rent indexes by tracking the prices for rentals on their platforms, treating a unit as rented for the last-listed price as it leaves the platform, and then comparing against previous rental prices for the same unit. It is thus capturing what is called the spot rent (this is the amount renters would pay to sign a new lease today i.e., just rents on newly signed leases).
Again, the government data also considers existing rentals, while Zillow.com only examines prices for new leases to capture current market conditions. Since rents typically change when leases expire, which tends to happen annually, this can lead to a lag in government data. The point of Chart 3 (and Chart 2) is that CPI Shelter is going to remain high (between 8.0% and 4.0%) in the second half of 2023.