The Rent Is Still
Too Darn High:
Housing-Related Inflation Misconceptions
Housing, particularly the “shelter” component of the Consumer Price Index, remains a central cause of U.S. inflation in 2023. The consensus view is that housing will soon drag down overall inflation. However, misconceptions about housing and inflation may lead investors astray. We write to provide some clarity.
“The rent is too darn high.”
This phrase, famously uttered by a New York State gubernatorial candidate in 2010 to describe a critical campaign plank, is still as true in 2023 as it was back then.1 “Shelter” accounts for a third of the U.S. Consumer Price Index (CPI). Within the shelter component, “rents” account for around 7.6% of overall CPI, while a less understood category, “owners’ equivalent rent” (OER), accounts for 25.6%.2 Overall, the shelter index was up 7.2% in the 12 months ending in August 2023, accounting for a large share of the overall increase in inflation. Yes, the rent is too darn high.
But despite the shelter inflation surge (or perhaps because of it), investors' overwhelming consensus is that overall inflation will soon recede, led by a slowdown in housing-driven inflation.
As for the rationale, most investors point to private sector rent surveys, which show rental prices decelerating, and conclude that government statistics will soon reflect the trend. (“It’s baked in already!” friends and colleagues assert.) Further, many colleagues disapprove of how “shelter” is accounted for in the CPI, whether how rent is calculated or how homeowners’ costs are covered via the OER component.
We hope our colleagues are correct about overall inflation trends, but the misconceptions we see on the subject give us pause. Below, we write to address and correct several shelter-related inflation misconceptions.
We write to address and correct several shelter-related inflation misconceptions.
First, and perhaps most importantly, CPI is a “cost-of-living” measurement exercise. The CPI rent component is not meant to be an up-to-the-minute gauge of price trends in the metro area apartment rental markets. The CPI aims to estimate the average U.S. household’s cost of a basket of goods and services consumed—including the shelter costs experienced by all occupants.
The “all occupants” portion is vital. Most, if not all, private sector rent metrics calculate rental rates in more expensive housing for a new or marginal tenant. However, the average renter did not experience as rapid an increase in rents in 2021 as many private sector surveys showed (consistent with the government figures!). Conversely, many renters may not see the same rate of deceleration in rents in 2023 as private metrics now suggest (see Figure 1). As a result, inflation may be stickier than expected in 2023 and 2024.
Further, we’ve heard cries that “the CPI methodology doesn’t make any sense!” or “the way the CPI measures housing is not updated frequently enough!” The Bureau of Labor Statistics (BLS) indeed queries cohorts of renters every six months to create the rent figures. They do this because most rental agreements are for six or 12 months, and government statisticians want to account for rental turnover and renewed leases.3 Your rent doesn’t change daily, so we don’t need daily updates on rent prices for the average person.
You may rejoice at the private sector metrics if you, dear reader, are trying to sign a new lease. But if you are someone forecasting CPI, focusing on the new or marginal renter may mislead you.
At this point, you might say, “Okay, I understand the marginal versus average mistake, but on a long enough time horizon, as enough units turn over, the marginal will become the average!”
Not so fast.
There is an additional wrinkle to the marginal new tenant approach taken by many private sector surveys. If the fraction of households that move varies significantly over time or is highly concentrated, a focus on the marginal renter would capture price changes for an ever-changing portion of the market (i.e., a specific group of people prone to moving), not the overall average rental cost in the marketplace. Consider this: While Bay Area rents have slowed along with layoffs in the tech sector, other regions have seen rent growth, even on the private sector rent surveys (see Figure 2).
Further, as mentioned above, the CPI measures rent growth for all occupants, while the alternative measures track rent growth for new tenants. However, the differences go further. For example, only the CPI survey adjusts for “aging, structural changes, and changes in utilities provision” (to capture when landlords pay for various utility costs, e.g., water).
Additionally, the BLS Housing Survey, which underlies the CPI rent component, is a random sample that “is fully representative of the rental housing stock in U.S. cities.” By contrast, two of the most popular private metrics, the Zillow Observed Rent Index (ZORI) and the CoreLogic Single-Family Rent Index (SFRI), use samples of mainly higher-tier detached rental units that advertise in the Multiple Listing Service (MLS). According to the Census Bureau’s 2018 Rental Housing Finance Survey, only 11% of single-unit rental properties are listed using a real estate agent (and thus listed in the MLS). Meanwhile, the Ambrose, Coulson, and Yoshida (ACY) Marginal Rent Index (MRI), another commonly used metric, covers larger apartment complexes in a few cities.4
Again, from a business perspective, these alternative data purveyors may want to measure a different sample of renters. Still, the BLS approach is a more encompassing way to gauge average household costs. More importantly, one sample (e.g., the private sector metrics) may not “lead” the government data as many investors expect; they are, in fact, apples and oranges.
On a related note, while rents grab the lion’s share of media attention, they make up just 7% of CPI. The primary subcomponent of shelter, “owners’ equivalent rent,” or OER as it is commonly referred to, makes up 25%. Some friends and colleagues worry that such an arcane concept (“Owners’ equivalent what?” they ask) obscures actual housing costs. The few colleagues who have looked into the matter ask, “How would the BLS know what I can rent my house for?”
Well, first off, most households in the U.S. own rather than rent their houses (last we checked, roughly two-thirds of Americans own their homes), which is why owners’ expenses are considered a much bigger weight in CPI. Remember, CPI measures inflation for the average U.S. household.
Second, as we highlighted above, the BLS attempts to calculate the cost of housing services—that is, the cost of the shelter services provided. Even if you are lucky enough to own your home, you still experience housing services costs. But how should statisticians measure it?
Since the BLS is trying to estimate the cost burden of housing services, using house prices would not solve the problem. For example, if a family bought an $800,000 house in Los Angeles and lived there for a decade, the cost is NOT $800,000 (in year one or amortized over ten years). Instead, the housing services cost is what the family expects to spend to consume a similar amount of housing services elsewhere.
As a solution, the BLS uses rents on comparable housing in the same region to estimate the cost (“imputed rent” or “owners’ equivalent rent”). Difficulties arise because it may be impossible to find a comparable rental home like yours in your area, especially if it’s in a large, gated community, of which there are few examples.
Admittedly, the methodology is not universally accepted: “11 of 17 Organisation for Economic Co-operation and Development (OECD) countries do not include owner-occupied housing in their CPIs at all”!5
Moreover, it is worth noting that the BLS methodology has evolved and will continue to do so. Before 1983, the BLS measured the services of owner-occupied housing by using the current mortgage rate offered on new mortgages! The index assumed that the mortgage interest home buyers agreed to pay over future years was part of their cost of living today. We’re not sure inflation-wary investors would be happy with the results had the BLS stuck with the methodology through 2023 (hint: inflation would still be very high)!
There are other alternatives to the “imputed rent” approach being considered that could be included in future versions of the CPI. For example, the Canadian CPI tracks actual monthly mortgage payments to measure housing services costs.6
However, rather than being an obscure methodology, the imputed rent approach should also be extended to other products and services. For example, for items such as automobiles or houses, or yes, even mobile phones, a lease or rental rate for autos and telephones may provide a better average cost than the one-time purchase price of such products.7 Think of it this way: What is the true inflation burden of your mobile phone, the one-time upfront cost or the ongoing, monthly services cost of its use? We argue it’s the latter, suggesting that it does make sense to rely on estimates of service costs after all.
One final point is worth considering: Forecasting inflation based on a bottom-up approach has an inherent danger. While it’s tempting to find a commodity or service and then extrapolate that to an overall price trend, one will find many moving parts once one delves into rent and inflation. At the very least, this complexity should make investors wary of grand predictions in either direction.
For example, a tight labor market with robust wage growth overwhelmingly determines rents (See Figure 3). To the extent the labor market remains tight, one would expect that demand for rental housing will remain strong, especially with home-buying affordability limited.
Our job is to challenge the consensus views driving markets. Above, we presented some reasons to be cautious when basing your inflation outlook on the “it’s baked in” theory of disinflation due to the shelter components in CPI. Karl Popper once wrote, “Whenever a theory appears to you as the only possible one, take this as a sign that you have neither understood the theory nor the problem which it was intended to solve.”
That the rent may still be too darn high well into 2024 is a theory worth considering.
Endnotes
1. Rent is Too Damn High. (2023, September). In Wikipedia. https://en.m.wikipedia.org/wiki/Rent_Is_Too_Damn_High_Party 2. Shares are from the BLS: https://www.bls.gov/news.release/cpi.t01.htm. Also, while the article covers CPI, shelter is a relatively smaller component at 24% of the Fed’s preferred gauge, the personal consumption expenditures (PCE). PCE covers a wider array of spending, hence the smaller shelter share.3. National Academies of Sciences, Engineering, and Medicine. 2022. Modernizing the Consumer Price Index for the 21st Century. Washington, DC: The National Academies Press. https://doi.org/10.17226/26485.4. “Disentangling Rent Index Differences: Data, Methods, and Scope∗.” Brian Adams , Lara Loewenstein , Hugh Montag , and Randal Verbrugge. US Bureau of Labor Statistics, Federal Reserve Bank of Cleveland, December 5, 20225. National Academies of Sciences, Engineering, and Medicine. 2022. Modernizing the Consumer Price Index for the 21st Century. Washington, DC: The National Academies Press. https://doi.org/10.17226/26485.6. ibid7. ibid
