Top
Reference

US CPI OER Background

This page is about the “Owner’s equivalent rent” (OER) component of the US Consumer Price Index (CPI).

Summary

4 Jan 2010

The OER component of CPI is an estimate of the inflation an owner would see in rent, if s/he were renting.

Cost of shelter is a major component of CPI, and it needs to be measured for homeowners as well as for renters. House prices are partly speculative and do not provide a good measurement of the cost of shelter alone. Hence the BLS was motivated to define the OER as, “the amount a homeowner would pay to rent, or would earn from renting, his or her home in a competitive market,” and to estimate inflation of OER.

In short, the BLS estimates inflation in the OER using inflation in market rents of nearby rental units; nearby, because their research indicates that changes in cost of shelter depends primarily on location. In a little more detail:

  1. The only primary data from owners is a one-off best guess of rent, used only to determine the OER weighting among all categories of consumer expenditure.
  2. Actual rents are collected in a large number of small geographic areas.
  3. Rents are adjusted to remove utilities.
  4. Monthly movement in the OER index is based on ratios of weighted averages of rents; the weight for each small area is determined mainly by the the proportion of owners.

Highlights

(Clippings below covered through Oct 2007)

Goal and rationale (4 Jan 2010) The definition of OER is, “the amount a homeowner would pay to rent, or would earn from renting, his or her home in a competitive market.” Cost of shelter is an important component of the CPI, and the main goal is to measure inflation in the OER, i.e., inflation in what homeowners are paying for shelter. The rationale for not just using house prices is that houses are highly leveraged investments in a cyclical market, and house prices consequently confound investment and cost-of-shelter.

Basic assumptions (10 Sep 2008) The basic requirement is that there are houses in the renter sample “which are similar in their most important aspects to those that are owner-occupied.” Here ‘similar’ just means that the change in the cost of shelter for a homeowner is well approximated by the change in the rent of a comparable property. BLS research suggests that by far the most important determinant of the required similarity is nearby location.

Methodology (10 Sep 2008) In a word, the BLS estimates inflation in the homeowner rents using inflation in market rents of nearby rental units. In just a little more detail, (1) Primary data from owners is only a one-off best guess of rent, used only to determine OER as a fraction of total consumer expenditure. (2) Actual rents are collected in a large number of small geographic areas. (3) Rents are adjusted to remove utilities. (4) Monthly movement in the OER index are based on ratios of weighted averages of rents; the weights are primarily geographically determined, where each small area is weighted by the number of owners.

Primary sources

See also

Clippings

Below are the clippings, from outside sources, that were used in constructing this page.

BLS white paper on OER

9 Dec 2005. Paper prepared for the Federal Economic Statistics Advisory Committee, from the BLS website.

http://www.bls.gov/bls/fesacp1120905.pdf

“Treatment of Owner-Occupied Housing in the CPI. Robert Poole, Frank Ptacek, Randal Verbrugge”

“A recent Organization for Economic Development and Cooperation (OECD) report shows that this approach is the one used by a plurality of countries (13 out of 31 countries studied), with the next chosen alternative simply leaving owner occupied housing out of the national CPI (9 countries). Only Australia and New Zealand use an acquisitions or house price approach. … The rental equivalence technique only required that there are enough houses in the renter sample “which are similar in their most important aspects to those that are owner-occupied.” The CPI sample of rental housing could serve that purpose. … the rental equivalence approach actually involves making a weaker assumption. Instead of assuming that the homeowner’s rental equivalent cost equals the market rent of a comparable property, what is rather assumed in the construction of the CPI shelter indices is that the change in the cost of shelter for a homeowner is equal to the change in the rent of a comparable property. … BLS research (including the most recent work, Verbrugge et al., 2005) has consistently come to the conclusion that, aside from location, it is difficult to find any convincing predictor of rent change, including initial rent level (which is presumably a fairly good proxy for quality). … based on actual market rents collected from a sample of renter-occupied housing units that are representative of owner-occupied housing. … Full technical documentation on the procedure was published in the CPI Detailed Report for January 1983. …

During the period 1983 through 1986, the rental equivalence measure was derived by reweighting the renter sample to represent owners. The BLS had not yet developed an owner sample needed to implement owner-renter matching, which was then thought to be the best approach to rental equivalence. The renter sample was augmented in heavily owner-occupied areas to better represent the owners. The renter weight was recalculated in augmented areas and the owner weights were derived for each area. Owner weights were derived by inflating or deflating the renter weights using the number of owners relative to the number of renters in the Enumeration District (ED). (An ED refers to the area assigned to a single Census-taker to count persons and prepare schedules within one Census period.) Since the EDs were selected to represent renters, the reweighted sample was inefficient. Economic rents were used for both Rent and OER. Economic rents are basically monthly cash rents paid to the landlord plus subsidies received by the landlord plus in kind services performed in lieu of cash that have been adjusted for changes in quality. … BLS subtracts from the Economic Rent an estimate of the embedded utility cost to yield Pure Rent. … [for a while an explicit owner-renter matching was maintained, but this was too difficult] … Homeowners are asked the often-cited question: If someone were to rent your home today, how much do you think it would rent for monthly, unfurnished and without utilities? This is the only place where the answers to this question is used; in determining the share of the market basket. … The initial sample was designed to produce approximately 50,000 renter-occupied housing units in about 10,000 distinct segments (small geographically-contiguous areas comprised of Census blocks) allocated across the 87 Primary Sampling Units (PSUs) included in the CPI. … Prior to sampling segments within the PSUs, each PSU was divided into 6 geographic strata, each strata representing approximately 1/6th of the total housing expenditure within the PSU. … Each segment is assigned two weights, one for Rent and one for OER, and each housing unit in the segment uses the same weights. Put simply, the Rent weight represents the relative importance of the sampled housing unit in the calculation of the Rent price relative and the Owner weight represents the relative importance of that same sampled housing unit in the calculation of the OER price relative.”

[The basic requirement is that there are houses in the renter sample “which are similar in their most important aspects to those that are owner-occupied.” Here 'similar' just means that the change in the cost of shelter for a homeowner is equal to the change in the rent of a comparable property. BLS research suggests that by far the most important determinant of the required similarity is nearby location. Hence rents are collected in a large number of small geographic areas, and each area is weighted by the number of renters for the rent index, and the number of owners for the OER index.]

[This paper also contains an extensive history of how the current methodology came about.]

BLS backgrounder on Owner's equivalent rent

9 Feb 2007. Factsheet on BLS website.

http://www.bls.gov/cpi/cpifact6.htm

“Consumer Price Indexes for Rent and Rental Equivalence”

“the expenditure weight in the CPI for rental equivalence is obtained by directly asking sampled owner households the following question: If someone were to rent your home today, how much do you think it would rent for monthly, unfurnished and without utilities? … The CPI Housing Survey is the source of the data on residential rents used to calculate changes in rents for the Rent of primary residence index. The Housing survey also uses these rent data in calculating changes in the rental value of owned homes for the Owners’ equivalent rent of primary residence index. …

Rent—Background: … The primary housing sample is a stratified cluster sample … Primary sample selection is done with every decennial census … Starting in 1978 … augmented the primary housing sample with housing units constructed since the most recent census. … The housing sample is divided into six subsamples called panels. Each panel is priced in consecutive order, so that every panel is priced twice a year. … Because each panel is representative of the entire sample and there is never an off-cycle month for the Housing survey, a panel of data provides sufficient information for monthly publication of the rent and rental equivalence indexes.

Rental Equivalence—Background: … Because the asset price method can lead to inappropriate results for goods that are purchased largely for investment reasons, the CPI implemented the rental equivalence approach to measuring price change for owner-occupied housing. It was implemented for the CPI-U in January 1983 … Rental equivalence measures the change in the implicit rent, which is the amount a homeowner would pay to rent, or would earn from renting, his or her home in a competitive market. … calculated by reweighting the rent sample to represent owner-occupied units. … Because the rent and rental equivalence indexes measure the change in the price of the shelter service provided by both renter-occupied and owner-occupied housing, segments were selected with probability proportional to estimated total expenditures for rent and implicit rent. … Each renter unit in the sample represents many renter units in the renter universe and many owners in the owner universe. The renter and owner costs of housing in the segment (see “Weighting during segment sample selection” above) are the basis of the renter and owner weights used by the price change calculation for the segment. … BLS uses pure rents and the owner weights to estimate the changes for the rental equivalence index.”

Additional details on the BLS weighting of rents in calculating OER

Oct 2007. Working paper 410 on the BLS website.

http://www.bls.gov/osmr/pdf/ec070090.pdf

“Explaining the Rent-OER Inflation Divergence, 1999-2006. Robert Poole, Randal Verbrugge”

“How is OER produced? The exact market rent of an owned home is, of course, unobservable. However, a well-known and empirically valid rule of thumb in real estate pricing is “location, location, location.” This principle carries over to rents: internal BLS research — most recently in Verbrugge et al. (2007) — has consistently supported the notion that, outside of location, it is difficult to find any reliable predictor of rent inflation. Hence, the BLS estimates inflation in the homeowner rents using inflation in market rents of nearby rental units. (Furthermore, about one-quarter of the total BLS sample of rental units consists of detached units, so — even though there is only weak evidence that rents differ by shelter type — much of the rent sample that is used for measuring homeowner shelter cost inflation consists of the same structure type.) … Monthly movements in the OER index, and the Rent index, are based upon ratios of weighted averages of rents. [the weight given to rental unit i for the index j, where j is either 'rent' or 'OER', is denoted wij] … Although both BLS indexes are based upon a single rent survey, the rent measures … for the OER index must be pure-shelter-service prices and hence must exclude utilities. … wiRent = 0 for some units, namely those in a special OER augmentation sample, which are termed “helper segments.” … wiOER = 0 for rent-controlled units. … wiRent [is not equal to] wiOER in general, since (for example) the OER index will place higher relative weight on units located near heavily-owner-occupied regions.”