An official website of the United States government
The New Tenant Rent Index and All Tenant Regressed Rent Index are research index series that use data sourced from data collected in the Consumer Price Index (CPI) Housing Survey. The New Tenant Rent Index (R-CPI-NTR) measures prices renters would face if they changed housing units every period. The rent component of the official CPI measures the change in all rents, including new leases, renewals, and rents in the middle of a lease. In contrast, the New Tenant Rent Index uses only a subset of the data the official CPI uses, namely the first survey observations after new tenants move into their sampled housing units. The All Tenant Regressed Rent Index (R-CPI-ATR) is a measure with a scope similar to the CPI, but using methodology similar to the New Tenant Rent Index. The All Tenant Regressed Rent Index measures the rent paid by all renters, both new and continuing, and incorporates most of the survey data used for the CPI Rent of primary residence index. The All Tenant Regressed Rent Index is published alongside the New Tenant Rent Index to facilitate comparisons.
The New Tenant Rent Index uses a regression method that uses pairs of observed rents for the same housing unit. The change in the natural logarithm of rent is regressed on period indicators, that are set to -1 for the starting period in a pair of rent observations, 1 for the latter period in the pair, and 0 otherwise. Observation pairs bookend the tenure of a renter within a housing unit: the first date records when a new renter moved in, and the second date when the next renter moved into the same housing unit. For example, suppose a tenant moves into an apartment in the third quarter of 2015, moves out of the apartment in first quarter of 2017, and someone else moves into the apartment in second quarter of 2017. In this example, the indicator for 2015 quarter 3 has a value of -1, the indicator for 2017 quarter 2 is 1, and the indicators for all other dates have a value of 0. Regression coefficients on the period indicators are converted to a chained index level. For each pair of rent observations, both observations are the first price after a new tenant moves into the unit. Pairing rents from the same unit mitigates bias coming from the composition of the sample changing. Tenure lengths for tenants average about three years but vary. This unbalanced panel is one reason the New Tenant Rent uses its regression method instead of a methodology more like the CPI's.
The All Tenant Regressed Rent Index uses a similar regression method that involves pairs of observed rents for the same housing unit. Unlike the New Tenant Rent Index, the All Tenant Regressed Rent Index does not restrict its data pool to only new tenant observations. For each observation pair in the All Tenant Regressed Rent Index, the second observation represents the rent paid by a tenant and the first observations represents the rent paid by a (possibly different) tenant in the same housing unit the last time it responded to the survey. Most observation pairs are thus separated by the six months, since the BLS Housing Survey surveys housing units every six months.
The New Tenant Rent Index and All Tenant Regressed Rent Index series are currently calculated as quarterly indexes, with observations from three months of the housing survey pooled together. The two series begin in 2005. Every period, the entire series is re-estimated. In the regression method used, new observation pairs influences the index over the entire time spanned by the pair. The most recent periods of a repeat transaction index like the New Tenant Rent Index are prone to large revisions, because the rent observations spanning those periods accumulate gradually as tenants move. For example, if a tenant moves into a housing unit in December 2023 and the subsequent tenant moves into the housing unit in December 2024, then the value for the New Tenant Rent Index in 2023Q4 will be revised when the housing unit is next surveyed after December 2024. In contrast, the non-seasonally adjusted indexes for the CPI are seldom revised. Sample sizes for the New Tenant Rent Index are much smaller than for the All Tenant Regressed Rent Index or the official CPI; this is especially the case in the first and fourth quarters when fewer moves happen.
The New Tenant Rent Index and All Tenant Regressed Rent Index use the "economic rent" calculated for the official CPI-U Rent of Primary Residence Index. Economic rent adjusts contract rents to account for changes in utilities bundled with rent, for work credits, and for the aging of units (see Measuring Price Change in the CPI: Rent and Rental Equivalence for details). The New Tenant Rent Index and All Tenant Regressed Rent Index adds a timing adjustment and omits units known to have been remodeled. The New Tenant Rent Index and All Tenant Regressed Rent Index also drop observations in the top and bottom 1 percent of annualized rent changes. Estimated confidence intervals for the New Tenant Rent Index and All Tenant Regressed Rent Index are calculated using a bootstrap method.
The New Tenant Rent Index and All Tenant Regressed Rent Index were adapted from the "New Tenant Repeat Rent Index" and "All Tenant Repeat Rent Index" of the research article "Disentangling Rent Index Differences: Data, Methods, and Scope". The article further details the indexes' construction and uses the index to compare the CPI rent component to other rent indexes. The New Tenant Rent Index and All Tenant Regressed Rent Index are currently calculated as prototype research indexes. Their methodology may change as they continue to develop.
Topic | CPI-U Rent of Primary Residence Index | All Tenant Regressed Rent Index | New Tenant Rent Index |
---|---|---|---|
Estimation method | Uses a weighted average of the sixth root of the changes in the economic rent | Uses the repeat rent methodology as described in Case-Shiller (1989) | Uses the repeat rent methodology as described in Case-Shiller (1989) |
Coverage Scope | Incorporates new and continuing tenants | Incorporates new and continuing tenants | Includes new tenants only |
Observation Dates | Assigns observations to date survey was conducted | Assigns observations to the date of move-in or the closest six month multiple after the move-in | Assigns observations to date of move-in |
Revisions | Never revised (unless correction policy invoked) | Perpetually revised, with recent periods being prone to large revisions | Perpetually revised, with recent periods being prone to large revisions |
Frequency | Monthly | Quarterly | Quarterly |
Nonresponse | Imputes rents to account for nonresponses | Excludes nonresponse observations | Excludes nonresponse observations |
Vacancy | Imputes rents to correct for vacancy bias | If a housing unit is no longer sampled, drops all observations after the last time a new tenant moved in | N/A |
Outliers | Caps extreme monthly rent changes at 2000% or -95% | Excludes top and bottom 1% of rent changes for each geographic area | Excludes top and bottom 1% of rent changes for each geographic area |
Quality change | Adjusts rents for changes in the certain unit characteristics | Excludes any housing units with field messages indicating remodeling or renovations | Excludes any housing units with field messages indicating remodeling or renovations |
Data: