Repeat Sales Index Renovation Bias

A repeat sales index tracks prices for the same properties, but those properties can change between sales. Renovation bias arises when value added by improvements is interpreted as market appreciation. Matching an address does not establish unchanged housing quality; depreciation can also complicate the comparison, as FHFA's 2007 research explains.

What does a repeat-sales index hold constant?

A repeat-sales index compares successive prices for the same properties, then combines those observations to estimate market price changes. Matching each home to its own earlier sale controls for characteristics that remain fixed, rather than comparing it with a different house with more bedrooms or a larger lot.

This reduces the effect of a changing sales mix. It does not establish that each property's size or condition remained unchanged. FHFA's March 1996 technical description explains how repeated observations help control for differences across homes. “Constant quality” describes the statistical objective.

That distinction matters when comparing home-price measures: matching the property and measuring changes to the property are separate tasks.

How renovation changes the meaning of a price increase

Consider a hypothetical home bought for $400,000 and sold several years later for $520,000. Its owner spent $60,000 adding living space between sales. The observed price increase is $120,000, or 30%.

Suppose, solely for this illustration, that the home would have sold for $480,000 without the addition. The assumed market gain is then $80,000, or 20% of the purchase price, while the addition contributes $40,000 to the resale price. The two sale prices alone cannot reveal that split.

The $60,000 construction bill does not establish a $60,000 increase in market value. Subtracting renovation spending from the price gain answers a cash-accounting question; it does not recover the price of an unchanged house.

Depreciation creates the opposite complication: a home's condition can deteriorate between observations. The net direction of bias depends on the balance of quality changes, a qualification emphasized in the 2007 research.

What the FHFA renovation bias research measured

FHFA published Alexander Bogin and William Doerner's Working Paper 17-02 on March 24, 2017. It examined whether accounting for renovations changed repeat-sales estimates.

The authors linked listings from 110 multiple listing services across 36 states to FHFA purchase and refinance-appraisal records. They identified likely renovations through changes in room counts, living area and other listing attributes. Their research compared quarterly metropolitan-area and ZIP-code indexes with and without time-varying renovation indicators over 2000–2015.

The reported maximum gap was approximately 15% in selected local areas, with larger effects concentrated in central districts of large cities. Effects generally diminished farther from downtown and were generally negligible in the smaller cities studied. FHFA's published summary does not present that maximum as a nationwide adjustment.

The measurement was a cumulative gap between index levels from a 2000 base, not an annual percentage-point error. Metro-wide effects were generally muted. The paper also identifies limits to the analysis: coverage was incomplete, the indicators did not measure renovation costs or quality, and depreciation was not netted out. Refinance timing also complicated assigning improvements to transactions.

These findings give analysts a reason to investigate a local sample. They do not provide a 2026 estimate of bias or justify subtracting 15% from a property's value.

How the research differs from the published FHFA HPI

The research added renovation indicators to an index specification. That exercise should not be read as an announcement that FHFA adopted those indicators throughout its published index suite.

The March 1996 technical description describes statistical weights tied to estimated variability between repeat observations. Such weighting differs from observing each remodel and deducting its value.

The particular series also matters. FHFA's August 25, 2026 quarterly report distinguishes purchase-only indexes from all-transactions indexes, which add refinance appraisals, and expanded-data indexes, which add FHA and county-recorder sales information. Those sample differences are separate from renovation adjustment.

Why index growth is not an investment return

In the hypothetical sale, $520,000 minus the $400,000 purchase and $60,000 renovation leaves $60,000 before other costs. Calling the 30% sale-price increase the owner's investment return would omit money invested after purchase.

A return calculation also needs the timing of cash invested and received, transaction and ownership costs, financing, and any rental income. An index percentage supplies none of that property's cash-flow history.

For underwriting, keep three assumptions separate: market price growth, the value expected from improvements, and the cash required to complete them. Test each explicitly. For a property-specific valuation, ask an appraiser to assess current condition and comparable sales.

Questions to ask before comparing local indexes

Before using an index in a chart or investment memo, check its documentation:

• Which geography? Record the exact boundary. A ZIP code, city and metropolitan area should not be treated as interchangeable.
• Which transactions? Identify property types, financing coverage, and whether observations include refinance appraisals.
• How is quality change handled? Look for documented renovation indicators, exclusions, statistical weights or other controls. Ask what information the method cannot observe.
• Which comparison period? Match start and end dates, reporting frequency and seasonal-adjustment status. Separate cumulative growth from one-year growth.
• How dependable is the local sample? Check observation counts, published uncertainty measures and the data release used.

Use the home-price data and reports collection to locate the underlying series and its documentation. If a neighborhood index suggests unusually strong appreciation, inspect the property histories behind the comparison before applying that growth rate to an unchanged home.