A house price index measures residential property-price change over time; its meaning depends on the sample, method, geography, and adjustment basis.
A house price index (HPI) measures how residential property prices change over time for a defined market. It compares prices across periods after using a stated method to reduce, or at least describe, differences in the location, type, size, age, and quality of homes entering the sample.
The term is also called a residential property price index (RPPI). It does not refer to one universal dataset. An HPI may cover a country, region, city, or property segment and may be built from sale prices, mortgage records, appraisals, or other observations. The methodology and coverage determine what its movements actually mean.
An HPI estimates price movement within a specified target market. A defensible description of any index should answer five questions:
Without those details, “home prices rose 5%” is incomplete. The statement may describe existing single-family homes in selected large cities, all residential purchases nationwide, mortgage-financed properties below a loan limit, or a different population entirely.
Most HPIs are designed to measure a rate of change, not a representative home’s current price. An index level of 220 does not mean the average home costs $220,000.
A median or mean sale-price series answers a different question: what was the middle or average price among properties sold during the period? That figure can move because the mix of homes changed. If one quarter contains more luxury-home sales, the median can rise even when comparable individual homes did not appreciate.
An HPI attempts to separate market price movement from that composition effect. How successfully it does so depends on its data and quality-adjustment method.
Housing is difficult to index because every property is different and the same home may sell only occasionally. Index providers use several methods to create a more comparable series.
| Method | Basic approach | Main strength | Main limitation |
|---|---|---|---|
| Repeat sales | Compare later and earlier transactions involving the same property | Holds the property’s identity constant and needs relatively few property characteristics | Uses only properties with repeated observations and may miss renovations or deterioration |
| Hedonic regression | Estimate how observed characteristics such as location, size, age, and type contribute to price | Can use single-sale properties and explicitly adjust for measured quality differences | Results depend on model design and the availability and accuracy of property characteristics |
| Stratification or mix adjustment | Divide transactions into more comparable groups, then combine group-level price changes | Transparent and practical when detailed characteristic data are limited | Broad groups can leave quality differences within each stratum |
| Sales-price appraisal ratio (SPAR) | Compare current sale prices with earlier assessed or appraised values for the same properties | Can use more current sales than a strict repeat-sales method | Depends on the coverage, timing, and quality of the appraisal or assessment base |
| Hybrid method | Combine stratification, hedonic adjustment, repeat sales, or other techniques | Can address multiple data problems within one framework | More complex and harder for users to reproduce |
A simple average or median transaction price can still be useful, especially for understanding the price level of homes currently changing hands. It should not be treated as a constant-quality price index unless the methodology supports that interpretation.
A Repeat-Sales Index forms pairs from multiple observations on the same home. The price change between the two dates contributes evidence about market appreciation during that interval.
This approach reduces distortion from comparing fundamentally different properties. It does not automatically adjust for a renovated kitchen, an added floor, deferred maintenance, a zoning change, or neighborhood redevelopment. Long intervals between sales may therefore provide noisier evidence than short intervals.
The FHFA House Price Index and Case-Shiller Home Price Index are prominent U.S. examples, but their source data, coverage, weighting, products, and revision processes differ.
A hedonic model treats a home’s price as the result of observable characteristics. A compiler may model location, floor area, lot size, number of rooms, age, structure type, and other available attributes. It can then estimate price change while holding those measured characteristics constant.
Hedonic methods can use more transactions than repeat-sales methods because a property need not have sold before. However, omitted or poorly recorded characteristics can affect the estimate. Model specification also matters: relationships between characteristics and prices may change over time or differ across markets.
Stratification separates sales into groups such as region, property type, age, or price segment. The compiler calculates movement within each group and combines those results using explicit weights.
This reduces the effect of large changes in the sales mix across broad categories. It cannot eliminate composition changes within a group. A “single-family homes in one region” stratum can still contain meaningful variation in location, size, condition, and quality.
An index is normalized to a value in a reference period, often 100. Suppose an HPI has a base of 100 and later reaches 145:
1Cumulative change = (145 / 100 - 1) x 100 = 45%
The measured market price level is 45% above its base-period level. The result does not indicate a $45,000 gain, and it does not say every covered home increased 45%.
To compare any two periods in the same series:
1Percentage change = (Later index / Earlier index - 1) x 100
If the index moves from 240 to 252:
1(252 / 240 - 1) x 100 = 5%
Changing the base from 100 to another reference value rescales every level but leaves the measured percentage change unchanged.
House-price reports commonly present several growth measures at once.
| Measure | Comparison | Best interpreted as |
|---|---|---|
| Month-over-month | Current month vs. preceding month | Short-run momentum, often noisy |
| Quarter-over-quarter | Current quarter vs. preceding quarter | Near-term movement with some smoothing |
| Year-over-year | Current period vs. same period one year earlier | Annual change with less direct seasonal distortion |
| Cumulative | One selected index date vs. another | Total measured change across the chosen interval |
| Annual average | Average index level for one year vs. another | Change in average levels, not necessarily the latest year-end rate |
A positive year-over-year rate can coexist with a negative month-over-month rate. That means the current level remains above its level one year earlier even though the latest monthly comparison declined.
Slower appreciation is not the same as depreciation. If annual growth falls from 8% to 3%, prices are still rising on that measure, but at a slower rate. A decline requires a negative change over the stated period.
These adjustments answer different questions and should be labeled explicitly.
Housing activity often follows recurring seasonal patterns. A seasonally adjusted (SA) series estimates and removes those patterns, making adjacent months or quarters easier to compare. A not seasonally adjusted (NSA) series retains the observed seasonal movement.
Use values from the same adjustment basis in one calculation. Dividing an SA observation by an NSA observation produces a change that has no clean interpretation. Seasonal factors can also be revised as new data arrive.
A nominal HPI shows house-price movement in current money. A real HPI deflates the nominal series with a selected general price index. A simplified calculation is:
1Real change = (1 + nominal house-price change) / (1 + inflation rate) - 1
If nominal house prices rise 6% while the selected inflation measure rises 4%:
1(1.06 / 1.04) - 1 = 1.92%
The approximate real increase is 1.92%, not 2% exactly and not 6%. The answer depends on using compatible periods and an appropriate deflator. The Consumer Price Index (CPI) is one possible reference, but different analytical purposes may call for another measure.
Different results do not necessarily mean one index is wrong. Compare the construction before comparing the headline number.
For a valid comparison, use matching periods and document each series’ population, method, adjustment basis, and release vintage.
The following examples illustrate why the label “HPI” is not sufficient on its own.
| Index family | Core scope | Important boundary |
|---|---|---|
| FHFA HPI | U.S. single-family home values, with multiple repeat-transaction datasets | The flagship purchase-only series centers on eligible conventional, conforming mortgages acquired by Fannie Mae or Freddie Mac |
| Case-Shiller | Existing U.S. single-family homes using matched public-record sales | The family includes national, composite, and selected metro indexes and excludes several property types |
| Eurostat HPI | New and existing residential properties purchased by households across participating European countries | National compilers use harmonized concepts but may apply methods suited to their available transaction data |
| BIS residential property price data | Detailed and selected series assembled across jurisdictions | Source, property type, frequency, area, method, and adjustment can differ by country |
The correct series depends on the question. A lender monitoring U.S. conforming-mortgage collateral, a policymaker comparing countries, and an analyst studying new-home transactions may need different indexes.
Residential property is both a household asset and common loan collateral. House-price movement can therefore affect several financial decisions:
An HPI is normally one input rather than a conclusion. Sales volume, inventory, mortgage rates, rents, household income, delinquency, construction, population, and employment help explain whether measured price movement is broad, sustainable, affordable, or financially material.
A house price index is not automatically:
For example, an HPI can rise while affordability worsens because mortgage rates or prices rise faster than household income. It can also rise while transaction volume collapses. The index describes price movement within its sample, not every dimension of market health.
Before using an HPI, verify:
This article is educational and does not provide an appraisal, lending decision, investment recommendation, or individualized financial advice.