Skip to content
Menu
Finabase Learn

Read the economic data Lesson 14 of 16

Economics · Economic indicators

Home price indexes: FHFA, Case-Shiller and real house prices

How US home price indexes work, why repeat-sales indexes beat the median sale price, and what real house prices did in 2006–2012 and 2020–2022.

On this page
  1. Why a median price is not a price index
  2. The FHFA House Price Index
  3. The Case-Shiller indexes
  4. Nominal and real house prices
  5. What the data show
  6. Pitfalls and limitations
  7. Comparing the measures
  8. How Finabase does it
  9. Frequently asked questions
  10. Summary
  11. Check your understanding
  12. Exercise

"Home prices are up 5%" sounds like a simple fact, but it depends on which homes, which sales and which method. The median price of homes sold in a quarter can fall while the value of almost every house rises, simply because cheaper homes made up more of the sales. That is why economists track house prices with repeat-sales indexes, which compare the same house with itself across two sales.

The two main US indexes of that kind are the FHFA House Price Index, published by the federal agency that regulates Fannie Mae and Freddie Mac, and the S&P Cotality Case-Shiller indexes (formerly S&P CoreLogic Case-Shiller). Both are cited every month, they usually tell the same story, and they differ in coverage and weighting.

This guide explains how each works, why the median price misleads, how to turn a nominal index into real house prices, and what the data show for the 2006–2012 bust and the 2020–2022 surge. The monthly flow data on construction and sales, starts, permits and new home sales, are in Housing data: starts, permits and new home sales explained.

Why a median price is not a price index

A median sale price is the middle price of the homes that happened to sell in a period. If more small condos or starter homes sell this quarter, the median falls even if every home is worth more than last quarter. The Census Bureau's median sales price of new houses sold (FRED MSPUS) is a good example: it tracks only new construction, so it moves with what builders choose to build as well as with prices.

A price index tries to hold quality constant. There are two common ways:

  • Repeat sales. Find homes that sold twice and measure how much each one's price changed between sales. Average those changes. The house is its own control for size, location and quality, except for renovations and wear.
  • Hedonic models. Estimate a price for each characteristic (bedrooms, square footage, location) and track the price of a fixed set of characteristics. The Census Bureau publishes a constant-quality index for new homes this way.

The FHFA House Price Index

FHFA describes its index as a weighted, repeat-sales index covering single-family homes, built from mortgage transactions on properties whose loans were purchased or securitized by Fannie Mae or Freddie Mac since January 1975. It covers the nation, all 50 states and more than 400 cities. There are three main versions:

Version Uses Frequency
Purchase-only Sale prices only Monthly, quarterly
All-transactions Sales plus refinance appraisals Quarterly
Expanded-data Sales plus FHA loans and county records Quarterly

The purchase-only index (FRED HPIPONM226S, January 1991 = 100) is the one most often quoted. The all-transactions index (FRED USSTHPI, 1980 Q1 = 100) goes back to 1975 but includes appraisals, which tend to lag and smooth turning points.

What FHFA does not cover matters. Homes bought with cash, with jumbo loans above the conforming limit, or with government-insured loans are outside the purchase-only index. That leaves out much of the top of the market. FHFA publishes monthly and quarterly; its next quarterly release, covering September 2026 and 2026 Q3, is scheduled for November 24, 2026, and the October 2026 monthly figures for December 29, 2026.

The Case-Shiller indexes

The S&P Cotality Case-Shiller indexes use the same repeat-sales idea, developed by economists Karl Case and Robert Shiller in the 1980s, but differ in three ways:

  • Coverage. They use public records of sales of single-family homes, whatever the mortgage, including cash and jumbo purchases. There are indexes for 20 metro areas, 10- and 20-city composites and a national index.
  • Weighting. Pairs and metro areas are combined by market value, so expensive homes and big markets count for more. FHFA's index is closer to an average of homes, Case-Shiller to an average of housing value.
  • Smoothing and lag. Each month's figure is a three-month moving average of sales pairs, and the data are published about two months after the reference month.

S&P's index data are licensed, so this page describes the method but does not chart them. In practice the two families move together over long periods; Case-Shiller tends to swing more in booms and busts because it covers coastal metro areas and the top of the market that the conforming-loan index misses.

Nominal and real house prices

Home price indexes are nominal. To ask whether housing became more expensive relative to everything else, divide by a price index such as the CPI.

One caution: CPI's own housing component uses rents and owners' equivalent rent, not house prices. Deflating house prices by the CPI compares them with the cost of living, which includes rent, and is a choice, not the only correct deflator.

What the data show

FHFA purchase-only house price index, nominal and real (CPI), Jan 1991 = 100 100 200 300 400 Real peak Mar 2006 Jan 2012: −29% real Feb 2020→Jul 2022: +37.5% Nominal Real (deflated by CPI-U) 1995 2000 2005 2010 2015 2020 2025
FHFA purchase-only house price index, seasonally adjusted, January 1991 = 100, nominal and deflated by CPI-U (rebased to January 1991 = 100), January 1991 to August 2026. October 2025 has no CPI reading and is skipped in the real line. Shaded: NBER recessions. Sources: FHFA (FRED HPIPONM226S) and BLS (FRED CPIAUCSL).

2000 to 2012: the boom and bust

From January 2000 to April 2006 the purchase-only index rose 61%, or 36% after inflation. The real peak came in March 2006; the nominal peak in April 2007. Then prices fell for years:

Measure (peak → trough) Peak Trough Fall
FHFA purchase-only, nominal Apr 2007 May 2011 −21%
FHFA purchase-only, real (CPI) Mar 2006 Jan 2012 −29%
FHFA all-transactions, nominal 2007 Q1 2012 Q2 −19%
Median price of new houses, MSPUS 2007 Q1 2009 Q1 −19%

Nominal prices did not regain their 2007 peak until May 2016; real prices took until July 2020. The fall in house prices, and the mortgage losses that came with it, were at the center of the 2007–2009 recession; see The business cycle: expansions, peaks, recessions and troughs.

2020 to 2022: the pandemic surge

From February 2020 to July 2022 the purchase-only index rose 37.5%. Low mortgage rates, demand for space and a short supply of homes for sale combined. Unlike 2006, the rise was not followed by a broad fall: the index flattened in late 2022 and early 2023 as mortgage rates rose, then kept climbing slowly. By August 2026 it was 59% above February 2020 in nominal terms. With CPI up 29% from January 2020, real prices were 24% higher than in January 2020.

Three US home price measures, quarterly, 2000 Q1 = 100 100 200 300 FHFA purchase-only (quarterly mean) FHFA all-transactions Median sales price, new houses (MSPUS) 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023 2025
Three home price measures, quarterly, 2000 Q1 = 100: FHFA purchase-only (quarterly mean of the monthly index), FHFA all-transactions, and the median sales price of new houses sold. Shaded: NBER recessions. Sources: FHFA (FRED HPIPONM226S, USSTHPI), Census Bureau and HUD (FRED MSPUS).

The chart shows the median problem in practice. The median price of new houses sold peaked at $442,600 in 2022 Q4 and was $393,300 in 2026 Q3, down 11%, and down 4.1% on the year. Over the same period both FHFA indexes rose. The fall reflects which new homes were sold, a mix that builders control; the same houses were not getting cheaper.

Pitfalls and limitations

  • Median prices mix composition with price. A shift toward cheaper or smaller homes lowers the median without any home losing value.
  • Repeat sales miss new homes and renovations. A house enters the index only on its second sale, and a kitchen remodel between sales looks like price growth.
  • Coverage differs. FHFA's purchase-only index excludes cash, jumbo and FHA purchases; the Case-Shiller indexes cover only their metro areas and the national composite.
  • Lags and smoothing. Case-Shiller is a three-month average published about two months later; FHFA's all-transactions index includes appraisals. Turning points show up late.
  • Seasonality. Prices rise in spring and summer. Compare seasonally adjusted data, or the same month a year apart.
  • Real depends on the deflator. CPI already contains rents; deflating by it answers "versus the cost of living", not "versus incomes". Price-to-income ratios answer a different question.

Comparing the measures

Measure Method Covers Frequency
FHFA purchase-only Repeat sales Fannie/Freddie-financed sales Monthly
FHFA all-transactions Repeat sales Sales plus refinance appraisals Quarterly
S&P Cotality Case-Shiller Repeat sales All recorded sales, 20 metros Monthly
Median price, new houses Median New single-family houses sold Quarterly

Mortgage rates connect house prices to the bond market; they follow longer-term Treasury yields, see The 10-year Treasury yield: why it matters and what drives it, and the real return on borrowing is discussed in Real interest rates: nominal rates minus inflation, explained.

How Finabase does it

Finabase does not chart house price indexes. The FHFA and Case-Shiller releases can appear on the economic calendar of the Economy page when the calendar's data provider lists them, with actual, forecast and previous figures. The charts and numbers on this page are ours, built from FHFA, Census Bureau and BLS data distributed through FRED.

For the market side, the Real estate sector page lists the S&P 500's real estate stocks with their sector index, homebuilder ETFs such as the SPDR S&P Homebuilders ETF have pages like XHB, and the Treasury page shows the yields that mortgage rates follow.

Frequently asked questions

What is the difference between the FHFA index and Case-Shiller?

Both are repeat-sales indexes. FHFA uses homes financed with Fannie Mae or Freddie Mac loans and weights homes roughly equally; Case-Shiller uses all recorded sales in its metro areas, weights by value, and averages three months.

Why did the median home price fall while indexes rose?

Because the median depends on which homes sold. The median price of new houses fell 11% from late 2022 to 2026 Q3 because the mix of new homes sold changed, while repeat-sales indexes, which compare the same houses, kept rising.

How much did home prices fall in the 2008 crash?

On FHFA's purchase-only index, 21% in nominal terms from April 2007 to May 2011, and 29% after inflation from March 2006 to January 2012. Indexes that cover the top of the market and the hardest-hit metro areas fell more.

Are home prices higher than in 2006 after inflation?

Yes. In August 2026 FHFA's purchase-only index, deflated by CPI, was about 21% above its March 2006 real peak.

What is a real home price?

It is a nominal price index divided by a general price index such as the CPI. It shows whether houses became more or less expensive relative to other goods and services.

Summary

Repeat-sales indexes, FHFA's for conforming-loan homes and the S&P Cotality Case-Shiller indexes for all recorded sales in major metro areas, measure how the price of the same homes changes, so they are not fooled by the mix of homes sold the way a median price is. On FHFA's index, prices fell 29% after inflation from 2006 to 2012, rose 37.5% from February 2020 to July 2022, and in the year to August 2026 rose 2.4%, which was a small real decline with CPI up 3.4%. Read them seasonally adjusted, in real terms when comparing decades, and with their coverage in mind.

Check your understanding

4 questions on this lesson. Each answer comes with the reason, and nothing is scored.

  1. In a quarter, the median price of homes sold falls 3%, while a repeat-sales index for the same area rises 1%. What most likely happened?

    1. Homes lost value, and the index is wrong
    2. More lower-priced homes sold than usual, so the median fell while the same homes gained value
    3. The index uses asking prices, not sale prices
    Show the answer

    B. More lower-priced homes sold than usual, so the median fell while the same homes gained value A median depends on the mix of homes sold. A repeat-sales index compares each home with itself, so a shift toward cheaper homes moves the median but not the index.

  2. A house price index rose 4% over a year in which CPI rose 5%. What happened to real house prices?

    1. They rose about 4%
    2. They fell about 1%
    3. They rose about 9%
    Show the answer

    B. They fell about 1% Real change ≈ 1.04 ÷ 1.05 − 1, about −1%. Homes became slightly cheaper relative to the general price level even though nominal prices hit a record.

  3. Prices fall sharply for luxury homes bought with cash and jumbo loans in coastal cities, while ordinary homes elsewhere hold steady. Which index is likely to show the bigger decline?

    1. FHFA purchase-only, because it covers more states
    2. Case-Shiller, because it covers all recorded sales in its metro areas and weights by value
    3. Both will show the same decline
    Show the answer

    B. Case-Shiller, because it covers all recorded sales in its metro areas and weights by value FHFA’s purchase-only index uses Fannie Mae and Freddie Mac loans, so cash and jumbo purchases are outside it. Case-Shiller includes them and gives expensive homes more weight.

  4. Why can the turning point of a housing downturn show up late in a Case-Shiller release?

    1. Because the index only uses new homes
    2. Because each month is a three-month moving average of sales pairs, published about two months later
    3. Because it is published once a year
    Show the answer

    B. Because each month is a three-month moving average of sales pairs, published about two months later Averaging three months of sales and publishing with a lag smooths the series and delays turns, so the index confirms a change in direction well after it starts.

Exercise

One task to do in Finabase. Nothing is graded or saved.

Open the Economy page with All selected and look for a house price release: the FHFA House Price Index or the S&P/Case-Shiller index, usually in the last week of the month. Read the year-over-year Actual. Then read the latest CPI year-over-year value from the chart under "Inflation, jobs and rates" and work out the real change: (1 + house prices) ÷ (1 + CPI) − 1. Are real home prices rising or falling? If no house price row is listed, use the lesson’s 2.4% for August 2026.

Open the Economy page
What you should find

When house prices rise more slowly than CPI, real prices are falling even while the nominal index sets records, as in the year to August 2026 (2.4% against 3.4%, about −1%). Case-Shiller and FHFA can differ in size because of coverage and weighting. CPI contains rents, so this compares house prices with the cost of living, not with incomes.

Sources and further reading

  1. Federal Housing Finance Agency — House Price Index (methodology, schedule)
  2. S&P Dow Jones Indices — S&P Cotality Case-Shiller Home Price Indices methodology (described only; no index data used)
  3. FHFA — Purchase-only house price index for the US, SA (FRED series HPIPONM226S), downloaded 2026-11-18
  4. FHFA — All-transactions house price index for the US (FRED series USSTHPI), downloaded 2026-11-18
  5. Census Bureau and HUD — Median sales price of houses sold (FRED series MSPUS), downloaded 2026-10-30
  6. BLS — CPI-U all items, SA (FRED series CPIAUCSL), downloaded 2026-10-30