Home prices hit bottom. Rent held up.

Price index

FHFA's index follows the prices of the same homes each time they sell or are refinanced, so it measures how much prices changed, not what a home cost. A county uses its own index when its record starts by 2006, so it covers the housing crisis, and shows a downturn of at least 4 years. Otherwise it uses its metro area's index (37 counties) or its state's (462). These 499 counties are home to 2% of Americans; the largest is Allegheny County, Pennsylvania, which uses Pennsylvania's. FHFA publishes metro and state indexes by quarter; we average each year's four quarters.

Not in the study

32 counties in South Dakota: neither their own record, their metro's nor their state's shows a downturn lasting 4 years. Puerto Rico and the US territories are not included.

The downturn

A period when prices fell below an earlier peak and stayed below it for at least 4 years. Shorter dips are left out, because thin early data swings up and down within a year or two. A downturn still under way counts however short it is so far: for 60 counties, home to 0.2% of Americans, the hardest downturn began in 2021 or later and hasn't ended.

The hardest downturn

For each downturn we follow prices for 10 years from the peak and compare each year's price with where it would have been at the county's usual pace, its average yearly growth across its whole record. For this ranking, we assume prices grow at the usual pace once they are back at the peak, or from their latest level where the record ends sooner, so the years lost still count. Adding up the yearly gaps gives the downturn's shortfall, and the hardest downturn is the one with the largest. Example: in Will County, Illinois, prices fell 28.9% from 1980 to 1982 and were back by 1986. They fell 28.7% from 2007 to 2012 and weren't back until 2021. The second's shortfall was nearly twice as large, so it is Will County's hardest downturn, though the first fell slightly further. This is the same downturn the PriceDNA Report™ replays.

The fall

From the year prices peaked to the lowest year that followed.

Recovered

The first year prices were back at the peak, in dollars. The typical time is 12 years from the peak, or 6 from the bottom. Counting the 60 unfinished downturns as not yet recovered still gives 12.

Rent

HUD's estimate of the 40th percentile of rents paid, utilities included, by people who recently moved into a 2-bedroom home. It is a benchmark for a modest rental, not what every renter pays. HUD's fiscal year 2008 rent is paired with the 2008 price. HUD published no rents for 1984; we use the midpoint growth between 1983 and 1985.

How HUD builds rents

Each year's rent starts from Census survey rents collected over earlier years, which HUD moves forward with inflation in rents; today the newest survey year is 3 years before the rent year (HUD, 2026 method). HUD adopted a 2000 Census base in 2005, began using the Census Bureau's yearly survey to update rents in 2008 (HUD, 2008 notice), and moved to a five-year survey base in 2012 (HUD, 2012 notice). The Census survey check below compares HUD's rents with the rents people actually paid.

Rent against its usual pace

A county's yearly rent growth during its downturn, against its average yearly rent growth in all its other years from 1983. Earlier decades had higher inflation, so we also compare the two after inflation, using only years the Consumer Price Index covers.

Missing rent

153 counties, home to 5% of Americans, have no rent figure, because HUD's rent record for the county starts after its downturn began. Most started before 1983, when HUD's record begins, such as Harris County, Texas, whose hardest downturn began in 1982. The rest are Alaska areas whose HUD record starts later, after they were redrawn. The rent findings cover the other 2,959 counties.

Population

Each county's people, from the US Census Bureau's American Community Survey 2020 to 2024 (table B01003). It shows where people live now, not where they lived during each downturn, and is used only to count people. Shares of all Americans count everyone in the 50 states and DC, including the counties left out of the study.

Inflation

The Bureau of Labor Statistics' Consumer Price Index for all items, US city average, yearly average (series CUUR0000SA0).

HUD's method changes

HUD documented changes to how it sets Fair Market Rents in 24 of the 45 years from 1983 to 2027, from rebuilding rents on new Census data to small changes in how it updates them. We use HUD's rents as published. Choosing which changes to correct would mean judging HUD's methods year by year, and correcting all 24 would replace most local rent data with a national average.

If, in the years of HUD's 9 major changes (new Census base data, the switch to the 40th percentile, and new metro areas), we replaced every county's rent change with the national change in rent from the Consumer Price Index, the typical rent change would be +17.6% instead of +14.1%, and 93.5% of Americans instead of 90.5% would live where rent ended higher. The main finding holds with either series. Replacing those years also removes real local movement, so the difference is not HUD's methods alone.

See the 24 years and HUD's notices

Years HUD documented a change to its Fair Market Rent method. ● marks the 9 major changes: new Census base data, the switch to the 40th percentile, and new metro areas.

Year	What HUD changed	Source
1983	Moved the standard from the 50th to the 45th percentile of recent movers' rents (described in HUD's 1986 notice; we did not find the 1983 notice itself)	HUD
1986 ●	Rebuilt rents on the 1980 Census	HUD
1994 ●	Rebuilt rents on the 1990 Census and adopted new metro areas	HUD
1995 ●	Moved the standard from the 45th to the 40th percentile	HUD
1996	Set a statewide minimum rent for low-rent rural areas	HUD
2001	Raised 39 metro areas to the 50th percentile	HUD
2002	Made a one-time adjustment for the natural gas price spike	HUD
2005 ●	Rebuilt rents on the 2000 Census	HUD
2006 ●	Adopted new metro areas, split some into separate rent areas, and brought back the state minimum	HUD
2007	Split more metros into separate rent areas	HUD
2008	Updated rents with the Census Bureau's new yearly survey for the first time	HUD
2009	Changed how that survey's updates were applied in some areas	HUD
2010	Added 3-year survey data where yearly data was thin	HUD
2012 ●	Rebuilt rents on the Census Bureau's 5-year survey	HUD
2013	Changed how rents are moved forward to the rent year	HUD
2014	Changed Puerto Rico's data sources	HUD
2016 ●	Switched to a forecast to move rents forward, and adopted new metro areas	HUD
2018	Limited any yearly fall to 10%	HUD
2020	Used local and regional forecasts instead of one national one; ended the 50th percentile areas	HUD
2022 ●	Adopted new metro areas	HUD
2023	Added private-sector rent data to the yearly update	HUD
2024	Counted recent movers as those who moved in the last 11 months, not 23	HUD
2026 ●	Adopted new metro areas; Connecticut moved to planning regions	HUD
2027	Changed how utility costs are updated	HUD

We found a documented change in 24 years. For the other years we found no notice of a change, except 1985 and 1989 to 1993, whose notices survive only as scans we could not read, so a change in those years is possible. That window falls outside the housing crisis. If you know of a change we missed, tell us and we'll add it and rerun the study.

Does the result change under other choices?

We reran the rent findings with each choice a reasonable person could make differently. Among Americans with a rent figure, 95% to 99% live where rent ended higher in every row. The typical rent change runs from +12.1% to +17.6%, and how HUD's method changes are treated moves it the most.

"Replaced" means every county's rent change in those years is swapped for the national change in rent (Consumer Price Index for rent of primary residence, US city average).

"Deepest fall" picks earlier downturns in some counties, more of which began before HUD's rent record, so fewer Americans have a rent figure in that row.

Prices don't depend on these rent choices. The typical fall is 20.6% with today's population and 20.2% with the 2010 Census.