Home › Research › Housing and autos
Housing and autos are slowing together. Here is what history says comes next.
Housing and car sales move together: over 50 years their annual changes have a correlation of about 0.6. Both are softening now, as mortgage rates climb back above 7%. But the slowdown is lopsided. Housing is well into a multi-year downturn. Auto sales have barely dipped. In past cycles, the break in auto sales is what separated a housing slump from a recession.
Bottom line
- The link is real. Since 1977, yearly changes in single-family housing starts and light-vehicle sales have a correlation of 0.63. Both depend on the same things: interest rates, credit and household confidence.
- Housing tends to move first, but only slightly. New-home sales lead auto sales by about three months on average. The lead is small and inconsistent from cycle to cycle.
- This correction is housing-heavy. Single-family construction is 21% below its 2022 high on a 12-month basis. Auto sales are just 2.5% below their recent high.
- History is a coin flip on recession, not a verdict. Joint housing-and-auto dips like this one were followed by a recession within two years about half the time. The ones that turned into recessions all had something today's data don't yet show: a sharp break in auto sales.
- Our base case is a grind, not a crash: housing volumes drift somewhat lower, prices stay flat to slightly down in nominal terms, and autos hold near 15.5–16 million. The signal that would change that view is below.
Where things stand
| Indicator | Latest | Change from a year ago |
|---|---|---|
| Existing-home sales (Sep) | 3.99M annual rate | −2.2% |
| Existing-home median price (Sep) | $417,500 | +1.3% (39th straight yearly gain) |
| New-home sales (Sep) | 676K annual rate | −5.3% |
| New-home median price (Sep) | $400,400 | −3.9% |
| Housing starts, total (Sep) | 1.221M annual rate | −7.4% |
| Case-Shiller national index (Aug) | — | +2.1% (15th month of real, inflation-adjusted decline) |
| Light-vehicle sales (Sep) | 16.0M annual rate | −3.6% |
| Light-vehicle sales (Oct forecast, J.D. Power) | 15.6M annual rate | Flat |
| 30-year mortgage rate (Oct 29) | 7.24% | Up about 1 point |
So far this is a volume correction more than a price correction. Nationally, home prices are still rising slightly in nominal terms but falling after inflation. New-home prices, where builders can cut, are down about 4% from a year ago. Prices are falling outright in a few metros, including Seattle (−2.0%), Las Vegas (−1.3%) and Denver (−0.8%).
Auto buyers face their own squeeze. The average new-car payment hit a September record of $821, 13.9% of new loans run 84 months or longer, and the New York Fed reports serious auto-loan delinquency at its highest level since 2010.
Do they really move together?
Yes. The lines above track each other through every major cycle since the late 1970s. Measured as year-over-year changes in 12-month averages, the correlation is 0.63. Using monthly data (3-month averages), it is 0.60 for total starts and 0.51 for new-home sales.
The timing evidence is weaker. Correlation peaks with no lag, but it falls off more slowly when housing is shifted earlier. New-home sales line up best with auto sales three months later (0.54 vs. 0.51 with no lag). That fits the story that a home purchase pulls other big purchases along with it, but it is not a reliable early warning on its own.
What happened after past slowdowns
In each of the three housing-led recessions of the past 50 years (1980–82, 1990–91 and 2007–09), housing fell 33% to 75% and auto sales fell 23% to 41%. The 2021–23 episode is the important exception. A rate shock cut single-family starts by about 24%, but auto sales were recovering from the chip shortage, and the economy did not enter a recession.
Two base rates, both calculated from the Census, BEA and Freddie Mac data:
| Setup | Times since 1978 | What followed |
|---|---|---|
| 30-year mortgage rate up 1 point or more from a year earlier (like now) | 10 | Single-family starts fell over the next 12 months 7 times; median change −5.7%. Auto sales were roughly flat (median +0.2%). A recession began within 24 months 4 times (1980, 1990, 2001, 2007). |
| Housing starts and auto sales both down from a year earlier (3-month averages) | 11 (excluding the 2020–21 pandemic swings) | A recession began within 24 months 5 times (two of them ahead of the same 1980 recession), or 6 counting the 2020 pandemic recession. The other half faded without one. |
Three paths from here
1. Grinding slowdown (base case). This is the most common outcome after a mortgage-rate jump. Housing volumes drift another few percent lower over the next year, home prices go roughly flat nationally and keep falling after inflation, and auto sales hold around 15.5–16 million. Builders and auto lenders feel it; the broader economy mostly does not.
2. Contained correction. The 1994–95 and 2022–23 path. Mortgage rates ease back below 7%, housing finds a floor at today's low level, and autos hold up. In both precedents, housing fell 19% to 24% without a recession following.
3. Housing-led recession. The 1979, 1989 and 2006 path. Rates stay high, households pull back on big purchases, and auto sales break down. About 4 in 10 mortgage-rate shocks of this size were followed by a recession within two years, though not all were housing-led. In the three that were, the auto break came first.
What to watch
- Auto sales below about 15 million for several months, or a 3-month average down more than 8% from a year earlier
- The 30-year mortgage rate: sustained above 7.5% leans toward path 3; back under 7% leans toward path 2
- The Fed's Dec. 8–9 meeting. It held at 3.75%–4.00% on Oct. 28, but two members voted for a hike.
- Auto-loan delinquencies, already at their highest since 2010, and subprime lending
- New-home prices: deeper builder price cuts would show the slowdown moving from volume to price
Data caveats
- Correlations and base rates are our own calculations from Census Bureau housing starts and new-home sales, BEA light-vehicle sales and Freddie Mac weekly mortgage rates, with NBER recession dates. Samples are small (3 housing-led recessions, 10 rate shocks), so treat the frequencies as rough guides, not odds.
- Recent auto comparisons are affected by the March–April 2025 tariff-driven buying rush, which makes spring 2026 look weaker than it was.
- The October auto figure is a J.D. Power forecast; official BEA data comes in early November.
- Monthly housing data is revised and has wide margins of error. Census says the September new-home sales changes were not statistically significant.
Not investment advice. Sources: Census Bureau (housing starts), Census Bureau (new-home sales), BEA (light-vehicle sales), Freddie Mac (mortgage rates), NBER (recession dates), NAR (existing-home sales), S&P Cotality Case-Shiller, Mortgage News Daily (September starts), WRE News (September new-home sales), Mercer Capital (September auto sales), CollisionWeek (October forecast), GoodCarBadCar (NY Fed auto credit), Federal Reserve (Oct 28 statement), Kiplinger (Fed calendar). Published Oct 30, 2026.