Using Income Changes to Forecast Home Prices by County

We use the strong association between income growth and home price appreciation to identify counties that are under and overvalued.

Our prior work at Home Economics has identified income growth as one of the most important determinants of home prices^a. We have used this finding to argue, for example, that the ‘Housing Crisis’ is driven primarily by rising and increasingly unequal incomes, especially in large cities.

We now use this finding in another capacity: to identify parts of the country where home prices have not risen as much as we might have expected, based on strong income appreciation, or where they are now likely to stagnate, having appreciated more than is justified by incomes.

There is strong empirical and academic evidence for using changes in income to predict changes in home prices.

• Abraham & Hendershott (1996), analysing U.S. metropolitan areas, estimated that a one-percentage-point rise in real income growth lifts prices by about 0.28 percentage points in the same year and a further 0.56 pp in the following year—roughly two-thirds of the adjustment arriving with a one-to-two-year lag.
• Oikarinen (2009) finds a similar pattern for Finland: only ~20 % of the long-run adjustment shows up within the first year, and ~35 % within two years, confirming that price responses to income shocks are sluggish but predictable.

Taken together, these studies support our screening logic: markets where incomes have surged ahead of prices are statistically prone to “catch-up” appreciation over the next couple of years, while the reverse is true for markets that have already overshot their income base.

Incomes and Prices: 2019-2023

We measured incomes and prices at the county level^b, leveraging income data from the National Historical Geographic Information System (NHGIS) for incomes, and the FHFA Home Price Index for home prices^c. The FHFA is a repeat-sales index based on more than six million purchase-mortgage transactions bought or guaranteed by Fannie Mae and Freddie Mac since 1975. It is the same index used by, for example, the Abraham & Hendershott research cited above.

As expected, we find a strong association between changes in incomes and home prices measured from just before the pandemic (2019) to after it (2023). This relationship is weak if we use all counties, but gains strength as we limit our analysis to the largest counties^d.

Price changes vs income changes

Prices in Travis County, TX (Austin), for example, rose 45%—a steep rise over only four years to be sure, but entirely justified by higher incomes (21% increase). Prices in Maricopa County, AZ, soared by 60%, fueled by a 27% increase incomes. Up and down the spectrum, from smaller income and price gains in places like Baltimore (+32% prices, +13% incomes) up to Miami, FL (+66% prices, +31% incomes), there is a clear and strong relationship between income and home price changes.

Even more interesting than the counties along the trendline are the relatively few that deviate. In Kings County, NY (Brooklyn), between 2019 and 2023, incomes rose a modest 15% but home prices severely lagged comparables, rising only 15% (for context, incomes in Orange County, CA, rose a similar amount but prices rose almost three times more). San Francisco is an even more extreme example: incomes and prices rose only 3%.

At the other end of the spectrum, places like Lee County, FL, massively outperformed: incomes rose 15% (ie, similar to Brooklyn), but home prices soared by 72%. Incomes in Fort Bend County, TX, rose only 4% (slightly more than those in SF), but prices skyrocketed by 44%.

These deviations from predicted performance provide opportunities for developers and investors.

Under and Overvalued Counties

Based on our analysis, the most undervalued and overvalued counties—and one county we think is particularly undervalued as of June 2025—are…

Username or E-mail

Password

Remember Me

 

 

Forgot Password

Don’t have an account? Sign up here.

1. see ‘Why is Housing so Expensive, here[↩]
2. We wanted to do this analysis at the Census tract level, but income data is not available from the Census Bureau for single year estimates at such a granular level[↩]
3. The advantage of using the HPI versus other more common sources for home price information (Zillow, for example), is that the HPI is more complete, in terms of both breadth and history, and also more granular, going down to the Census tract level (Zillow data is also granularly provided at the neighborhood level, but is not easily joined to Census information since there are no universal identifiers associated with it, eg FIPS codes or geometry data) [↩]
4. the R-squared for all counties is a paltry 8%, but rises to 36% when we look only at the largest 10% of counties[↩]