Identifying Undervalued ZIPs

Home prices track incomes

The strongest and most consistent predictor of home prices is local incomes. Where incomes are high, so are prices—and when incomes rise in a given area, home prices usually follow.

This is because higher-income households have a greater ability and willingness to pay for desirable locations—those with good schools, safety, and short commutes. The sorting of geographies into high amenity / high income / high price, versus low amenity / low income / low price, operates both across cities (see chart inset) and within them. There is an extensive body of research corroborating and explaining this observation^a.

Once high-income households concentrate in an area, the effect reinforces itself. Their presence attracts better public goods, more services, lower crime, and higher-end retail—all of which are desirable amenities in their own right. Over time, these feedback loops deepen the link between incomes and home prices.

Using the home price-income link for valuation

The strong price-income relationship provides a foundation for identifying over- and undervalued geographies: those with unusually low prices relative to local incomes may be poised for appreciation^b.

Our work here follows earlier research Home Economics has written about using incomes to identify under and over-valued housing markets^c.

The map below plots average household incomes across ZIP codes in the New York–New Jersey–Pennsylvania metro area. ZIPs shaded black have lower incomes, while those shaded blue have higher incomes. Paying subscribers can view this map for any of the 30 largest U.S. cities using the dropdown menu.

The scatterplot in the bottom right shows where each ZIP falls in the city’s price-income relationship. The black line is the trend; the dotted lines mark ±30% deviations around the trend line. ZIPs above the upper line appear overvalued; those below the lower line, undervalued.

Users can toggle yellow outlines around undervalued areas using the “Identify undervalued ZIPs” switch in the bottom-left box. They can also draw custom boundaries—for example, separating Manhattan from New Jersey—to make more meaningful, apples-to-apples comparisons (the price-income relationship is very different between two).

Watch a video demonstration here. View the tool in fullscreen here.

---

Methodology notes

Home prices

We use Zillow data. For consistency, we rely on ZIP-level prices for three-bedroom homes, which have the most complete coverage across major cities. In ZIPs where Zillow lacks three-bedroom data, we use its broader “all homes” series. The ZIPs where we rely on this backup data are indicated with squares in the scatter plots, and excluded from the ‘undervalued ZIPs’ highlighting (they are often cheaper but at least partly due to being smaller).

Incomes

Ideally, we would use median household income for homeowners by ZIP, but this data isn’t published. Census microdata is available only for PUMAs, which don’t align with ZIPs, and published ZIP-level homeowner income data is top-coded at $250,000.

Instead, we use Census tables that report aggregate homeowner earnings by ZIP and divide these figures by the number of owner-occupied units in each ZIP to estimate mean homeowner income.

These means are much higher than median income levels, but because every ZIP is treated consistently, relative comparisons should, for the most part, remain valid^d.

1. A broad empirical literature confirms that local incomes are the dominant long-run determinant of housing prices, and that deviations from this relationship can signal over- or undervaluation. Gallin (2003) first showed that U.S. metro-level house prices and incomes are cointegrated over time, establishing income as a fundamental anchor. Li (2014) demonstrated that both labor income and wealth drive city-level price variation, reinforcing the income side of valuation models. Oikarinen et al. (2023) extended this analysis to 70 metros, finding that while house prices remain strongly correlated with incomes across cities, the slope and intercept of that relationship shift with supply elasticity and amenity growth—implying that current price-income residuals capture relative valuation rather than a stationary equilibrium. Louie et al. (2025) further show that income growth continues to predict house-price growth even after controlling for supply constraints, underscoring the robustness of income as the primary demand-side driver. Together, these studies support using price-income deviations as a practical, income-anchored way to identify potentially mispriced geographies, while recognizing that such gaps may persist where structural factors justify them.[↩]
2. Of course, ZIPs that sit well below the fitted income–price line may be priced for poorer amenities, looser supply, or weak future income growth — or they may be temporarily undervalued. This analysis is merely a starting point—though hopefully a very solid one—for further analysis.[↩]
3. In ‘Why is Housing so Expensive?’, we identified increasingly unequal incomes within cities a the primary driver of the housing crisis. In ‘Using Income Changes to Forecast Home Prices by County’, we used deviations from the home price change vs income change relationship to identify under and overvalued counties.

   Our work here looks at ZIP codes within cities, and uses levels rather than changes to identify mispricings. There are benefits to this approach (e.g.: doesn’t suffer as much from delayed income data), but also drawbacks (e.g.: may confound structurally misvalued geographies with cyclically mispriced ones) [↩]

4. There is some argument to be made, though, that means are especially high versus medians in wealthier ZIPs[↩]