Sections

Research

How racial bias in appraisals affects the devaluation of homes in majority-Black neighborhoods

Photo of a white realtor showing a Black couple information about their house.

Contents



Summary

Previous Brookings research found that homes are undervalued in majority-Black neighborhoods, yet could not identify why. Recently, the Federal Housing Finance Agency released new neighborhood-level data on home appraisals. Using this data in combination with other sources, this report draws the following conclusions on home appraisal bias’s effect on housing markets in majority-Black neighborhoods: 

We continue to find that homes in Black neighborhoods are valued roughly 21% to 23% below what their valuations would be in non-Black neighborhoods. Neighborhoods with a majority of Latino or Hispanic, Asian American, or white residents do not experience home price devaluation, using the same model. Appraisal bias explains a fraction of the devaluation of homes in Black neighborhoods: approximately 9% to 19%, depending on modeling approaches. The vast majority of homes in majority-Black neighborhoods and throughout the country are appraised at or above their contract price, leaving much of the variance unaccounted for. The cost of devaluation across the 113 metro areas in the U.S. with at least one majority-Black neighborhood is approximately $162 billion. 

Adjusting for characteristics of homes and neighborhoods, we find that appraisal transactions in majority-Black neighborhoods are 1.9 times more likely to be appraised under the contract price than homes in majority-white neighborhoods. Specifically, an additional 6% (13% versus 7%) of homes in Black neighborhoods are under-appraised relative to non-Black neighborhoods. After adjusting for characteristics of the homes and neighborhoods, this bias against homes in Black neighborhoods persists at a level of 4.4%. The bias is also found in over-appraisal; in majority-Black neighborhoods, an additional 5.2% of homes are appraised at or below the contract price when they would have been appraised above it if the home had been in a non-Black neighborhood. Taking these estimates together, we find that 10% of appraisals in majority-Black neighborhoods are valued on the wrong side of the contract price, compared to what would be expected in the absence of racial bias. Neighborhoods with a majority of Latino or Hispanic and Asian American residents are also more likely than majority-white neighborhoods to experience under-appraisal. 

We estimate that the median appraisal is 15% lower in majority-Black neighborhoods compared to homes in neighborhoods where less than 1% of the population is Black. This result is estimated by modeling appraisal values as a function of actual sales prices as well as neighborhood and home characteristics.  

Our overall conclusion is that at least 10% of homes are at risk of under-appraisal in majority-Black neighborhoods, and this has a modest but meaningful effect on overall valuations and final sales prices—limiting wealth accumulation for homeowners in majority-Black neighborhoods. Under-valuations also delay or cancel transactions, which reduces mutually beneficial market exchanges in Black communities. 

At the same time, those seeking to make housing markets work more effectively in majority-Black neighborhoods will have to look beyond appraisal reform. Appraisal bias appears to explain less than 20% of overall devaluation, and most appraisals in majority-Black neighborhoods are above the contract price. Likely, lending and underwriting practices as well as consumer bias lower home values and reduce demand in majority-Black neighborhoods below what we would see if racism was absent.

Understanding devaluation

Our previous research has shown that owner-occupied homes are substantially devalued in majority-Black neighborhoods. After considering the characteristics of homes and neighborhoods that are most important to buyers—age and size of the home, test scores at local elementary and secondary schools, proximity to stores, commuting times, and crime—we find that homes in Black neighborhoods are valued roughly 23% lower than similar homes in neighborhoods with few Black residents. 

We and others have speculated that bias in the appraisal process may account for some of this discrepancy. Consistent with this expectation, economists at Freddie Mac estimated that homes in majority-Black neighborhoods are 1.7 times more likely to receive an appraisal below the final contract price compared to homes in predominantly white neighborhoods.  

Still, the available evidence suggests appraisal effects are unlikely to explain most of the devaluation in Black neighborhoods. Most homes appraise at or above the final contract price. While a low appraisal could depress the final price, it more likely confirms a reasonable compromise between buyer and seller. A seller can also challenge a low appraisal. Additionally, our research on commercial real estate valuations shows that they are also devalued, despite the different actors and agencies involved. 

Alternatively, participants in the housing market—including buyers—may exhibit systemic bias against Black neighborhoods, even in the absence of racist beliefs or discriminatory behaviors toward Black individuals. This bias may be the result of ignorance, prejudice, or both. 

New data on appraisal bias

In October, the U.S. Federal Housing Finance Agency (FHFA) released a novel database that helps examine concerns about appraisal bias. The Uniform Appraisal Dataset (UAD) relies on a standardized appraisal form sent to Fannie Mae or Freddie Mac for single-family homes. FHFA staff constructed the database using approximately 46 million final appraisals. 

The data is available for census tracts (which are akin to neighborhoods), and the FHFA reports the summary statistics on the value of appraisals, including the percentage of appraisals that were valued at, below, and above the contract price. The appraised value itself contains no information on discrimination; nor does the sale price. The value of a home is determined by many factors, but when analyzed alongside the contract price, the appraised value does contain insightful information. 

Since the contract price is the price agreed to in the actual real estate market, it is a more relevant estimate of what the home is worth compared to the appraisal. If the potential contract price was too low, the seller would have a strong incentive to hold out for a better offer; if it was too high, the buyer would have a strong incentive to walk away. The appraiser’s job is to provide the lender an accurate assessment of the financial value of the home, so that it can weigh the risks of buying the home on behalf of the borrower and convey those risks to secondary market participants such as Freddie Mac and Fannie Mae, who buy mortgages from smaller banks and sell them to investment banks and other investors. An appraiser who consistently over-values homes can be seen as injecting poor information and credit risk, as economist Paul Calem and co-authors have argued, but banks make money off transactions and many mortgages are resold in secondary markets. Thus, most banks have an incentive for the appraisers’ valuation to match or exceed the contract price, so that the transaction proceeds and generates fees for the bank and a product to sell. 

In practice, most homes (usually above 60%) appraise above the contract price. This has been the case since 2013 (see Figure 1), but it has dipped somewhat in recent years—perhaps because the pace of inflation has caught some appraisers by surprise. The share of homes that under-appraise is typically below 10%, but that rose sharply in 2021 and stands near 16% in the middle of 2022. Only around one-quarter of homes appraise at the contract price. 

Appraisal values relative to contract price by quarter, 2013 Q1-2022 Q2A key question for our analysis is: Do appraisers treat homes in Black neighborhoods differently than homes in other neighborhoods?  

Analyzing the UAD data on the percentage of homes that are over- or under-valued relative to the contract price can give some insight into this question. This test has the advantage of classifying homes into simple bins (under-valued, over-valued, and appropriately valued), but only in relation to whether or not the appraisal matches the contract price (which only happens in about 25% of cases). An important limitation of this analysis is that it says nothing about the extent of under-valuation, and it cannot be used to estimate the percentage of homes under-valued compared to a hypothetical case of no discrimination. However, it does provide a lower-bound estimate by allowing directional calculations of appraisal values relative to the contract price. 

Imagine that homes in majority-Black neighborhoods were just as likely to be under-valued as homes in majority-white neighborhoods—but every under-valued home in a majority-Black neighborhood was undervalued by 50%, whereas under-valued homes in white neighborhoods were only undervalued by 1%. The “directional” analysis would fail to detect this bias. 

With this in mind, we do find important evidence of appraisal bias, consistent with previous work from Freddie Mac. Without any adjustments for home or neighborhood characteristics, homes in majority-Black neighborhoods are about 6 percentage points (ppt) more likely to be under-appraised relative to the contract price compared to homes in neighborhoods with a Black population share of less than 1% (see Figure 2).  

We know homes and neighborhoods have important differences in underlying characteristics that are valuable to market participants. Thus, as in our previous work, we gathered data by census tract for the size, age, and other characteristics of homes in each tract, as well as neighborhood features such as test scores and proximity to amenities. We analyzed 2016-2020 appraisals using 2016-2020 census data. 

After adjusting for home and neighborhood characteristics—using the same variables from our 2018 paper—the estimated under-appraisal effect falls only slightly, to 4.4 ppt. At the same time, homes in majority-Black neighborhoods are 5.2 ppt less likely to be over-appraised, using the same model but replacing the percent under-appraised with the percent over-appraised.  

Taken together, about 10% of homes in majority-Black neighborhoods are affected by the direction of appraisal bias compared to homes in neighborhoods with few Black residents. But this does not tell us the extent of the bias. If under-valuation is more severe in majority-Black neighborhoods than non-Black neighborhoods when it occurs, this result will be obscured by the directional analysis. 

Percent of homes appraised below final sale price by neighborhood Black populationWe ran the above analysis separately for each large racial and ethnic group. Neighborhoods with a majority of Latino or Hispanic or Asian American residents are also more likely to experience below-contract price appraisals than neighborhoods with few of those residents. The estimated effect is slightly lower than that found in majority-Black neighborhoods. Meanwhile, majority-white neighborhoods are much less likely to experience an under-appraisal compared to neighborhoods with few white residents. The predicted effect on under-appraisal for majority-white neighborhoods is -6.1 ppt. When applied to the mean for all tracts, this predicts that only 2.2% of appraisals in majority-white neighborhoods are under the contract price, compared to 12.6% in majority-Black neighborhoods, 11.5% in majority-Latino or -Hispanic neighborhoods, and 11.1% in majority-Asian-American neighborhoods (Table 1). 

Estimated effect on appraisal on majority race-ethnicity group status at the neighborhood levelMajority-white neighborhoods differ from neighborhoods with single-group non-white majorities in that they are experiencing slower housing price appreciation (according to 2016 to 2020 Redfin sales data) and are less likely to sell homes above asking. This suggests that majority-white neighborhoods are “cooler” markets, which may make appraisals more predictable. But this does not seem to explain the results. Adding controls for price growth and the percentage of homes sold above asking has little effect on the above results. Likewise, if homes values are easier to predict in majority-white neighborhoods, then homes should not be any more likely to be over-appraised—but we find that they are. 

Estimating lost appraisal value

The UAD does not provide data on the contract price—only the appraisal value and the ratio of the appraised value to contract price. Sale price data is available from other sources. To quantify the appraisal penalty to Black neighborhoods in dollar terms, we combined the UAD with data on sales from Redfin, a national real estate brokerage.  

Conditional on the median sale price, the Black population share should make no difference to the median appraised value if the appraisal market was working fairly. But our analysis shows otherwise. Using our preferred list of control variables from our earlier research and including the median sale price from Redfin, we find a large negative effect of Black population shares on the median appraisal value. Homes in majority-Black census tracts receive a median appraisal 15.2% lower than sales prices would predict, adjusting for other factors (see Figure 3). The model predicts 89% of the variation in median appraisals, so it is unlikely that omitted variables are driving this result, though it is certainly possible. Appraisals are expected to be lower when the quality of the home—including whether and how it was renovated—is lower, but the sale price already takes those conditions into account. 

Predicted median appraisal of homes given median sale price by Black population shareThus, even though 90% of homes in majority-Black neighborhoods are appraised “in the same direction” as homes in non-Black neighborhoods, the extent of the penalty appears to be much higher and more pervasive. The distribution of appraisals in majority-Black neighborhoods is out of alignment with the market. 

Re-estimating the devaluation of homes in Black neighborhoods

Appraisals below contract can have large effects on the market. A 2016 paper by Hamilton Fout and Vincent Yao found that an appraisal below contract price increases the odds that the sale price is reduced by 8% to 51%, and the probability that a sale is delayed or canceled rises from 25% to 32%. Estimates from that research suggest that under-appraisal reduces a home’s value by around 3.5% in large metropolitan areas, with negligible results on the overall market, as most homes are not under-appraised relative to the contract price. 

Our results are consistent with Fout and Yao’s findings. A standard deviation in the share of properties valued below contract price predicts a 4.5% lower price per square foot in our preferred model that otherwise omits the racial composition of the neighborhood. Given our estimates above that homes in majority-Black neighborhoods are 4.4 ppt more likely to be under-valued, we can calculate how sales prices are affected in Black neighborhoods. We estimate that under-appraisal bias results in a 3.3% reduction in the sales price per square foot in majority-Black neighborhoods. We estimate a 4.5% reduction using the final sale price rather than the price per square foot.  

The lost value resulting from appraisal bias can be used to estimate how overall devaluation is affected. We return to our 2018 work and re-estimate devaluation in our new database. Doing so and using the above results, we find that appraisal bias explains roughly 15% to 19% of overall devaluation in Black neighborhoods. Using an alternative method described below, we find that appraisal bias explains 9% to 14% of overall devaluation and as much as 17% of lower appraisals.  

Interactive map

We restrict this analysis to the 102 metropolitan areas with at least one majority Black neighborhood. We also give extra weight in the analysis to metro areas with larger Black populations to reduce the influence of measurement error; as such, the estimates should be thought of as characterizing the experience of the average Black person living in different types of metropolitan areas.

To reach these results, we first replicate our 2018 analysis and predict home values as a function of the Black population share and a rich set of housing and neighborhood characteristics. This produces results that match our previous work, despite using different measures of home values over a different time period (2016 to 2020 instead of 2012 to 2016). Using self-reported census valuations, a home in a neighborhood that is 50% Black is valued 23% lower than a home in a neighborhood with no Black residents. Using sales prices adjusted for square footage, the value is 21% lower. Based on this estimate, the cost of devaluation is approximately $162 billion. 

To see how appraisal bias affects these results, we include it directly in the regression model. Because appraisal bias is correlated with the Black population share, the devaluation estimate falls (as predicted) by 9% to 14% of the total. Thus, these results suggest that even in the absence of appraisal bias, homes in Black neighborhoods will be undervalued by approximately 20%.

The devaluation of appraisal values in Black neighborhoods is very similar to the devaluation in sales prices and self-reported home values, suggesting that the overall housing market—not just the appraisal process—is driving the results. Even adjusting for low appraisals, we predict that appraisals would undervalue homes in Black neighborhoods by approximately 20%, with slightly higher devaluation at the top end of the market.

Estimates for the devaluation of homes in Black neighborhoods using different analytic strategies and accounting for appraisal bias, 2016-2020

Among the large racial and ethnic groups in the United States, devaluation is unique to Black neighborhoods. Using our preferred outcome (median sale price per square foot), we re-ran the analysis above using Latino or Hispanic, Asian American, and white neighborhoods, and compared majority-group status with group shares below 1%. Despite the appraisal bias documented above, we find little to no evidence of devaluation in Latino or Hispanic and Asian American neighborhoods. Size-adjusted sales prices are no higher or lower in majority-Latino or -Hispanic neighborhoods than in neighborhoods with a Latino or Hispanic population share lower than 1%. Prices are slightly higher in majority-Asian-American neighborhoods (6%) and much higher in majority-white neighborhoods (35%), relative to the 32% devaluation in majority-Black neighborhoods. Note, this 32% figure for devaluation in majority-Black neighborhoods is higher than the 21% from the table above because it includes all neighborhoods with 50% or more Black population shares, whereas the table reports the estimated penalty when the neighborhood is exactly 50% Black.

Conclusions and limitations

The new FHFA data on appraisals have allowed for novel analysis of the appraisal process and how it functions in majority-Black and other neighborhoods. Appraisals usually come in above the final market price, and the banks who purchase appraisal services have an incentive for this outcome. But whether the appraisal comes in below, above, or at the contract price should not vary by neighborhood demographics, unless there is something about the neighborhood that affects the appraisal more than the sale price. Our analysis controls for the characteristics real estate agents and housing market participants consider, and still finds that race has a strong effect on the direction of appraisals away from the contract price. Moreover, appraisal values are systematically lower (by 15%) in majority-Black neighborhoods, even given sale prices and other attributes. Altogether, the evidence strongly suggests that appraisers introduce systemic bias that favors white neighborhoods at the expense of Black, Latino or Hispanic, and Asian American neighborhoods.

Still, there are several major limitations to this analysis.

One is that our measure of appraisal bias (from FHFA data) is benchmarked against the contract price and limited to reporting the percentage of homes that fall at, above, or below that price. This is likely to understate bias because it ignores the degree of bias within each group. The UAD contains a ratio that measures the contract price divided by the appraisal price, and this data is consistent with the above analysis. But this appears to be a noisy measure of bias, as it is less correlated with the final sale price than the “percent appraised below contract” variable. Theoretically, the ratio could easily be distorted by outliers. For example, most homes sold in a neighborhood could be appraised well below their actual value, but if a few relatively expensive homes are appraised well above, the ratio for the neighborhood will look unbiased.

A more fundamental problem with measuring appraisal bias by comparing the contract price to the appraisal value is that the contract price is not necessarily a measure of a home’s “true” value, for several reasons.

First, if sellers anticipate a low appraisal, they have a strong incentive to lower the contract price, so that the deal does not fall through. Many homebuyers make their offer contingent upon appraisal to avoid being stuck with an unexpected payment that they cannot finance, in the event of a low appraisal.

More importantly, the broader real estate market may devalue homes in majority-Black neighborhoods, including buyers. If appraisals were removed from the real estate market process, we estimate that homes would still be devalued in majority-Black neighborhoods by around 20%—but not in Latino or Hispanic, Asian American, or white neighborhoods. Along these lines, we find that appraisal bias can only explain 9% to 19% of overall devaluation of homes in Black neighborhoods. Even in majority-Black neighborhoods, most homes (87%, without adjustment) appraise at or above their contract price, with the majority (61%) appraised above.

Rooting out discrimination demands that scrutiny goes beyond appraisals and appraisers. Early in his administration, President Joe Biden announced the formation of the Interagency Task Force on Property Appraisal and Valuation Equity (PAVE) to address racial bias in home appraisals. In March 2022, PAVE released its final action plan—the most wide-ranging set of reforms ever put forward to advance equity in the home appraisal process. The plan lays out 21 actions to be taken by 13 federal agencies, including the development of new rules to remove discrimination from every stage of the home valuation process as well as efforts to build a more diverse home appraiser workforce.

In 2018, we found a 23% value difference between majority-Black neighborhoods and places where the Black population share is less than 1%, but we could not explain why. Now, the Federal Housing Finance Agency’s new dataset gives researchers the ability to see the extent that appraisal bias devalues homes. Still, if upward of 20% of the value gap can be attributed to appraisal bias, then what about the other 80%? In future work, we will maintain a focus on home values and investigate the fundamental reasons why consumers and other market actors undervalue homes in majority-Black neighborhoods.

This post was produced through a partnership between Brookings Metro and the NAACP Empowerment Programs.

Metropolitan area dashboard

Select a metropolitan area using the dropdown menu below to view its dashboard of indicators. The summary metrics provide top-level information about the metropolitan area while the neighborhood characteristics are broken out by the share of the neighborhood population that is Black.

Dashboard

  • Footnotes
    1. For a neighborhood with a Black population share of exactly 50%, the appraisal devaluation is approximately -8%. This linear analysis uses the same method as reported below. The underlying model uses the Black population share as the key explanatory variable, whereas the text above and the figures use binary variables for the share of Black population, reporting the coefficient on whether the share is at or above 50%. Both approaches have their strengths and weaknesses. The linear approach uses more information by including values between 0% and 100% Black shares. The non-linear approach allows for categorical jumps in the value relationships and better characterizes all majority-Black neighborhoods, though is limited to comparing majority-Black neighborhoods to those with less than 1% Black population shares.