Racial Discrimination in the U.S. Housing and Mortgage Lending Markets: A Quantitative Review of Trends, 1976–2016


We examine trends in racial and ethnic discrimination in U.S. housing and mortgage lending markets through a quantitative review of studies. We code and analyze as a time series results from 16 field experiments of housing discrimination and 19 observational studies of mortgage lending discrimination. Consistent with prior research, we find evidence of a decline in housing discrimination from the late 1970s to the present. Our results show that this trend holds in both the national audits sponsored by the U.S. Department of Housing and Urban Development (HUD) and in non-HUD studies. The decline in discrimination is strongest for discrimination that involves direct denial of housing availability, for which discrimination has declined to low levels. The downward trend in discrimination is weaker for measures reflecting the number of units recommended and inspected, and significant discrimination remains for these outcomes. In the mortgage market, we find that racial gaps in loan denial have declined only slightly, and racial gaps in mortgage cost have not declined at all, suggesting persistent racial discrimination. We discuss the implications of these trends for housing inequality, racial segregation, and racial disparities in household wealth.

More than 50 years ago, President Lyndon B. Johnson signed the Fair Housing Act as a crucial step toward reducing the pervasive racial discrimination in the U.S. housing sector by prohibiting discrimination based on race, color, national origin, religion, sex, familial status, or disability. The Equal Credit Opportunity Act (ECOA), enacted about 6 years later, extended many of the same prohibitions of discrimination to credit markets, including mortgage markets. The ECOA also prohibits discriminating on the basis of a neighborhood’s racial composition—a practice known as “redlining”—and some courts have interpreted the Fair Housing Act in the same manner (LaCour-Little 1999).

Despite the enactment of these laws, much evidence shows that discrimination persists in the housing sector. The persistence of discrimination has been vividly illustrated by Newsday, which conducted 86 matched tests in which pairs of equally qualified white and minority auditors approached real estate agents asking to see homes for purchase on Long Island, New York (Choi et al. 2019). In about 40% of tests, minority auditors received disparate treatment compared to their white audit partner. Disparities in treatment included agent steering toward own-race neighborhoods, making racial comments about neighborhoods to white but not minority auditors, and giving more listings to the white auditor. In some cases the disparities in treatment were egregious. For instance, in a few audits the minority auditor was told by their realtor that they would only be shown listings if they first received a lender pre-qualification letter to certify that they could afford homes in their target price range, while the white auditor was taken by the same agent directly to see properties without any pre-qualification. The Newsday study results are highly consistent with larger-scale governmental and academic studies that have documented recent discrimination in housing markets and mortgage markets (e.g., Delis and Papadopoulos 2018; Turner et al. 2013). These results lead to the question we consider in this paper: how has discrimination in the housing sector changed since the passage of the Fair Housing Act in 1968?

In this analysis, we examine how discrimination in housing and mortgage lending against blacks, Latinos, and Asians has changed over the last 40 years. To do this, we perform a meta-analysis of existing studies assessing discrimination in housing and mortgage lending markets since the late 1970s to the present.

Understanding trends in housing discrimination is important for helping policymakers to assess progress in meeting the key goal of the Fair Housing Act: to curtail discrimination in the housing sector. More broadly, knowing the trend in housing discrimination is a first step to understanding how anti-discrimination enforcement, shifts in racial attitudes, and changes in the mortgage market have impacted racial discrimination in housing and mortgage lending. Finally, the trend in housing discrimination is relevant for understanding how the significance of race has changed (or not changed) in American society over the last 40 years.

Discrimination in housing and mortgage markets is important because of its effects on individuals and the communities to which they belong. Housing and mortgage discrimination contributes to segregation by funneling demand toward racially homogeneous neighborhoods. In addition, housing discrimination contributes to housing insecurity by making it more difficult for members of marginalized groups to find suitable housing. Finally, mortgage discrimination has important impacts on racial disparities in household wealth, since home equity is a major source of wealth accumulation.


Changing Racial Inequality in Housing Outcomes

Over the last 40 years, housing inequalities among whites, blacks, and Latinos have declined. Residential segregation has steadily declined since 1970 (Vigdor 2013), and racial gaps in neighborhood income have also declined (Firebaugh and Farrell 2016). Black and Latino rates of homeownership have increased (Bostic and Surette 2001). A significant share of the black and Latino middle classes have found housing opportunities in mostly white neighborhoods (Sharkey 2014).

Yet the reductions in racial inequality in housing have been slow, uneven, and fragile. Indices of segregation have declined, but only slowly, and residence remains highly racially segregated especially in larger and older metropolitan areas such as Boston or Chicago (Logan 2013). Many of the gains in black and Latino ownership and home equity relative to whites were lost during the recent financial crisis (Faber and Ellen 2016). Blacks and Latinos on average continue to live in much poorer neighborhoods than white Americans (Logan 2011). In sum, a good argument can be made that stability from the past is a better description of racial inequalities in housing than meaningful change.

The persistence of housing inequalities suggests that housing and mortgage discrimination remain significant in spite of legal remedies like the Fair Housing Act. At the same time, racial gaps in housing and neighborhood conditions by themselves cannot be taken as indicating discrimination, because these outcomes are affected by other conditions such as racial inequalities in household income, credit history, and general conditions in the housing market.

Racial discrimination in housing is also influenced by racial attitudes and racism in the U.S. Data on discrimination, prejudice, and racism in the U.S. paint a complex picture of change. Studies of attitudes suggest that the most open and extreme forms of racism and prejudice have declined; these studies suggest that few American now openly support white preference and white supremacy as a principle (Schuman et al. 1997). However, several authors argue that these explicit forms of discrimination have been replaced by other forms of discrimination that are more subtle and covert (Sears 1988; Bobo et al. 1997; Bonilla-Silva 2009). A recent study of discrimination in the labor market finds no change in levels of discrimination against African Americans over the last 25 years, despite changes in racial attitudes during this period (Quillian et al. 2017).

To assess the trend in housing and mortgage discrimination, we create a time series of results from studies of discrimination in the housing sector. We focus our analysis on differential treatment on the basis of race that disadvantages a racial group, the most direct form of discrimination, sometimes called “differential treatment” discrimination in case law (National Research Council 2004).

Housing Market Discrimination

By housing market discrimination, we refer to the discrimination that is carried out by market agents such as landlords, homeowners, and real estate or rental agents. Key outcomes include the refusal to show or rent properties to minorities, or worse treatment of minorities in the housing search process (e.g., recommending fewer units).

The most prominent source of data on discrimination in the U.S. housing market comes from a series of large national fair housing audits conducted by the U.S. Department of Housing and Urban Development in 1977, 1989, 2000, and 2012. These studies are notable for their scale, including covering a large number of metropolitan areas and many distinct measures of treatment. For instance, in the 2000 HUD study, 2264 black–white rental and sales tests were carried out in 16 metro areas with large African American/black populations, such as Atlanta, Chicago, Detroit, Philadelphia, and New York CityFootnote 1 (Turner et al. 2002, Exhibit 2–3). In the 2012 HUD study, 2999 black–white tests took place in 26 metro areas, including many of the same cities sampled in the 2000 study (Turner et al. 2013, p. 15).

In the HUD studies, trained testers, who are white, black, Hispanic/Latino, Asian, and Native American, pose as prospective renters or home buyers (e.g., Turner et al. 2013). In most audits, two auditors (one white, one minority) apply for the same advertised housing unit. The members of each audit pair are given profiles that make them equally desirable renters or home buyers. For example, in the 2000 HUD study, the auditors were matched with respect to age, gender, marital status, number of children, length of time at the current residence, reasons for moving, household income, and credit history (Turner et al. 2002; Turner and Ross 2003a, b). After each audit is completed, the white and minority auditors independently kept track of how they were treated: e.g., whether the advertised unit was still available, whether other similar units were available, the number of available units recommended, etc. (Turner et al. 2002). The authors of the HUD studies developed estimates of discrimination by comparing treatment across the white and minority auditors.

The HUD audit studies have shown that the most egregious forms of discrimination have declined over time, but they have also not disappeared. The 1977 and 1989 audits found much higher discrimination than the 2000 and 2012 audits, especially in basic forms of access like being told that units are available and being allowed to see the advertised unit (Wienk et al. 1979; Turner et al. 1991, 2002, 2013; Turner and Ross 2003a, b). In the 1977 HUD study, only the white tester was told that the advertised apartment unit is available in 30% of cases, while only the black tester was told that the unit is available in 11% of cases, for a net differential treatment measure of 19% (Wienk et al. 1979). By 2000, the net differential treatment measure for this outcome declined to 4%, with whites favored in 12% of cases and blacks favored in 8% of cases (Turner et al. 2002). There has been less of a decline in some of the more subtle forms of discrimination, such as in the number of units that auditors are shown (Turner et al. 2013).

The HUD studies are a crucial resource. Nevertheless, there are reasons to check the conclusions of the HUD audits by using the results of field experiments conducted independently of the HUD studies. The HUD studies use face-to-face audits, which involve trained human testers who interact with real market agents (e.g., landlords, homeowners). A concern with these audits is the possibility that auditors may, perhaps even subconsciously, act in ways that help produce the expected outcome (of discrimination against the minority auditors). As Heckman and Siegelman (1993) have noted, this sort of motivation may inflate observed levels of discrimination in studies that employ in-person auditors.

More generally, we know that the results of field experiments can be highly sensitive to the specific procedures employed by researchers (Gerber and Green 2012). It is difficult to estimate the generalizability of findings from the HUD studies, which use similar research designs, without also assessing the findings of studies that employed different research designs and procedures. There is now a significant body of studies that were not conducted by HUD that can be used to examine levels of discrimination in the housing market (e.g., Fang et al. 2018; Friedman et al. 2010; Hanson and Hawley 2011; Hanson and Santas 2014). Many of these studies are conducted as correspondence tests, which do not involve the use of human auditors. Instead, researchers often submit inquiries regarding rental ads or homes that are listed for sale (e.g., whether the unit or home is still available, whether they can see the unit); race is often signaled via the use of a racially/ethnically distinctive name in an email (e.g., Carpusor and Loges 2006) and/or accent in a phone call (Massey and Lundy 2001). Some of the non-HUD studies also include many applications and metropolitan areas; for instance, Hanson and Hawley (2011) submit inquiries from fictitious individuals to landlords posting rental ads on Craigslist in cities including Atlanta, Boston, Chicago, Dallas, and San Francisco. In short, the non-HUD studies allow for an important check on results from the HUD discrimination audits.

Mortgage Discrimination

In the mortgage discrimination literature, there are only a few field experiments focusing on racial or ethnic discrimination (Hanson et al. 2016; Ross et al. 2008). Because applying for a loan requires the use of detailed financial information (e.g., credit scores) that can be publicly verified, it is difficult to perform a face-to-face audit with fictitious white and minority borrowers which goes beyond the initial phase of inquiry about loans.

Mortgage discrimination studies have predominately estimated discrimination using survey and administrative data to model the approval or cost of a loan as a function of borrower characteristics (e.g., Black et al. 1978; Delis and Papadopoulos 2018; Holloway 1998; Miller 1988). Studies examine if borrowers from different races and ethnicities with similar financial characteristics have had their loans approved with equal frequency and equivalent terms (e.g., interest rates). Racial gaps among comparable or equally qualified borrowers are taken as evidence of discrimination—a method often called the “residual method” to measure racial and ethnic discrimination. This method is not as good as the field experimental method, which allows for experimental control of confounding variables, because it is difficult to ensure that all relevant non-race characteristics have been accounted for (Gerber and Green 2012; National Research Council 2004). But in cases where the controls are fairly complete, this is still a good method that has the important advantage that it can often be applied in situations where field experiments are difficult or impossible, and to outcomes closer to the outcomes of interest (e.g., loan approval/denial).

Most of the studies in this literature draw heavily on a large body of loan-level data which include borrower characteristics (e.g., race/ethnicity, income) that lenders are required to gather and report under the 1975 Home Mortgage Disclosure Act (HMDA). The 1989 revision of the law requires reporting on the disposition of individual applicants. HMDA reporting standards require the following information: (1) date of application; (2) loan amount; (3) Census Tract of property; (4) if the property is owner occupied; (5) purpose of the loan (purchase, improvement, or refinancing); (6) loan guarantee (conventional, FHA, or VA); (7) loan disposition (approved, approved but withdrawn, no lender action taken, or denied); (8) race; (9) gender; and (10) applicant income.

Unfortunately, the HMDA data lack some important indicators of creditworthiness such as the credit scores of borrowers. This is a problem if credit history is correlated with both race/ethnicity and application outcome. To address this issue, studies have often linked the loan-level data in the HMDA (e.g., race, loan outcome) with other loan characteristics (e.g., credit history, unit type) available in proprietary datasets such as Dataquick, Loan Performance, and others (Bayer et al. 2018; Bocian et al. 2008; Haughwout et al. 2009). HMDA data have shown that black applicants are denied loans at twice the rate of white applicants, controlling for income and gender; however, when the models account for additional indicators of credit worthiness (debt, down payments, and credit scores), black applicants are still 60% more likely to be denied (Cloud and Galster 1993). While the disadvantage shrinks, it still remains at significant levels.

As far as we know, no studies have combined the results across mortgage discrimination studies to look at long-term trends in mortgage discrimination. Delis and Papadopoulos (2018) analyze HMDA data to assess whether levels of discrimination in loan approval and pricing changed before and after the financial crisis. In the pre-crisis period (2004–2007), they found evidence that whites received loans at a lower cost than minorities did, although this effect disappears in the post-crisis period (2008–2013).

Past Reviews of Changes in Housing and Mortgage Discrimination

Several narrative reviews summarize the literature on housing discrimination and discuss trends, considering trends by comparing results from the four major HUD audits (e.g., Turner et al. 2013; Oh and Yinger 2015). We know of only four efforts at quantitative summaries of the literature on housing discrimination. Two of these are recent studies that use formal meta-analytic methods and are international in scale. Two earlier reviews of discrimination studies by Galster (1990a, b) employ elements of the quantitative review approach, although he does not employ a model of study results. We know of no prior meta-analyses focused on mortgage lending discrimination.

Flage (2018) examines housing discrimination based on 25 correspondence studies in North America, Europe, and Israel from 2006 to 2017. His data and results largely reflect the European context: 70% of their studies and about 90% of discrimination estimates in their analyses are from European studies. Moreover, his study does not examine change over time; he treats his data as representing one period.

Auspurg et al. (2019) examine housing discrimination in Europe and North America in 71 field experiments. Their survey covers many countries and a time range of about 40 years. The authors find evidence of declining discrimination overall in their pooled sample. Looking at North America (Canada and the U.S.) their point estimates suggests declining discrimination, although this decline is not statistically significant. A limitation of their research design is their use of a composite measure of discrimination, which pools measures of discrimination from many different outcomes (e.g., racial differences in positive replies, reply length, nature of treatment, etc.). This may mask differences in trends across distinct outcomes. As suggested by some theories of racism, it is plausible that more explicit or overt forms of discrimination have declined (e.g., refusing to show units to minority applicants), whereas more subtle forms of discrimination have not (e.g., racial differences in the number of units recommended).

Two earlier reviews by Galster (1990a, b) employ elements of a formal quantitative review without a formal meta-analysis model. Galster reviewed a number of small-scale discrimination audits conducted by fair housing organizations in the 1970s and 1980s and summarizes their results numerically. He finds significant discrimination, evidence that discrimination declined from the 1970s to the 1980s, and great variability in findings across audits.Footnote 2

Understanding Housing and Mortgage Discrimination in the U.S., 1976–2016

We aim to answer the following questions:

  1. (1)

    What is the average level of discrimination in housing and mortgage markets?

  2. (2)

    How do levels of discrimination vary over target groups and outcomes?

  3. (3)

    What are the time trends in these forms of discrimination?

In contrast to the Auspurg et al. (2019) meta-analysis that examined trends in housing discrimination internationally, we focus on assessing time trends in the more limited context of the U.S. housing market. This has the advantage of much greater homogeneity in terms of context and outcomes than the Auspurg meta-analysis, which averaged discrimination across many types of outcomes and countries. To the best of our knowledge, our study is the first meta-analysis to assess time trends in discrimination in the U.S. mortgage lending market.


To address these questions, we combine results from a large number of studies of housing and mortgage lending discrimination using meta-analysis statistical methods. Compared to a traditional literature review, the quantitative meta-analysis approach has the advantage in that it puts the emphasis on the magnitude of effects for reasonably similar outcomes across studies, while narrative reviews tend to emphasize patterns of statistical significance and the direction of effects (a procedure often called “vote counting” in the meta-analysis literature; see Hedges and Olkin 1980 for a discussion of problems inherent in vote counting). We can also use the data to examine trends over time by forming a time series of study results, an approach that has been used previously to study trends in employment discrimination (see Quillian et al. 2017; for a commentary on the method, see Ross 2017).


In the housing discrimination literature, we sought to include all studies that used field experimental methods to study racial and ethnic discrimination in the U.S. housing market. In the mortgage discrimination literature, we found that there were only two experimental studies (Hanson et al. 2016; Ross et al. 2008). Because this is too few to assess trends over time, we instead compiled analyses of observational data that looked at racial disparities in mortgage outcomes (loan approval or cost) while controlling for non-racial borrower characteristics. To locate studies meeting these criteria, we first conducted searches in the bibliographic databases Google Scholar, ProQuest Sociological Abstracts, ProQuest Dissertations, and NBER Working Papers. For the housing literature, we used search terms including “racial discrimination in housing,” “housing discrimination,” and “racial discrimination in apartment renting.” For the mortgage literature, we used search terms including “racial discrimination in mortgage loans” and “racial discrimination in mortgage lending.” As a second method to find studies, we examined the references cited in studies we located through bibliographic search and in several reviews and meta-analyses we were aware of. Through these search methods we located 16 field experiments of housing discrimination and 19 observational studies of discrimination in the mortgage market that are the basis of our analysis.Footnote 3

Coding and Outcome Measures

To quantitatively code these studies, we first read a selection of studies to develop a coding rubric that defined basic variables used in our analysis. We then coded each study using the rubric and performed our analysis.

For purposes of our analysis, we coded African American and black as a single category and Latino, Chicano, and Hispanic as a single category. Anglo or Caucasian was coded as white. We use African American and black, and Latino and Hispanic interchangeably.

For the housing discrimination analysis, the basic measure we use as the outcome is the difference in the percentage of positive responses received by the majority group minus the percentage of positive responses received by the minority group in the field experiment (equivalently, it is the percentage of pairs in which the white auditor is favored minus the percentage in which the minority auditor is favored). This is a measure widely used in housing discrimination studies, including the HUD studies (e.g., Wienk et al. 1979; Turner et al. 2002). Intuitively, it can be interpreted as the net percentage of applications in which the majority group received an advantage. Because of the experimental design of studies in our housing discrimination analysis, further controls are not needed for this measure to be an unbiased estimate of racial discrimination.

Formally, let \(c_{\text{w}}\) be the number of positive responses received by native whites, and \(c_{\text{m}}\) be the number of positive responses received by the target minority groups (e.g. African Americans), and \(n_{\text{w}}\) be the number of inquiries submitted by white applicants, and \(n_{\text{m}}\) be the number of inquiries submitted by minority group members. The discrimination difference (y) is \(\frac{{c_{\text{w}} }}{{n_{\text{w}} }} - \frac{{c_{\text{m}} }}{{n_{\text{m}} }}\).Footnote 4

For the mortgage discrimination analysis, we use studies that examine racial disparities in loan approval rates and in the cost of receiving mortgages. We examine how the mortgage outcome (loan denial or measures of loan cost) is related to race and ethnicity net of controls. A measure that is recommended in the meta-analysis literature for this purpose is the partial correlation. In our case this is the partial correlation of the race variable and the outcome variable (i.e., the correlation between the two variables net of controls). This can be calculated from most tables of regression results using the formula (see Stanley and Doucouliagos 2012; Aloe 2014):

$$r = \frac{t}{{\sqrt {t^{2} + {\text{d}}f} }},$$

where “t” is the t-statistic for a regression coefficient indicating race, “df” is the degrees of freedom of the regression, and r is the partial correlation measure for race. Larger partial correlations indicate more discrimination (race more strongly predicts loan rejection or a high-cost mortgage net of controls). The standard error of this statistic can also be calculated from results commonly available in regression tables (see Stanley and Doucouliagos 2012, chapter 2).Footnote 5

Most of the studies in our sample use linear regression or linear probability models, for which the method above gives the partial correlation directly. Some studies use logistic or probit regression. In these cases we also apply this formula above to the regression results (substituting z score for t score when z scores are reported). This may be interpreted as an estimate of the partial correlation of the logit or probit transformed probability of the outcome. The independent variables we use include the year the data were gathered and the race or ethnicity of the target group. When year covers a range of years in a study, we use the average year of the range for analysis purposes.

Specific Outcomes

In the housing discrimination analysis, we examine four specific outcomes: whether the housing agent or landlord responds to an initial request for information or an appointment, whether the advertised unit is still available for rent or purchase, the number of units or homes recommended to the renter or buyer, and the number of units or homes inspected. Table 1 provides a breakdown of the number of studies including each outcome for each target group in the rental and sales markets.Footnote 6

Table 1 Counts of studies by outcome and minority group for housing discrimination analysis

In the mortgage analysis, we classify outcomes into two categories: loan denial/approval and loan cost. Loan denial is based on regressions predicting whether a loan applied for was denied (vs. approved). Loan cost is based on regressions in which interest rate or high-cost mortgage, conditional on loan approval, is the outcome. We generally coded results from the most complete model or the model that seemed to be the author’s preferred model based on the text. Most studies use the HMDA data that have information on loan and borrower characteristics like income, age, and race. Some studies also add merged information on borrower credit characteristics like credit scores, allowing for matching on more attributes by race (Bayer et al. 2018; Bocian et al. 2008; Haughwout et al. 2009). We estimate results both overall and separately for studies with credit scores. Table 2 shows the count of all studies and of studies including borrower cost information.

Table 2 Counts of studies by outcome and minority group for mortgage discrimination analysis

Statistical Model

We use methods from the meta-analysis literature, the branch of statistics concerned with combining results across multiple studies, to conduct our analysis (for an introduction see Borenstein et al. 2009). We model the outcome of housing or mortgage discrimination as a function of the target group (black, Hispanic/Latino, Asian, other) and measures for the year.

We use random effects meta-analysis methods for our statistical analysis (see Raudenbush 2009). Random effects incorporates study-level random factors in the model. This means that models both use data from the individual cases in each study (e.g., people applying for housing and mortgages) and also have a second level capturing influences at the study level. For instance, this would capture factors like differences in the exact cities used in housing discrimination studies.

In practice, using a random-effect model has the effect of substantially increasing standard errors and making tests of significance more conservative compared to methods that ignore random influences at the study level. To reach significance, a random-effect model requires both reasonably large within study sample size and reasonable consistency across studies.

More formally, random effects meta-analysis models the outcome by assuming that it is normally distributed around the population mean level of the outcome, θ. If \(y_{i}\) is the discrimination measure (the net racial difference in treatment for the housing market or the partial correlation with race for the mortgage market) for the ith study, then the meta-analysis model is:

$$y_{i} = \theta + u_{i} + e_{i} , \;{\text{where}}\;u_{i} \sim N\left( {0, \tau^{2} } \right)\;{\text{and}}\;e_{i} \sim N\left( {0, \sigma_{i}^{2} } \right)$$

There are no predictor variables in this base model, just an average effect and a random effect for study-level variation. Here \(\tau^{2}\) is the between study variance, estimated as part of the meta-analysis model based on variation in discrimination across studies, while \(\sigma_{i}^{2}\) is the variance of the effect (outcome) in the ith study. The variance of the study effect (outcome) measure is estimated from statistics reported in each study. For the discrimination difference, the effect variance is estimated using outcome counts and formulas for the variance of the difference in proportions (Borenstein et al. 2009, formula 5.16), adjusted for paired applications when appropriate. For the partial correlation, the effect variance is estimated using formulas in Stanley and Doucouliagos (2012) section 2.3.3. The between study variance is estimated from the residual variation in study outcomes not accounted for by random sampling variation; estimation is by restricted maximum likelihood. The estimated average effect across studies (\(\theta\)) is estimated by averaging study effects with study weights equal to \(1/\left( {\tau^{2} + \sigma_{i}^{2} } \right)\), giving less weight to studies with noisy estimates.

Meta-regression adds predictors to this framework (Raudenbush 2009; Borenstein et al. 2009). It allows us to model the discrimination outcome as a function of one or more characteristics of the studies plus (in the random-effects specification) residual study-level heterogeneity (between study variance not explained by the covariates). In this paper, we use study year as the main predictor, allowing us to estimate the average change over time in the outcome from studies done in various years.

The model assumes that the study-level heterogeneity follows a normal distribution around the linear predictor:

$$y_{i} = \beta_{0} + \beta_{1} x_{i} + u_{i} + e_{i} , \;{\text{where}}\;u_{i} \sim N\left( {0, \tau^{2} } \right)\;{\text{and}}\;e_{i} \sim N\left( {0, \sigma_{i}^{2} } \right)$$

where \(\beta_{0}\) is the intercept, \(x_{i}\) is year for the ith study and \(\beta_{1}\) is the slope of year. The estimation is by restricted maximum likelihood.

Interpreting Statistical Significance

The point estimates we present are based on combining quantitative measures of outcomes across studies. They can be regarded as giving best estimates from the literature about the extent of discrimination in the housing and mortgage markets on average and whether it has declined over time.

We also present inferential statistics for these estimates. These are based on a random effects meta-analysis model relating the data to an underlying hypothetical population of studies of mortgage and housing discrimination, from which our studies are randomly selected. Statistically significant results suggest a high level of confidence that the null hypothesis (e.g., of no change in discrimination) can be rejected in this larger body of hypothetical studies. Even without significance, however, the point estimates represent a “best guess” based on the existing body of studies, and thus we still view them as informative.


The Housing Market

We begin by documenting the extent of discrimination in the housing market: discrimination by landlords, real estate agents, and sellers. This has been the focus of the HUD studies and other experimental studies.

Table 3 shows the levels of discrimination by target group and by outcome, averaged with random effects meta-analysis weights over all the studies in our data. The outcome measure is the discrimination difference, or the percentage of audits in which the white applicant is favored in the outcome minus the percentage in which the minority applicant is favored (equivalently, the success rate for whites minus the success rate for minorities). A discrimination difference of zero means no discrimination, and larger numbers (to the maximum of 100%) indicate more net discrimination against minorities. Negative numbers mean that the minority applicants have a net advantage.

Table 3 Mean discrimination difference in housing outcomes by group and market

The results are shown across four outcomes: landlord or agent responds to inquiry, advertised unit is available, number of units recommended, and number of units inspected. The first two outcomes are available from a variety of studies, including the HUD studies of housing discrimination but also several non-HUD studies (some of which are correspondence studies). The second two outcomes are available from the HUD studies and a few other in-person audits somewhat similar to the HUD studies.

Comparing across groups, Table 3 shows that housing discrimination is generally high against blacks, moderate against Hispanics, and relatively low against Asians. There is also very high discrimination against Arab Americans, although this estimate is based on only two studies. All four minority groups appear to experience some discrimination. For example, compared to equally qualified white applicants, the probability of receiving a response to an initial inquiry is 8% points lower among blacks, 4% points lower among Hispanics, and 3% points lower among Asians. In both the rental and sales markets, the level of discrimination is usually lower for more exclusionary forms of discrimination (not receiving a response and availability of advertised unit), compared to more subtle forms of differential treatment (e.g., in number of units recommended/inspected).Footnote 7

How has discrimination in the housing market changed over time? Table 4 and Figs. 1, 2, and 3 show results to address this question. We estimate change over time in two ways. First, we divide the data into two periods: 2005 or earlier and after 2005. We show the mean level of discrimination in each period. This divides the data roughly in half in terms of studies, and allows for a clear description of averages in the two halves of the data. Second, we compute the slope coefficient of year as a predictor of discrimination from a random effects meta-regression. The slope gives average change in discrimination per year across the entire period.Footnote 8

Table 4 Change in housing discrimination over time
Fig. 1

a Discrimination against blacks in the rental market before and after 2005. b Discrimination against blacks in the sales market before and after 2005

Fig. 2

a Discrimination against Hispanics in the rental market before and after 2005. b Discrimination against Hispanics in the sales market before and after 2005

Fig. 3

a Discrimination against Asians in the rental market before and after 2005. b Discrimination against Asians in the sales market before and after 2005

The results indicate that housing discrimination has decreased over time. In general, the post-2005 estimates are considerably lower than the pre-2005 estimates. For the response and availability outcomes, the declines are by about 10% points in the rental market and 5% points in the sales market, leaving us with low levels of discrimination in the post-2005 period.

Discrimination generally remains at high levels post-2005 in the number of units recommended and inspected, which indicates that there has been a smaller decline in discrimination for these outcomes. The slopes showing average yearly change generally show a similar pattern. For African Americans after 2005, there remains a gap of 10% or more in the number of units or homes recommended compared to whites.

These trends are generally similar to the trends that are observed when the results of only HUD studies are compared across time (e.g., Turner et al. 2013). Our results confirm that this holds when combined with the body of information from other studies—non-HUD studies make up most of the studies for the response and availability outcomes in the rental market.

To further confirm the similarity of the patterns in the HUD and non-HUD studies, we calculated the slope of annual change using only the non-HUD studies for the response and availability outcomes for the rental market. This is shown (when sufficient data are available) in the far-right column of Table 4. The general pattern of decline holds when using only the non-HUD studies, and the results are fairly similar to those of HUD studies. Both the HUD and non-HUD studies present similar patterns of change over time in housing discrimination.

The Mortgage Market

What are the trends in discrimination in the mortgage market? In Table 5 we address this question. There are two outcomes: whether a mortgage application is denied and the cost of the mortgage received, conditional on approval. The results are not based on experimental data for reasons explained previously, but instead rely on the “residual method” of using racial or ethnic gaps after controls for relevant non-racial factors (e.g., income) as measures of discrimination.

Table 5 Mean mortgage discrimination outcomes by group

Table 5 shows the measures of discrimination by race or ethnicity of the borrower. The number shown is the partial correlation between minority race and the outcome (controlling for other variables) times 100 (putting it on a 0–100 scale), with higher numbers indicating more discrimination. The left column shows the measure for all available studies in the analysis.

Results are shown for blacks, Hispanics, blacks and Hispanics pooled (some of the older studies combined black and Hispanics into one group; e.g., Tootell 1996), and Asians. There is evidence of anti-minority discrimination in both loan denial and mortgage cost. Discrimination appears to be more severe against blacks than against Hispanics, with low discrimination against Asians.Footnote 9

The right columns show estimates only for the smaller number of studies that control for loan-level data on borrower race with official credit information like borrower credit scores. The measures of discrimination are on average lower compared to the studies without these controls, but only by a little.

Table 6 shows changes in the mortgage discrimination outcomes over time. The results are graphed in Fig. 4. Results are shown for blacks, Hispanics, and merging all black, Hispanic, and combined black-Hispanic effect sizes together.Footnote 10 Results are shown for the full set of studies; the studies including borrower credit information are insufficient to allow an analysis of change over time separately for these studies.Footnote 11

Table 6 Change in mortgage discrimination black and hispanic borrowers, all studies
Fig. 4

Racial disparity in loan denial and cost before and after 2005

The results show some evidence of a decline in the racial gap in loan approval rates over time. However, the decline is relatively small; it is much smaller than the observed declines in housing discrimination. For Hispanics, the racial difference in the loan approval rate relative to whites in our data increased after 2005.

For mortgage cost, the before–after 2005 breakdown shows either no change or a slight increase in racial disparities in the cost of a mortgage. For discrimination against blacks and blacks and Hispanics (when pooled), the slope coefficients of year are negative, but they are not statistically significant.

Discrimination in the mortgage market primarily shows stability over time, with modest or no declines in racial differences for loan approval/rejection or mortgage cost. There is much less change in discrimination than in the housing market. Black and Hispanic home seekers continue to be rejected at higher rates than whites with similar characteristics and are also more likely to receive high-cost mortgage products.


We find a sharp decline in two measures of discrimination in housing access: racial differences in responses to inquiries and availability of the unit advertised. The decline in these forms of discrimination is evident in both the HUD and non-HUD studies. In the 1980s and 1990s, it was quite common for black and Hispanic auditors to be told that a unit is no longer available and then for a white auditor to be subsequently shown the unit. This stark form of discrimination, which involves lying to the applicant about what is available, now occurs much less often than it did in the 1980s and 1990s.

We find evidence of some decline, but to a much lesser degree, in the types of discrimination that involve comparative differences in treatment among white, black, and Hispanic auditors. White auditors are still generally recommended more units and succeed in inspecting more units than their equally qualified minority counterparts; this suggests that while anti-discrimination efforts have been successful in reducing more overt forms of discrimination, they may have been less effective at curbing more subtle differences in how white and minority home seekers are treated.

In the mortgage market, we find that black and Hispanic borrowers are more likely to be rejected when they apply for a loan and are more likely to receive a high-cost mortgage, conditional on loan approval. Our meta-analysis of studies of racial disparities in the mortgage market suggests that discrimination in loan denial and cost has not declined much over the previous 30–40 years, a disturbing finding.

A limitation of our mortgage analysis is the reliance on studies that use observational data and the residual method to assess discrimination. It is possible that other factors not in these studies may account for some or all of the remaining racial gap (for a discussion of this problem in the context of one prominent study, see Ross and Yinger 2002, chapter 5). However, contrasting studies with more and less complete data on borrowers in our analysis did not produce much change in discrimination estimates (see Table 5). The lingering racial disparities could be partly accounted for by patterns like the heavier marketing of high-cost mortgage products by lenders in minority communities (which could be viewed as resulting from discrimination in lender targeting of clients).

The reduction in the most exclusionary forms of housing discrimination suggests that in most cases discrimination will not block persistent efforts by black or Hispanic households to move into white or affluent neighborhoods. However, we believe that more subtle forms of discrimination will steer households with weaker neighborhood preferences (which we suspect are most households) toward own-race neighborhoods, helping to maintain residential segregation. Housing discrimination restricts the housing choices of black and Hispanic households and increases their level of housing insecurity.

In the mortgage market, persistent discrimination means that black and Hispanic households are less likely to be able to secure mortgages and are more likely to pay higher mortgage costs (e.g., Bayer et al. 2018; Bocian et al. 2008). This contributes to the reduced accumulation of home equity for black and Hispanic households (Faber and Ellen 2016). In a recent case study of Baltimore by Rugh et al. (2015), the authors found that black borrowers in their sample paid an additional 5–11% more in monthly payments, and that this contributed to the loss of more than $2 million in home equity from foreclosures. Because home equity is the major source of wealth accumulation for most households in the U.S., this form of discrimination inhibits the upward social mobility of minorities and exacerbates large racial disparities in wealth.

These results also indicate that it is incorrect to view all types of discrimination as following the same time trends. The data reveal a more complex picture where the most egregious forms of discrimination that involve acts of categorical exclusion have declined—like a real estate agent or landlord telling an applicant that a unit is no longer available when in fact it is available. More subtle forms of discrimination have declined as well, but less so, and remain at significant levels. This is consistent with theoretical accounts that emphasize persistent covert and subtle forms of discrimination even as the most explicit forms of discrimination have declined. To the extent that anti-discrimination enforcement is one factor accounting for the decline of explicit forms of discrimination, the Fair Housing Act has been successful. However, white applicants are still given more options and overall better treatment in the housing search process, and their advantages in mortgage pricing and availability have not decreased. In sum, the results suggest that anti-discrimination enforcement in the housing and mortgage markets should continue, and efforts should be increased to ensure that all home seekers receive equal treatment regardless of their race or ethnic background.

Data Availability

The data and code used in this paper are available online at: http://sites.northwestern.edu/dmap.


  1. 1.

    There were 1152 black–white rental tests and 1112 black–white sales tests (Turner et al. 2002, Exhibit 2–3).

  2. 2.

    We include two studies that were published separately from Galster’s summaries. We were not able to include other audits he summarizes either because the outcomes reported are not comparable to other studies we consider or because critical information like sample sizes are not reported.

  3. 3.

    A bibliographic list of these studies is included in the Online Appendix.

  4. 4.

    Another option would be the ratio of white success rates to minority success rates, a number described as the “discrimination ratio” in Quillian et al. (2019). There are two reasons we use the difference measure in this manuscript rather than the discrimination ratio. First, several studies in the housing literature do not differentiate between “neither received a positive response” and “both received a positive response” in reporting results of paired studies, and without this information the ratio measure cannot be calculated. Second, because the rate of positive responses (to both groups) in housing studies tends to be much higher than in employment studies, the ratio runs into upper bound issues when the overall response rate is high. This is less of a problem for difference measures.

  5. 5.

    Often when correlations are outcomes in meta-analyses they are transformed by taking Fischer’s z-transformation to give them a more symmetric distribution. However, Stanley and Doucouliagos (2012) argue this is only necessary when some of the correlations are large, getting close to − 1 or 1. None of our correlations in absolute value are above .2, and so we perform the meta-analysis using the more interpretable raw correlations.

  6. 6.

    We do not include steering toward own-race neighborhoods as an outcome because this outcome is available in too few of the discrimination audits outside of the HUD studies.

  7. 7.

    We also performed standard publication bias tests using the trim- and-fill procedure (see Borenstein et al. 2009, chapter 30). The results showed evidence of publication bias only for a couple of outcomes for Hispanics. The procedure is limited by the small sample of studies.

  8. 8.

    Scatterplots showing the regression lines are shown in the Online Appendix.

  9. 9.

    We also performed a standard publication bias test using the trim- and-fill procedure (see Borenstein et al. 2009, chapter 30). The results showed evidence of publication bias for the mortgage cost outcome but not the loan denial outcome. Breaking down the results before and after 2005 did not produce evidence of publication bias in mortgage cost for either sub-period.

  10. 10.

    The number of studies is sometimes larger for the pooled analysis than the sum of blacks and Hispanics separately because this includes both studies focused on blacks or Hispanics and some studies using a combined black-Hispanic group.

  11. 11.

    This is due to the small number of studies with credit information and the fact that studies including borrower credit are all concentrated in a short period of time. The studies with borrower credit characteristics that examine loan denial are from 1990 or earlier, while those that examine mortgage cost use data from 2000 to 2008.


  1. Aloe, A. M. (2014). An empirical investigation of partial effect sizes in meta-analysis of correlational data. The Journal of General Psychology,141(1), 47–64. https://doi.org/10.1080/00221309.2013.853021.

    Article  Google Scholar 

  2. Auspurg, K., Schneck, A., & Hinz, T. (2019). Closed doors everywhere? A meta-analysis of field experiments on ethnic discrimination in rental housing markets. Journal of Ethnic and Migration Studies,45(1), 95–114. https://doi.org/10.1080/1369183X.2018.1489223.

    Article  Google Scholar 

  3. Bayer, P., Ferreira, F., & Ross, S. L. (2018). What drives racial and ethnic differences in high-cost mortgages? The role of high-risk lenders. The Review of Financial Studies,31(1), 175–205.

    Article  Google Scholar 

  4. Black, H., Schweitzer, R. L., & Mandell, L. (1978). Discrimination in mortgage lending. The American Economic Review,68(2), 186–191.

    Google Scholar 

  5. Bobo, L., Kluegel, J. R., & Smith, R. A. (1997). Laissez-Faire racism: The crystallization of a kinder, gentler, antiblack ideology. In S. A. Tuch & J. K. Martin (Eds.), Racial attitudes in the 1990s: Continuity and change. Westport, CT: Praeger Publishers.

    Google Scholar 

  6. Bocian, D. G., Ernst, K. S., & Li, W. (2008). Race, ethnicity and subprime home loan pricing. Journal of Economics and Business,60(1–2), 110–124.

    Article  Google Scholar 

  7. Bonilla-Silva, E. (2009). Racism without racists: Color-blind racism and the persistence of racial inequality in America. Lanham, MD: Rowman & Littlefield Publishers.

    Google Scholar 

  8. Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to meta-analysis. West Sussex, UK: John Wiley & Sons.

    Google Scholar 

  9. Bostic, R. W., & Surette, B. J. (2001). Have the doors opened wider? Trends in homeownership rates by race and income. The Journal of Real Estate Finance and Economics,23(3), 411–434.

    Article  Google Scholar 

  10. Carpusor, A. G., & Loges, W. E. (2006). Rental discrimination and ethnicity in names. Journal of Applied Social Psychology,36(4), 934–952. https://doi.org/10.1111/j.0021-9029.2006.00050.x.

    Article  Google Scholar 

  11. Choi, A., Herbert, K. Winslow, O., & Browne, A. (2019). Long Island divided. Newsday (Long Island), November 17, 2019. https://projects.newsday.com/long-island/real-estate-agents-investigation/

  12. Cloud, C., & Galster, G. (1993). What do we know about racial discrimination in mortgage markets? The Review of Black Political Economy,22(1), 101–120. https://doi.org/10.1007/BF02689922.

    Article  Google Scholar 

  13. Delis, M. D., & Papadopoulos, P. (2018). Mortgage lending discrimination across the U.S.: New methodology and new evidence. Journal of Financial Services Research. https://doi.org/10.1007/s10693-018-0290-0

    Article  Google Scholar 

  14. Faber, J. W., & Ellen, I. G. (2016). Race and the housing cycle: Differences in home equity trends among long-term homeowners. Housing Policy Debate,26(3), 456–473. https://doi.org/10.1080/10511482.2015.1128959.

    Article  Google Scholar 

  15. Fang, A. H., Guess, A. M., & Humphreys, M. (2018). Can the government deter discrimination? Evidence from a randomized intervention in New York City. The Journal of Politics,81(1), 127–141. https://doi.org/10.1086/700107.

    Article  Google Scholar 

  16. Firebaugh, G., & Farrell, C. R. (2016). Still large, but narrowing: The sizable decline in racial neighborhood inequality in metropolitan America, 1980–2010. Demography,53(1), 139–164. https://doi.org/10.1007/s13524-015-0447-5.

    Article  Google Scholar 

  17. Flage, A. (2018). Ethnic and gender discrimination in the rental housing market: Evidence from a meta-analysis of correspondence tests, 2006–2017. Journal of Housing Economics,41, 251–273. https://doi.org/10.1016/j.jhe.2018.07.003.

    Article  Google Scholar 

  18. Friedman, S., Squires, G. D., & Galvan, C. (2010). Cybersegregation in Boston and Dallas: Is Neil a more desirable tenant than tyrone or jorge?. Washington, DC: Department of Sociology, University at Albany, SUNY and Department of Sociology, George Washington University.

    Google Scholar 

  19. Galster, G. (1990a). Racial discrimination in housing markets during the 1980s: A review of the audit evidence. Journal of Planning Education and Research,9(3), 165–175.

    Article  Google Scholar 

  20. Galster, G. (1990b). Racial steering in urban housing markets: A review of the audit evidence. The Review of Black Political Economy,18(3), 105–129.

    Article  Google Scholar 

  21. Gerber, A. S., & Green, D. P. (2012). Field experiments: Design, analysis, and interpretation (1st ed.). New York: W. W. Norton.

    Google Scholar 

  22. Hanson, A., & Hawley, Z. (2011). Do landlords discriminate in the rental housing market? Evidence from an internet field experiment in US cities. Journal of Urban Economics,70(2), 99–114. https://doi.org/10.1016/j.jue.2011.02.003.

    Article  Google Scholar 

  23. Hanson, A., Hawley, Z., Martin, H., & Liu, B. (2016). Discrimination in mortgage lending: Evidence from a correspondence experiment. Journal of Urban Economics,92, 48–65. https://doi.org/10.1016/j.jue.2015.12.004.

    Article  Google Scholar 

  24. Hanson, A., & Santas, M. (2014). Field experiment tests for discrimination against hispanics in the U.S. rental housing market. Southern Economic Journal,81(1), 135–167. https://doi.org/10.4284/0038-4038-2012.231.

    Article  Google Scholar 

  25. Haughwout, A., Mayer, C., & Tracy, J. (2009). Subprime mortgage pricing: the impact of race, ethnicity, and gender on the cost of borrowing. Brookings-Wharton Papers on Urban Affairs, 33–63.

  26. Heckman, J. J., & Siegelman, P. (1993). The Urban Institute audit studies: Their methods and finding. In M. Fix & R. J. Struy (Eds.), Clear and convincing evidence: Measurement of discrimination in America (pp. 187–258). Lanham, MD: Urban Institute Press.

    Google Scholar 

  27. Hedges, L. V., & Olkin, I. (1980). Vote-counting methods in research synthesis. Psychological Bulletin,88(2), 359–369.

    Article  Google Scholar 

  28. Holloway, S. R. (1998). Exploring the neighborhood contingency of race discrimination in mortgage lending in Columbus, Ohio. Annals of the Association of American Geographers,88(2), 252–276. https://doi.org/10.1111/1467-8306.00093.

    Article  Google Scholar 

  29. LaCour-Little, M. (1999). Discrimination in mortgage lending: A critical review of the literature. Journal of Real Estate Literature,7(1), 15–49. https://doi.org/10.5555/reli.7.1.v4m86h814810j233.

    Article  Google Scholar 

  30. Logan, J. R. (2011). Separate and unequal: The neighborhood gap for blacks, hispanics and Asians in metropolitan America. NYC: Russell Sage Foundation.

    Google Scholar 

  31. Logan, J. R. (2013). The persistence of segregation in the 21st century metropolis. City & Community,12(2), 160–168. https://doi.org/10.1111/cico.12021.

    Article  Google Scholar 

  32. Massey, D. S., & Lundy, G. (2001). Use of Black English and racial discrimination in urban housing markets: New methods and findings. Urban Affairs Review,36(4), 452–469. https://doi.org/10.1177/10780870122184957.

    Article  Google Scholar 

  33. Miller, S. M. (1988). Discrimination in the home-mortgage accept-reject decision: The case of hartford. Housing Finance Review,7, 19–29.

    Google Scholar 

  34. National Research Council. (2004). Measuring racial discrimination. Washington, D.C.: The National Academies Press.

    Google Scholar 

  35. Oh, S. J., & Yinger, J. (2015). What have we learned from paired testing in housing markets? Cityscape,17(3), 15–59.

    Google Scholar 

  36. Quillian, L., Heath, A., Pager, D., Midtbøen, A., Fleischman, F., & Hexel, O. (2019). Do some countries discriminate more than others? Evidence from 97 field experiments of racial discrimination in hiring. Sociological Science,6, 467–496.

    Article  Google Scholar 

  37. Quillian, L., Pager, D., Hexel, O., & Midtbøen, A. H. (2017). Meta-analysis of field experiments shows no change in racial discrimination in hiring over time. Proceedings of the National Academy of Sciences,114(41), 10870–10875.

    Article  Google Scholar 

  38. Raudenbush, S. (2009). Analyzing effects sizes: Random effects coding. In H. M. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis. New York, NY: Russell Sage Foundation.

    Google Scholar 

  39. Ross, S. L. (2017). Measuring trends in discrimination with field experiment data. Proceedings of the National Academy of Sciences,114(41), 10815. https://doi.org/10.1073/pnas.1715243114.

    Article  Google Scholar 

  40. Ross, S. L., Turner, M. A., Godfrey, E., & Smith, R. R. (2008). Mortgage lending in Chicago and Los Angeles: A paired testing study of the pre-application process. Journal of Urban Economics,63(3), 902–919. https://doi.org/10.1016/j.jue.2007.07.006.

    Article  Google Scholar 

  41. Ross, S. L., & Yinger, J. (2002). The color of credit: Mortgage discrimination, research methodology, and fair-lending enforcement. Cambridge, MA: MIT Press.

    Google Scholar 

  42. Rugh, J. S., Albright, L., & Massey, D. S. (2015). Race, space, and cumulative disadvantage: A case study of the subprime lending collapse. Social Problems,62(2), 186–218. https://doi.org/10.1093/socpro/spv002.

    Article  Google Scholar 

  43. Schuman, H., Steeh, C., Bobo, L. D., & Krysan, M. (1997). Racial attitudes in America: Trends and interpretations. Cambridge, MA: Harvard University Press.

    Google Scholar 

  44. Sears, D. O. (1988). Symbolic racism. In P. A. Katz & D. A. Taylor (Eds.), Eliminating racism. Boston: Springer.

    Google Scholar 

  45. Sharkey, P. (2014). Spatial segmentation and the black middle class. American Journal of Sociology,119(4), 903–954. https://doi.org/10.1086/674561.

    Article  Google Scholar 

  46. Stanley, T. D., & Doucouliagos, H. (2012). Meta-regression analysis in economics and business (1st ed.). London: Routledge.

    Google Scholar 

  47. Tootell, G. M. B. (1996). Redlining in Boston: Do mortgage lenders discriminate against neighborhoods? The Quarterly Journal of Economics,111(4), 1049–1079. https://doi.org/10.2307/2946707.

    Article  Google Scholar 

  48. Turner, M., & Ross, S. (2003a). Discrimination in metropolitan housing markets: Phase 2—Asians and Pacific Islanders. Washington, D.C.: U.S. Department of Housing and Urban Development.

    Google Scholar 

  49. Turner, M., & Ross, S. (2003b). Discrimination in metropolitan housing markets: Phase 3—Native Americans. Washington, D.C.: U.S. Department of Housing and Urban Development.

    Google Scholar 

  50. Turner, M., Ross, S., Galster, G., & Yinger, J. (2002). Discrimination in metropolitan housing markets: Phase 1. Washington, D.C.: U.S. Department of Housing and Urban Development.

    Google Scholar 

  51. Turner, M. A., Santos, R., Levy, D. K., Wissoker, D., Aranda, C., & Pitingolo, R. (2013). Housing discrimination against racial and ethnic minorities 2012. Washington, D.C.: Urban Institute.

    Google Scholar 

  52. Turner, M., Struyk, R., & Yinger, J. (1991). Housing discrimination study synthesis. Washington, DC: U.S. Department of Housing and Urban Development.

    Google Scholar 

  53. Vigdor, J. L. (2013). Weighing and measuring the decline in residential segregation. City & Community,12(2), 169–177. https://doi.org/10.1111/cico.12023.

    Article  Google Scholar 

  54. Wienk, R. E., Reid, C. E., Simonson, J. C., & Eggers, F. J. (1979). Measuring discrimination in American Housing Markets: The housing market practices survey. Washington, D.C.: U.S. Department of Housing and Urban Development.

    Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Lincoln Quillian.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Supplementary material 1 (DOCX 112 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Quillian, L., Lee, J.J. & Honoré, B. Racial Discrimination in the U.S. Housing and Mortgage Lending Markets: A Quantitative Review of Trends, 1976–2016. Race Soc Probl 12, 13–28 (2020). https://doi.org/10.1007/s12552-019-09276-x

Download citation


  • Discrimination
  • Housing
  • Mortgage
  • Race
  • Trends