The Price of Denial: Examining the Health Consequences of Mortgage Lending Bias in Michigan Metro Areas.

The Price of Denial: Examining the Health Consequences of Mortgage Lending Bias in Michigan Metro Areas - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice J Urban Health . 2025 Jun 11;102(3):483–494. doi: 10.1007/s11524-025-00985-w The Price of Denial: Examining the Health Consequences of Mortgage Lending Bias in Michigan Metro Areas Jamison Koeman 1 , Wenchu Pan 2 , Alexa Eisenberg 1 , Lu Wang 2 , Roshanak Mehdipanah 1, ✉ Author information Article notes Copyright and License information 1 Department of Health Behavior and Health Equity, School of Public Health, University of Michigan, 1415 Washington Heights, Ann Arbor, MI 48109-2029 USA 2 Department of Biostatistics, School of Public Health, University of Michigan, Ann Arbor, MI USA ✉ Corresponding author. Accepted 2025 May 6; Issue date 2025 Jun. © The New York Academy of Medicine 2025. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. PMC Copyright notice PMCID: PMC12279676  PMID: 40500567 Abstract Anti-Black mortgage lending discrimination (MLD) has resulted in homeownership disparities among White and Black households. This study examines the association between MLD and health opportunities, based on various neighborhood determinants of health (e.g., employment, education, and housing status), in Michigan’s three largest metropolitan areas. Multilevel models using data from the Home Mortgage Disclosure Act and Census from 2015 to 2019 were used to examine the Black-White odds ratio of mortgage denial. An association analysis was performed to examine the relationship between the Black-White odds ratio of mortgage denial and neighborhood health opportunities. Loan denial for Black mortgage applicants was 1.99 (95% CI1.87, 2.12) times more likely than for White applicants. Neighborhoods with greater anti-Black MLD had fewer harmful environmental exposures, greater health insurance coverage, and better housing and socio-economic conditions. Anti-Black MLD remains a common practice and supports the assertion that Black home-seekers experience greater MLD in neighborhoods with better opportunities for health compared to White home-seekers, who access these benefits more freely. Greater protections against MLD—including increasing oversight of the data reporting requirements and improving mortgage access for Black households—are needed to promote health equity. Keywords: Mortgage lending, Discrimination, Racism, Social determinants of health, Health equity Background Established in 1968, the Fair Housing Act outlawed discrimination by direct providers of housing (e.g., landlords, real estate companies, banks or other lending institutes) based on race, religion, sex, familial status, disability, and national origin [ 1 ]. However, weak enforcement mechanisms have enabled housing discrimination to continue despite the passage of laws like the Home Mortgage Disclosure Act of 1975 requiring mortgage lending reporting [ 1 , 2 ]. Widespread anti-Black mortgage lending discrimination persists in the U.S. [ 3 ]. In 2020, across the U.S., Black mortgage applicants were about twice as likely to get denied as White applicants [ 4 ]. Studies at the metropolitan area level, including Atlanta, Los Angeles, and New York, estimated that, compared to White mortgage applicants, Black applicants were about 1.46 to 4.94 times as likely to have their application denied [ 5 , 7 ]. Besides having lower access to mortgage lending in general, Black households are more likely than White households to receive higher cost, higher risk loans, which make Black households more vulnerable to foreclosure [ 8 , 9 ]. Mortgage discrimination has contributed to the growing Black-White homeownership gap, resulting from structural anti-Black racism in the U.S., including a legacy of housing policies that advanced homeownership for White Americans while denying it to Black Americans [ 10 ]. Homeownership, important in the U.S. as a vehicle for wealth and health opportunities, is far more accessible to White households than Black households [ 10 ]. Compared to a White homeownership rate of about 70%, the Black homeownership rate is around 40% [ 11 , 12 ], and this 30-point White-Black homeownership gap has been nearly stagnant since the 1960 s [ 4 ]. Similarly, in 2016, the median White family owned ten times as much wealth as the median Black family [ 13 ]. Mortgage Lending and Health Equity A mortgage loan is one of the primary pathways to homeownership. Homeownership not only has economic benefits, but also potential health benefits. Globally, studies suggest that owning a home is associated with better mental health and child behavioral development than renting [ 14 , 17 ]. Pollack found that renting unaffordable housing had a higher strain on health than owning unaffordable housing, suggesting that homeownership may have a protective effect against the negative health impacts of unaffordability [ 15 ]. Housing discrimination, including mortgage lending discrimination, has been linked to barriers in education, employment and economic growth, neighborhood segregation, exposure to environmental hazards, and reduction of housing choice to areas with fewer opportunities, including decreased access to healthy foods, schools, and healthcare centers [ 18 , 20 ]. In addition, higher neighborhood-level anti-Black mortgage lending discrimination has been associated with worse overall and Black-specific health outcomes. Areas where Black households had a higher odds of loan denial than White households had worse overall and Black-specific cancer outcomes [ 21 , 22 ]. Other studies showed areas with higher anti-Black mortgage lending discrimination had lower overall breast cancer mortality [ 5 ] and lower Black-specific preterm birth [ 23 ], and two studies show no association with pregnancy health and breast cancer incidence [ 6 , 24 ]. Mixed results are likely due to the presence of both adverse and protective characteristics in areas with higher levels of discrimination, which are more likely to be higher income and predominantly White. Alternatively, these outcomes can also be influenced by uncertainty regarding the most impactful life-course periods of exposure to discrimination. Discriminatory experiences and their effects on health may not be accurately captured in cross-sectional analyses. For instance, the Uncertain Geographic Context Problem refers to the difficulty and potential for bias in studying the effects of place-based factors on individual health outcomes due to complex spatial and temporal dimensions of those contexts [ 25 ]. More affluent neighborhoods, for example, may increase one’s proximity to health-promoting amenities (e.g., green spaces, healthier food options, well-funded schools). However, they can also expose Black households to psychosocial stressors (e.g., unfair treatment, othering) while reducing access to social resources that are critical for countering the effects of racism [ 6 ]. This is consistent with the rental housing discrimination literature, where one study identified that the potential mental health benefits of using a housing choice voucher to move into a lower poverty neighborhood could be offset by housing discrimination [ 26 ]. This study adapted existing methodology developed by Mendez and colleagues to evaluate anti-Black mortgage lending discrimination across the three largest metropolitan areas in Michigan and is significant for two main reasons [ 24 ]. First, while existing studies at the neighborhood level are limited to Milwaukee, Atlanta, Philadelphia, and California, no literature exists on neighborhood-level anti-Black mortgage lending discrimination across the Detroit, Lansing, and Grand Rapids metropolitan areas, which are home to 5,980,829 households or 60% of the state’s households [ 27 ]. The redlining practices of the 1930 s were instrumental in shaping the development of American cities, including the establishment of suburbs in the 1960 s and 1970 s [ 10 ]. The discriminatory lending practices and others that followed led to greater investment into the development of predominantly White suburban areas, while denying loans and mortgages to largely Black communities in urban centers, including Lansing, Grand Rapids, and Detroit [ 10 ]. From there, deindustrialization and globalization contributed to a decline in manufacturing jobs, and remaining employment opportunities moved to the suburbs, taking economic resources away from the cities. This process continued to deny homeownership to Black families while creating opportunities for White families. The large homeownership gap in Michigan, as well as the history of racial discrimination among these metropolitan areas, makes this location appropriate for our study. As a result, these areas also account for nearly 80% of Black households in a state where the Black-White homeownership gap—43 vs. 78%—is larger than that of the nation as a whole [ 11 , 12 ]. In particular, Detroit, the most populous city in Michigan and the second largest majority Black city in the country, was hit especially hard by the foreclosure crisis [ 28 ]. Between 2009 and 2013, fewer than 375 mortgages were issued per year [ 28 ]. Second, prior studies on anti-Black mortgage lending discrimination and health have expressed a need for more research to understand mediating factors between anti-Black mortgage lending discrimination and health outcomes [ 5 , 22 , 24 ]. This study seeks to explain the lack of consensus about the relationship between anti-Black mortgage lending discrimination and health present in the current literature. The goal was to better understand whether anti-Black mortgage lending discrimination takes away health opportunities from individuals by denying them the chance to live in areas that are more conducive to good health. The results of this study are significant for policymakers, the real estate community, and housing advocates. Methods Data Source Our study sample includes the Lansing-East Lansing, Grand Rapids-Kentwood, and Detroit-Warren-Dearborn Metropolitan Statistical Areas (MSAs). We used the Home Mortgage Disclosure Act (HMDA) data from the Consumer Financial Protection Bureau (CFPB) website from 2015 to 2019 to analyze mortgage denial rates across the census tracts in the Lansing, Grand Rapids, and Detroit MSAs [ 29 ]. With all but the smallest institutions required to report, HMDA data captures about 90% of mortgage lending activity [ 29 ] and includes details of each mortgage loan application, demographic and financial information about the applicant. We also used the 2019 5-Year American Community Survey (ACS) data from the census tract where the applicant was seeking to purchase a home [ 30 ]. In addition, to examine social determinants of health factors, we used census-tract data from the 2021 Environmental Justice Screening and Mapping Tool (EJSCREEN) data, which includes the 2019 National Air Toxics assessment data [ 31 ]. Table 1 provides all of the variables included, along with their source, year, and description, as well as the descriptive analyses described below. We used census tract level as a proxy for neighborhood to capture the variability across geographic areas within our metropolitan areas, as this was the finest spatial scale at which all data in this analysis was available. Table 1. Description, source, and year of variables included in this study Characteristic Source and year Description Homeownership ACS a 2015–2019 Percentage of houses occupied by owners (among all housing occupied by either renter or owner) Occupied housing ACS 2015–2019 Percentage of occupied homes (among all housing units either vacant or occupied) Employment ACS 2015–2019 Percentage of employed residents (among total labor force) High school education ACS 2015–2019 Percentage of residents with a high school diploma (among total population) Health insurance coverage ACS 2015–2019 Percentage of adults with public or private insurance (among total adult population) Commute time ACS 2015–2019 Percentage of workers with a commute time less than 30 min (among all workers who commute to work) Diesel PM exposure NATA b 2019 PM10 emissions from on-road and nonroad mobile sources burning diesel or residual fuels (US EPA 2019). Based on annual average ambient outdoor concentration, human activity patterns, demographic features, and microenvironmental factors (US EPA 2019) Ozone exposure EJSCREEN c 2018 Ozone summer seasonal average of daily maximum 8-h concentration in air in parts per billion PM 2.5 exposure EJSCREEN 2018 PM2.5 levels in air, ug/m3 annual average House built pre-1960 EJSCREEN 2016–2020 Percent of housing units built pre-1960, as indicator of potential lead paint exposure Traffic Proximity EJSCREEN 2019 Count of vehicles (AADT, average annual daily traffic) at major roads within 500 m divided by distance in meters Wastewater discharge EJSCREEN 2019 RSEI modeled toxic concentrations at stream segments within 500 m, divided by distance in kilometers Superfund proximity EJSCREEN 2022 Count of proposed or listed NPL – also known as superfund – sites within 5 km (or nearest one beyond 5 km), each divided by distance in kilometers RMP facility proximity EJSCREEN 2022 Count of Risk Management Plan (RMP) (potential chemical accident management plan) facilities within 5 km (or nearest one beyond 5 km), each divided by distance in kilometers Hazardous waste proximity EJSCREEN 2022 Count of hazardous waste facilities (TSDFs and LQGs) within 5 km (or nearest beyond 5 km), each divided by distance in kilometers Underground storage tanks EJSCREEN 2022 Count of LUSTs (multiplied by a factor of 7.7) and the number of USTs within a 1,500-foot buffered block group Open in a new tab a ACS—American Community Survey; b NATA—National Air Toxics Assessment; c EJSCREEN—Environmental Justice Screening and Mapping Tool; Inclusion and Exclusion Criteria The HMDA dataset included 1,588,152 initial observations from 2015 to 2019. We only included mortgage applications that were for home purchase loans of one-to-four family dwellings, secured by first lien, and owner-occupied as a principal dwelling and excluded loans purchased by a lending institution. Because our analysis compares denial rates for Black and White applicants, we only included mortgage applications where the primary applicant was either non-Hispanic Black/African American or non-Hispanic White, referred to as Black or White hereafter. We removed the 1.5% of observations with missing data (unknown gender is considered missing), because the missing pattern was random, and the missing percentage was low. To improve the quality of our mortgage lending index, we only used data from census tracts with at least five White and five Black primary applicants studied, because in census tracts with too few observations, the contrast between White and Black applicants was not sufficient to provide useful information in identifying the mortgage lending discrimination. A sensitivity analysis was done to ensure that the inclusion criteria did not change the significance of any of the results. Our final dataset included 252,863 observations from 921 census tracts. Statistical Modeling To analyze anti-Black mortgage lending discrimination, we developed a multilevel logistic regression model with mixed effects to account for within-census tract correlation, where we evaluate the effects of individual-level features, census-tract-level features, and the interaction between the applicant’s race and census-tract-level features. The full model was as follows: log p ij 1 - p ij = β 0 + μ 0 i + β 1 + μ 1 i Rac e ij + β 2 + μ 2 i Income _ Loan _ Rati o ij + β 3 Gender ij + β 4 Co _ Applicant ij + β 5 Economy _ Statu s i + β 6 Segregation _ Inde x i + β 7 A vailable _ Housing i + α 1 Race ij ∗ Segregation Inde x i + α 2 Race ij ∗ Economy Statu s i , where subscript i represents the label of census tract, j represents the jth applicants within the census tract, p ij is the probability that a mortgage application gets denied for the jth applicant in census tract i , and the μ ’s represent the census-tract specific effect shift from overall average that counts for those unmeasured confounder features. In the HMDA data, we considered a mortgage loan application denied only when the variable named “action_taken” was equal to 3, meaning “application denied.” The individual-level features in the model included race, income-loan ratio (winsorized at the 1 st and 99 th percentile), gender, and whether having a co-applicant. Census-tract-level features included racial composition (represented by the percentage of Black population within the tract, categorized by quantile to reflect potential non-monotone effect), local economy (represented by the ratio of median annual income comparing census tract to MSA), and available housing (represented by the number of 1-to-4 family units in the census tract, in the scale of 1000 family units per unit). Mortgage Lending Discrimination Index To summarize census-tract level status on mortgage lending discrimination, we used the result from the above model to create a mortgage lending discrimination index, which was defined as the odds ratio of getting denied in the mortgage loan application comparing a Black primary applicant to White primary applicant in a certain census tract, controlling for income-loan ratio, gender, co-applicant status, and available housing. The formula of calculating the mortgage discrimination index for a census tract is β 1 + μ 1 i + α 1 Segregation _ Index i + α 2 Economy _ Status i Association Between Mortgage Lending Discrimination Index and Social Determinants of Health Variables To determine risk factors for neighborhood-level anti-Black mortgage lending discrimination and provide evidence for the pathway between anti-Black mortgage lending discrimination and health opportunities, we fit a linear regression model to analyze the association between the mortgage lending discrimination index and social determinants of health variables, which is equivalent to an ANOVA F-test. These variables were included due to their relationship with health. We categorized these features by state quantile (worse than state median, within state 50–75th percentile, better than state 75th percentile) for the association analysis. Statistical significance was set at 0.05 for all analytic models. Data were analyzed using R version 4.2.0, where we used dplyr, flextable, gtsummary, and ggplot2 for the plots; lme4 and nlme for the model; and leaflet, usmap, and tidycensus for the interactive map. Results Table 2 shows the summary statistics of individual-level features, including loan-associated features (loan outcome and amount) and applicant features (income, gender and whether having a co-applicant), stratified by the race of the applicant, Black or White. There were more White applicants than Black applicants overall. Among Black applicants, the proportion of women to men was half, whereas among White applicants, women made up one third of the applicant pool compared to men. White applicants generally had more successful loan applications, applied for higher loan amounts, had higher gross annual income, and had a higher probability of applying with a co-applicant. Table 2. Descriptive statistics for individual-level features, stratified by MSA/MD Year Overall Lansing, East Lansing Detroit, Warren, Dearborn Grand Rapids, Wyoming Race White N = 2 22,605 1 Black/African American N = 30,258 1 White N = 16,008 1 Black/African American N = 1432 1 White N = 162,965 1 Black/African American N = 26,599 1 White N = 43,632 1 Black/African American N = 2227 1 Loan outcome Obtained conventional mortgage 206,515.0 (92.8%) 25,104.0 (83.0%) 15,067.0 (94.1%) 1251.0 (87.4%) 149,918.0 (92.0%) 21,888.0 (82.3%) 41,530.0 (95.2%) 1965.0 (88.2%) Obtained high-cost mortgage 117.0 (0.1%) 12.0 (0.0%) 8.0 (0.0%) 0.0 (0.0%) 96.0 (0.1%) 10.0 (0.0%) 13.0 (0.0%) 2.0 (0.1%) Poor credit denial 10,234.0 (4.6%) 3444.0 (11.4%) 615.0 (3.8%) 128.0 (8.9%) 8400.0 (5.2%) 3,134.0 (11.8%) 1219.0 (2.8%) 182.0 (8.2%) Denial for other reasons 5739.0 (2.6%) 1,698.0 (5.6%) 318.0 (2.0%) 53.0 (3.7%) 4,551.0 (2.8%) 1567.0 (5.9%) 870.0 (2.0%) 78.0 (3.5%) Loan amount (thousand dollars) 160.0 (120.0) 135.0 (100.0) 135.0 (102.0) 125.0 (100.0) 163.0 (124.0) 135.0 (101.0) 165.0 (101.0) 145.0 (77.0) Gross annual income (thousand dollars) 68.0 (57.0) 58.0 (41.0) 64.0 (54.0) 56.0 (43.2) 68.0 (59.0) 59.0 (42.0) 65.0 (55.0) 50.0 (37.0) Gender Female 71,057 (31.9%) 15,235 (50.4%) 6015 (37.6%) 612 (42.7%) 51,822 (31.8%) 13,739 (51.7%) 13,220 (30.3%) 884 (39.7%) Male 151,548 (68.1%) 15,023 (49.6%) 9993 (62.4%) 820 (57.3%) 111,143 (68.2%) 12,860 (48.3%) 30,412 (69.7%) 1343 (60.3%) Coapplicant No 141,867 (63.7%) 24,789 (81.9%) 9839 (61.5%) 1087 (75.9%) 107,556 (66.0%) 22,129 (83.2%) 24,472 (56.1%) 1573 (70.6%) Yes 80,738 (36.3%) 5469 (18.1%) 6169 (38.5%) 345 (24.1%) 55,409 (34.0%) 4470 (16.8%) 19,160 (43.9%) 654 (29.4%) Open in a new tab 1 n (%); median (IQR) Results of the multivariable multilevel mixed effects regression model (Table 3 ) demonstrate how the odds (OR) of mortgage loan application denial are associated with each covariate in the model, controlling for other confounders. Lower odds of mortgage application denial were found significantly for applicants who are White, have a higher income-loan ratio, have a co-applicant, live in a census tract with higher local income, or live in a census tract with more available housing. The odds of mortgage loan denial for Black applicants was 98.9% higher (OR = 1.704 × 1.167 = 1.989; 95% confidence interval (CI)) 1.865, 2.121) than those for White applicants, for census tracts with a population that is less than 25% Black and for which the Census-to-MSA annual median income ratio is 1, controlling for gender, income-loan ratio, co-applicant status, and the number of 1 to 4 family units. Table 3. Result of the multivariable mixed-effect regression model; bold indicates significance Characteristics OR 1 SE 1 95% CI 1 p -value 2 (Intercept) 0.165 0.010 (0.146, 0.185) < 0.001*** Race Black vs. White 1.704 0.115 (1.493, 1.945) < 0.001*** Income-loan ratio 0.646 0.028 (0.593, 0.704) < 0.001*** Gender Female vs. male 0.976 0.015 (0.947, 1.006) 0.119 Co-applicant 0.879 0.015 (0.850, 0.910) < 0.001*** Census-to-MSA/MD income 0.758 0.026 (0.708, 0.811) < 0.001*** Black population percentage 25–50% vs. 0–25% 0.956 0.031 (0.897, 1.018) 0.160 50–75% vs. 0–25% 1.259 0.081 (1.110, 1.427) < 0.001*** 75–100% vs. 0–25% 1.724 0.118 (1.507, 1.973) < 0.001*** Number of 1–4 family units 0.886 0.021 (0.846, 0.928) < 0.001*** Race × census-to-MSA/MD income Black × census-to-MSA/MD income 1.167 0.060 (1.054, 1.291) 0.003** Race × Black population percentage Black × 25–50% vs. 0–25% 1.139 0.055 (1.036, 1.251) 0.007** Black × 50–75% vs. 0–25% 0.968 0.070 (0.840, 1.116) 0.658 Black × 75–100% vs. 0–25% 0.913 0.068 (0.790, 1.056) 0.220 Open in a new tab 1 OR odds ratio, SE standard error, CI confidence interval 2 * p < 0.05; ** p < 0.01; *** p < 0.001 When considering the interaction, the odds ratio for mortgage loan denial related to race (Black vs. White) is further increased among the census tracts whose Black population is within 25–50% by 13.9% ( p = 0.007), which indicates a stronger race discrimination in mortgage lending than the census tracts whose Black population was less than 25%. More specifically, controlling for the other variables in the model, for census tracts whose census-to-MSA annual median income ratio is 1, where the population is 0–25%, 25–50%, 50–75%, and 75–100% Black, the Black-White odds ratio of mortgage loan denial was 1.989, 2.265, 1.925, and 1.816, respectively. This indicates that before reaching a threshold where the White population percentage is too high, anti-Black mortgage lending discrimination is more common when the share of the White population is higher. Further, there is an effect modification by the ratio of annual median income comparing census tract to MSA on the Black-White odds ratio of mortgage loan denial, suggesting that anti-Black mortgage lending discrimination is more common in higher income areas. More specifically, for every 1 unit increase in the ratio of annual median income comparing census tract to MSA, the odds of mortgage loan denial for Black applicants increased by 16.7% (OR = 1.167, p = 0.003). Mortgage Lending Discrimination Index Figure 1 a shows the demographic distribution of the mortgage lending discrimination index. The gray regions are census tracts that did not meet our inclusion criteria of at least five White applicants and five Black applicants. For the 921 census tracts that are included in the final analysis, the mortgage lending discrimination index ranges between 1.37 and 3.2, with a mean of 2.07 and a median of 2.04. Figure 1 b presents the median income ratio and Fig. 1 c shows the percentage of the Black population. Similar trends between anti-Black mortgage lending discrimination, mortgage lending discrimination, income, and race were highlighted in these figures, and our analysis demonstrates that anti-Black mortgage lending discrimination is common across Michigan’s metropolitan areas. Fig. 1. Open in a new tab Geographical distribution of census-level mortgage-lending discrimination index ( a ), median income ratio ( b ), and percentage of Black population ( c ) Association Analysis Results of association analysis between the mortgage lending discrimination index and census-tract level social determinants of health (Table 4 ) show that, except for the count of risk management plan (RMP) facility proximity, all features are significantly associated with the mortgage lending discrimination index. Neighborhoods with higher anti-Black mortgage lending discrimination had fewer harmful environmental exposures (including traffic proximity, wastewater discharge, ozone exposure, underground storage tanks, homes built pre-1960), greater health insurance coverage, and better housing and social conditions (including homeownership, occupied housing, education, employment). Table 4. Summary result of association between discrimination index and census-tract level determinants of health variables Characteristic Quantile Quantile range Mean of index Sd p -value d Homeownership Worse than state median 0–0.75 2.00 0.31 4.43E-15 Within state 50–75 th percentile 0.75–0.88 2.11 0.31 Better than state 75 th percentile 0.88–1.00 2.21 0.29 Occupied housing Worse than state median 0–0.97 1.97 0.29 1.31E-18 Within state 50–75 th percentile 0.97–0.99 2.10 0.30 Better than state 75 th percentile 0.99–1.00 2.18 0.32 Employment Worse than state median 0.46–0.95 1.94 0.28 8.02E-28 Within state 50–75 th percentile 0.95–0.97 2.14 0.31 Better than state 75 th percentile 0.97–1.00 2.19 0.29 High school education Worse than state median 0.38–0.92 1.93 0.28 5.29E-51 Within state 50–75 th percentile 0.92–0.95 2.11 0.27 Better than state 75 th percentile 0.95–1.00 2.27 0.29 Health insurance coverage Worse than state median 0–0.94 1.97 0.29 1.20E-20 Within state 50–75 th percentile 0.94–0.96 2.09 0.31 Better than state 75 th percentile 0.96–1.00 2.20 0.30 Commute time Worse than state median 0–0.67 2.05 0.30 1.21E-17 Within state 50–75 th percentile 0.67–0.77 1.98 0.29 Better than state 75 th percentile 0.77–1.00 2.25 0.30 Diesel PM exposure Worse than state median 0.09–0.28 2.05 0.31 4.18E-07 Within state 50–75 th percentile 0.06–0.09 2.19 0.29 Better than state 75 th percentile 0.01–0.06 1.97 0.17 Ozone exposure Worse than state median 44.65–46.88 2.02 0.30 1.76E-16 Within state 50–75 th percentile 43.51–44.65 2.14 0.32 Better than state 75 th percentile 32.12–43.51 2.36 0.23 PM 2.5 exposure Worse than state median 8.85–10.24 2.03 0.30 6.41E-17 Within state 50–75 th percentile 8.31–8.85 2.29 0.28 Better than state 75 th percentile 5.19–8.31 2.19 0.22 House built pre-1960 Worse than state median 0.34–1.00 1.97 0.30 1.89E-23 Within state 50–75 th percentile 0.18–0.34 2.08 0.27 Better than state 75 th percentile 0–0.18 2.20 0.30 Traffic proximity Worse than state median 561.2–12,283.4 2.03 0.31 1.10E-10 Within state 50–75 th percentile 129.9–561.2 2.18 0.30 Better than state 75 th percentile 1.6e-3–129.9 2.19 0.29 Wastewater discharge Worse than state median 6.5e-4–89.2 2.05 0.29 0.02 Within state 50–75 th percentile 3.1e-5–6.5e-4 2.10 0.32 Better than state 75 th percentile 6.4e-13–3.1e-5 2.14 0.30 Superfund proximity Worse than state median 0.06–2.54 2.07 0.31 8.04E-04 Within state 50–75 th percentile 0.04–0.06 2.01 0.30 Better than state 75 th percentile 0–0.04 2.15 0.33 RMP facility proximity Worse than state median 0.22–7.92 2.06 0.32 0.53 Within state 50–75 th percentile 0.11–0.22 2.09 0.31 Better than state 75 th percentile 0–0.11 2.08 0.25 Hazardous waste proximity Worse than state median 0.71–8.31 2.04 0.31 1.65E-05 Within state 50–75 th percentile 0.13–0.71 2.15 0.32 Better than state 75 th percentile 0–0.13 2.11 0.22 Underground storage tanks Worse than state median 4.54–84.78 2.01 0.30 3.06E-18 Within state 50–75 th percentile 0.70–4.54 2.17 0.29 Better than state 75 th percentile 0–0.70 2.27 0.31 Open in a new tab d the p -values are from the ANOVA F -test Discussion Our results demonstrate that anti-Black mortgage lending discrimination has remained a common practice for more than 55 years after the passage of the Fair Housing Act. The odds of mortgage denial were nearly twice as high for Black applicants compared to White applicants across the three major Metropolitan Areas in Michigan. Further, the mortgage lending discrimination index varied significantly by census tract, suggesting that some neighborhoods in Michigan have more anti-Black mortgage lending discrimination than others. While these findings support a wealth of literature describing the persistence of anti-Black mortgage lending discrimination [ 3 , 5 , 7 ], our study also shows that anti-Black mortgage lending discrimination could reduce access to various determinants of health that help promote better health among residents. Various approaches to using HMDA data have been developed in public health literature. In a 2024 systematic review, researchers showed that the majority of studies utilizing HMDA data typically used multilevel logistic regression models, similar to ours, with loan denials as the dependent variable and race/ethnicity as the primary independent variable, adjusting for confounders such as gender and income-to-loan ratio [ 32 ]. Major differences in these indices were how they defined their neighborhoods, from census tracts [ 24 , 33 ] to those created using adaptive spatial filtering to derive measures outside of administrative boundaries [ 21 ]. This was largely driven by the geographic availability of their health outcomes [ 32 ]. Our study followed Mendez and colleagues’ approach to focus on census tracts, which is the smallest spatial area provided by HMDA data, given our focus on determinants of health that are available at this level. To begin to understand the variance by census tract, the study identified several factors associated with higher neighborhood-level anti-Black mortgage lending discrimination. Our analysis highlights several potential mediating factors and generally supports the literature describing that higher anti-Black mortgage lending discrimination largely occurs in neighborhoods with better health and health-promoting factors [ 5 , 24 ]. In particular, our findings support the assertion that Black home-seekers experience higher mortgage discrimination in neighborhoods with fewer environmental health risks and better access to health care, housing, education, and employment, and are therefore more often denied opportunities for health and health-determining choices compared to White home-seekers, who can access these benefits more freely. However, there is inconsistency in the literature, and further study is needed to understand this complex relationship that includes the potential that higher exposure to discrimination in health-promoting neighborhoods could offset the health benefits for Black populations [ 6 , 26 ]. This study provides further evidence that systemic anti-Black racism is a structural driver of health inequity. One study by Lynch and colleagues showed that the ongoing legacies of historic redlining, sponsored by the government, have been demonstrated in current forms of lending disinvestment [ 34 ]. We also know that neighborhoods that were historically redlined had present-day determinants of health that were less conducive to health than neighborhoods that were not historically redlined [ 35 ]. Furthermore, neighborhoods that experienced historical redlining have been associated with poorer health outcomes, including preterm birth and stroke [ 36 , 37 ]. Therefore, contemporary anti-Black mortgage lending discrimination contributes to a legacy of poor consequences for health, socioeconomic welfare, and neighborhood stability. Limitations One major limitation of the study was our inability to include many census tracts due to low amounts of mortgage lending. This offers an important finding as well. As is clearly the case with Detroit, census tracts near the center of cities in our sample received very few mortgages despite having large populations. Neighborhoods in these areas tend to have a higher non-White population and lower homeownership than other parts of the metropolitan area. However, barriers to mortgage access exist beyond low homeownership and a limited demand for mortgages. In Detroit, credit score requirements are a barrier to a population from whom credit is repeatedly withheld, the quality of the housing stock deters investors, and low incomes limit purchasing power [ 28 , 35 ]. As a result, homeownership is often obtained without a mortgage through alternative means, including land contracts. A second limitation was that our neighborhood-level focus limits our understanding of individual factors that may influence the relationship between mortgage discrimination and health equity. Since the HMDA data do not include information about an applicant’s housing location during the application process or ultimate residence upon denial, we discuss loss of access to health determining factors based on the characteristics of the denied neighborhood, as opposed to the potential health impacts experienced because of the denial. We also recognized that neighborhood characteristics are not simply confounders but are conceptually interlinked with the historical and systemic factors we study. Future analyses may explicitly explore these pathways or utilize mediation analysis to better capture the complex interplay between mortgage lending discrimination, neighborhood characteristics, and health opportunities. Furthermore, the application process requires far more inputs, including credit score and housing condition, than what we have controlled for in our analysis. Credit score, however, is known to be a discriminatory metric due to a lack of access to credit and has been linked to health through various pathways, including health care access and socioeconomic status [ 38 ]. There are also far more opportunities for racism in the mortgage lending process than simply the evaluation of the mortgage application, including racially motivated steering, application discouragement, and referral to other lenders [ 1 , 18 ]. Our analysis did not account for the type of mortgage product, but research has demonstrated that higher risk loans are geographically concentrated in predominantly non-White neighborhoods [ 9 ]. Despite the limitations of our analysis, it is likely that anti-Black mortgage lending discrimination is widely occurring across Michigan’s three largest metropolitan areas. Public Health Implications Our findings show that anti-Black mortgage lending discrimination limits Black applicants’ access to neighborhoods with better health and health-promoting factors. Therefore, greater protections against mortgage lending discrimination—including increasing oversight of the HMDA data reporting requirements and improving mortgage access for Black households—are needed to promote health equity. To improve mortgage access for prospective Black homeowners, implementing measures such as alternative ways to measure creditworthiness, small-dollar mortgage loan access, rate-and-term refinancing, and credit and downpayment assistance programs are some potential approaches. Policies such as these are necessary to begin to dismantle systemic racism in the housing market and promote health equity. Acknowledgements We would like to thank the School of Public Health and the Poverty Solutions initiative for supporting this work. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. Massey DS. The legacy of the 1968 fair housing act. Sociol Forum . 2015;30(S1):571–88. 10.1111/socf.12178. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. McCoy P. The home mortgage disclosure act: a synopsis and recent legislative history. J Real Estate Res . 2007;29(4):381–98. 10.1080/10835547.2007.12091207. [ Google Scholar ] 3. Namin S, Zhou Y, Xu W, et al. Persistence of mortgage lending bias in the United States: 80 years after the Home Owners’ Loan Corporation security maps. J Race Ethn City . 2022;3(1):70–94. 10.1080/26884674.2021.2019568. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Choi JH, Mattingly PJ. What different denial rates can tell us about racial disparities in the mortgage market | Urban Institute . Urban Institute; 2022. Accessed April 8, 2024. https://www.urban.org/urban-wire/what-different-denial-rates-can-tell-us-about-racial-disparities-mortgage-market 5. Collin LJ, Gaglioti AH, Beyer KM, et al. Neighborhood-level redlining and lending bias are associated with breast cancer mortality in a large and diverse metropolitan area. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cosponsored Am Soc Prev Oncol . 2021;30(1):53–60. 10.1158/1055-9965.EPI-20-1038. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Michaels EK, Canchola AJ, Beyer KMM, Zhou Y, Shariff-Marco S, Gomez SL. Home mortgage discrimination and incidence of triple-negative and Luminal A breast cancer among non-Hispanic Black and non-Hispanic White females in California, 2006–2015. Cancer Causes Control . 2022;33(5):727–35. 10.1007/s10552-022-01557-y. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Beyer KMM, Laud PW, Zhou Y, Nattinger AB. Housing discrimination and racial cancer disparities among the 100 largest US metropolitan areas. Cancer . 2019;125(21):3818–27. 10.1002/cncr.32358. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Loya J, Flippen C. The Great Recession and ethno-racial disparities in access to mortgage credit. Soc Probl . 2021;68(4):1026–50. 10.1093/socpro/spaa040. [ Google Scholar ] 9. Calem PS, Gillen K, Wachter S. The neighborhood distribution of subprime mortgage lending. J Real Estate Finance Econ . 2004;29(4):393–410. 10.1023/B:REAL.0000044020.67401.51. [ Google Scholar ] 10. Rothstein R. The color of law: a forgotten history of how our government segregated America . Liveright Publishing Corporation; 2017. Accessed July 20, 2021. https://www.epi.org/publication/the-color-of-law-a-forgotten-history-of-how-our-government-segregated-america/ 11. Census Bureau. B25003B: Tenure (Black or African American alone household). 2024. Accessed April 16, 2024. https://data.census.gov/table?q=B25003B 12. Census Bureau. B25003A: Tenure (White alone household). 2024. Accessed April 16, 2024. https://data.census.gov/table?q=B25003A 13. McIntosh K, Moss E, Nunn R, Shambaugh J. Examining the Black-white wealth gap. Brookings. February 27, 2020. Accessed April 8, 2024. https://www.brookings.edu/articles/examining-the-black-white-wealth-gap/ 14. Ghimire J, Carswell AT, Ghimire R, Turner PR. The impact of U.S. housing type and residential living situations on mental health during COVID-19. Int J Environ Res Public Health . 2021;18(16):8281. 10.3390/ijerph18168281 [ DOI ] [ PMC free article ] [ PubMed ] 15. Pollack CE, Griffin BA, Lynch J. Housing affordability and health among homeowners and renters. Am J Prev Med . 2010;39(6):515–21. 10.1016/j.amepre.2010.08.002. [ DOI ] [ PubMed ] [ Google Scholar ] 16. O’Donnell J, Kingsley M. The relationship between housing and children’s socio-emotional and behavioral development in Australia. Child Youth Serv Rev . 2020;117: 105290. 10.1016/j.childyouth.2020.105290. [ Google Scholar ] 17. Park GR, Seo BK. Revisiting the relationship among housing tenure, affordability and mental health: do dwelling conditions matter? Health Soc Care Community . 2020;28(6):2225–32. 10.1111/hsc.13035. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Mehdipanah R. Without affordable, accessible, and adequate housing, health has no foundation. Milbank Q . 2023;101(S1):419–43. 10.1111/1468-0009.12626. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Turner MA, Santos R, Levy DK, Wissoker DA. Housing discrimination against racial and ethnic minorities 2012: full report. Urban Institute. June 4, 2016. Accessed February 10, 2018. https://www.urban.org/research/publication/housing-discrimination-against-racial-and-ethnic-minorities-2012-full-report 20. Christensen P, Sarmiento-Barbieri I, Timmins C. Housing discrimination and the toxics exposure gap in the United States: evidence from the rental market. Rev Econ Stat . 2022;104(4):807–18. 10.1162/rest_a_00992. [ Google Scholar ] 21. Beyer KMM, Zhou Y, Matthews K, Bemanian A, Laud PW, Nattinger AB. New spatially continuous indices of redlining and racial bias in mortgage lending: links to survival after breast cancer diagnosis and implications for health disparities research. Health Place . 2016;40:34–43. 10.1016/j.healthplace.2016.04.014. [ DOI ] [ PubMed ] [ Google Scholar ] 22. Zhou Y, Bemanian A, Beyer KMM. Housing discrimination, residential racial segregation, and colorectal cancer survival in southeastern Wisconsin. Cancer Epidemiol Biomarkers Prev . 2017;26(4):561–8. 10.1158/1055-9965.EPI-16-0929. [ DOI ] [ PubMed ] [ Google Scholar ] 23. Mendez DD, Hogan VK, Culhane JF. Institutional racism, neighborhood factors, stress, and preterm birth. Ethn Health . 2014;19(5):479–99. 10.1080/13557858.2013.846300. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Mendez DD, Hogan VK, Culhane J. Institutional racism and pregnancy health: using home mortgage disclosure act data to develop an index for mortgage discrimination at the community level. Public Health Rep . 2011;126(3_suppl):102–114. 10.1177/00333549111260S315 [ DOI ] [ PMC free article ] [ PubMed ] 25. Kwan MP. The uncertain geographic context problem. Ann Assoc Am Geogr . 2012;102(5):958–68. 10.1080/00045608.2012.687349. [ Google Scholar ] 26. Osypuk TL, Schmidt NM, Kehm RD, Tchetgen Tchetgen EJ, Glymour MM. The price of admission: does moving to a low-poverty neighborhood increase discriminatory experiences and influence mental health? Soc Psychiatry Psychiatr Epidemiol . 2019;54(2):181–90. 10.1007/s00127-018-1592-0. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Census Bureau. S0201 selected population profile in the United States. 2024. Accessed April 17, 2024. https://data.census.gov/table?q=total%20population&g=040XX00US26 28. Dewar M, Seymour E, Druță O. Disinvesting in the city: the role of tax foreclosure in Detroit. Urban Aff Rev . 2015;51(5):587–615. 10.1177/1078087414551717. [ Google Scholar ] 29. Consumer Financial Protection Bureau. Download HMDA data. Consumer Financial Protection Bureau. Accessed April 8, 2024. https://www.consumerfinance.gov/data-research/hmda/historic-data/ 30. United States Department of Commerce. The American Community Survey. Published online 2019. https://www2.census.gov/programs-surveys/acs/methodology/questionnaires/2019/quest19.pdf . Accessed 8 Apr 2024 31. US EPA O. EJScreen: environmental justice screening and mapping tool. September 3, 2014. Accessed April 8, 2024. https://www.epa.gov/ejscreen 32. Edwards JR, Ong C, Barber S, et al. Methodologic strategies for quantifying associations of historical and contemporary mortgage discrimination on population health equity: a systematic review. J Racial Ethn Health Disparities . Published online September 17, 2024:1–12. 10.1007/s40615-024-02137-3 [ DOI ] [ PMC free article ] [ PubMed ] 33. Gee GC. A multilevel analysis of the relationship between institutional and individual racial discrimination and health status. Am J Public Health . 2008;98(Supplement_1):S48-S56. 10.2105/AJPH.98.Supplement_1.S48 [ DOI ] [ PMC free article ] [ PubMed ] 34. Lynch EE, Malcoe LH, Laurent SE, Richardson J, Mitchell BC, Meier HCS. The legacy of structural racism: associations between historic redlining, current mortgage lending, and health. SSM - Popul Health . 2021;14: 100793. 10.1016/j.ssmph.2021.100793. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Mehdipanah R, McVay KR, Schulz AJ. Historic redlining practices and contemporary determinants of health in the Detroit metropolitan area. Am J Public Health . 2023;113(S1):S49–57. 10.2105/AJPH.2022.307162. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Krieger N, Van Wye G, Huynh M, et al. Structural racism, historical redlining, and risk of preterm birth in New York City, 2013–2017. Am J Public Health . 2020;110(7):1046–53. 10.2105/AJPH.2020.305656. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Jadow BM, Hu L, Zou J, et al. Historical redlining, social determinants of health, and stroke prevalence in communities in New York City. JAMA Netw Open . 2023;6(4): e235875. 10.1001/jamanetworkopen.2023.5875. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Dean LT, Nicholas LH. Using credit scores to understand predictors and consequences of disease. Am J Public Health . 2018;108(11):1503–5. 10.2105/AJPH.2018.304705. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Articles from Journal of Urban Health : Bulletin of the New York Academy of Medicine are provided here courtesy of New York Academy of Medicine ACTIONS View on publisher site PDF (900.1 KB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Modeling the dynamics of access to healthy food. The case of Cleveland, Ohio#1015755
• 12 U.S.C. § 2279aa-1 — Federal Agricultural Mortgage Corporation#564008
• Nearly 99% of 2025 Illinois Wesleyan Grads Employed or in Graduate Programs#810006
• Observation and Reflection of Neighborhood and Community 2026: Observation and Reflection of Disparities in Access to Healthy Foods, Education, and Stable Healthcare#169126
• 43 U.S.C. § 512 — Release of Government liens after contract with irrigation districts#596476