Historic redlining and the siting of oil and gas wells in the United States

Oil and gas development in residential areas exposes nearby residents to air and water pollutants, noise, and other stressors linked to cardiovascular disease, impaired lung function, adverse mental health, and adverse birth outcomes. Approximately 17 million U.S. residents live within 1.6 km of at least one active well, underscoring widespread potential exposure. Prior research documents higher concentrations of ozone, fine particulate matter, and volatile organic compounds near wells, as well as water contamination. Multiple studies report that racially and socioeconomically marginalized communities face heightened risks and disproportionately high exposure to oil and gas infrastructure and hazardous materials. The HOLC redlining process in the 1930s, which graded neighborhoods A (best) through D (hazardous, redlined), systematically penalized neighborhoods with Black and immigrant residents and those near industrial uses, contributing to persistent segregation and health disparities. Although worse HOLC grades have been associated with numerous adverse health outcomes and environmental disadvantages (e.g., less greenspace, higher heat), it has been unclear whether historic redlining influenced the siting of oil and gas wells or whether wells were disproportionately sited in neighborhoods that HOLC later graded poorly. This study asks whether neighborhoods that received worse HOLC grades have higher exposure to oil and gas wells compared to better-graded neighborhoods.

The paper synthesizes evidence showing: (1) proximity to oil and gas wells is associated with increased air pollution (ozone, PM2.5, VOCs) and water contamination; (2) proximity is linked to elevated risks for cardiovascular disease, respiratory impairment, anxiety, depression, preterm birth, and impaired fetal growth; and (3) these risks and exposures are often higher among racially and socioeconomically marginalized groups. It also summarizes research on HOLC redlining, which assigned worse grades to neighborhoods with Black and immigrant residents even after accounting for housing value/condition, and links poorer grades to adverse health outcomes (asthma, cancer, cardiovascular disease, heat-related illness, COVID-19, preterm birth, low birthweight) and environmental disadvantages (reduced greenspace, higher urban heat). Prior environmental justice studies have documented disproportionate siting of oil and gas infrastructure and flaring in marginalized communities, inverse associations with income for wastewater injection wells, and localized disparities in regions such as Los Angeles, Kern County (CA), Ohio, and Texas. The literature suggests structural racism shapes environmental exposures but had not conclusively linked HOLC grades to urban oil and gas well siting across multiple U.S. cities, motivating the present study.

Study design: The authors conducted a retrospective, cross-sectional analysis linking digitized HOLC appraisal maps (Mapping Inequality Project) with a national dataset of oil and gas wells from Enverus Drilling (coverage 1898–2021). The primary objective was to test whether neighborhoods with worse HOLC grades were exposed to more wells. Secondary analyses leveraged temporal variation relative to the city-specific HOLC appraisal date to examine wells drilled/operated before versus after appraisal. Study area: 33 U.S. cities across 13 states with urban oil and gas wells; a subset of 17 cities (n=1,695 HOLC neighborhoods) had 1940 census data for propensity score analyses. Exposure assessment: The unit of analysis was HOLC-graded neighborhoods. For each neighborhood, the authors counted wells of any type (oil, gas, oil and gas, injection, unknown) located inside the boundary or within 100 m of the boundary to capture nearby sources without subsuming adjacent HOLC neighborhoods. They also computed counts by well type and conducted temporal assessments: pre-appraisal (wells with at least one production date before the appraisal year in that city) and post-appraisal (wells drilled/operated after appraisal through 2021). Wells lacking any usable drilling/production dates were excluded from pre/post analyses but included in the primary cumulative analyses. Propensity score restriction and matching: For the 17-city subset with 1940 census data, HOLC neighborhoods were assigned 1940 sociodemographic characteristics (apportioned from census tracts), including total population; proportions of Black, foreign-born, and non-White employed residents; proportion with high school completion; number of homes; median home value; proportion of homes needing major repair; proportion of homes with radios; and persons per home. Propensity scores (probability of receiving the poorer adjacent grade) were estimated using an ensemble machine learning approach (SuperLearner in R) with GLM, Bayesian GLM, multivariate adaptive regression splines, and GAMs. Pairwise adjacent-grade comparisons were constructed (A vs B, B vs C, C vs D). Neighborhoods above the 90th percentile propensity to receive the poorer grade were excluded to ensure overlap; remaining neighborhoods were matched to the nearest neighbor on propensity score with replacement. Statistical analysis: The authors described spatial distributions and used ANOVA to compare exposure across grades in the full set of HOLC neighborhoods. For matched analyses, they used targeted maximum likelihood estimation (TMLE) to estimate differences in well exposure between adjacent grades for three exposure definitions: (1) all wells (including those without dates), (2) wells drilled/operated before appraisal, and (3) wells drilled/operated after appraisal. Sensitivity analyses included restricting exposure to wells inside neighborhood boundaries (no 100 m buffer) and analyzing well density (wells per km²) rather than counts to account for heterogeneous neighborhood sizes. Analyses used alpha=0.05 and R v4.0. Ethical considerations: Only publicly available, aggregated sociodemographic data were used; no human subjects data were involved.

- Across all included cities, redlined D-graded neighborhoods had higher well density than A-graded neighborhoods: 12.2 ± 27.2 vs 6.8 ± 8.9 wells km−2, nearly double. - Temporal patterns: • Pre-appraisal: D-graded neighborhoods (later designated D) had more wells before appraisal (n ≈ 1,421) than all other grades combined (n ≈ 440). Mean pre-appraisal well density was 14.1 ± 24.9 wells km−2 in D neighborhoods versus 2.5–2.8 wells km−2 in A–C. In the subset with 1940 census data, pre-appraisal density was 7.8 ± 10.5 (D) vs 2.4 ± 4.3 (A) (difference not statistically significant). • Post-appraisal: After appraisal, well counts were higher in worse-graded neighborhoods (e.g., A 192, B 835, C 1,590, D 2,977). Mean post-appraisal density was 6.5 ± 11.1 wells km−2 in D neighborhoods vs 3.7–4.8 in A–C (significantly higher in D). - Propensity score restricted and matched TMLE analyses (adjacent-grade contrasts): • B vs A: +0.9 wells (95% CI: 0.6, 1.3). • C vs B: +1.6 wells (95% CI: 1.0, 2.1). • D vs C: +0.3 wells (95% CI: 0.1, 0.6). • Pre-appraisal D vs C: +0.3 wells (95% CI: 0.1, 0.6). • Post-appraisal: worse grades consistently associated with significantly more wells. - Aggregate matched result: Redlined neighborhoods had 2.0 additional wells (95% CI: 1.3, 2.7) compared to comparable better-graded neighborhoods. - Findings were robust in sensitivity analyses using within-boundary exposure only and using well density instead of counts. - By well type, the majority of oil, gas, and injection wells were concentrated in D-graded neighborhoods; many wells with missing dates were also in D-graded areas (notably Los Angeles).

The study demonstrates that neighborhoods given worse HOLC grades—particularly redlined D neighborhoods—had more oil and gas wells and higher well densities than better-graded neighborhoods. This pattern held both before HOLC appraisal (suggesting pre-existing disparities) and after appraisal (consistent with subsequent disproportionate siting or continued development in marginalized areas), and remained significant after propensity score restriction and matching on 1940 sociodemographic characteristics. These findings support the hypothesis that structural racism embedded in federal housing and lending policies contributed to environmental exposure disparities, specifically in the distribution of oil and gas wells in urban areas. The consistency of results across well types and robustness to sensitivity analyses strengthens causal interpretation that HOLC grading and associated racialized practices are linked to disproportionate exposure. The implications are substantial for environmental justice and public health, given evidence that higher exposure to oil and gas operations is associated with adverse outcomes (e.g., preterm birth, cardiopulmonary morbidity, mental health effects). The study suggests that redlining may be one mechanism by which marginalized communities experienced greater harmful exposures, contributing to persistent health inequities.

This multi-city analysis links historic HOLC redlining to disproportionate exposure to oil and gas wells: redlined neighborhoods had higher well counts and densities both before and after appraisal, and matched analyses show significantly more wells in worse-graded neighborhoods. The work advances understanding of how structural racism in historic housing policy relates to contemporary environmental burdens. Future research should: (1) incorporate additional cities and regions lacking 1940 census alignment (especially in the U.S. South) to improve generalizability; (2) integrate more granular historical sociodemographic data (with better racial/ethnic disaggregation) and geologic controls; (3) examine causal pathways linking policy decisions to siting over time; and (4) evaluate health impacts linked to measured exposure gradients from active and abandoned wells in historically redlined communities. Policy efforts should address legacy environmental inequities associated with historic redlining.

- Data limitations in 1940 census: lack of racial/ethnic disaggregation for certain groups (e.g., Mexican-American, Chinese-American) and missing tract-level data for some cities with substantial well exposure (e.g., San Antonio, Erie), leading to exclusion from propensity score analyses and potential underestimation of associations, particularly in the U.S. South. - Incomplete and misaligned oil and gas records: many wells lacked production dates, especially older wells, limiting pre/post appraisal analyses and potentially undercounting pre-appraisal wells in C and D neighborhoods. - HOLC maps reflect one instrument of historic housing discrimination and may not capture other contemporaneous or subsequent discriminatory practices; results may not generalize to non-HOLC areas or later periods. - Possible residual confounding despite propensity score restriction/matching; unmeasured geophysical/geographic determinants of drilling were not fully captured. - Heterogeneous neighborhood sizes may influence exposure metrics, though sensitivity analyses using density were conducted. - State-level production data and classifications may introduce bias. - The fixed historical window and reliance on available archival data may limit causal inference about siting decisions and their drivers across time and locales.