Ambient air pollution is a significant public health concern in the United States, causing over 100,000 premature deaths annually. Disparities in exposure exist, with Black, Hispanic, and Asian communities experiencing higher levels of hazardous air pollutants and greater proximity to polluting industries, even when accounting for socioeconomic factors. Fossil fuel power plants are major contributors to air pollution, emitting particulate matter (PM), nitrogen oxides (NOx), sulfur oxides (from coal plants), and mercury. Studies have shown that communities of color bear a disproportionate burden of PM emissions from coal-fired plants, and recent emission reductions have benefited White Americans more. However, the historical factors driving these inequalities are not well understood, and the role of structural racism remains under-explored. The Home Owners' Loan Corporation (HOLC) maps, created in the 1930s, graded neighborhoods based on perceived lending risk (A-D). This practice, known as 'redlining,' assigned lower grades (particularly 'D' or 'hazardous') to neighborhoods with significant Black, East Asian, Filipino, and foreign-born populations, reflecting existing racist beliefs about property values. Redlining is believed to have exacerbated racial segregation and hindered homeownership opportunities for people of color. Recent research has linked historical redlining to various present-day environmental and health disparities, including less urban vegetation, higher temperatures, higher air pollution levels, oil and gas well siting, and increased rates of preterm birth, asthma, and cancer. This study investigates the relationship between HOLC grades and the subsequent siting of fossil fuel power plants, and resulting disparities in present-day (2018-2019) emissions of NOx, SO2, and PM2.5. The primary objective is to determine if redlining contributed to disparities in power plant siting. The secondary objective is to assess if these historical decisions led to persistent differences in present-day emissions burdens, showcasing the enduring impact of redlining on neighborhood-level environmental inequalities.

Existing research has established a link between historical redlining and a variety of negative present-day environmental and health outcomes. Studies have shown correlations between redlined neighborhoods and lower levels of urban vegetation, increased surface temperatures, elevated air pollution levels, the siting of oil and gas wells, and higher rates of adverse birth outcomes, asthma, and cancer. However, most studies have been cross-sectional, lacking historical data on these outcomes and hindering understanding of how redlining shaped neighborhood trajectories over time. Furthermore, few studies have adequately controlled for pre-existing neighborhood differences that predate the HOLC maps, potentially confounding the relationship between redlining and present-day conditions. Prior work has also indicated that the presence of people of color and socioeconomic status can predict the siting of hazardous waste facilities and power plants, suggesting potential mechanisms through which redlining might have influenced plant location. Some research also highlights the awareness of industry and government officials regarding public opposition and demographic factors when deciding on facility placement to avoid local resistance.

This study used digitized HOLC neighborhood boundaries as the spatial unit of analysis. Neighborhoods were considered 'exposed' if they were downwind of a power plant within 5 km, incorporating both distance and wind direction to focus on plants most likely to impact air quality. Sensitivity analyses were performed using proximity alone to assess the impact of wind direction on the results. The siting of new fossil fuel power plants was examined across three time periods: 1940-1969 (post-war), 1970-1999 (post-Clean Air Act), and 2000-2019 (modern era). Coal/oil-fired and peaker plants were analyzed separately due to their typically higher emissions rates. All plants operational since 1940 were included, including those since retired. Cumulative annual emissions from all upwind plants within 5 km ('emissions burden') were analyzed. This metric reflects both the number of plants and their emission intensity. Since historical emissions data were limited, emissions analysis focused only on plants operational in 2019. To account for pre-existing differences between HOLC grades, the study primarily compared adjacent grades (D vs. C, C vs. B, B vs. A). This was done because A and D grades already showed substantial differences at baseline. The analyses controlled for U.S. census region and the presence of power plants before 1940. A sensitivity analysis additionally controlled for 1940 population demographic and socioeconomic characteristics to isolate the effect of redlining. The study included 8,871 HOLC-graded areas from 196 urban areas and 3,284 fossil fuel power plants. Multivariate regression models were used to analyze the associations between HOLC grade and power plant siting (Poisson and negative binomial models for presence and number of plants, respectively), and between HOLC grade and present-day emissions (log-linear models). The models controlled for pre-existing power plants, census region, and—in sensitivity analyses—1940 population size and socioeconomic characteristics. Sensitivity analyses also used a 10 km radius instead of 5 km to define proximity to plants. Wind direction data from the North American Land Data Assimilation System (NLDAS-2) was used to determine downwind areas.

The study found a significant association between HOLC grade and the siting of fossil fuel power plants. Compared to C-graded neighborhoods, D-graded ('redlined') neighborhoods had a higher risk of a fossil fuel plant being sited within 5 km during all three time periods (1940-1969: 72% higher risk; 1970-1999: 20% higher risk; 2000-2019: 31% higher risk). This association was stronger for coal/oil plants and peaker plants. The association between redlining and power plant siting weakened over time, but persisted into the 21st century. Similar patterns were observed, though less consistently and with smaller effect sizes, when comparing other adjacent HOLC grades. Analysis of present-day emissions showed that neighborhoods with D-grades had significantly higher average annual emissions of NOx (82% higher), SO2 (38% higher), and PM2.5 (63% higher) compared to C-graded neighborhoods, even after controlling for pre-existing power plants and census region. These associations were also apparent, although less strongly, when comparing C vs. B grades. Sensitivity analyses, controlling for 1940 population size and socioeconomic characteristics or using a 10km radius, generally supported these findings, although precision decreased in some instances due to missing data for 1940 data. The patterns of increased emissions burdens were less pronounced for SO2 compared to NOx and PM2.5. The study also found a higher prevalence of peaker plants upwind of worse-graded neighborhoods, particularly during the 1940-1969 period.

The findings strongly suggest that historical racism, as reflected in the HOLC redlining maps, played a significant role in shaping the location of fossil fuel power plants and the distribution of present-day air pollutant emissions. The association between redlining and power plant siting, while strongest during the post-war era, persisted for over seven decades. The higher emissions burdens in redlined neighborhoods are likely a consequence of past siting decisions, the long lifespan of power plants, and the higher emission rates of older plants. While the study controlled for pre-existing conditions, it acknowledges the limitations in fully isolating the causal relationship. However, the persistent association, even after controlling for relevant factors, strongly points to the enduring legacy of redlining. The mechanisms linking redlining to power plant siting are likely multi-faceted. Depressed land values, reduced homeownership rates, and lower household income in redlined areas may have made them more attractive sites for power plants due to reduced land costs and less resistance from residents. Redlining also deepened racial segregation and disinvestment, potentially reducing the political influence of affected communities to resist power plant siting. The study notes that the HOLC maps merely reflected, rather than created, existing racist biases in property valuations.

This study provides compelling evidence of the long-term impact of historical redlining on environmental inequalities in the United States. The association between redlining and the siting of fossil fuel power plants, resulting in present-day disparities in air pollutant emissions, highlights the enduring legacy of systemic racism. Future research should investigate the specific mechanisms through which redlining influenced power plant siting, considering local historical contexts and variations across cities and regions. Further investigation should assess the broader impact of redlining on the siting of other polluting infrastructure. This research underscores the need for regulatory and political interventions to address environmental racism and achieve environmental justice.

While the study employed rigorous methods and controlled for several confounders, limitations exist. The study could not fully isolate the causal relationship between redlining and power plant siting due to the lack of data on all potential confounders, despite controlling for several key factors. The availability of data also introduced limitations; missing data on 1940 socioeconomic and demographic characteristics affected the precision of some findings, while missing data in the emissions datasets may have led to some underestimation of emission magnitudes. The analysis focused solely on power plants; other sources of air pollution and other infrastructure siting decisions might also be affected by redlining. The study also acknowledges the possibility that the correlation between pre-existing industrialization and worse HOLC grades might partially explain the observed results. Finally, the study did not directly assess racial/ethnic disparities in power plant proximity over time.