Coal Mines and Multi-Faceted Risks in the United States: On a Path Toward a Sustainable Future or Emptying Out?

Abstract

Coal is a major source of fuel in many countries, but its mining and use increase the impacts on human and environmental health. Given the wide variation in coal use by nation, we focused on one—the United States. Specifically, we examined the public health, community, and demographic characteristics of 130 U.S. counties with underground and/or surface coal mines that exemplify a land use that leads to a combination of natural and anthropogenic risks. We compared these 130 to other U.S. counties, finding that the coal counties had poorer health outcomes and behaviors, fewer community assets, lower socioeconomic status, and higher population losses compared to the remaining counties in their host states and other U.S. counties. Next, we looked for differences within the 130 coal counties. Counties with the most coal mines and production had the worst outcomes, especially when located in rural areas. We then examined federal programs to assist these coal communities, observing that the largest federal government programs appear to be sending more resources to the most distressed coal counties compared to the less distressed ones. The daunting challenge for the most heavily coal-dependent counties, their state governments, and federal government supporters is to jointly develop plans that are realistic and affordable, take advantage of local natural and human resources, and offer a path to sustainable existence. If the efforts do not succeed, these places run the risk of becoming politically invisible and their populations are likely to empty out.

1. Introduction

Walter Rostow’s The Stages of Economic Growth: A Non-Communist Manifesto [1] proposed stages of economic growth leading to a vibrant economy. The key, he asserted, was to use natural resources, especially coal and iron ore, to create steel, which would lead to more jobs, wealth creation, and the expansion of wealth-building infrastructure. Coal was first mined in the United States in 1701 near Richmond, Virginia. U.S. coal production markedly rose globally during the twentieth century, even while the fuel markets changed, the type of coal changed, and coal mining locations shifted. Today, there is increasing concern about coal’s contribution to climate change; yet, there are arguments for maintaining it as a major source of electrical power generation [2,3,4,5]. For example, Kolstad [6] concludes that coal may be more valuable than many believe because it is a stand-in for failures in other energy sources (especially solar and wind). Similarly, the Trump administration has signaled it is “all in” on fossil fuels to meet the surging power needs of manufacturing and support for the massive data centers required to grow artificial intelligence [7].

Coal production in the United States decreased by about 50 percent in the twenty-first century, particularly after 2012. Employment in mines also plummeted. In 2014, 25 states produced coal. In 2023, the number fell to 21, with continued decreases predicted. The decrease to date has been less in the surface mines of Wyoming and Montana than it is in the Midwest and Appalachia, areas with a long history of underground mining [8]. In 2015, four publicly traded U.S. coal companies filed for Chapter 11 bankruptcy protection, and the outlook did not improve with additional companies filing thereafter [9]. Long-term trends for a continued decline are predicted for the U.S. coal industry [10,11]. However, this may change as the new administration will likely continue to support coal by reducing environmental regulations that target the industry [12,13].

In 2017, coal accounted for 38 percent of the electrical generation capacity of the world. While its use plummeted during the pandemic, it set a record in 2024 at 8.77 billion tons. The current major coal-producing countries are Australia, China, Germany, India, Indonesia, Kazakhstan, Poland, Russia, South Africa, and the United States. The largest consumer of coal, however, is China [14]. Coal dust and miners’ health issues are universal problems that each of these countries must deal with. Underground coal miners have high rates of injury and death due to explosions, equipment failures, mining collapse, and exposure to toxic chemicals, vapors, and carcinogens [15]. Workers in both surface and deep mining are at risk of developing coal workers’ pneumoconiosis and silicosis [16,17]. Mental health and stress-related problems plague both miners and their families [18], and the effects of noise pollution [19], pre-term births [20], and other poor health outcomes are linked to serious environmental and social problems in coal-producing areas [21,22]. How nations approach these issues may differ widely as solutions to the interconnected economic, environmental, health, and social problems are not simple. Regional sustainable development theory tells us solutions must include institutional, political, and social supports [23] that are tailored to fit the conditions of individual locales [24]. While nations may choose different paths to manage their coal legacy, a growing body of international literature calls for a just transition policy for workers, their families, and host communities. This literature and extensions will be the focus of the final section of this paper.

This paper asks how current U.S. policies are serving to address the sustainable development needs of its coal-producing counties. It addresses four questions:

• How different are the health, community asset, and demographic measures for counties producing coal from counties in the same states without active coal mines, and from non-coal-producing counties in the remainder of the United States?

We expect coal-producing counties to demonstrate poorer health outcomes and behaviors, relatively fewer local assets, and higher poverty rates than non-coal-producing counties.

2.

How different are coal-producing counties from each other for these same measures?

We expect high-coal-mine and coal production counties to demonstrate poorer outcomes than mid- or low-coal-mine and coal production counties.

3.

Do current U.S. government programs to reclaim and redevelop coal mining areas appear to be distributed to the neediest places?

We expect that several U.S. government programs provide more support for counties with more coal mines and coal production and less support for those with fewer coal mines and less production.

4.

Do the current levels of investment and types of investment allow these areas to rebound and build sustainable communities?

We expect the data to show a substantial gap between what is called for and what is needed, partly due to the absence of a definitive standard for what constitutes a just transition and endpoints for these places.

The paper is divided into five parts: This introduction is followed by a literature review that concentrates on the local health, environmental, and demographic risks associated with coal mining, and the government programs to reduce risks and provide local benefits. The third section summarizes the data and methods used in the study. Section 4 answers the first three research questions. The conclusion responds to question 4, examining the moral and political challenges the U.S. government faces in its response to the results of the declining coal mining communities.

The major contribution of this study is to delve into locales that represent the U.S. coal legacy and identify their different levels of community risk and distress. Risk is not equal across the areas with coal mines. This leads to an evaluation of current U.S. government programs that direct assistance to these communities, their adequacy, and a discussion of the moral and political challenges for the sustainable development of local coal mining areas.

2. Context

Questions 1 and 2 focus on local areas with coal mines. As coal mining was key to growing the economy and political power of the United States, the national response to the negative consequences of the industry did not come quickly. In the twentieth century, four pieces of legislation mark these efforts: (1) the creation of the Bureau of Mines in 1910, (2) the Federal Coal Mine Safety Act of 1952 requiring annual safety inspections of underground mines, (3) the Federal Coal Mine Health and Safety Act of 1969 which expanded safety for both underground and surface mines, and the (4) Surface Mining Control and Reclamation Act (SMCA) of 1977 which focused on the environmental consequences of surface mining and the reclamation of both surface and underground mines. Overall, policies were directed at improving worker safety, compensating and providing health care to miners with black lung disease, and offering support to coal-mining areas losing jobs.

The human health evidence supporting these directives is strong [14,25,26,27,28,29]. The National Institute of Occupational Safety and Health (NIOSH) has the responsibility to report on workers’ health. In 2011, NIOSH reiterated that exposure to coal mine dust leads to coal workers’ pneumoconiosis, chronic obstructive pulmonary disease (COPD), and various other pulmonary diseases leading to respiratory impairment, disability, and premature death [25]. The report notes that the prevalence of coal-related diseases has increased, especially among those under 50 years old. The worst outcomes were noted for the central Appalachia underground mines in Kentucky, Virginia, and West Virginia. Explanations for this range from increased mining in small mines with high silica environments to higher smoking rates among miners. This information has been in the literature for many decades, with the authors concluding that the government needs to develop ways of reducing exposure to crystalline silica dust [14,26,27,28,29].

A second body of literature focuses on the environmental consequences of coal, including decades-long underground fires and emissions into communities [30,31,32,33,34]. Underground mining often leads to unstable surface conditions and, in the case of surface mining, removes layers of rock, soil, and vegetation that leave a scarred surrounding landscape. The legacy of leaking minerals from mines and loss of vegetation can severely degrade local water quality, especially when the population depends on underground water and has no potable surface water alternative. Similarly, in the latter half of the twentieth century, acid rain was considered to be a major source of acidification of lakes. In this century, burning coal is seen as a major contributor to worldwide climate change.

Question 3 focuses on the U.S. federal government support for coal mining areas. Coal production has consequences for miners but also for their families and neighbors. When coal mining jobs decrease, worker families and communities lose income, local taxes, and workers’ buying power. These negative economic consequences cast a pall over an area, leading to out-migration, a brain drain, and the shrinking of local social capital [35,36,37,38,39]. Coal areas can become stigmatized, leading to a loss of local investment. The net effect is the loss of resources that fund schools, police and fire services, housing, and a long list of public health services required for maintaining a basic quality of life. The result is that those who can leave often will. As described below in detail, we used quality of life and community data developed by Niche, Inc. (Pittsburgh, PA, USA) to examine these issues in a set of the coal counties with the most distressing symptoms.

Overall, the literature points to the poorer health outcomes and behaviors, fewer local amenities, and lower socioeconomic status of residents of coal areas. Coal states and the U.S. federal government are well aware of these consequences. Glenda Owens [40], Deputy Director of the Office of Surface Mining Reclamation and Enforcement (OSMRE) of the U.S. Department of the Interior, reviewed national programs in her testimony before the Senate Committee on Energy and Natural Resources in November 2023. Eleven of the twenty senators on this committee, including chair Joseph Manchin (West Virginia), were from states with active coal mines in 2024. Owens summarized the federal government actions to close mine shafts and openings, eliminate highwalls, remove spoils and fix embankments, restore land and streams, replace polluted water supplies, and provide funds for redevelopment [41,42,43,44].

Funding for the reclamation of abandoned mines is through a tax on coal production by mine operators. One program is the Abandoned Mine Reclamation Fund (AML Fund), which pays for mine remediation. Some of the funds go to states with recognized reclamation programs, and the remaining resources are assigned by OSMRE to conduct remediation and reclamation in states without approved programs. The Abandoned Mine Land Economic Revitalization (AMLER) program constitutes a second one that distributes funds to six states that are part of Appalachia. It has explicit economic development programs for the coal areas (Alabama, Kentucky, Ohio, Pennsylvania, Virginia, and West Virginia). Overall, the AML Fund focuses on reclamation, whereas AMLER focuses on economic revitalization and associated remediation. Question 3 examines the geographical distribution of these funds used for the environmental AML Fund and AMLER [44].

3. Materials and Methods

3.1. Places to Study

The first important data decision was to use the county scale. The state scale is too broad to capture the differences among local areas. The municipal government level, census tract, and block scales might show intra-county variation, but these scales are not practical because there are so much missing data in rural areas. Counties serve as local governments in many rural areas. Thus, data for all U.S. counties, county equivalents, and the District of Columbia were used. This included 3145 counties and county equivalents (excluding non-state entities considered U.S. territories and possessions, such as Puerto Rico, Guam, and the U.S. Virgin Islands that are administered differently).

Annual U.S. government reports [9] indicate the number of mines by county and state. The 2024 report [8] listed 130 counties with surface and/or underground coal mines. A total of 130 U.S. counties had at least one coal mine, compared to 1497 counties which did not have a mine but were located in coal states, and 1518 counties in states with no coal mines. However, we assumed that differences existed among these counties, and we developed the classification method described below to divide the 130 coal counties into low coal presence, medium coal presence, and high coal presence. We also used another dataset to divide the high coal counties into two categories based on their location in a metropolitan region or not (see below for more details). In other words, thinking of the U.S. as an onion, we began by examining the outer layers, including all the counties in states and gradually peeling away the layers until we arrived at the core of coal counties that appear to be most burdened by multiple risks associated with coal mining.

The selection of counties as the analytical unit raises statistical challenges. One is ecological fallacy, or assessing individual risk based on group data. For example, without individual health records, we cannot be sure that coal miners are the people who disproportionately suffer injury-related deaths in those counties. Recognition of this problem and clearly stating it as a limitation in the results is imperative. The second challenge is the small numbers problem, which occurs because some places have few residents and a few more or less outcomes can mean much higher or lower rates. The net effect is to weaken results because the highest and lowest values will be disproportionately in places with the smallest populations. The third challenge is the modifiable area unit problem, or when the issue being studied is disconnected from the data-gathering level. For example, coal regions in this study cross county boundaries; some counties have many mines, whereas others have only one. Moreover, workers may not live in the county where the mines are located. Recognizing this problem, we classified each coal county based on the number of mines and the magnitude of production rather than population. However, to fully address this issue, worker data by residence would have to be collected in a detailed epidemiological study. Recognizing these statistical limitations, we proceeded with selecting databases and indicators that would allow us to answer our research questions [45,46].

3.2. Variables to Study

More than one hundred variables were available in federal government databases. We picked indicators that were consistent with the literature and had the fewest missing cases in their respective databases. For example, we eliminated drug overdose, firearm fatalities, HIV prevalence, homicide, and suicide from our analyses. While these are important risks, they were missing 15 percent or more of the time for many counties. However, premature mortality, injury death rate, and life expectancy were available and selected for use. We emphasize that the comparisons for question 1 are between counties, not individual states or the United States as a whole. Comparisons with larger units would be heavily biased by urban areas like New York City, Los Angeles, Chicago, and other major urban centers.

Table 1. Variables and data sources used in the study.

Variable	Data Source
Coal mining county, 2023	Annual Coal Report. U.S. Energy Information Administration [47]
Health Outcomes & Behaviors (n = 11)
Premature mortality (Years of potential life lost), 2019–2021	* National Center for Health Statistics
% Poor or fair health, 2021	* Behavioral Risk Factor Surveillance System
% Poor physical health days, 2021	* Behavioral Risk Factor Surveillance System
% Low birthweight babies, 2019–2021	* Behavioral Risk Factor Surveillance System
% Adults reporting currently smoking, 2021	* Behavioral Risk Factor Surveillance System
% Excessive drinking, 2021	* Behavioral Risk Factor Surveillance System
Teen births per 1000 15–19-year-old females, 2016–2021	* Behavioral Risk Factor Surveillance System
Injury death rate, 2017–2021	* Behavioral Risk Factor Surveillance System
Yrs life expectancy, 2017–2021	* Behavioral Risk Factor Surveillance System
% Annual mammogram, 2021	* Medicare
% Flu vaccinated, 2021	* Medicare
Community Characteristics (n = 14)
Population, 2023	U.S. Census Bureau [48]
Population change, 2010–2023, %	U.S. Census Bureau
% Access to broadband, 2018–2022	* American Community Survey
Food environmental index, 2019 & 2021	* Indicator of Access to Health Foods
School funding adequacy, 2021	* School Funding Indicators Database
% Home owners, 2018–2022	* American Community Survey
% Access to exercise opportunities, 2010, 2022, 2023	* U.S. Census Bureau
Primary care physician rate, 2022	* Area Health Research File
Child care centers per 1000 children, 2010–2022	* Homeland Infrastructure Foundation
Mental health provider rate, 2023	* Center for Medicare and Medicaid Services, National Provider Identification
% Voter turnout, 2018–2022	* American Community Survey
% Census participation, 2020	* Census Operational Quality Metrics
Social association rate per 10,000 residents, 2020	* County Business Patterns
Religious congregations per 100,000, 2020	Association of Statisticians of American Religious Bodies [49]
Demographics (n = 11)
Median Household Income, $, 2018–2022	* American Community Survey
% Children in Poverty, 2018–2022	* American Community Survey
Income ratio, or income inequality, 2018–2022	* American Community Survey
% Children in single parent households, 2018–2022	* American Community Survey
% Some college, 2018–2022	* American Community Survey
% 18 years and younger, 2022	* U.S. Census Bureau
% Persons 65 years, 2022	* U.S. Census Bureau
% Non-Hispanic Black, 2022	* U.S. Census Bureau
% Hispanic, 2022	* U.S. Census Bureau
% American Indian or Alaska Native, 2022	* U.S. Census Bureau
% Non-Hispanic White, 2022	* U.S. Census Bureau

* Accessed via County Health Rankings and Roadmaps [50].

lists 37 indicators used to answer questions 1 and 2 at the county scale. Among the 37 indicators, the income ratio (measuring income inequality) may be the least familiar to some readers. The variable is the ratio of household income at the 80th percentile to income at the 20th percentile. The expectation is that higher income ratios would be associated with less civic engagement.

Given the history of coal mining in the United States, we expected large differences among the health outcomes and behavior indicators. Substantial differences were also expected in community characteristics and demographic indicators, especially socioeconomic status. Furthermore, the longer period of exposure in the Appalachian regions (centered in Kentucky, Pennsylvania, Virginia, and West Virginia) led us to expect larger legacies than we would find in the Midwest (centered in Illinois, Indiana, and Ohio) and western U.S. coal areas (Montana, New Mexico, North Dakota, and Wyoming).

In addition to the datasets in Table 1, we used a dataset created by Niche, Inc. [51,52] to compare the sixteen counties with the strongest manifestation of underground and surface coal mines (see method below). As noted above, we began with the full set of counties in the United States, and then analyzed the 130 coal counties. Exploring these 130, we found one set of sixteen with the highest presence of coal mines and production. To further explore them, we used Niche ratings according to twelve dimensions: commute, cost of living, diversity, good for families, health and fitness, housing, jobs, nightlife, outdoor activities, public schools, weather, and an overall rating. Scores range from D to A+ (A+ is the highest rating). Our goal was to see if we could find anticipated differences among these sixteen counties in community-level quality-of-life indicators beyond those available in Table 1.

The final datasets were U.S. government records of how many Abandoned Mine Land Fund and AMLER [44,53] grants were given to states and counties.

3.3. Methods

For question 1, we employed one-way ANOVA with post hoc Tukey test to compare the 130 coal counties, the 1497 non-coal counties in coal states, and 1518 other counties in the U.S. without coal production.

To address question 2, we used discriminant analysis to examine the relationship between coal mining presence and multiple indicators of local health outcomes and behaviors, community assets, and demographic characteristics among the 130 counties. We expected to find high coal presence to be associated with poorer health outcomes and behaviors, fewer community assets, and low socioeconomic status. The discriminant analysis process takes a categorical indicator and creates new statistical variables that help distinguish among places, thus predicting an outcome for each place that can be compared to the actual outcome (see Section 4.2 for further explanation). Multinomial regression and ordinal regression were alternative methods for a categorical variable but were not selected because nearly all our predictor variables were interval data, which can be difficult to interpret with these methods.

To address question 3, we compared U.S. government records of grants to states and counties to see if the counties classified as having the most coal presence received the most federal grants and funding. We used Spearman rank correlation and compared average values received by counties for this analysis.

Question 4 reflects on the answers to questions 1 to 3. It examines the options available as part of a coal transition program to assist different levels of risk and distress among the 130 coal counties. These range from many coal counties that are not markedly different from non-coal counties in their host states to sixteen counties that require special attention and assistance.

The following set of eight bullets summarizes the overall research design:

• Define research questions and expected outcomes;
• Identify all coal counties and organize them into groups based on the number of mines and the amount of coal production (coal presence);
• Identify health outcomes and behaviors, community characteristics, and demographic measures to compare the sets of counties;
• Compare coal counties (n = 130), non-coal counties in coal states (n = 1497), and other U.S. counties (n = 1518) using ANOVA, and post hoc tests;
• Look for differences among 130 coal counties in health, community, and demographic characteristics using discriminant analysis;
• Look for differences among the 16 counties with the most coal presence using Niche data and a metropolitan versus non-metropolitan distinction;
• Examine the distribution of two federal funds for areas with coal mines by state using rank correlation, and then compare intra-state difference by county in the amount of dollars received by extent of coal presence in two states that have many coal mines;
• Consider the need for transition funding, especially for counties with markers of high coal presence and distress measured by the extent of poor health outcomes, limited community and demographic assets, and low quality of life measures.

4. Results

4.1. Preliminary Results

Two preliminary analyses were performed to help shape the research. The first was about the wide variation in the location of coal mines and the magnitude of those variations.

Table 2. Location of coal mines by state, 2023.

State	Number of Mines	Thousand Short Tons	Number of Counties with Mines	Number of Counties with Underground Mines	Number of Counties with Surface Mines
Alabama	20	12,043	5	3	5
Alaska	1	1009	1	0	1
Colorado	9	12,371	6	5	1
Illinois	14	36,865	10	8	4
Indiana	17	23,782	8	2	8
Kentucky	112	28,346	21	13	18
Louisiana	1	289	1	0	1
Maryland	12	1280	2	1	2
Mississippi	1	2685	1	0	1
Missouri	1	140	1	0	1
Montana	4	29,072	3	1	2
New Mexico	2	7987	2	0	2
North Dakota	4	24,087	3	0	3
Ohio	9	1947	5	1	5
Oklahoma	1	2	1	0	1
Pennsylvania	113	42,647	21	11	18
Texas	4	13,815	4	0	4
Utah	6	6921	4	3	1
Virginia	41	10,548	5	4	5
West Virginia	165	84,560	22	19	15
Wyoming	15	237,261	4	0	4
Total	552	577,657	130	71	102

Data source: U.S. Energy Information Administration Annual Coal Report, 2023 [8].

lists the states hosting the 130 counties with at least one underground or surface coal mine. Pennsylvania and Virginia have almost half of the mines. However, Wyoming produces almost half of the coal. In contrast, Alaska, Louisiana, Mississippi, Missouri, and Oklahoma each have only one mine. Kentucky, Pennsylvania, and West Virginia each have double-digit counts of both underground and surface mines. This means that the counties do not fit a simple “coal” county label.

To account for the differences in the number of mines, the type of mine, and the amount of coal produced, we created a classification system. A county could receive 0–4 points. It received a point if it had two or more underground mines and another point if it had two or more surface mines. A county received another point if it produced at least 300 short tons of coal from an underground mine in 2023 and another point if it produced 300 or more short tons of coal from a surface mine (about 40 percent of counties produced at least 300 short tons (a “short ton” is 2000 pounds in the U.S. and equivalent to 907.18 kg)). As there is no classification system for mines based on their production of tonnage, three hundred short tons was selected because there was a break in the data at approximately 300 short tons for both surface and underground mines. For example, beginning with Alabama, Jefferson County has three underground and five surface mines (2 points), and both types produced more than 300 short tons (2 points). Jefferson received a score of 4. In contrast, Dekalb County, Alabama, had no underground mine and only one surface mine that produced eight short tons. Therefore, De Kalb received no points. Overall, 16 of the 130 counties received scores of 4, and 26 had scores of 0. All of the counties with scores of 4 were in West Virginia (8), Kentucky (3), Virginia (2), Alabama (2), and Pennsylvania (1). With only 130 coal-producing counties, we faced the small numbers problem described earlier in distinguishing among the 130 counties and needed a way of exploring their differences. We thus reduced the five groups (0, 1, 2, 3, or 4 points) to three as follows: (1) least-coal-presence group (LeastCP): 0 or 1 point; (2) middle-coal-presence (MidCP): 2 or 3 points; and (3) most-coal-presence (MostCP): 4 points. We expected more distress in the MostCP counties with 4 points, and the least distress in LeastCP counties with 0 or 1 point, with intermediate results in the MidCP counties with 2 or 3 points. This simple scoring system may understate the impacts of large surface mines in Wyoming, Montana, and other Western states, which we looked for in the results.

A second preliminary analysis explored the difference between the coal counties and their host states. We expected the coal counties to be relatively poor and have poor health outcomes, but some of their host states had the same characteristics. Thus, we used persistent poverty and health outcome indicators at the state scale for this analysis. As part of the American Recovery and Reinvestment Act of 2009, the U.S. Census Bureau defined “persistent poverty” counties and census tracts as places with ≥20 percent of their population in poverty for at least 30 years [54]. The 21 coal mining states had 53 percent of all U.S. counties and 67 percent of the persistent poverty counties. The 130 coal counties had almost 13 percent of the persistent poverty counties compared to 4 percent of the U.S. counties. In other words, persistent poverty is more prevalent in the coal counties than in the coal states as a whole, but many persistent poverty counties in coal states are not coal counties. For health outcomes, seven of the ten highest years of premature life lost rates are among coal states and none of the fifteen lowest rates are among coal states [55]. Reflecting on these preliminary analyses, we compared the coal counties to the non-coal-producing counties in the coal states, and also to all the counties in the non-coal-producing states.

4.2. Question 1: Comparisons of Coal Counties, Non-Coal Counties in Coal States, and Other U.S. Counties

The results of the one-way ANOVA and Tukey post hoc tests comparing the three sets of counties were not subtle. In 10 of the 11 health outcomes and behavior comparisons, the coal counties had the worst outcomes, and the non-coal states’ U.S. counties had the best ones. The non-coal counties in coal states were in the middle in every comparison. The lone exception was that the U.S. non-coal counties had the highest excessive drinking rates and the coal counties the lowest. The coal county numbers were significantly different from the other two groups in every case (p < 0.05). The median difference between the two poles was 17 percent, and the median difference between the county coal and state coal counties was 6 percent. The biggest differences were the measures of premature mortality found in the premature years of life lost and injury deaths.

The exception of the excessive drinking comparison is interesting. The presence and use of intoxicating beverages (also narcotics) are not allowed in both surface and underground mines because of the added risk it implies [56]. This practice may help explain the exception, albeit without direct evidence; it would be an ecological fallacy to conclude that the coal miners are less likely to be intoxicated than other residents. Indeed, an analysis of the U.S. Behavioral Risk Factor Survey concludes that miners have higher alcohol consumption rates than gas and oil workers and non-manual labor workers [57]. Furthermore, tobacco smoking also adds to the risk in mines and, yet, is more prevalent in coal mining counties. The finding that there is less excess alcohol consumption in coal mining counties is questionable.

In terms of community characteristics, 8 of the 14 indicators listed in Table 1 follow the pattern of the best results in the non-coal counties and worst results in the 130 coal counties. The coal counties as a whole experienced population loss, and had the lowest broadband access, the least access to local exercise opportunities, the fewest childcare centers, and the lowest voter turnout and census participation rates compared to the non-coal counties. The counties in the non-coal states had the best outcomes for all of these. But, for other measures, the coal counties did not have the worst outcomes. For demographic measures, for example, the coal counties had the lowest household income, highest percentages of children in poverty, highest income ratio, and lowest record of some college attendance variables. Yet, they did not have the most unfavorable results for all the demographic indicators. The results are inconsistent.

Overall, we found the 130 coal counties are distinct from the non-coal counties in the United States, especially for health behaviors and outcomes, some indicators of community assets, and low socioeconomic status. A full table of the results for the three groups of counties by the 36 variables is available from the authors.

4.3. Question 2: Differences Among the Coal Counties

Are there major differences within the set of 130 coal counties? We expected major differences, which we examined using the three coal county groups described earlier. The previous analysis used data on over 3000 counties. This one had only 130 coal counties. The implication is that the number of variables had to be limited to ascertain which variables most strongly differentiated among the counties. Accordingly, each of the three sets of variables (health, community, and demographic) was analyzed with discriminant analysis. Several variables were dropped because too many counties had missing data, and some made little contribution to the statistical analysis. Two counties were not included because of missing data. After these analyses, we had 128 of the 130 counties and 16 variables. As noted in the methods section, discriminant analysis is a commonly used tool for discerning the differences between distinct groups of a categorical variable.

The results of the discriminant analysis appear in

Table 3. Comparisons of coal counties in the U.S. states (n = 128).

Variable	Function 1: Group with Four Coal Production Points vs. Group with One or No Major Coal Production Points	Function 2: Group with Two or Three Coal Production Sites vs. Group with Four Coal Production Points
Demographic (n = 3 of 11)
Median Household Income, $	−0.368
% Children in Poverty	0.335
Income ratio	0.211	−0.265
Community Characteristics (n = 4 of 14)
Population change, 2010–2023, %	−0.444	−0.217
% Access to exercise opportunities	−0.260	0.305
% Voter turnout	−0.367
% Census participation	−0.253	0.412
Health Outcomes & Behaviors (n = 10 of 11)
Premature mortality	0.534
% Poor or fair health	0.389	−0.219
% Poor physical health days	0.461
Low birthweight babies	0.437
% Excessive drinking	−0.385	0.342
% Adults reporting current smoking	0.372
Teen births per 1000 15–19-year-old females	0.319
% Annual mammograms	−0.260	0.367
Injury death rate	0.582
Life expectancy	−0.554

. Discriminant analysis creates n − 1 number of new statistical variables. The trichotomy of low, medium, and high coal mining county presence was reduced to two statistical variables called discriminant functions. The first function contrasts the counties with four points (MostCP) with counties with 0–1 points (LeastCP). The high-coal-presence counties (MostCP) have high injury death rates (R = 0.582), a low life expectancy (R = −0.554), and relatively many years lost to premature death (R = 0.558). These counties disproportionately reported many poor physical health days, poor or fair health, many low-birthweight babies, higher adult smoking rates, and declining populations. In contrast, group counties with the least coal presence (LeastCP) have better health outcomes and behaviors, more citizen participation, and more favorable socioeconomic characteristics. This strong statistical function had a medium to high canonical correlation of 0.621. The average standardized Z-score for the high-coal-presence county function was 1.80 compared to −0.51 for the least-coal-presence group. In other words, the average distance between the two sets of counties is a standardized z score of 2.32, which is a large difference. In essence, this function says that health outcomes and behaviors are the main distinctions between the most- and least-coal-presence counties.

The second function distinguished between the MidCP group (2–3 points) and the MostCP group (4 points). The canonical correlation was 0.438, and the average z-score for these two groups was −0.619 and 0.812, respectively, with an average z-score distance of 1.4. The highest correlations were related to civic engagement, with the highest correlations for census participation, access to exercise opportunities, and mammography rates for the mid-coal-presence counties and lowest values for the most-coal-presence counties. Consistent with this observation is the relatively lower income ratios in the middle group than in the high-coal-presence group. This means that the income segregation is less marked in the middle- than the high-coal-presence group, which is consistent with the literature that shows more civic engagement with less income differentiation [58]. The differences identified between the medium- and high-coal-presence counties are important. They identify the high-coal-presence counties as lacking access to exercise activities and mammograms, and having less interaction with government officials than the middle-coal-presence counties. These correlations are moderate and they represent only a few measures of service needs. Nevertheless, they are concerning because they are part of the assets local areas need for a healthy sustainable future, and should be considered as part of a community transition program.

Discriminant analysis predicts a group classification for each county, which can be compared to the actual group; that is, a county may be a county with the most coal presence and the discriminant model will predict the probability for it belonging to the most-, mid-, and least-coal-presence-county groups. The model accurately predicted 72 percent of the cases (74 percent of the LeastCP group, 68 percent of the MidCP group, and 75 percent of the MostCP counties. Overall, a county with underground and surface mines, two or more of them, and more coal production is strongly associated with poor health outcomes, fewer community assets, and poor socioeconomic status.

We were concerned that high-surface-mine counties with no underground mining (LeastCP) might not be accurately predicted. This was not the case. Nearly all were accurately predicted or were predicted to be in the MidCP group.

4.4. Differences Among the Sixteen High-Coal-Presence (HCP) Counties

To further differentiate among the coal counties, we used Niche’s ratings to compare the 16 MostCP counties across twelve dimensions developed by Niche (see Materials and Methods). The results listed in

Table 4. Niche rating of the sixteen high-coal-presence counties by metropolitan region designation.

Variables	MostCP Counties in a Metropolitan Region (n = 7)	MostCP Counties Not in a Metropolitan Region (n = 9)
Median number of A+, A, and A− ratings	2.5	1
Median number of C+, C, C−, and D+ ratings	3	7
Median county population, 2023, thousands	74	23

Data Source: Niche [52,60].

are striking. The median number of high (A+, A, and A−) and low (C+, C, C−, and D+) ratings are similar for the seven counties located in metropolitan regions, 2.5 and 3, respectively. This compares to 1 and 7, respectively, among the nine counties not located in metropolitan regions. In other words, the six more populated and metropolitan region counties have much higher ratings than their ten less populated counterparts located outside metropolitan regions. The exception is also telling. Nine of the ten non-metropolitan counties had A− ratings for cost of living. Niche measures affordability with housing costs, food prices, median income, and taxes. A rating of A+, A, or A− means a lower cost of living in that area. On the flip side, nearly every one of the non-metropolitan counties had job and public school ratings of C or D. These findings are consistent with a study of the U.S. Environmental Protection Agency’s former national priority list of mining sites that include some coal mines but also many former gold, zinc, lead, and other mineral mines [59]. We view these results as a marker of a stigmatization of rural areas that face extreme difficulty in rebuilding a more sustainable future (see Discussion and Conclusion).

4.5. Question 3: Distribution of Funds Among Coal States and Counties

Do federal funds go to the places that have the most coal presence? Federal and state data suggest a correlation between the distribution of funds and needs. The AML Fund program has provided more than 130 grants. States with many coal counties received most of them. Specifically, the Spearman rank correlation between the number of coal counties in 21 states and the number of grants was Rho = 0.72 (p < 0.05). Pennsylvania, Virginia, Indiana, West Virginia, Wyoming, and New Mexico received almost half of these grants and about the same proportion of the coal counties. Only New Mexico among these five states appears to have a disproportionate number of grants and Illinois has a relatively low share.

As of 30 September 2022, the Abandoned Mine Land Reclamation Fund (AML Fund) had collected over $12 billion, including interest earned, through a fee assessed on each produced ton of coal. About half has been distributed to local areas for reclamation, with about $2 billion to worker health and retirement programs, and another $2 billion used for emergency responses and OSMRE operating expenses. This leaves about $2 billion unappropriated [61].

The AMLER funds are targeted to counties and tribal groups. Hence, we were able to compare intrastate distributions for the six eligible states (Alabama, Kentucky, Ohio, Pennsylvania, Virginia, and West Virginia) and to the CROW, the HOPI, and the Navajo Nation. AMLER funds, which were first distributed in FY 2016, can be used to build infrastructure, and provide jobs and skills, as well as for building tourism and other economic opportunities. Kentucky and West Virginia have published county-specific distributions. Kentucky, for example, distributed about $200 million across 87 projects (an average of $2.3 million per project). Using the group’s coal presence classification, we found that more funds were made available to the MostCP counties compared to the MidCP and LeastCP ones. Specifically, the average MostCP county in Kentucky received 8.4 percent of the cumulative AMLER funds committed to Kentucky compared to 5.4 percent for the MidCP county, and 2.8 percent for a LeastCP county [62].

Like Kentucky, West Virginia also has many coal counties. The MostCP counties received an average of 8.9 funded projects during the period 2016–2024. The MidCP group received an average of 4.3 and the LeastCP counties an average of 3.3 grants [63].

5. Discussion

We started with four questions with multiple parts. Beginning with all U.S. counties, we went through a step-by-step process to identify counties with the most evidence of coal production. We first compared the differences between the 130 coal counties, the non-coal counties in their 21 host states, and the remaining U.S. counties in states that had no coal production sites. As expected, the coal counties had less healthy outcomes and behaviors. For community assets and demographics, the largest differences between the coal and non-coal counties were in socioeconomic status, population size, and population change, as well as access to local exercise sites, the presence of childcare centers, voter turnout, and several other indicators.

The second research question probed the differences within the set of 130 coal counties. We found that sixteen counties with multiple surface and underground mines and high coal production rates (MostCP) had notably higher premature death rates, especially injury-related, more poor and fair health days, higher adult smoking rates, and almost always poorer health outcomes and behaviors than counties with less coal presence. These sixteen counties also were more likely to lose population than their counterparts. They fit the label of high-hazard and high-risk places.

Yet, there were differences within the sixteen. A dataset built to rate places for potential migrants and investors showed that nine of the sixteen MostCP counties, all located in rural areas, had the attribute of a low living cost, but marked by few jobs and poor public schools. These nine appear to be the most disadvantaged among all the 130 coal counties. Seven of the nine are classified as persistent poverty counties by the U.S. Census Bureau [54].

This study faced two data availability limitations for addressing the first two questions. While we had a good deal of data, with a few exceptions, it centered around the last decade because much of the health and community data have only been publicly accessible for that period. The biggest gap is for environmental measures of air, water, and land pollution. Environmental data are available in the U.S. EPA’s EJScreen database but are limited in the number of indicators and hampered by the reality that much of it is estimated rather than actual measures at the site [64].

Question 3 asked about the distribution of federal funds to coal counties. At the state scale, the states with the largest number of coal mines have, to date, received the most AML Funds. The fit is not perfect, but we would be hard-pressed to conclude that the states with the most mines and production have been disproportionately deprived. The AMLER funds are limited to six states and three tribal groups. Kentucky and West Virginia, among these six states, have 43 of the 130 coal counties and 11 of the 16 MostCP counties. In both states, the counties with more coal prominence received a notably larger share of the grants and resources. The conclusion is that U.S. efforts to address the coal legacy appear to be consistent with where the resources are needed. However, it would be a valuable exercise to conduct detailed case studies by state to compare future use goals and plans with resource allocations from the federal and state governments.

The fourth research question focuses on public policy implications. Do current levels of investment and types of investment allow these areas to rebound and build sustainable communities? Our research shows large disparities among the coal counties in outcomes that current public policies can only partly address. The nine rural coal counties have a wide variety of markers of risk, distress, and stigmatization. A smaller set of counties with prominent markers of coal presence are located in metropolitan regions and should be more able to rebuild rather than become a sacrifice zone. The remaining coal counties have less coal presence than these two sets. More than 80 of the 130 have only one or a small yield mine and they show less distress than the other coal counties. We conclude that current programs appear to be giving priority to the places that most need it. Yet, we reiterate that much more detailed case analyses are required to determine if our initial findings match on the ground realities at these local scales.

6. Conclusions

Our research could not address whether there is enough support going to the most distressed coal-producing counties in the United States. It does, however, suggest that some counties need a great deal more support than others. Various federal and state reports suggest that considerably more funding could be used to aid these communities [31,52,53,54], and future research should add a qualitative component (such as case studies or interviews with policymakers) to address this issue.

The United States has a long history of “just transition” policies for returning military since the American Revolutionary War and continuing through many others [65]. The U.S. government provided land, scholarships, loans for housing, health benefits, and a variety of other benefits to former military personnel. Another prominent transition program is directed at the sites that built, developed, and tested nuclear weapons in about one hundred places across the nation. The federal government spent hundreds of billions of dollars to remediate radioactive contamination at a dozen of these sites, and it will be spending hundreds of billions more for at least the next five decades [66,67]. This Department of Energy program at over $400 billion is the largest debt owed by the U.S. Government. The U.S. Defense Department spent about $50 billion to destroy the U.S. chemical weapon stockpile [68]. These were both war-related programs that the federal government created and is responsible for the weapons and remediation. Indeed, Hazel O’Leary, former Energy Secretary, labeled the cleanup program as “closing the circle on the atom” [69].

Stepping back from military-related programs where there is a clear moral and legal responsibility for massive programs directed by the federal government, we recognize a set of programs related to business transition. The United States lost millions of manufacturing jobs, including those in Rostow’s iron and steel industry. Steel and automobile workers, businesses, and host communities such as Akron (OH), Pittsburgh (PA), Gary (IN), Detroit, and Flint (MI) experienced massive losses in jobs, population, and taxes when those industries declined. The cities received assistance from federally sponsored programs to help rebuild, and the government used tariffs, subsidies, and other devices to strengthen the declining industries. Yet, the government did not take over the industries, nor did it assume responsibility for saving either them or their employees [70,71,72]. Steel and autos have continued to be produced in the United States, but the locations of their production have shifted. An important difference between coal on the one hand and steel and automobiles on the other is the impact they leave on land. For example, Pittsburgh (PA) and Detroit (MI) were the most prominent steel and automobile manufacturing centers in the world, respectively. A massive amount of air, water, and land contamination occurred in these places. The U.S. government and responsible parties spent enormous amounts of money to control the contamination and to allow for the reuse of the sites where the contamination occurred. Arguably, they could have spent more. However, almost every steel plant in the Pittsburgh area and the automobile production plants in Detroit are now used for different purposes, including housing, where higher levels of remediation were required. Both cities have lost population, revenue, and services in their painful transitions. However, their relative losses pale in comparison to those endured by coal mine areas, especially those we categorized as high-coal-presence in non-metropolitan areas.

There is no permanent federal commitment to coal as there has been for returning soldiers and nuclear and chemical weapons sites. Should there be? It is prudent to begin by examining the idea of just transition from coal to other energy sources from an international perspective. The Stockholm Environment Institute (SEI) [73] makes the case for strong and just transition programs across the globe [74,75,76,77,78,79,80,81,82]. The papers and reports identify three major policy option categories. One involves compensating workers for lost wages and pensions, communities for lost taxes, and companies for devalued assets. Similarly, the World Bank reviewed 11 of its funded projects, noting that a just transition from coal production means that governments must prepare well in advance of coal mine closures and implement strong safety nets for workers ahead of job losses. As global coal mining has shifted from West to East, future coal mine closures and associated job losses will be concentrated in Asia. It is anticipated that the top three global coal producers, China, India, and Indonesia, will be affected the most by mine closures. As many of these areas may be unable to create new job opportunities, governments can implement labor mobility schemes, enabling coal miners and their families to move to areas with strong economies and new job prospects [83].

The argument for just transition options is that the world is asking coal and other fossil fuel providers to downsize for the benefit of current and future generations. Compensation is required to allow the people and places affected by the transition to obtain opportunities to adjust and diversify. In other words, some argue that the world is fighting a war against climate change, and that coal miners and their communities should be seen as part of the casualties of this war. This is not the same driver as the competition that led to the loss of iron and steel production and many other manufactured products. Is the United States, the second largest user of coal, morally obligated to provide extended benefits to people and places that are losing a lot in this war? Coal-dependent communities, states, and regions face just transition challenges that vary greatly by location [78,81]. Remote and rural communities in the American West are particularly vulnerable to the social and economic impacts of the coal transition [79].

Some of these ideas are part of the federal government’s efforts to support workers. A second option is to re-train former coal workers to fulfill other jobs. However, this option is likely to increase the brain drain in coal counties, leaving them with huge gaps in their best-trained younger populations. Another option is to help high-risk communities by providing resources to support local cultural and environmental programs. The AMLER program is illustrative in this regard.

The United States has a truly polarized political environment. Politically, 16 of the 21 states with coal counties voted for Donald Trump in the 2024 presidential election. The new administration will likely renew its previous efforts to widen the market for coal, appoint directors to environmental programs with ties to the coal industry, and attempt to weaken environmental regulations [84]. While this could lead to more support for coal counties, will the most hard-pressed counties we identified be able to have a sustainable quality of life, or will these places continue to slide toward emptied-out ghost towns? It is not a fait accompli that they will. Some former mining places have carved out new economic bases. In the United States, two of the most famous successes are Pitkin and Routt counties, both in Colorado. The first began as a silver mining area and the second as a gold mining one. They now have attractive ski and resettlement areas (Aspen and Steamboat Springs, respectively). The rebound of former coal mining areas is less striking, but Scranton (PA) and several places in the Kanawha coal area of West Virginia have reconfigured their mines as museums, places to grow mushrooms, generate electricity, and, otherwise, use existing local natural and human resources to rebuild. In contrast, Thurmond (WV) in Fayette County was once a thriving coal town. Now, it is a ghost town (with an estimated population of five). Thurmond has served as a recreation site, owned by the National Park Service for the New River Gorge National Park and Preserve. The historic railway depot is the visitor center, and the town is a designated historic district on the National Register of Historic Places. In other words, the policy decisions for making a just transition are complicated, involving the local governments and the states to design realistic and affordable plans that merit thoughtful analytical work and communications with residents in the most vulnerable places. We hope that this research emphasizes that the differences among U.S. coal counties and the multifaceted risks they face demand different types of support to help them achieve a more sustainable quality of life.

This paper focused on the United States as one of the major coal producers and users. We believe the approach used might be reasonably applied to other coal-using countries if accessible databases allow for comparisons of health and environmental impacts at various scales. Where such databases are not available, we suggest a review paper that compares the health and environmental impacts of coal in several countries and the policies those nations have created to address the challenges. Because of data availability and language limitations, we suspect that residents of these countries who are familiar with their databases and policies are the appropriate analysts for such an effort. Perhaps the editors of this journal and this special issue could facilitate such an effort.