Prisons as Panacea or Pariah? The Countervailing Consequences of the Prison Boom on the Political Economy of Rural Towns

Abstract

The nascent literature on prison proliferation in the United States typically reveals negative impacts for communities of color. Given that Southern rural communities were the most likely to build during the prison boom (1970–2010), however, a more nuanced understanding of prison impact is warranted. Using a dataset matching and geocoding all 1663 U.S. prisons with their Census-appointed place, this study explores the countervailing consequences of the prison boom on rural towns across multiple periods. For example, locales that adopted prisons at earlier stages of the prison boom era received a short-term boom compared to those that did not, but these effects were not lasting. Furthermore, later in the boom, prison-building protected towns against additional economic decline. Thus, neither entirely pariah nor panacea, the prison functions as a state-sponsored public works program for disadvantaged rural communities but also supports perverse economic incentives for prison proliferation. Methodological, substantive, theoretical, and policy implications regarding the intersection of race and punishment are explored.

1. Introduction

Scholarly attention to punishment has centered on the causes and consequences of U.S. mass imprisonment being rooted in maintaining racial, political, and economic hegemony [1,2,3,4]. The impact of mass imprisonment on continued racial and economic stratification in the United States cannot be overstated [1,2,3,4,5,6,7,8,9,10]. For over a decade, more than 2.2 million Americans have been incarcerated annually. It is well documented that the United States imprisons its citizens at a higher rate than any other developed nation, with the burden of imprisonment falling disproportionately on communities of color [1,2,3,4,5,6]. While research on mass imprisonment has made significant theoretical and empirical contributions to improving sociological understanding of punishment, studies of mass imprisonment primarily focus on northern, poor, urban, minority communities and families [4,7,8,9,10,11]. However, we cannot forget that mass imprisonment is not limited to northern urban areas—rural towns also experience mass imprisonment [12,13].

Mass imprisonment is a salient issue calling together conservatives and liberals. For example, Rand Paul and Corey Booker have teamed up to head a special U.S. Senate committee on the topic [14]. To address mass incarceration, it is necessary to better understand what scholars [15,16] refer to as prison proliferation, often conceptualized as the widespread construction of prison facilities throughout the United States. In 1970, there were 511 prisons scattered across the USA. Since then, the USA has experienced a prison boom, building 1152 facilities at a cost of more than $23 billion and spanning 580 square miles. If one aligned each facility built during the prison boom together in a contiguous landmass, it would span more than half the state of Rhode Island. A better understanding of prison proliferation can also reveal the nuanced ways that racial and economic inequality intersect with punishment. For instance, while there are more than 400,000 corrections officers in the USA, racial minorities continue to be overrepresented in this occupational field [17].

Due to recent budget shortfalls, many states are considering closing prison facilities as a cost-saving measure. Many well-intentioned activists and scholars are similarly advocating for prison abolition [2,18,19]. However, a closer examination of the persistence of prison-building—or constructing or erecting a prison within a municipality or county—can shed light on why rural communities might resist closing, thwarting attempts to roll back mass imprisonment. As we have witnessed with other crime control policies (e.g., the “war on drugs”), there are collateral consequences. Rushing to close prisons without considering the consequences may also produce unanticipated repercussions. Regretfully, rural communities are often placed between the proverbial rock and a hard place given that prisons are disproportionately built in rural communities with larger African-American and Latino populations [20]. Given where prisons are built, prison abolition efforts may inadvertently harm vulnerable populations within African-American and Latino communities. Despite theoretical and substantive advances in mass imprisonment scholarship, Western [4] maintains that mass imprisonment is invisible in the lives of mainstream white Americans and means very little to that segment of society. In contrast, for poor, uneducated African-Americans and Latinos, the criminal justice system has become omnipresent, shaping nearly every facet of their existence [7,11,21,22]. For many rural poor communities, prison-building represents another shift in the purpose of prisons—the prison is now a state-sponsored public works program.

Contemporaneously, yet distinct from the increase in the “supply” of prisoners, a confluence of political, cultural, and economic factors also drive prison “demand.” The five-fold increase in the imprisonment rate has been accompanied by a tripling in the number of prisons. Although mass imprisonment and the prison boom occurred simultaneously, there is considerable variation in the number of prisons built in each state and region during this massive expansion of the criminal justice system over the past 45 years. Thus, prison proliferation is not simply a function of mass imprisonment. For example, Illinois, Georgia, and Ohio have roughly 50,000 inmates each in state prison, but each respective state houses prisoners in a number of different facilities (from Illinois’s 55 prisons to Georgia’s 82 prisons). During this era of mass imprisonment, the United States has leapt ahead of all developed nations in the per capita number of citizens it incarcerates [4,5,6,23]. In fact, Louisiana has become the world leader in incarcerating its residents, with over 35,000 citizens annually or roughly 1600 per 100,000 persons serving time [23]. This rate of imprisonment is especially high given that Louisiana has only 16 prisons housing over 41,000 prisoners annually. In comparison, Texas has 121 prison facilities for the more than 100,000 inmates it incarcerates annually. This means that Texas has an average of 826 inmates per facility, whereas Louisiana has an average of 2500 per prison facility. This vast difference in the respective number of prisons and prisoners per facility demonstrates the variability of prison-building across states and within regions. This also suggests that the same logic driving mass imprisonment may not apply to prison proliferation as the two have distinctly different functions.

While the sociology of punishment incorporates the collateral consequences of mass imprisonment [4,5,10], the core function of imprisonment is social control. That is not to say that political motivations are absent from mass imprisonment [24,25,26,27,28,29]. However, unlike imprisonment, prison-building is not primarily an element of social control. Prison-building is fundamentally a process involving the coordination of, and across, the political economy of local, state, and sometimes federal actors and agencies. It is also an inherently political process given that the vast majority of prisons are constructed and operated by state legislatures. In fact, some [26,29] argue that prison-overcrowding litigation produced increased prison-building distinctly from increased imprisonment. Therefore, theories explaining mass imprisonment should not be extended wholesale in explaining the impact prisons have on communities.

The nascent literature on the prison boom consistently portrays the negative impacts of prison-building for urban communities of color [2,3,30] and rural communities in general [2,31,32,33,34,35]. Implicitly or explicitly, this literature is based on the prison–industrial complex (PIC). Eric Schlosser (3) defines the PIC as “a set of bureaucratic, political, and economic interests that encourage in- creased spending on imprisonment, regardless of the actual need” [3]. The PIC can be summarized as: (1) politicians exploit crime legislation to secure votes; (2) private companies seek profits by serving or operating prisons; and (3) rural town leaders use prisons for economic development. From the PIC perspective, racist profit-making motives explicate both mass imprisonment and the prison boom. These studies suggest that the causes and consequences of the prison boom extend far beyond the typical purposes of prisons. From this perspective, prison-building is primarily a profit-generating venture. Classic works of the criminological literature posit that the institutional purpose of the prison is incapacitation, rehabilitation, deterrence, and retribution. While Wacquant [9] asserts that the modern prison has been repurposed to define, confine, and control urban minority populations, others argue that the prison now serves as a state-sponsored economic development project similar to a public works program [2,3]. Yet, the line of inquiry raised by Schoenfeld [29] and Guetzkow and Schoon [26], regarding the role of prison overcrowding causing the prison boom, could provide a plausible alternative explanation for prison-building.

Understanding the impact of prisons may, however, require a more nuanced approach given that Southern rural communities with larger proportions of people of color are most likely to build prisons [20,36]. It also stands to reason that if prisons did not provide some benefits and were only detrimental to rural communities, there would be less “demand” for this type of facility. However, because a baseline or average effect across different periods of the prison boom has not been established at the town level, it is impossible to know if prison-building benefits rural, low-income African-Americans or whites. To better understand the challenges and opportunities confronting rural communities, Eason [37] introduces a Weberian, positive (less normative) perspective on prison-building. Along with this perspective, he developed a lexicon of prison-building. I will refer to these concepts throughout this piece in creating a new way to conceive of prisons’ impact. The difficulty in establishing a baseline is partly due to the theoretical limitations of the current literature, but also due to data limitations and methodological challenges in analyzing prison-building at the town level.

While knowledge on the prison boom is increasing, an empirically driven theoretical understanding of prison impact—the economic, political, and social benefits/costs for a host community as a result of prison placement—has yet to be established. This is an important knowledge gap, as the literature does not establish what prison-building means for racial and economic inequality in rural communities. I add to this burgeoning literature by considering the effects of prison-building on the political economy of rural towns. I bring an innovative, multi-method approach by analyzing novel longitudinal data taken from the Prison Proliferation Project1, in order to investigate the challenges and opportunities presented by prison-building for rural towns across different time periods of the prison boom (1970–2000). Figure 1 below represents the universe of U.S. prison. Each facility is represented by a point indicating size and latitude longitude coordinate in 2010. This study uses municipal-level, national data, including geocoded matches of the entire prison universe, from 1663 U.S. prisons across each U.S. Census place that has ever constructed a prison. According to the U.S. Department of Commerce, a U.S. Census-designated place is defined as “the statistical counterparts of incorporated places, and are delineated to provide data for settled concentrations of population that are identifiable by name but are not legally incorporated under the laws of the state in which they are located” [38]. In addition to the original data, this study also uses multilevel fixed effects analysis and propensity score matching (as a robustness check) to explore the impact of prison-building on the political economy of all rural U.S. Census places during the prison boom.

The current study makes a substantive contribution to the literature on the prison boom by demonstrating that prison-building can be a mixed bag for rural communities. I call for a reexamination of existing paradigms of prison placement and impact from the perspective of Southern, rural towns, where prison proliferation is most prominent. Lastly, this study highlights the salience of prison-building to the political economy of depressed rural communities. This paper proceeds as follows. First, I begin by constructing a theoretical framework for measuring prison impact by chronicling the existing literature on the prison boom. Second, using these conceptual arguments, I examine additional approaches to measuring prison impact. Third, before presenting results, I outline hypotheses and describe my research design, including methods and data. Next, I highlight the results of prison-building on the political economy of rural towns across four dependent variables: median family income, median home value, poverty, and unemployment. Finally, I conclude by exploring the implications of the findings.

2. Towards a Theory of Prison Impact

I propose a perspective to deepen our understanding of what theorist David Garland [39] refers to as the New Iron Cage, where justifications for prison-building have been replaced “with the system taking on a life of its own, giving rise to adaptive behavior serving secondary interests.” The most insidious form of the new iron cage is the penal-industrial complex—“the set of bureaucratic and private institutions that produce and manage jobs around prison building” ([39], p. 179). I redefine the penal-industrial complex as the economic, social, and political institutions related to the causes and consequences of the prison boom. Using this perspective allows us to understand the prison as a complex institution that is linked to the political economy of rural towns, functioning as neither panacea nor pariah. The penal-industrial complex can be differentiated from the prison–industrial complex because of its positive focus on punishment. I improve on the penal-industrial complex by expanding beyond its focus on profits for corporations and jobs for rural communities in order to explain how prisons impact rural communities and why it may prove difficult to reduce prison-building and, ultimately, the number of prisoners.

While employment is a strong impetus for prison “demand,” poverty reduction and increases in family income/median home value may provide more than an economic boost. Poverty reduction and increases in family income/median home value demonstrate the potential benefits prisons can have for families in prison towns. A prison town is a nonmetropolitan municipality that has secured and constructed a prison run by federal, state, or private operators. More importantly, these benefits may be directly attributable to new prison facilities and the political coalitions within towns that seek these public works reward. Understanding how prisons benefit places can better explain why disadvantaged towns marshal limited resources to acquire these stigmatized institutions.

This reframing allows us to better understand the bureaucratic function of punishment and its role in the prison boom. To flesh out this new theoretical frame, I will engage the existing literature on the prison boom to conceptualize each step in the prison-building process. Recently, Campbell and Schoenfeld [29] proposed an institutional approach to analyzing state and federal policies across multiple periods to differentiate the contributing factors of mass imprisonment. Similarly, I differentiate steps in the building process that will clarify the importance of prison impact across multiple periods of the boom and across different geographical spaces.

2.1. Understanding the Prison Boom

In a seminal piece on the prison boom, Lawrence and Travis and (2004) claim that prisons have forever altered the physical, social, economic, and political landscape of rural America. Surprisingly, we still know little about prison-building. With few exceptions, scholars do not often consider where and why prisons are built [2,12,20,36,37,39,40,41,42]; how the prison boom impacts these places [31,32,33,34,35,42,43]; or local perceptions of prison impact [44,45,46,47,48]. Thus, while studies of prison towns are increasing, the causes and consequences of the prison boom in the rural United States remains underinvestigated. A central reason for this lacuna is sociologists’ focus on punishment in the urban north, rendering the prison boom a byproduct of racist practices connected to mass incarceration [2,3,5,30].

While studies of prison-building are increasing, these studies outline different processes related to the prison boom. Eason [37] advocates for scholars to examine the prison boom by providing a comprehensive conceptual framework detailing the multiple, and often conflicting, political, social, and economic choices across multiple actors and units of analysis involved in the process of prison-building. I borrow several key terms here to elucidate the prison-building process. Prison placement is the process associated with the political economy of prison-building within a municipality, paying particular attention to the role of civic leaders and the local political elite in securing the facility. To date, the few empirical studies of prison placement have mapped key social, political, and economic traits of rural communities that predict the arrival of a new prison. Building on past studies [42,49], Eason [20] completed the most comprehensive national town-level analysis predicting prison-building to date. This study found that prison placement is a rare event, with less than 5% of all U.S. rural towns receiving a prison during the height of the prison boom. Prison placement is correlated with higher proportions of racial and economic disadvantage in rural towns in the South. In fact, 45% of all prison towns are in the South. Despite the considerable attention given to private prisons, this study also demonstrates that the majority of prisons (82%) are built by state legislatures [20]. Given that prison placement is not a random event, predictors of prison-building need to be improved in order to develop better controls for the propensity to build prisons. In other words, to better understand the social, political, and economic “demand” factors of prison placement, research must distinguish those towns most likely to build prisons.

Prison siting is the process associated with the political economy of prison-building, emphasizing the role of the state. Gilmore [2] provides arguably the most advanced theoretical understanding of prison siting in a study that encompasses state and local levels of community impact. Gilmore [2] argues that the prison boom in California came about through a relative surplus in financial capital, political capital, and rural land. This Keynesian approach conceptualizes political forces as the drivers of prison expansion, with the surplus population of chronically unemployed residents as grist for the mill in prison expansion.

Hoyman and Weinberg [49] examine prison siting at the county level. Their study uses proportional hazard modeling to examine how 24 prisons were sited in North Carolina from 1970 to 2000. This study’s county-level predictors for prison siting included: “Motivation” (poverty, unemployment); “resources” (college graduates, owner occupants); “Not in my backyard” (NIMBY) (owner occupants, population density, African-Americans); and propinquity (adjacent siting). While propinquity was positively correlated with future prison siting for both counties with or without prior established prisons, the coefficients in this analysis were insignificant. These findings suggest that adjacent prison placement should be incorporated into future analysis. The study concludes that population density, owner occupants, and college graduates are negatively correlated with prison siting, with “the presence of African-Americans is not an important predictor” ([49], p. 107). The authors’ major theoretical contribution is that demography, not just socioeconomics and race, predicts prison siting. I will also take these factors into account when considering prison impact. While this study provides valuable insight into potential indicators for prison siting, the findings cannot be generalized beyond prison siting in North Carolina at the county level. I expand on these insights by conducting a national analysis over time.

2.2. Understanding and Measuring Prison Impact

Understanding the context of prison placement also improves our knowledge of the potential ways prisons impact places. To date, prison impact studies have primarily focused on the advantages or disadvantages provided by the placement of a prison in rural communities [32,33,34,35,43]. Measures of employment growth and poverty alleviation are central economic indicators in the few existing empirical studies on prison impact. While Glasmeier and Farrigan [43] find that prisons provide nominal positive economic impact on the most disadvantaged communities, many scholars agree that prisons do not boost the economy of the average rural community [32,33,35]. In fact, some work suggests (implicitly or explicitly) that prison stigma negatively impacts rural communities by limiting other types of development beyond future prison facilities [2,3,32,33,34,35].

Rural prison placement and impact are often understood as an outcome of processes of imprisonment of urban minorities that benefit rural whites [2,3,5]. In this scenario, the complicated stratification processes of prison placement and the subsequent impact on rural communities are reduced to a zero-sum game along race and community dimensions, whereby rural white communities win prisoners at the expense of urban African-American/Latino neighborhoods. From this perspective, one could argue that development in rural white communities is a linear outcome of the urban arrests of African-Americans and Latinos [1,2,18,19]. These perspectives, however, do not fully account for the local context of decision-makers in rural communities. In this respect, recent qualitative studies of prison towns (primarily by geographers) have begun to tease out the nuances of prison impact [46,47,48,50]. However, negative perceptions of prison impact still dominate the scholarship in this area.

There are several reasons to question the depth and degree of negative impact of prisons on rural communities. First, studies demonstrate that prisons harm rural communities using data from one state over a short period of time [2,35]. Second, most analyses limit their consideration of effects to economic growth [32,33,35,43]. This limited approach does not provide clarity on the more crucial policy question of whether or not prisons slow economic decline. Third, the unit of analysis is not standardized in quantitative studies of prison towns—some measure counties [32,33,34] while others measure U.S. Census places [43].

In short, these studies fail to consider whether prisons and the new jobs they bring slow economic decline, on the one hand, or the way prisons may benefit rural communities, on the other. Although prisons may not solve all the economic woes facing most disadvantaged rural communities, they may offer a reprieve from an economic downturn. More to the point, we lack empirically grounded theories to conceptualize the links between the prison boom and disadvantage in rural towns.

Expanding how we think about prison impact broadens our ability to explore the link between prison proliferation and inequality. Prison proliferation is a pressing question not only for scholars of crime but also for scholars of rural economic development and urban poverty. In other words, prison proliferation is not simply about where prisoners are from or where prisons are constructed. Instead, the prison boom should also be understood as a product of the power structures that exist between and within rural and urban communities. By examining prison impact on a national scale, we can better grasp the mechanisms and processes of racial and economic stratification across the rural-urban continuum. This approach allows us to consider how the modern prison has become more than a means of social control but also a form of poverty management, exploitation, and stigma. Additionally, exploring how prisons benefit and challenge rural communities has implications for the prison abolition movement, as rural communities will be affected by a move away from an overreliance on imprisonment, along with a possible decrease in prison-building and potential prison closings.

Until recently, qualitative studies [46,47,48,50] on prison impact lacked methodological diversity. In a key piece on prison impact, Blankenship and Yanarella [51] review prison town literature and subsequently provide several components for measuring the shortcomings and merits of prison-building for communities. They began formulating a conceptual basis for towns to provide policy alternatives to building new prisons. Then they proposed that when towns are deciding whether to place a prison within their towns, the prison should: (1) Create high-quality jobs within the local economy; (2) preserve local capital; (3) improve local public infrastructure and increase the local tax base; and (4) improve or maintain the local social capital and civic culture. To date, no study has taken this holistic approach as a framework for measuring the pros and cons of prison-building—this piece is an attempt to demonstrate both.

Building from here, Hooks et al. [32,33] provide a county-level spatial economic analysis of prison impact over time for all U.S. prisons (rural and urban), including controls for established prisons. By accounting for variables that are central to conceptualizing local economic development and using a time series of data, this study introduces innovations in data and analysis of prison placement. Perhaps what is most innovative in this study is the incorporation of the Land-Deane spatial analysis to measure the impact of new prisons across surrounding communities. Its most significant contribution is the finding that “prisons impede economic growth in counties growing at the slowest pace, [which] flies in the face of the widely held view that prison construction can assist struggling local areas” ([33], p. 51). These findings support the claim that prior prisons create an agglomerative prison economy, driving away other types of future developments. I build on this line of reasoning by investigating the role of prior prison placement in future prison-building. Despite the advances this study provides in examining prison impact, there are limitations. First, the data limitations plaguing the field are prevalent with Hooks et al. [32,33] given the sources of data used and process they described to geocode prisons within counties. Second, studies of prison towns vary widely in their unit of analysis. Although many studies use the county as the unit of analysis [32,33,34], an increasing number use the town or U.S. Census place as the unit of analysis [20,43]. Each unit of analysis presents unique advantages and challenges for interpreting outcomes. While county-level analyses have provided better and more reliable data over time, they tend to obfuscate prisons’ impact on extremely disadvantaged rural towns. Therefore, using the U.S. Census place as the unit of analysis can provide insight into how prisons impact the political economy of rural towns. Third, there are limitations in using cross-sectional analysis to compare prison impact on all counties—the most serious is that comparisons cannot be drawn between the communities most likely to build. Given that prison-building is a rare event, using quasi-experimental methods to compare the places most likely to build is best for making substantive theoretical and empirical claims about prison-building. I also improve on Hooks et al. [32,33] by using a fixed-effects model and propensity score matching as a robustness check.

While recent studies provide county-level descriptors, there remains room for statistical improvement in prison town studies. An analysis of town-level demographics offers a more nuanced assessment of the prison boom by using more methodologically rigorous techniques, as encouraged by McShane, Williams, and Wagoner [52]. Besser and Hanson [31] address this mandate by using a place-based national analysis to argue that towns with prisons experienced negative impacts such as increased poverty, unemployment, and a rise in minority residents. While groundbreaking, the study’s results were far from conclusive as the methodology did not control for prior levels of its measured effects. Moreover, it reproduced measurement error in its use of mailing address data from the American Correctional Association (ACA).

Glasmeier and Farrigan [43] improved on Besser and Hanson’s [31] study by providing national descriptive statistics for prison inmates, prison employees, and the largest prison-building states. Moreover, their study used town-level predictors to measure the economic impact of prisons. Glasmeier and Farrigan [43] measure prison siting as a treatment by using propensity scores to compare the likelihood that a town would receive a prison. Their contribution of clear, standardized measures of propinquity to prior prisons makes this an innovative approach to studying rural prison impact. Nevertheless, the absence of controls for region, state, and race are problematic given that other studies have found these factors to be significant predictors of prison-building. These initial studies provide a good description of overall trends but do not provide a strong theoretical context for these trends.

Although U.S. Census places are usually considered too limited in size to capture employment and well-being outcomes, they nevertheless allow us to better understand the relationship between concentrated disadvantage and the prison boom. The U.S. Census place, or rural town, allows for an appropriate spatial scale for this analysis. Spatial scales, according to Lobao, Hooks, and Tickamyer [3,53] are “the geographic levels at which social processes work themselves out, are conceptualized, and are studied.” Using the appropriate scale, we can make sense of prison impact on rural towns by identifying “how and why spatial context contributes to inequality” ([53], p. 3). Much like neighborhoods in urban areas, rural U.S. Census places refract extreme concentrations of racial and economic disadvantage, evidenced by high rates of concentrated poverty [54] and residential segregation [55]. Given that studies of concentrated disadvantage often focus on urban neighborhoods [56], I instead use U.S. Census places (towns) to better understand how the prison boom has impacted disadvantaged rural communities.

Past studies have suggested that race, region, population density, poverty, unemployment, degree of rurality, and prior prison placement are correlated with prison placement [20]. Therefore, I take these factors into consideration when predicting prison impact. While Hoyman and Weinberg [49] measure poverty and Hooks et al. [32,33] measure employment as outcomes, I improve on past studies by examining both. Furthermore, I improve on the ACA data used by Besser and Hanson [31] by using an extensive verification process in assigning each prison to the U.S. Census place identified by the Bureau of Prisons and the state department of corrections for each state in the contiguous United States. This study provides a framework based on rigorous empirical analyses to show how prisons impact rural communities. To this end, this study measures how newly constructed prisons, across different periods of the prison boom, affect changes in median family income, median home value, unemployment, and poverty.

3. Data, Measures, and Analytic Strategy

3.1. Data: Prison Proliferation Project

The data for the analyses are from the Prison Proliferation Project. Given the high potential for error in most prison-building data, I compiled, cleaned, and geocoded the Prison Proliferation Project data from several sources over a five-year period. The primary sources of the prison data are the 2010 Directory of Adult and Juvenile Correctional Departments, Institutions, Agencies, and Probation and Parole Authorities ([57]; hereafter, the ACA directory) and the Inter-University Consortium for Political and Social Research (ICPSR) data holdings for the 2005 Census of State and Federal Adult Correctional Facilities (CSFACF) provided by the United States Department of Justice, Bureau of Justice Statistics [58]. The latter includes a listing of the 1600-plus U.S. prison facilities by latitude and longitude coordinates, U.S. Census place, name of facility, year of facility construction, and a limited sample of facility renovations by year. The ACA directory is a compilation of data on institutions throughout the United States, its territories, and military facilities overseas. The directory includes data on facility name; street and mailing addresses; opening date; capacity; security level; average daily population; gender of population; adult or juvenile indicators; number of full-time and part-time staff; cost of care per day; and other information. This source was recommended as the “gold standard” of prison locations by a contact at the United States Bureau of Prisons. Even so, its use of mailing addresses introduces the potential for error. ACA data alone cannot be used to geocode prisons because more than 150 prisons list addresses located five or more miles from the actual prison facility. I have improved on the ACA and CSFACF data by using an extensive verification process assigning each prison to the U.S. Census place identified by the Bureau of Prisons or state department of corrections.

Entries from the ACA directory and CSFACF holdings were reviewed by trained coders and entered into a database. The current data includes adult facilities operated by federal and state governments, Native American governments, and private contract facilities. The database entries were then checked in their entirety for errors and duplicate entries. Due to differences in record keeping across jurisdictions, entries in the ACA directory exhibit some inconsistencies with respect to the types of data provided and the format of those data. Where data were missing, researchers contacted either the facility itself or the state department of correction attempt to obtain proper figures. I also used the CSFACF data to reconcile some latitude/longitude coordinates and used Google Maps to obtain the latitude and longitude coordinates for facilities with adequate information and checked street addresses again. For those ACA entries that provided only a mailing or P.O. Box address, researchers again used Google Maps with satellite view to obtain visual confirmation of the facility and to collect the latitude and longitude coordinates. The location of each facility was then verified using the Coding Accuracy Support System, a location verification system used by institutions like the United States Postal Service, and each case was reconciled a third and final time.

These data were augmented and merged with Geolytics’ decennial U.S. Census demographic and economic data using GIS software [59,60,61,62]. To date, this process has resulted in a dataset that includes every U.S. Census place from 1970 to 2000, normalized to 2000 decennial boundaries (N = 25,150) and 1663 prisons spread across all 50 states. Unlike most data on prisons, these data are not a sample of prisons or towns—the dataset includes every adult prison facility and every municipality in the USA. These data were then merged with files containing U.S. state-level economic and program transfer data covering the years 1980–2011, as maintained by the University of Kentucky Center for Poverty Research (UKCPR). The UKCPR data include state-level predictors such as political party affiliation of governor and both chambers of the legislature, as well as state-level demographic, poverty, and employment variables.

The data are additionally delineated along the rural-urban continuum according to the Beale Rural Urban Classification Codes (RUCC) as displayed in

Table 1. Beale Rural Urban Classification Codes (RUCC) 1973, 1983, 1993, and 2003.

Code	Metropolitan Type	Beale Code
1	Metropolitan	Counties in metro areas of 1 million population or more
2	Metropolitan	Counties in metro areas of 250,000 to 1 million
3	Metropolitan	Counties in metro areas of fewer than 250,000
4	Non-metro	Urban population of 20,000 or more, adjacent to a metropolitan area
5	Non-metro	Urban population of 20,000 or more, not adjacent to a metropolitan area
6	Non-metro	Urban population of 2500 to 19,999, adjacent to a metropolitan area
7	Non-metro	Urban population of 2500 to 19,999, not adjacent to a metropolitan area
8	Non-metro	Completely rural or less than 2500 urban population, adjacent to a metropolitan area
9	Non-metro	Completely rural or less than 2500 urban population, not adjacent to a metropolitan area

Source: Prison Proliferation Project.

. The RUCC designates the degree of rurality (with 9 being the most rural) or urbanity (with 1 being the most urban) for each census location in every period of the analysis, with 1974, 1983, 1993, and 2003 as designated years for determining rural–urban classifications. Because of the potential for prison populations to tip the RUCC designation from rural to urban, I use 1974 as a baseline for this analysis.

3.2. Measures

3.2.1. Dependent Variables

The dependent variables used to measure the impact of newly constructed prisons from 1970 to 2000 at the U.S. Census place (town) level are as follows: (1) Change in median value of homeowner property in 1980, 1990, and 2000; (2) change in median family income in 1980, 1990, and 2000; (3) change in poverty rate in 1980, 1990, and 2000; and (4) change in unemployment rate in 1980, 1990, and 2000.

3.2.2. Explanatory Variables

The key explanatory variable is prison-building. Because each prison is geocoded into a Census place according to the date each facility opened, I created four dichotomous indicators denoting periods of prison openings across each place: (1) Whether or not a prison was built at that location prior to 1969; (2) whether or not a prison was built at that place between 1969 and 1978; (3) whether or not a prison was built at that place between 1979 and 1988; and (4) whether or not a prison was built in that town between 1989 and 1998. I offset each decennial prison-building indicator by one year—for example, coding prisons built between 1969 and 1978 instead of 1970 to 1979, working under the assumption that it will take a minimum of two years for a prison to impact the dependent variables measured during each decennial census year [43]. Structuring the data in this way allows me to parse out the current and prior period effects of prison-building.

3.2.3. Control Variables

For each census year in the set of 1980, 1990, and 2000, I employ several place-level census variables as controls in the models. These variables include the log of the total population (and its square); the percent of the population age 25 and over with a bachelor’s degree; the percent African-American (and its square when appropriate); the percent Latino (and its square when appropriate); the percent of housing units that are owner-occupied; and the percent of residents that reside in town from the prior census. Also included within each analysis are the prior decade’s dependent variables. For example, when measuring unemployment, the percentage of those unemployed in the prior decade would be a control variable in the analysis.

3.2.4. Inappropriate Controls

The perceived impact of prisoners on outcome variables further complicates the analysis. Prisoners should be counted in their last known residence prior to imprisonment. However, if prisoners are counted where they are serving their sentence during the census, their presence could inappropriately increase the poverty or unemployment rates of any given location. To minimize any potential bias caused by prisoners, I exclude them from the analysis for several reasons. First, and most importantly, according to the census, the poverty status of prisoners cannot be determined [63]. Second, the multiple ways prisoners can be counted can bias analysis. Third, the census’s rules for counting prisoners changed dramatically between 1970 and 2000, further complicating place-level measures of institutionalized populations [30,63]. The best way to account for prisoner impact in this analysis would be to acquire the average daily population for each prison in the year that it opened. In completing the analysis for the current study, I have built the most comprehensive dataset on U.S. prisons. I did not, however, include data on the average daily population for each prison in the year that it opened because of the difficulty of compiling such data. While these data are important to understand prison impact trends overall, they are not central to the questions asked in this study and are extremely difficult to procure.

Additionally, prison-building can obscure racial demography—yet this process of how prisons obscure racial composition in rural towns with prison is largely unknown. The ways prisons alter racial composition is a key piece of that puzzle best answered using highly sensitive micro data available through the U.S. Census Bureau. Given that the majority of prison-building does not occur in towns that already have prisons, one way to control for the potential racial biases of prison-building would be to control for group quarters population. This gets incredibly complicated because group quarters are not standardized across decades of the prison boom and one could not tell if they were controlling for senior homes, colleges, or prisons in certain decades (especially in rural towns) without using micro data from the U.S. Census. In addition to these challenges, group quarters in a correctional facility in a given period is highly correlated with prior prison-building. Furthermore, group quarters in a correctional facility also includes jails. Because this investigation is about prisons, controlling for jail populations will skew the findings. Given the messiness of these measures over time and place, prison-building in the prior period is the best control.

3.2.5. Propensity to Build Prisons

The non-random nature of prison placement raises concerns about selection processes [64]. Although the classic solution to this problem is a Heckman Selection Model [65], I chose propensity score matching [65] as the primary robustness check for the fixed-effects models across each period because it has become more common in criminal justice analyses, especially for countervailing findings [66,67,68]. To gauge the propensity to build, this analysis will use demographic predictors of prison town studies [20] including U.S. region, population size of town, degree of rurality by town, percent below poverty in a town, percent Africa-American population in a town, and percent Latino in a town. Propensity score matching will help mitigate measurement error resulting from selection bias. I use a one to many Kernel matching algorithm to create the propensity score. As in similar analyses, I use the demographic characteristics of prison towns in a given period to estimate the propensity scores for that period. These results are available upon request.

3.3. Analytic Strategy

Although I use a similar approach to Hooks et al. [32], the analysis here is quite improved as I focus on how prisons slow economic decline by controlling for the propensity to build prisons. Given that prison-building is not a random occurrence, the fixed-effects and propensity matching models provide a more rigorous test of selection than the cross-sectional analysis used in prior research. Additionally, the primary unit of analysis is the U.S. Census place, as opposed to the county. As previously argued, U.S. Census places provide an improved way to measure local stratification at the municipal level. To this end, I consider how newly constructed prisons impact U.S. Census place (1) median home values; (2) median family income; (3) poverty; (4) and unemployment. Each analysis on existing and newly constructed rural prisons will run across the following periods: Whether or not a town built a prison prior to 1969; whether or not a town built a prison between 1969 and 1978; whether or not a town built a prison between 1979 and 1988; and whether or not a town built a prison between 1989 and 1998. Outcomes will be presented in each period as prison town versus non-prison town across the South and the nation. The estimates reported for median home value and family income are adjusted for inflation based on the consumer price index for 2014.

The initial stage of the analytical plan provides descriptive statistics of prison towns versus all other towns within each period. To begin, means of independent variables of towns receiving prisons versus all other rural towns will be compared on a national and regional basis to differentiate prison impact. These statistics provide a baseline for comparing prison and non-prison towns. Next, I examine predictors of prison-building across U.S. regions, with a focus on the South. In comparing means by region, I expect variation across race, poverty, and total population of towns. In the second analytical stage, I estimate each dependent variable, beginning with median home value as a function of prison-building across each period. The primary analysis is a state-level fixed effects regression measuring for state-level variation in prison impact at the U.S. Census place. Like other studies that control for state level variation in imprisonment rates [69,70], controlling for variation in state effects enhances the regression model because state prison-building policies differ widely. Thus, we should expect variation for state-level fixed effects models. Given that the majority of work on prison impact finds a negative relationship with prison-building, I triangulate the entire population of rural prisons across ordinary least-squared regression (OLS), propensity scores model, and a state-level fixed-effects model to verify results. Propensity score matching is used in regression models to account for differences across and within similar levels of analysis. Using propensity scores matching provides a robustness check of the outcomes in the fixed-effects analysis. Detailed results for each analysis are available upon request. Only the main effects for the state-level fixed effects models across each period appear in the final table.

4. Results

4.1. The Prison Boom

In 2010, there were 1663 U.S. prison facilities with an average daily population of 758 prisoners and 231 staff. Of these, roughly 81% are state-run, 9% are federal (including Native American and military prisons), and 10% are private. Nearly 70% of these facilities are located in non-metropolitan communities. Prison-building is not a random event; some towns are more conducive to prison-building than others. For instance, Cañon City, Colorado, has seven of the 34, or 21%, of prison facilities in that state. Texas built 133 facilities, or 8% of all U.S. prisons. Huntsville, Texas is home to seven prisons, while Beaumont and Gatesville, Texas house six facilities each. Through 1969, there were only 523 prisons constructed across the United States (see

Table 2. Descriptive statistics of all variables included in the analysis from 1970 to 1980.

Variables	1970				1980
	Prison Towns		Rural Towns		Prison Towns		Rural Towns
	All	South	All	South	All	South	All	South
Dependent Variables
Median Home Value ($)	…	…	…	…	106,695	993,034	91,721	80,074
Median Family Income ($)	…	…	…	…	47,919	45,396	45,657	41,857
Poverty (%)	13	19.7	15.7	21.3	17.74	19.09	18.42	22.07
Unemployment (%)	4.3	3.7	4.6	4.3	8.01	6.8	7.48	6.52
Explanatory Variable(s)
Prison-Building, 1969–1978	…	…	…	…	85	42	…	…
Prison Built before 1969	233	120	…	…	…	…	…	…
Control Variables
Total population	16,170	18,000	13,990	15,843	13,571	19,498	8892	9021
Black (%)	13.7	26.2	13.3	22.9	13.8	25.72	15.28	26.37
Hispanic (%)	2.8	2.9	6.7	7.1	3.89	4.06	5.74	4.94
Owner-Occupied (%)	61.9	58.6	63.6	63.3	65.8	66.99	69.36	70.6
College Degree (%)	10.2	10	9.2	8.7	13.3	11.25	10.96	9.9
Same Residence (as prior census %)	54	54.4	55.8	56.1	54.6	56.11	55.86	58.6

Source: Prison Proliferation Project.

).

Figure 2 demonstrates the unevenness of the prison boom over time. This period stands in stark contrast to building patterns during any other period in U.S. history. Between 1970 and 2010, 1140 correctional institutions were built across the United States. The rate of prison-building over this period varies as well. Between 1970 and 1979, 185 prisons were constructed. This decade stands apart as the first to average annual double-digit openings, with the high-water mark occurring in 1976, with 37 new prisons. Overall, the 1970s mark an annual increase in prison-building of 2.6%. During the 1980s, this annual rate increased to 3.1% as 323 new facilities were constructed. The 1990s stand out within the prison boom, witnessing the creation of 462 new prisons. During the apex of the prison boom, from 1990 to 1999, nearly a quarter of all prisons in U.S. history were constructed. After 2000, building slowed just below the prison boom rate, with only 170 facilities coming online, for an annual average of 17 during that decade. This slowdown is not much less, however, than the 18.5 average annual rate of prison-building during the 1970s. The limited sample of facility renovations shows that while new prison-building was slowing down in the 2000s, renovations were ramping up, with over 250 facilities undergoing renovations between 2000 and 2010.

During the prison boom between 1970 and 2010, every state built at least one facility. Of the 1140 prisons built during this era, Texas alone accounted for over 11%, building 133 new facilities. Indeed, Texas’s prison-building pattern is representative of the overall boom. For example, from 1970 to 1979, Texas built four prisons. This rate quintupled from 1980 to 1989, as 21 new facilities were erected. Similarly, from 1990 to 1999, 80 new prisons were constructed in this state, with 30 being opened in 1995 alone. Since 2000, Texas has only constructed five new prisons. Following Texas, the most prolific prison-building states are Georgia, Florida, Oklahoma, and New York. Between 1970 and 2010, Florida, New York, Oklahoma, and Georgia constructed 69, 58, 55, and 52 new facilities, respectively. These four states collectively account for nearly 21% of prisons constructed during the boom. Rounding out the top prison-building states during the boom are Illinois (46), California (43), North Carolina (38), Virginia (35), and Michigan (32). Taken together, these 10 states account for 49% of U.S. prison-building during the boom.

4.2. Prison Towns vs. Rural Towns

Table 3. Descriptive statistics of all variables included in the analysis from 1990 to 2000.

Variables	1990				2000
	Prison Towns		Rural Towns		Prison Towns		Rural Towns
	All	South	All	South	All	South	All	South
Dependent Variables
Median Home Value ($)	86,643	76,779	88,987	80,617	85,662	76,180	96,846	87,558
Median Family Income ($)	47,545	42,046	46,807	43,873	48,125	44,515	51,315	47,155
Poverty (%)	22.85	26.99	21.71	24.53	20.62	23.81	20.1	23.3
Unemployment (%)	9.83	9.11	8.58	8.4	7.78	7.92	8.8	8.59
Explanatory Variable(s)
Prison-Building, 1989–1998	…	…	…	…	314	196	…	…
Prison-Building, 1979–1988	159	73	…	…	…	…	…	…
Control Variables
Total population	10,122	10,693	9599	10,059	9875	10,334	10,634	10,774
Black (%)	18.55	35.13	16.6	27.67	19.87	29.16	17.86	33.28
Hispanic (%)	5.98	3.7	6.5	6.12	12.19	13.15	6.78	5.51
Owner-Occupied (%)	64.02	64.87	66.05	67.4	64.2	63.93	62.11	62.32
College Degree (%)	12.4	11.55	12.52	11.53	13.22	12.78	15.08	14.02
Same Residence (as prior census %)	52.46	54.56	55.41	57.62	53.68	54.66	54.11	55.44

Source: Prison Proliferation Project.

provides descriptive statistics of all variables in the analysis in nonmetropolitan U.S. Census Places with a RUCC > 4 in 1974 (see Table 1). These variables are further segmented along several key axes, demonstrating substantial differences between towns that constructed prisons in any given period versus those that did not. Given that the South has built more prisons in every period, I pay special attention to that region in describing prison-building trends. Almost two-thirds (62.4%) of the 314 prisons built in rural towns between 1989 and 1998 were erected in Southern states.

There was not a period during the prison boom where the South built less than 45% of all rural U.S. prisons. Prison-building is a rare event in any period. Prior to 1969, only 177, or roughly 1.3%, of all rural towns sited prisons. In the next decade alone, 67 different towns gained a prison. Between 1979 and 1988, roughly 95% of rural towns built a prison. This rate more than doubled between 1989 and 1998, with 1.9% of all rural towns building at least one prison. The rate of prison-building fell to a pre-boom low between 1999 and 2008, as only 108 rural towns constructed prisons. Prior to 1969 rural towns that built a prison were, on average, poorer, had larger total populations, and more Latinos when compared to rural towns nationally and within the South. This period was also marked by similarities between prison towns’ and rural towns’ rates of unemployment, same residence as the prior census, and residents with college degrees across the South and nationally.

Towns that built prisons between 1969 and 1978 had, on average, a median home value $5200 higher than that of towns that did not construct a prison. Furthermore, the average population of a town building a prison during that same period was 4679 greater than other rural towns. The difference in population was even more pronounced in the South, with Southern prison towns having, on average, twice the total population of other rural towns. Curiously, however, in this period the percent of African-Americans and Latinos in towns building prisons was lower when compared to towns that did not site prisons.

Towns that constructed prisons between 1979 and 1988 resembled rural towns that did not build prisons more than in any other period. The most significant change in towns siting prisons during this period was the rate median home value change between towns that did and did not build prisons. In the prior period, towns that did not site a prison had, on average, a lower median home value. This period witnessed a reversal in the trend. This change was most notable in the South where, on average, median home values increased by 38% from the prior period for all rural Southern towns, compared to a 24% increase for rural Southern prison towns.

While poverty increased in all rural towns from the first to the second period of prison-building, the shift was greater in towns that built prisons. Southern towns building prisons experienced, on average, nearly an 8% increase in poverty from the prior period, whereas towns that did not witnessed a modest 2.5% increase in poverty. One would expect an adverse selection of prison placement by income and poverty, given that prisons are more likely to be sited in towns with lower median family incomes and increased poverty rates after 1990. The racial composition of towns building prisons also changed between these periods, with increases in African-American and Latino populations (the only exception being Southern rural towns that did not site a prison). Similar to trends in poverty, the rate of minority populations grew faster in towns that constructed a prison than in towns that did not. Most notably, the percent of African Americans between 1969 and 1978 in Southern towns that built prisons was 25.72%. During the 1979 to 1988 period, on average, 35.13% of the population was African-American in Southern towns that constructed a prison.

In the period prior to 1980, there were nominal differences in median family income between prison towns and other rural towns, with prison towns on average, having higher median incomes. However, between 1989 and 1998, this trend shifted, with prison towns having, on average, lower median family incomes. With respect to racial composition, the average rural town and prison town witnessed a loss in its white population and gains in its African-American and Latino populations during the prison boom. The fastest growth was witnessed in prison towns, where the percent of Latinos tripled since the first period. The average percent of African-Americans in rural towns only increased 2.58% between period one and period three. On the other hand, the percent of African-Americans in prison towns increased during that same period from nearly 14% to 20%. Regarding unemployment, there was no descriptive difference in any period between prison towns and rural towns.

4.3. Impact on Rural Towns across the Prison Boom

4.3.1. Estimating Prison Impact

Table 4. State-level fixed-effects estimates of rural property value, family income, poverty, and unemployment as a function of recent prison-building from 1980 to 2000.

Variables	1980		1990		2000
	All Rural Towns	Southern Towns	All Rural Towns	Southern Towns	All Rural Towns	Southern Towns
	FE	FE	FE	FE	FE	FE
Median Home Value ($)
Prison-Building, 1989–1998	…	…	…	…	−936	−492
Prison-Building, 1979–1988	…	…	−504	−4488 *	1237	2570
Prison-Building, 1969–1978	7355 *	10,139 *	4631 *	−2483	−1989	786
Prison Built before 1969	4384 *	1534	475	878	2180	1244
Median Family Income ($)
Prison-Building, 1989–1998	…	…	…	…	−418	−492
Prison-Building, 1979–1988	…	…	−63	−614	1065	2571
Prison-Building, 1969–1978	1893	3542 **	583	1380	−2246	786
Prison Built before 1969	574	−373	831	860	1720	1244
Poverty (%)
Prison-Building, 1989–1998	…	…	…	…	0.89	0.94
Prison–Building, 1979–1988	…	…	1.67 *	1.01	0.94	1.69
Prison-Building, 1969–1978	−0.21	−1.71	0.6	−0.24	0.89	0.71
Prison Built before 1969	0.05	1.2	0.3	0.01	0.84	1.26
Unemployment (%)
Prison-Building, 1989–1998	…	…	…	…	−1.01 *	−0.49 *
Prison-Building, 1979–1988	…	…	0.62	−0.12	−0.56	0.4
Prison-Building, 1969–1978	0.05	0.04	−0.42	−0.3	0.19	0.19
Prison Built before 1969	0.06	0.77 *	0.13	−0.69	−0.42	0.59

Source: Prison Proliferation data; * p < 0.05; ** p < 0.01; *** p < 0.001.

presents multivariate fixed effects regression results estimating the impact of prison-building on the political economy of rural places from 1970 to 2000. Please see the Table S2 in the Supplementary Materials for robustness checks. The specific variables examined here include U.S. Census place median family income, median home value, unemployment, and poverty. Again, median home value and family income were adjusted for inflation based on the 2014 consumer price index. There are three periods of prison-building used to estimate the effects prisons have on rural towns in the decennial census years of 1980, 1990, and 2000. The first period of prison-building between 1969 and 1978 corresponds with outcomes in 1980; the second, 1979–1988 corresponds with outcomes in 1990; and third, 1989–1999 corresponds with outcomes in 2000. Given the numerous robustness checks and control variables used across the multiple analysis, I present fixed-effects coefficients for recent prison-building in each row of the table below.

4.3.2. Estimating Prison Impact in 1980

I begin by considering the impact of prison-building on median home values. Controlling for both U.S. Census place-level demography—including total population, percent of residents residing in town from the previous census, poverty, percent African-American, percent Latino, and owner occupancy—and prison-building in prior periods, prison-building was associated with an increase of in median home value in 1980. This increase is not a trivial amount in rural areas where poverty is higher and median family income is lower. Prison-building results in an estimated increase median home value of more than $7300. This analysis also establishes that prison-building before 1969 had a positive and statistically significant impact on median home values in 1980. The small coefficient from the fixed-effects model highlights the importance of state-level variance in prison-building. In 1980, 80% of the variance in median home values can be accounted for by state-level differences. Next, I consider median family income controlling for total population; percent of adults above age 25 completing a bachelor’s degree; percent African-American; and percent Latino. There was a statistically significant association between recent prison-building between 1969 and 1978 and an increase of over $4300 in median family income in Southern rural towns in 1980. Estimates of poverty are consistent with median family income and median home value. Overall, the impact of newly constructed prisons is not associated with poverty in 1980. However, after controlling for the prior decade’s poverty, percent African-American, and percent Latino, prison-building reduced poverty in rural Southern towns by around 2% on average. Prison-building prior to 1969 also had a positive and statistically significant association with median home values in 1980, suggesting that the towns most likely to build a prison prior to the prison boom were already on a downward economic trajectory. Unemployment is the final outcome considered. I control for total population, prior decade’s unemployment, percent African-American, and percent Latino. The fixed-effects model shows no statistically significant relationship between prison-building and unemployment in this period. Prior research shows that there is a large, positive, and statistically significant relationship between prison-building and unemployment [31,32,33]. Results from this period demonstrate that even when prisons have negative impacts on rural communities, the impact is not as devastating as suggested by prior research.

4.3.3. Estimating Prison Impact in 1990

Here, I consider the relationship between prison-building from 1979 to 1988 and the political economy of rural towns in 1990. Similar to Section 4.3.2, I begin with median home values. I control for prison-building in prior periods, total population, prior decade’s unemployment, median home value, percent of residents that are in the same residence as the prior census, percent African-American, and percent Latino across all models for median home value. Comparisons between towns in this period to those that adopted prisons in the earlier period demonstrates a drastic shift from prisons providing a boost in median home value to decreasing median home value in rural Southern towns.

While early adopters reaped benefits, prison-building offered little economic reprieve to towns siting prisons during this period. Overall, there is little association between prison-building and outcomes in this period. The state-level, fixed-effects models suggests that prison-building may harm locals by increasing poverty by 1.67% in towns that sited a prison. Towns that site a prison in this period appear to be more economically challenged, on average, before getting a prison when compared to other rural towns. Prison-building in this period does not do much to alleviate these challenges—the only silver lining for towns building prisons in this period is that rural Southern towns, on average, receive a statistically significant reduction in unemployment. While this relationship does not pan out across the fixed-effects and propensity score models, this finding points to a shift in trends regarding how prison-building impacts unemployment. The move from a positive association of prisons and unemployment to a negative association is the most significant relationship in the final period of the prison boom.

4.3.4. Estimating Prison Impact in 2000

Given the limited impact of prison-building, I begin by discussing unemployment in this period. Controls used for measuring unemployment include prison-building in prior periods, total population, prior decade’s unemployment, percent African-American, and percent of adults above age 25 completing a bachelor’s degree. State-level fixed-effects models show that prison-building between 1989 and 1998 yields minimal unemployment reduction in rural towns. This suggests a more protective effect of prison-building as opposed to the rise in unemployment experienced in the first period. This continues the unemployment trend from the second period (1979 to 1988).

For median home values, the negative trend that began in the prior period continues. Again, for median home values, I control for prison-building in prior periods, total population, prior decade’s median home value, percent of residents that are in the same residence from the prior census, and percent African-American. Prison-building has no statistical association with poverty or median family income during this period.

5. Discussion and Conclusions

The findings here both support and challenge what we know about punishment. First, there are significant effects across periods of prison-building. The analysis here demonstrates that, on average, towns that adopted a prison in the first period of the prison boom (1969–1978) experienced an increase in median home value and median family income, and a reduction in poverty. Despite the slight increase in unemployment during this period, the evidence presented here suggests that prison-building may have been a key mechanism that slowed economic decline for towns that adopted prisons in the first period of the prison boom. Overall, prisons provided a short-term economic boon in some periods for rural communities that built prisons. While prison-building increases median home value, reduces unemployment, and eases poverty, these effects are not lasting.

Examining the effects of prisons across different periods of national economic hardship (the 1980s) and prosperity (the 1990s) allows us to discuss how broader economic trends impact rural towns. While prisons may increase or decrease unemployment depending on the period of building, the findings also suggest that prisons replaced jobs lost across struggling rural communities that built prisons. In other words, towns that built prisons between 1989 and 1998 experienced, on average, a less pronounced economic decline than similarly situated towns. Moreover, when economic conditions were more favorable, prisons provided a boost to local economies by reducing unemployment. However, the impact of a prison on unemployment does not seem to carry over from prior decades, suggesting a decaying effect. This decaying effect may be linked to broader downward economic shifts in rural communities. Because the impact of prison-building on poverty in 1980 was statistically insignificant in 1990 and 2000, the effect of prisons on poverty may decay over time as well. The lagged effects of mass imprisonment and prison-building help shape the context of early versus late adopters of prison placement. Prison-building not only lessened economic strain during trying economic times in the 1980s, it also improved conditions for towns that built them. The benefits of prison-building do not extend through the 1980s in large part because of deindustrialization and the mechanization of farming in rural communities. Moreover, there was a sustained economic downturn during this period. In addition, the fixed-effects analysis controls for factors that produce prison-building, so this analysis is more sensitive to the downward economic spiral many rural communities building prisons were affected by. Given that prisons may not necessarily produce dire consequences in rural communities [2,3,31,32,33,35], how should we make sense of prison impact?

If the prison boom can be explained as a function of economic and racial stratification (e.g., racial threat and the conflict perspective), how then can we reconcile the benefits prisons provide to communities of color? Hawkins [71,72] challenges the use of the conflict perspective in explaining mass imprisonment. Informed by these studies, I consider the conflict perspective as it applies to the prison boom. For example, in 2000, African-Americans and Latinos made up 22% and 7%, respectively, of all correctional officers in the USA [17]. Combined with the fact that prisons are more likely to be built in communities with higher percentages of African-Americans and Latinos [20], we can imagine a fair number of jobs in corrections going to people of color in rural communities. The presence of a significant number of African-Americans and Latinos in correctional officer positions in rural communities complicates a central theme of racial dominance that currently exists in the prison boom literature. In addition to social control, poverty management, racial exploitation, and stigma, prisons appear to offer jobs to disadvantaged rural communities with higher concentrations of people of color. Given this evidence, one can argue that the benefits and harms of prisons need to be understood according to the preferences of rural communities most likely to build.

Rural communities of color disproportionately building prisons suffer alienation before the facility arrives, an effect that is seldom considered. I suggest that we rethink the prior literature using Ward [17]:

Research has been less attentive to ethnoracial group agency in justice processes, including the ways race and ethnicity relate to policy interests, organizational cultures, levels of influence, and eventual distributions of services and sanctions within justice systems themselves. Imbalanced attention to the subjectivity of marginalized groups in justice processes and neglect of actual and potential group agency not only limits appreciation of the complex dynamics of radicalized social control but may also reinforce stereotypes about and its racial dimensions.

([17], pp. 67–68)

Unlike Wacquant (2010), I am not attempting to “slay the chimera of the ‘Prison Industrial Complex’” ([70,71]; [73], p. 607). Instead, I am constructing a more empirically based, theoretically driven, methodologically rigorous framework that takes a holistic approach to considering how race matters in advancing the study of prison proliferation. This line of research situates the study of prison impact within a theoretical perspective that anchors the rural prison town at the center. From this perspective, I suggest that the prison, as a complex institution, should be understood as neither panacea nor pariah for rural towns. Moreover, because the prison boom cannot be explained as simply a function of mass imprisonment, by using the sociology of punishment [74], we can consider how prison-building, as an object of punishment, impacts communities, and how communities in turn impact prisons. Specifically, using a revised penal–industrial complex framework, we can think about how the findings presented here also challenge the existing paradigms of the prison proliferation.

When considering the prison boom, scholars overwhelmingly find that prison-building is bad for communities of color [2,3,30,32,33,34,35]. Some even claim that prison-building is a form of environmental racism for rural places [34,35]. However, previous studies do not have national data on all prisons properly assigned to locales. These studies also do not consider the importance of region (e.g., the South) in their analysis. The Prison Proliferation Project is unique in that it contains every prison, geocoded into each U.S. Census place, over time. Furthermore, I deploy more sophisticated analysis, such as fixed-effects models and propensity score matching. These techniques are more sensitive to the counterintuitive findings on prison-building. This study serves as a starting point for investigating prison impact across multiple periods and places. Future studies should look deeper into the political process at work in prison-building. This means including more predictors at the local town and state levels, as well as other macro predictors. My intention here is to complicate and provide a more nuanced approach (e.g., racial threat and conflict perspectives) to the prison boom by suggesting that we need to probe this phenomenon across multiple scales of inequality.

I suggest that we rethink not only what and when we measure, but also the theoretical framework we use to understand prison-building. As Eason [20] demonstrates, implicitly or explicitly, many scholars use the Marxian exploitation narrative of the prison industrial complex as the dominant explanation for prison-building. This narrative influences empirical work that is chiefly concerned with documenting the inequality created by prison-building. While I admire this position (I originally came to study prison towns from this perspective), it is primarily informed by a scholar-activist agenda concerned with bringing an end to the prison-industrial complex. This is a highly normative framework reflected in the empirical findings of scholarship on this topic. If one only measures for negative consequences of key predictors on an outcome, chances are an analysis will yield a negative relationship. The prison-industrial complex oversimplifies the dynamics of race and class in accounting for the nuanced process of prison-building. Recently there has been a move towards a Weberian explanation of prison-building using the penal-industrial complex [37]. Building from this study, the penal-industrial complex should be understood as the economic, social, and political institutions related to the causes and consequences of the prison boom. This more neutral perspective allows scholars to explore the empirical challenges and opportunities posed by prison-building in rural communities. A Weberian perspective can also aid policy-makers in minimizing the unintended consequences of rolling back the prison boom because it allows for an assessment of the benefits and harms prisons cause to rural communities.

The causes and consequences of the prison boom provide a vital nexus to understand how criminal justice policy produces stratification in rural America and how rural communities in turn impact systems of punishment. By understanding how prisons impact the political economy of rural towns, the source of prison “demand” and continued dependence on prisons for rural communities can be directly linked to mechanisms that continue to drive the prison boom. Furthermore, understanding why rural towns “demand” prisons allows us to explain prison proliferation as more than just a function of mass imprisonment. It is also important to address the implications of sustaining the prison boom amidst continued economic uncertainty and state budget shortfalls. However, the implications of prison boom studies have primarily been focused on private prisons [3], despite most U.S. prisons being operated by states. The past 40 years of prison proliferation have fundamentally changed the purpose of the prison in rural America. The prison boom, like mass imprisonment, results in stratification through the redistribution of scarce state resources (e.g., prisons and prisoners) across rural and urban communities. However, unlike the corrosive effects of mass imprisonment, I assert that the prison now serves as a state-sponsored public works program for disadvantaged rural communities of color.

To be clear, I am not for advocating for prison-building as a poverty reduction or economic growth strategy for disadvantaged rural communities of color. Nor am I am calling for the pending rollback of mass imprisonment to be suspended. I am, however, cautioning that the discontinuation of prison-building, or an outright reduction in the number of prisons in rural communities, may have collateral consequences, including increasing poverty in rural communities with higher concentrations of people of color. Given the urban focus of research on penal expansion, the expansive literature documenting the deleterious impact of the criminal justice system on low-income communities of color, and the stigma associated with prisons, it should not be surprising that the positive impacts of prison-building on rural communities have not yet been unearthed. These benefits, however, are not trivial and could help explain prison expansion, despite considerable declines in crime and increases in the cost of imprisonment in a time of relative fiscal constraint, particularly in Southern states. I believe it also does a disservice to the cause of rolling back mass imprisonment to insist that prisons have negative effects on rural towns when that claim does not seem to align with rural residents’ experiences. Characterizing a stigmatized institution like a prison as beneficial may seem counterintuitive. However, given the dire straits, limited development options, and stigma associated with many disadvantaged rural communities, town officials may view a prison as the least-worst economic development option. To curb future rural prison “demand” and continued dependency, we must provide alternatives to the perverse economic incentives of prison-building. To better investigate why rural towns build prisons and the impact of these facilities on the town, I make several suggestions for future investigations. First, the economic effects across the racial divide of prison towns (e.g., white prison versus black or Latino town) should be differentiated. Furthermore, future investigations should consider the size and number of prison facilities. Specifically, we should examine whether larger prisons create a more sustained economic effect versus the effects of multiple prisons built in a single town.