SCUXXX10.1177/2329496520974006Social CurrentsMcGee et al. 974006research-article2020

Article

Social Currents 1 –15 Locked into Emissions: © The Southern Sociological Society 2020 Article reuse guidelines: sagepub.com/journals-permissions How Mass Incarceration DOI:https://doi.org/10.1177/2329496520974006 10.1177/2329496520974006 Contributes to Climate journals.sagepub.com/home/scu Change

Julius Alexander McGee1 , Patrick Trent Greiner2, and Carl Appleton3

Abstract The phenomenon of mass incarceration has dramatically altered the economic and infrastructural landscape of the United States. These changes have numerous implications regarding the use of fossil fuels, which are the single largest contributor to climate change. The present study argues that mass incarceration creates three social patterns that result in significant increases in industrial emissions. (1) Mass incarceration incentivizes further industrial development through the construction of new prisons and the continued maintenance of existing prisons to house prisoners. (2) The needs of the millions of individuals currently incarcerated in the United States incentivize industrial expansion through the production of goods and materials used inside prisons. (3) Incarcerated individuals are being used to reduce the cost of labor, which expands . We construct several fixed-effects panel regression models with robust standard errors predicting industrial emissions for U.S. states from 1997 to 2016 to assess how increases in the number of individuals in U.S. state, federal, and private prisons is correlated with industrial emissions over time. We find that increases in incarceration within states are associated with increases in industrial emissions, and that increases in incarceration lead to a more tightly coupled association between per capita and industrial emissions.

Keywords mass incarceration, climate change, prison industrial complex, environmental sociology

Introduction have been explored extensively (see Davis 2003; Davis and Barsamian 1999; Schlosser The number of individuals who are incarcer- 1998), little attention has been given to the ated in the United States is a source of revenue environmental consequences of the PIC. for over 3,100 private corporations (Urban

Justice Center 2018). Broadly speaking, these 1 3,100 corporations are a part of what is known Portland State University, OR, USA 2Vanderbilt University, Nashville, TN, USA as the prison industrial complex (PIC), a set of 3George Mason University, Fairfax, VA, USA political, bureaucratic, and economic interests that are driven by profit to increase imprison- Corresponding Author: Julius Alexander McGee, Portland State University, Post ment (Cooper et al. 2016). While the political, Office Box 751, Portland, OR 97207-0751, USA. economic, and social implications of the PIC Email: [email protected] 2 Social Currents 00(0)

Pellow (2018, 2019) has demonstrated that by exploring how changes in the percentage of prisons are loci for environmental injustice, people in the United States who are incarcer- exacerbating preexisting forms of social ated accelerates states’ contribution to climate inequality such that prisoners are dispropor- change, which accelerates the global depletion tionately exposed to environmental harm. of ecosystems. We use a series of fixed-effects There are numerous reasons to suspect that panel regression models with robust standard the PIC also contributes to environmental errors, and unit-specific intercepts for all 50 harm. On one hand, incarceration works as a U.S. states and the District of Columbia, as “locus for the coercion of demand and con- well as time dummies from 1997 to 2016 to sumption,” compelling those who would oth- estimate how the percent of individuals impris- erwise marginally participate in markets to oned in state, private, and federal penitentia- become active, perpetual consumers (Deckard ries is associated with industrial CO2 emissions. Delia 2017). The U.S. government on average We argue that as the percentage of people spends $485 million annually on capital for incarcerated increases over time, the demand incarceration, which goes toward funding con- to construct and maintain prisons will increase struction and renovation in prisons (Henrichson the amount of fossil fuels used in industrial and Delaney 2012). A number of corporations development. Furthermore, we argue the within the PIC are the beneficiaries of this increasing percentage of incarcerated people spending and directly contribute to industrial will result in more emissions from industrial emissions by burning fossil fuels to produce processes as economic activity grows, because products used to build and maintain prisons incarceration facilitates industrial expansion (Urban Justice Center 2018). through coerced of industrial In addition, some entities within the PIC goods and increased industrial activity within have benefited financially from prison work prisons. This process stimulates both economic programs that use prison labor to manufacture activity and emissions, resulting in a tighter industrial equipment for private companies, coupling of economic growth and industrial state governments, and the U.S. federal govern- emissions. The goal of our study is to encour- ment (Goodridge, Jantz, and Leslie Christian age further discussion on the environmental 2018). Many prison programs, consequences of mass incarceration by dem- such as the federal employment program onstrating one way in which incarceration UNICOR,1 pay prisoners as little as $0.23 to contributes to environmental degradation. $1.15 per hour (far below the federal minimum paid to non-incarcerated United States The Political of citizens) for their labor (UNICOR 2019), Mass Incarceration which helps to stimulate industrial growth by reducing the cost of labor. Mass incarceration fills a void left by a shrink- The consumption of industrial goods by ing welfare state onset by neoliberal policies incarcerated people and the use of incarcerated (Wacquant 2009). While impoverished com- labor for industrial development form the basis munities, particularly poor communities of of our hypothesize that the PIC contributes to color, have historically been disproportion- industrial emissions by perpetuating the tread- ately incarcerated in the United States, the era mill of production (henceforth ToP) (Gould, of mass incarceration is distinct in that it exerts Pellow, and Schnaiberg 2015; Schnaiberg a systematic influence on the livelihood of 1980). ToP is a phenomenon that occurs when poor people in general, and young Black men decreases in the social efficiency of natural in particular (Holzer, Offner, and Sorensen resource use within the productive sphere 2005; Western and Wildeman 2009). increases environmental degradation. The In the 1970s, employment in manufacturing present study seeks to explore how mass incar- began to decline in the United States due to ceration decreases the social efficiency of nat- outsourcing, rising , and interna- ural resource use within the productive sphere tional trade (Elwell 2004; Shapira and Teitz McGee et al. 3

2017). Harvey (2003) argued that these shifts unequal criminalization of drug use in Black were part of a neoliberal restructuring effort and brown communities resulted in the dispro- within global capitalism that saw corporate portionate use of militarized police forces, such elites reduce the power of labor and revolu- as SWAT teams, to terrorize poor Black and tionary movements through various political brown neighborhoods (Alexander 2012). and economic actions. This included a number During this same period, the American of international coups in favor of regimes that Legislative Exchange Council (ALEC) spent were sympathetic to the labor needs of multi- millions of dollars lobbying state and federal national corporations, and policies in the government(s) to change laws around non-vio- global North that reduced the size of the wel- lent crimes, arrestable offenses, and to extend fare state (Streeck 2014). sentencing periods for various crimes, which Reacting to the increased mobility of capi- functioned to increase the total number of incar- tal—itself a result of neoliberal governance cerated individuals (Cooper et al. 2016). The approaches—local and state governments use of mass surveillance in Black and brown reduced taxes on capital gains and large corpo- communities during this period—aimed at sup- rations to attract private investment in munici- pressing resistance—also contributed directly palities (Harvey 2003; Streeck 2014), which to the rise of the carceral state (Thompson 2010; ultimately increased state borrowing. As gov- Wacquant 2009; Wang 2018). ernment revenues from taxes diminished and In addition to the explicit criminalization debt from borrowing increased, states began to and terrorization of Black and brown commu- cut funding for supplement programs to nities, the carceral state also targeted the avoid fiscal crises (O’connor 2017), which dis- broader gamut of communities heavily affected proportionately impacted poor communities of by industrial divestment (Shelden and Vasiliev color. Individuals in communities that resisted 2017). This included a disproportionate num- these changes became the target of the emerg- ber of poor white individuals as well as physi- ing carceral state (Camp 2016). cally and mentally disabled people (Rembis Camp (2016) argued that the modern car- 2014; Shelden and Vasiliev 2017; Thompson ceral state emerged by constructing racial ene- 2010). Wang (2018) described this phenome- mies out of those resisting the changes of non as carceral capitalism, where individuals neoliberalism. This legitimized the state and are policed on a continuum based on “probabi- enabled it to concentrate on the maintenance of listic ranking of subjects according to risk” order during a period of economic restructur- (125). These risks are determined based on the ing. In general, the rise of the carceral state “likelihood of criminal behavior,” which jus- under neoliberalism represents a shift in how tifies the surveillance of marginalized com- the state responds to civil disobedience. munities, and increases the probability of Whereas in the past the state responded to dis- incarceration for individuals within such com- obedience by expanding welfare, under neolib- munities. While neoliberalism led to the initial eralism, the state uses incarceration (among acceleration of incarceration, the PIC repre- other symbolically violent tactics) to respond sents a unique political-economic response to to mass unrest by criminalizing community the problems of mass incarceration—one that responses to economic shocks (Fording 2001). facilitates further economic growth while also For example, in the early 1970s the war on contributing to industrial emissions. drugs disproportionately targeted poor commu- nities of color ravaged by industrial divestment The Prison Industrial (Mauer 2006). This worked by criminalizing Complex’s Contribution to drug use within poor Black and brown commu- Industrial Emissions nities hurt by deindustrialization, despite the fact that rates of drug use are not significantly Beginning in the mid-1990s, a number of pri- different across race and class (Alexander vate corporations and politicians began to ben- 2012; Saxe et al. 2001). Furthermore, the efit financially and politically from the 4 Social Currents 00(0) existence of large prison populations (Elk and Across the United States, mass incarcera- Sloan 2011; Gilmore 2007; Hallinan 2003; tion increases the use of industrial equipment Herival and Wright 2007; Thompson 2010, by generating a demand for the construction of 2012). These groups poured substantial prisons. Between 1980 and 2004, 936 prisons resources into lobbying for more punitive were built in the United States, as compared laws, fewer restrictions on the use of prison with the 711 prisons built between 1811 and labor, and reduced regulation of private prison 1979 (Lawrence and Travis 2004). Federal and management. Collectively, these efforts com- state governments fund the construction of prise the PIC—a collection of corporations, many new prisons by issuing bonds to private organizations, and groups with collective investors. These investors create profits by col- interests that benefit from the existence of lecting interest from these bonds and entering large prison populations. into contracts with private prison operators, While the PIC profits financially from who in turn create profits by reducing the per mass incarceration in a multitude of ways, diem cost of housing incarcerated people there are three specific forms of profit that (Henrichson and Delaney 2012). The materials result in increased fossil fuel use within U.S. and processes involved in the construction of states: (1) private contractors responsible for new prisons, as well as the renovation of exist- producing equipment used to construct and ing prisons, require substantial amounts of fos- maintain prisons benefit from the increased sil fuels (Huling 2002). The outcome of this is demand for prisoner housing; (2) private an overall increase in states’ use of fossil fuels. corporations responsible for producing the goods used by prisoners, such as beds and Increased Demand for Goods clothing, benefit from the coerced consump- Necessary for Day-to-Day Living tion of an ever-growing prison population; within Prison (3) federal, state, and private industries ben- efit from their contracts for cheap labor pro- Housing people in prison requires the consis- vided by prisoners. tent provision of numerous goods and services, many of which are necessary for day-to-day The Increased Demand for Prison living, and are produced using fossil fuels. The Construction United States spends over $80 billion per year on prisons and more than half of this money is The PIC encourages demand for goods used spent on transactions with the thousands of pri- to construct and maintain prisons by trans- vate corporations that provide services to and forming socioeconomic crises into economic within prisons (Urban Justice Center 2018). opportunities; this phenomenon is known as Prisoners require beds, clothing, hygiene prod- Disaster Capitalism (Klein and Peet 2008). ucts, and furniture among other goods, which Specifically, the socioeconomic conditions are produced by private corporations with the that lead to mass incarceration often benefit explicit aim of profiting from prison popula- private corporations who receive state and tions. The production of clothing and other federal funds to house prisoners. For example, textiles, some of which are used exclusively after inmate populations in New York had within prisons, rely on carbon intensive pro- fallen to record lows in the early 1970s, harsh cesses that contribute to climate change (Kerr drug laws doubled the prison population and Landry 2017). Private corporations in the within a decade—creating a crisis of over- United States are responsible for producing crowding. In response, the state turned to its these materials for prisoners, who represent to Urban Development Corporation—which had these corporations a source of revenue. For the authority to issue bonds without voter example, privately traded prison supply com- approval—to finance a 12-year, $7-billion panies manufacture and distribute mattresses, project to construct new prisons and prison uniforms, hygiene products, and other items cells across the state (Schlosser 1998). for prisons around the United States. These McGee et al. 5 companies have continuously expanded their protecting such companies from the highly- production over time in response to a growing politicized and unsavory optics associated demand brought about by consistent increases with moving production processes abroad in the size of the incarcerated population (Weiss 2001). States-level political elites and (Gargan 2016). industry actors have also benefited from prison labor via the indirect suppression of Prison Labor Used for Industrial wages. Access to prison labor has been used as Production a point of leverage in neoliberal efforts to dis- mantle labor unions, and to eliminate collec- The use of prison labor to produce industrial tive bargaining rights for private sector goods also contributes to industrial develop- workers more broadly (Thompson 2012). This ment and, as a result, industrial emissions. is especially true in Virginia, Ohio, New Prison labor is used to subsidize federal Jersey, Florida, and Georgia, where state expenditures by creating a monopoly on gov- inmates have been paid not in dollars, but in ernment contracts for textile production time off their sentences (UNICOR 2017). We (Reynolds 1997). For example, UNICOR, a note that a number of recent studies have federal work program, touts fair wages and found that union participation suppresses CO2 recidivism reduction, while their program’s emissions (Alvarez, McGee, and York 2019; explicit intention is to extract labor to navigate Hyde and Vachon 2019). Considering as budgetary challenges abundantly clear in their much, we argue that the use of incarcerated annual reports. As they state, “[i]n fiscal years labor to undermine unions may also reduce of 2018 and beyond, [Federal Prison the tendency of U.S. unions to curb Industries] will continue to face the challenge emissions. of shrinking federal budget and the resulting decline in discretionary spending” (UNICOR The Treadmill of Mass 2017). In reaction to these budgetary chal- Incarceration lenges, UNICOR devised a five-year plan to increase their workforce by 4,000 workers Mass incarceration and the PIC are part of the with the goal of growing sales by 8 percent in “Treadmill of Production” (Gould et al. 2015; 2018 and 5 percent per annum through 2022 Schnaiberg 1980). ToP is a recurring pattern of (UNICOR 2017). Programs such as UNICOR relations between capital, labor, and the state work with federal and local lawmakers to that perpetuates production through increased mandate government agencies purchase goods resource extraction on an ever-expanding scale. from federal prison factories (Helfenbein Many elements of ToP directly correspond to 2017). This is especially true in the textile the political economic circumstances discussed industry, where Federal Prison Industries above, offering further insights into the rela- (FPI) sold over $126 million in clothing and tionship between the PIC and emissions. textiles during the 2017 fiscal year, the major- In general, ToP analyzes the relationship ity of which went to the Department of between capital, labor, and the state as fol- Defense (UNICOR 2017). The United States lows—as investments of capital into research textile industry is the fifth largest emitter of and technology that reduce labor time increase,

CO2 within the United States (Energy positions available to laborers decrease. In Information Association [EIA] 2018). response, laborers turn to the state to imple- Furthermore, prison labor acts to incentivize ment policies that expand economic opportu- private sector companies who may be tempted nities for workers. Similarly, investors look to to move their production abroad or who the state to support private capital accumula- already have done so to open factories in the tion through increased spending and deregula- United States. This is done by enticing compa- tion. Because of these dynamics, capital, labor, nies to take advantage of cheap domestic and the state are caught in a “treadmill of pro- prison labor within the United States, while duction,” where surplus generated by increased 6 Social Currents 00(0) production through resource extraction is con- 2013, p. 92). In the context of the present tinually used to support and maintain the needs study, economic restructuring sparked by neo- of capital investment, labor, and the state. liberal policies—which shifted a large portion Under ideal conditions of the treadmill, of industrial production in the United States to increased production is met with increased the global South (Harvey 2003)—in addition consumption, new investment in physical capi- to reduced funding for wage supplement tal, and a redistribution of excess surplus by programs (Korteweg 2003)—which served to the state into wage supplement programs, such stabilize economic conditions in the past— as unemployment and social security. However, led to social disorganization in communities Schnaiberg (1980) acknowledged that the ideal that were made to be economically reliant on conditions of the treadmill are not always industrial production and the welfare state. realized, which creates tensions between each These structural changes created the condi- facet of the treadmill. For example, wage sup- tions for mass incarceration, and the carceral plement programs alleviate tension between state emerged in reaction to this social capital and labor by lifting people out of disorganization. poverty, putting more money into the hands Mass incarceration intercedes in the tread- of consumers, and increasing consumption mill by alleviating many of the tensions that (Danziger, Haveman, and Plotnick 1981). drive treadmill expansion, particularly under However, they also pressure employers to neoliberalism. In this way, it also creates space increase wages to incentivize work for entry- for the logic of accumulation that characterizes level employees, which raises the cost of pro- the treadmill process, allowing it to continue duction (Piven and Cloward 1993). Under unhindered. This process is facilitated by the neoliberalism, the state facilitates the accumu- state, which alleviates the tension between lation of capital at the expense of labor (Harvey labor and capital through enhanced policing 2003). Specifically, taxes on capital are tactics that target the communities and indi- reduced, limiting the state’s ability to fund viduals most harmed by capital’s divestment wage supplement programs. The carceral state from labor and reduced wage supplements represents a shift in the balance between capi- from the state (Camp 2016; Fording 2001). As tal and labor. With limited taxes to fund wage rates of incarceration increase over time, pris- supplement programs, the state uses violence ons and incarcerated people become a site for to facilitate the accumulation of capital capital investment through expanded resource (Kohler-Hausmann 2015). extraction (Deckard Delia 2017). Mass incar- Stretesky, Long, and Lynch (2013) expand ceration creates an increasing number of indi- ToP by acknowledging the dialectical relation- viduals in need of food, housing, and other ship between crime and ecology. Building on amenities necessary for day-to-day living, social disorganization theories in criminology, which are provided through processes of capi- which claim that crime is the result of penuri- tal investment within the PIC. Government ous economic conditions, poor social bonds, spending on incarceration, around $30,000 per and changing cultural values (Sampson and prisoner per year, stimulates a much more reli- Groves 1989; Shaw and McKay 1942), able source of demand than expenditures on Stretesky et al. (2013) argued that expanding wage-supplement programs (Deckard Delia the treadmill can also cause social disorganiza- 2017). As Deckard (2017) wrote, “it [spending tion. Specifically, they contend that moments on incarceration] is compulsory, dependable of social disorganization “often stem from and efficient, and [it comes] without the wage exploitative relationships between the ecology inflationary consequences that previous forms and the economy, such as the rapid extraction of public expenditure on the impoverished cre- of natural resources, armed conflict over natu- ated” (p. 5). ral resources, and the potential impacts of pol- In addition to imprisoning the individuals lution on humans that weakens social bonds to most heavily affected by capital’s labor cost conventional ” (Stretesky et al. reduction strategies, incarceration circumvents McGee et al. 7 the contradictions embedded in the treadmill With respect to the first component of our by reducing the need for investment into labor modeling approach, we hypothesize that as the saving processes, and it does so by providing percentage of incarcerated people within states industries with cheap labor. Low cost prison increases over time, demand for goods used in labor is increasingly being used in industrial prisons will increase, which will in turn aug- manufacturing (see Goodridge et al. 2018). ment the amount of fossil fuel burned to pro- Unlike contract-based wage-labor rendered by duce these goods, leading to an overall increase choice, prison labor is coerced and less pro- in the amount of CO2 emitted from industrial tected by the state. This allows surplus to be production within states. With respect to the extracted from prison labor more efficiently second component of our modeling approach, than from wage-labor. In sum, mass incarcera- we hypothesize that as the percentage of incar- tion refashions the relationships between the cerated people increases within states, the cost facets of the treadmill by suppressing the con- of producing industrial goods will decrease tradictions derived from , due to the increased supply of prison labor. As while simultaneously perpetuating the ToP in a result, the overall productive capacity of creating a space for new, more reliable forms states will increase, which will increase indus- of capital investment. trial production. This will lead to an overall increase in fossil fuel use from industrial pro- Modeling Approach/Research duction, and will augment the association Question between gross domestic product (GDP) per capita and industrial emissions per capita. The goal of our modeling approach is two- The logic of the latter component of our fold: (1) analyze how the PIC contributes to approach is in line with previous studies that industrial emissions directly by measuring have explored the decoupling of economic the association between emissions and the growth and emissions (McGee and Greiner percentage of people in prisons and (2) ana- 2018; Thombs 2020; York and McGee 2017). lyze how incarceration influences the rela- Similar to these studies, our goal is to assess if tionship between changes in socioeconomic activity (in this and emissions. In general, this approach is case the percentage of individuals housed in intended to assess the degree to which incar- prisons) de/couples the association between ceration increases or decreases the social effi- GDP per capita and emissions. In these mod- ciency of natural resource use. To capture the els, decoupling refers to an outcome where essence of ToP, CO2 emissions from indus- the association between GDP per capita and trial production are used to measure the social emissions decreases when it exists in the social efficiency of natural resource use. The logic context of a state where a greater number of underlying this choice is as follows: fossil individuals have been incarcerated. Conversely, fuels are a natural resource that, when burned coupling refers to an outcome where the asso- during industrial development, contribute to ciation between GDP per capita and emissions climate change through the emission of CO2, increases when that association is set in a and as a result accelerate the depletion of social context where more individuals have ecosystems (Turner 2018). As Gould, Pellow, been incarcerated. and Schnaiberg (2004) noted, “decreased social efficiency of natural resource utiliza- Data and Methods tion produced . . . a shift towards vastly increased rates of ecosystem depletion.” In We constructed fixed-effects panel regression this sense, any significant, positive associa- models with robust standard errors that account tion between incarceration and emissions for clustering in 50 states and the District of constitutes a decrease in the social efficiency Columbia from 1997 to 2016. We chose years of natural resource use through the perpetual within states as our unit of analysis because depletion of ecosystems via climate change. state-level data provide the most reliable and 8 Social Currents 00(0)

robust estimates of CO2 emissions and rates of policies or allow private prisons to operate. We incarceration. Furthermore, mass incarceration then created a dummy code for each unit (year is a phenomenon specific to the United States within specific states) that either identified (see discussion above) that has accelerated the state as having cap and trade/private pris- over time within states more than it has accel- ons or not. When these variables are included, erated across states. Our model includes fixed- our standard errors increase; however, none effects estimators for states/political units and increase above a p of .05. Furthermore, years. Such an approach allows for the control the coefficients for these dummy codes all had of a number of contemporaneous (e.g., year to p values above .05, meaning private prisons year change in the cost of goods) and extempo- and cap and trade did not substantially change raneous (e.g., number of previously existing industrial emissions in the years measured industrial facilities, or long term state specific within our models. We also explored if the policies) factors that are not possible to include relationship between incarceration and indus- in analyses with the currently existing data. As trial emissions changed substantively in states such, our model is set up to capture how incar- with cap and trade or in states with private ceration is associated with changes in emis- prisons by interacting these dummy codes with sions over time within states (and the District our variable for incarceration. We found that of Columbia) because much of the variation there is no statistically significant difference in within our variables of interest appear to be the relationship between incarceration and structured this way. We also note that the emissions between states with cap and trade approach used in this study is in line with pre- policies, neither is there a statistically signifi- vious studies of inequality and emissions that cant difference in this same relationship within have used similar units of analysis (see states that allow private prisons. Jorgenson, Schor, and Huang 2017). We do not control for the age structure of the In line with the STIRPAT (Stochastic population, since our data on employment are a Impacts by Regression on Population, more accurate assessment of the size of the Affluence, and Technology) tradition, we working population, which is one of the pri- transformed of all our variables into natural mary intentions behind controlling for this log form. As a result, our model is multiplica- variable. We did not include a control for tive as opposed to additive (York, Rosa, and urbanization in the models presented here Dietz 2003). In addition, converting our vari- because we could not find consistent, reliable ables to natural log form allows our coeffi- data on urbanization at the state level from cients to express the elastic relationship 1997 to 2016. Specifically, the data for the per- between our independent variables and our centage of the urban residents are only avail- dependent variables, meaning their relation- able in the year 2000 and from 2010 to 2016. ship is expressed in relative proportions (i.e., Robustness checks that incorporated these data our coefficients express the relative associa- into our models found that the limited sample tion between a percent change in our depen- size (2000 and 2010–2016 as opposed to 1997– dent variable and a 1 percent increase in our 2016) increased our standard errors; however, independent variables). We chose to use GDP controlling for urbanization did not substan- per capita in chained dollars, as our models tially change the magnitude or direction of the assess change over time and chained dollars coefficients of interest in any models. allow for a more accurate assessment of eco- Each of our models use states as the unit of nomic growth. In additional robustness checks, analysis and include unit-specific intercepts as we created dummy codes for states that have well as time dummies to control for general cap and trade policies, as well as for states with period effects. Our panels are balanced. Our private prisons, to serve as state-level controls modeling approach controls for any effects that account for political factors that vary that are constant over the span of time exam- across states. These data were created by iden- ined for each state, such as geographic and tifying which states either have cap and trade geological characteristics, and any effects that McGee et al. 9 are constant across states for a given point in Independent Variables time. Fixed-effects allows us to estimate how changes in our independent variable correlate Our main indicator variable in this analysis is with changes in our dependent variable. The the percent of incarcerated individuals in state, benefit of this approach is that it controls for federal, and private prisons. These data were any unobserved time-variant and time-invari- obtained from the Bureau of Justice Statistics ant factors not directly accounted for in our (2018) National Prisoner Statistics (NPS). models. As a result, we only include variables NPS excludes data on people held in local in our models that are sourced from reliable jails and in other jurisdictions. Our data for data, and that have been shown to be associ- the number of incarcerated individuals in the ated with emissions. All reports of statistical District of Columbia stop after 2001. After significance are based on a .05 alpha level with 2001, responsibility for sentenced prisoners a two-tailed test. from the District of Columbia was transferred to the Federal Bureau of Prisons (Bureau of Justice Statistics 2018). Dependent Variables Each of our models incorporate a number of independent variables. In line with the All of our models use CO2 emissions from industrial production (all in million metric STIRPAT tradition, we control for the total tons) as a dependent variable. The data for this population of each state using data from the variable were obtained from the Energy U.S. Census Bureau (2018) and GDP per cap- Information Association (EIA) in December of ita (measured in millions of 2012 dollars) of 2018. According to the EIA (2018) “state-level each state using data from the Bureau of emissions estimates are based on energy con- Economic Analysis (BEA 2018). We also sumption data for the following fuel catego- control for other variables that have been ries: three categories of coal, natural gas, and found to be significantly associated with ten petroleum products including—asphalt emissions at the state level, such as the per- and road oil, aviation gasoline, distillate fuel, centage of individuals who are employed jet fuel, kerosene, hydrocarbon gas liquids (BEA 2018), and the percentage of GDP from (HGL), lubricants, motor gasoline, residual manufacturing (BEA 2018). Finally, we con- fuel, and other petroleum products.” Industrial trol for the percent of the population that is emissions measure both combustion and pro- below the poverty line (U.S. Census Bureau cess emissions from industrial development. 2018) because people in poverty are more This includes the consumption of asphalt and likely to be incarcerated. road oil, for which emissions are calculated based on the state where asphalt and road oil Results were consumed. While a number of green- house gases are known to contribute to the Table 1 shows the descriptive statistics of all geophysical phenomena of climate change, the variables used in our analysis in their raw here we choose to focus on CO2 emissions form (i.e., prior to their log transformation). from industrial processes because (1) no other The results from our analysis are presented in greenhouse gases contribute so greatly to the Table 2. In Model 1 of Table 2, both population threat of global climate change as carbon and GDP per capita are found to be associated dioxide emissions borne from socioeconomic with increases in emissions, and are signifi- processes (IPCC 2014), (2) no other green- cantly different from zero. These results are house gases emissions are as extensively esti- consistent with the findings of previous mated as CO2 (i.e., it offers the greatest data research (Jorgenson et al. 2017; York et al. coverage), and (3) no other greenhouse gas 2003). The independent variables for poverty, emissions are as intimately bound up with the employment, and manufacturing as a percent industrial processes that are impacted by the of GDP are not found to be significantly differ- growing PIC as CO2. ent from zero in Model 1, which indicates that 10 Social Currents 00(0)

Table 1. Descriptive Statistics—1,001 Observations in 50 States.

Covariates Mean Median SD Max. Min. Industrial emissions 19.04 11.30 30.07 238.80 0.10 (million metric tons) GDP per capita $0.048 $0.046 $0.010 $0.148 $0.005 (millions of 2012 U.S. dollars) Incarcerated population 0.422 0.404 0.165 1.517 0.111 (percent of total) Total population 4,226,018 5,977,525 6,588,466 39,300,000 489,451 Employed population 62.46 62.90 4.65 73.10 42.70 (percent of total) Impoverished population 40.45 40.15 5.55 62.25 22.78 (percent of total) Manufacturing 12.13 11.47 5.54 37.43 0.358 (percent of GDP)

Note. All emissions are reported in million metric tons. GDP measures are reported in millions of 2012 U.S. dollars. SD = standard deviation; GDP = gross domestic product.

Table 2. Fixed-Effects Regression Model with Robust Standard Errors of the Relationship between Prison Population, Economic Size, and Industrial Emissions from 1997 to 2016.

Covariates Model 1 Model 2 Incarcerated population 0.480*** 2.008** (percent of total) (.128) (.692) GDP per capita 0.748** 1.283*** (millions of 2012 U.S. dollars) (.247) (.379) Total population 0.555* 0.567* (.235) (.215) Employed population 0.205 0.366 (percent of total) (.639) (.616) Impoverished population −0.066 −0.017 (percent of total) (.214) (.209) Manufacturing 0.053 0.060 (percent of GDP) (.091) (.092) Incarcerated population × GDP _ 0.484* (.212) R2 within .323 .334 N/political units 1001/51 1001/51

Note. Standard errors in parentheses. Political unit groups include the 50 U.S. states and the District of Columbia. GDP = gross domestic product. *p < .05. **p < .01. ***p < .001 (two-tailed tests with 0 as null hypothesis). they are not significantly associated with each one unit increase in incarceration is asso- industrial emissions. In models not shown ciated with a .481 percent increase in industrial here, we estimated the association between emissions. incarceration and industrial emissions without Model 2 in Table 2 includes the same vari- these variables and found the results to be sub- ables as Model 1, but also interacts percent stantively unchanged. Model 1 in Table 2 also incarceration and GDP per capita. Here the shows that incarceration is positively associ- interaction of percent incarceration and GDP ated with industrial emissions. Specifically, per capita is positive and significant. This McGee et al. 11

Table 3. Estimated GDP Per Capita Slope Coefficients for Table 2, Model 2.

Incarcerated population values GDP per capita 1st percentile 0.158 (0.09%) (.291) 25th percentile 0.495* (.196%) (.226) Median 0.730** (.319%) (.232) 75th percentile 0.889*** (.444%) (.260) 99th percentile 1.114*** (.705%) (.322) Figure 1. Figure 1 displays the estimated

association of GDP per capita to CO2 emissions Note. Standard errors in parentheses. Model 2 (vertical axis) conditioned upon the percent of the coefficients represent the estimated association population that has been incarcerated (horizontal between GDP and industrial emissions, conditional on axis). incarcerated population as a percent of the total. GDP = gross domestic product. *p < .05. **p < .01. ***p < .001 (two-tailed tests with 0 population), GDP per capita has an elastic rela- as null hypothesis). tionship with industrial emissions (i.e., a 1 per- cent increase in GDP per capita is associated indicates that the associations of these two with a 1 percent increase in emissions). Figure variables with industrial emissions are inter- 1 further illustrates the findings of Model 2. twined—meaning that the association between Looking to Figure 1, we note that the GDP per capita and emissions is conditioned coefficient for GDP per capita is higher in upon the percent of the population that is incar- state-years where incarceration is higher. This cerated, and that incarceration’s association to indicates that incarceration leads to a tighter emissions is conditioned upon the value of coupling between GDP per capita and indus- GDP per capita. In general, this finding sug- trial emissions. gests that the association between GDP per capita and industrial emissions is higher in Discussion and Conclusion those contexts where a higher proportion of the population is incarcerated. Although mass incarceration accounts for a Table 3 helps to interpret the findings in small percentage of states’ overall population, Model 2, indicating the estimated association our findings suggest that increases in incarcer- between GDP per capita and emissions at the ation are significantly associated with increases 5th, 25th, 50th, 75th, and 95th percentiles of in industrial emissions, as well as a tightening incarceration. Upon examining Table 3, it is of the association between economic growth clear that the association between GDP per and industrial emissions within individual capita and emissions increases as the rate of states over time. It should be noted that our incarceration increases, and that this associa- models only assess how incarceration influ- tion is significant at the 25th, 50th, 75th, and ences changes in industrial emissions from 95th percentiles of incarceration. For instance, year-to-year. Thus, in a broader context we according to the results of Model 2, in a state argue that mass incarceration is escalating the where the rate of incarceration is equivalent to treadmill of production that characterizes the the median (.319 percent of its population), a 1 maintenance, expansion, and intensified use percent increase in GDP per capita is associ- of infrastructure that contributes to climate ated with a .730 percent increase in industrial change. Theoretically, these findings support emissions. Yet, in a state at the 95th percen- our hypothesis that mass incarceration is part tile of incarceration rates (.706 percent of its of the ToP—a cycle of economic development 12 Social Currents 00(0) that continually uses more natural resources to empirical sense, all we can truly say is that support capital accumulation (Gould et al. incarceration is associated with higher emis- 2015; Schnaiberg 1980). As incarceration sions. We believe that only further study increases, actors within the PIC are able to can help to properly identify the feasibility of generate higher profits by producing more of the mechanisms that we propose underlie the goods used to build and maintain prisons, these trends. There are additional implica- and industrial firms are provided with an ever- tions behind our results that future research larger U.S. labor source that offers globally should explore. As one reviewer noted, it is competitive wage costs. possible that there is a connection between The relationship between mass incarcera- the policies passed that support the PIC and tion and the PIC is most likely one of histori- policies that make it easier to emit industrial cal contingency, where mass incarceration emissions (e.g. environmental deregulation). develops as a state response to social change Thus, future research exploring how policies perpetuated by neoliberalism, and the PIC that have supported mass incarceration cor- develops as an economic response to mass respond with the rejection or enactment of incarceration. These structural changes are environmental legislation would prove bene- indicative of the dialectical pattern noted by ficial. For future researchers, the task at hand Stretesky et al. (2013) between ecological dis- is to further explore this phenomenon as our organization and social disorganization. To knowledge of the social processes underlying this end, one can understand the relationship the PIC deepens and more data, such as the between mass incarceration, the PIC, and the proportion of state GDP drawn from the PIC, environment as the result of converging social become available. To this end, we hope that patterns that are linked through a number of our findings set a precedent and provide moti- historical contingencies. Clearly, it is not a vation for assessing the relationship between natural law that growth in incarceration mass incarceration, the PIC, and climate will increase industrial emissions. However, change. historically, industrial manufacturing has exploited workers, consumers, and the envi- Declaration of Conflicting Interests ronment by continually reducing the cost of The author(s) declared no potential conflicts of labor, increasing the demand of industrial interest with respect to the research, authorship, goods, and increasing the use of fossil fuels. and/or publication of this article. Incarceration allows these patterns to continue unabated, and in many instances provides the Funding tools necessary to accelerate the pace at which The author(s) received no financial support for the such patterns recur. The state helps to facilitate research, authorship, and/or publication of this the exploitation of workers by providing cheap article. competitive labor to industrial manufacturers, while also supporting the accumulation of ORCID iD capital through the creation and exploitation of Julius Alexander McGee https://orcid.org/ an ever-increasing captive consuming class 0000-0002-2087-6865 (Deckard Delia 2017). Industrial manufactur- ers must continually use more fossil fuels to Note produce these goods, resulting in increases in 1. UNICOR is a Federal Prison Industry program emissions over time. (operating under the trade name UNICOR). It It should be noted that our explanations for was established in 1934 by an Executive Order these findings are strictly theoretical, though issued by President Franklin D. Roosevelt. On we argue that they fit within the conceptual January 1, 1935, FPI officially began opera- framework outlined above as well as could be tions as a wholly-owned corporation of the expected from any theoretical lens. In a strict, United States Government (UNICOR 2019). McGee et al. 13

References Competing Theories of the State.” American Political Science Review 95(1):115–30. Alexander, Michelle. 2012. The New Jim Crow. Gargan, Henry. 2016. “Detention Supplier Bob New York: The New Press. Barker Co. Plans $4.15M Expansion of Fuquay- Alvarez, Camila Huerta, Julius Alexander McGee, varina Facility.” The New Observer. Retrieved and Richard York. 2019. “Is Labor Green? A November 5, 2020 (https://www.newsobserver. Cross-national Panel Analysis of Unionization com/news/local/community/southwest-wake- and Carbon Dioxide Emissions.” Nature and Culture 14(1):17–38. news/article116652023.htm). Bureau of Economic Analysis. 2018. Retrieved Gilmore, Ruth Wilson. 2007. Golden Gulag: December 3, 2018 (https://www.bea.gov/). Prisons, Surplus, Crisis, and Opposition Bureau of Justice Statistics. 2018. Retrieved in Globalizing California. Berkeley, CA: November 15, 2018 (https://www.bjs.gov/ University of California Press. index.cfm?ty=nps). Goodridge, Julie Mari Schwartzer, Christine Jantz, Bureau of Labor Statistics. 2018. Retrieved and Leslie Christian. 2018. “Prison Labor in November 15, 2018 (https://www.bls.gov/). the United States: An Investors Perspective.” Camp, Jordan T. 2016. Incarcerating the Crisis: Menasha, WI: North Star Asset Management. Freedom Struggles and the Rise of the Gould, Kenneth A., David N. Pellow, and Allan Neoliberal State. Berkeley, CA: University of Schnaiberg. 2004. “Interrogating the Treadmill California Press. of Production: Everything You Wanted to Cooper, Rebecca, Caroline Heldman, Alissa Know about the Treadmill but Were Afraid Ackerman, and Victoria A. Farrar-Meyers. to Ask.” Organization & Environment 17(3): 2016. “Hidden Corporate Profits in the U.S. 296–316. Prison System: The Unorthodox Policy- Gould, Kenneth A., David N. Pellow, and Allan making of the American Legislative Exchange Schnaiberg 2015. Treadmill of Production: Council.” Contemporary Justice Review 19(3): Injustice and Unsustainability in the Global 380–400. Economy. New York: Routledge. Danziger, Sheldon, Robert Haveman, and Robert Hallinan, Joseph T. 2003. Going Up the River: Plotnick. 1981. “How Income Transfer Travels in a Prison Nation. New York: Random Programs Affect Work, Savings, and the Income House. Distribution: A Critical Review.” Journal of Harvey, David. 2003. The New Imperialism. Economic Literature 19(3):975–1028. Oxford, England: Oxford University Press. Davis, Angela Yvonne. 2003. Are Prisons Obsolete? Helfenbein, Rick. 2017. “Private Industry Can’t New York: Seven Stories Press. Compete If Government Prefers Prison Labor.” Davis, Angela Yvonne and David Barsamian. 1999. Retrieved March 3, 2019 (https://thehill.com/ The Prison Industrial Complex. Oakland, CA: blogs/pundits-blog/labor/314079-private- Ak Press. industry-cant-compete-if-govt-prefers-prison- Deckard Delia, Natalie. 2017. “Prison, Coerced labor). Demand, and the Importance of Incarcerated Henrichson, Christian and Ruth Delaney. 2012. The Bodies in Late Capitalism.” Social Currents of Prisons: What Incarceration Costs 4(1):3–12. Taxpayers. Washington, DC: VERA Institute Elk, Mike and Bob Sloan. 2011. “The Hidden of Justice. History of ALEC and Prison Labor.” The Herival, Tara and Paul Wright. 2007. Prison Nation, p. 1. Profiteers: Who Makes Money from Mass Elwell, Craig Kent. 2004. Deindustrialization Incarceration. New York: The New Press. of the US Economy: The Roles of Trade, Holzer, Harry J., Paul Offner, and Elaine Sorensen. Productivity, and Recession. Washington, DC: 2005. “Declining Employment Among Young Congressional Information Service, Library of Black Men: The Role of Incarceration and Congress. Child Support.” Journal of Policy Analysis and Energy Information Association. 2018. “State Management 24:329–50. Carbon Dioxide Emissions.” Retrieved Huling, Tracy. 2002. “Building a Prison Economy December 3, 2018 (https://www.eia.gov/envi- in Rural America.” Pp. 197–213 in Invisible ronment/emissions/state/). Punishment: The Collateral Consequences of Fording, Richard C. 2001. “The Political Response Mass Imprisonment, edited by M. Mauer and to Black Insurgency: A Critical Test of M. Chesney-Lind. New York: The New Press. 14 Social Currents 00(0)

Hyde, Allen and Todd E. Vachon. 2019. “Running Reynolds, Morgan. 1997. “The Economic Impact of with or against the Treadmill? Labor Unions, Prison Labor.” Retrieved May 13, 2019 (http:// Institutional Contexts, and Greenhouse Gas www.ncpathinktank.org/pub/ba245). Emissions in a Comparative Perspective.” Sampson, Robert J. and W. Byron Groves. 1989. Environmental Sociology 5(3):269–82. “Community Structure and Crime: Testing IPCC. 2014. Synthesis Report, edited by Core Social-disorganization Theory.” American Writing Team, R. K. Pachauri, and L. A. Meyer. Journal of Sociology 94(4):774–802. Jorgenson, Andrew K., Juliet B. Schor, and Xiaorui Saxe, Leonard, Charles Kadushin, Andrew Huang. 2017. “Income Inequality and Carbon Beveridge, David Livert, Elizabeth Tighe, Emissions in the United States: A State-level David Rindskopf, Julie Ford, and Archie Analysis, 1997–2012.” Ecological Brodsky. 2001. “The Visibility of Illicit 134:40–48. Drugs: Implications for Community-based Kerr, John and John Landry. 2017. “Pulse of the Drug Control Strategies.” American Journal of Fashion Industry.” Global Fashion Agenda. Public Health 91(12):1987–94. Retrieved January 7, 2020 (http://globalfashion Schlosser, Eric. 1998. “The Prison-industrial agenda.com/wp-content/uploads///Pulse-of- Complex.” The Atlantic Monthly 282(6):51–77. the-Fashion-Industry_.pdf2017052017). Schnaiberg, Allan. 1980. The Environment: From Klein, Naomi and Richard Peet. 2008. “The Shock Surplus to Scarcity. New York: Oxford Doctrine: The Rise of Disaster Capitalism.” University Press. Human Geography 1(2):130–33. Shapira, Philip and Michael Teitz. 2017. “Growth Kohler-Hausmann, Julilly. 2015. “Guns and Butter: and Turbulence in the California Economy. The Welfare State, the Carceral State, and the Regional Research Institute.” Pp. 81–281 in Politics of Exclusion in the Postwar United Deindustrialization and Regional Economic States.” The Journal of American History Transformation, edited by Lloyd Rodwin and 102(1):87–99. Hidehiko Sazanmi. New York: Routledge. Korteweg, Anna C. 2003. “Welfare Reform and Shaw, Clifford and Henry D. McKay. 1942. Social the Subject of the Working Mother: ‘Get a Disorganization Theory: Social Structure, Job, a Better Job, Then a Career.’” Theory and Communities, and Crime. Chicago, IL: Society 32(4):445–80. University of Chicago Press. Lawrence, Sarah and Jeremy Travis. 2004. “The Shelden, Randall G. and Pavel V. Vasiliev. 2017. New Landscape of Imprisonment: Mapping Controlling the Dangerous Classes: A History America’s Prison Expansion.” Washington, of Criminal Justice in America. Waveland Press. DC: Urban Institute, Justice Policy Center. Streeck, Wolfgang. 2014. Buying Time: The Mauer, Marc. 2006. Race to Incarcerate. New Delayed Crisis of Democratic Capitalism. New York: The New Press. York: Verso Books. McGee, Julius Alexander and Patrick Trent Greiner. Stretesky, Paul B., Michael A. Long, and Michael J. 2018. “Can Reducing Income Inequality Lynch 2013. The Treadmill of Crime: Political Decouple Economic Growth from CO2 Economy and Green Criminology. New York: Emissions?” Socius 4:2378023118772716. Routledge. O’connor, James. 2017. The Fiscal Crisis of the Thombs, Ryan P. 2020. “In-and-beyond State Power: State. New York: Routledge. How Political Equality Moderates the Economic Pellow, David N. 2018. “Political Prisoners and Growth-CO2 Emissions Relationship, 1990- Environmental Justice.” Capitalism Nature 2014.” The Sociological Quarterly 12:1–20. Socialism 29:1–20. Thompson, Heather Ann. 2010. “Why Mass Pellow, David N. 2019. “Struggles for Incarceration Matters: Rethinking Crisis, Environmental Justice in US Prisons and Jails.” Decline, and Transformation in Postwar Antipode 0:1–18. American History.” The Journal of American Piven, Frances Fox and Richard Cloward. 1993. History 97(3):703–34. Regulating the Poor: The Functions of Public Thompson, Heather Ann. 2012. “The Prison Industrial Welfare. New York: Vintage Books. Complex: A Growth Industry in a Shrinking Rembis, Michael. 2014. “The New Asylums: Economy.” New Labor Forum 21(3):39–47. Madness and Mass Incarceration in the Turner, C. 2018. Climate Change and Biodiversity. Neoliberal Era.” Pp. 139–59 in Disability Scientific e-Resources. Essex, UK: ED Tech Incarcerated. New York: Palgrave Macmillan. Press. McGee et al. 15

UNICOR. 2017. “2017 Annual Report.” Retrieved Wang, Jackie. 2018. Carceral Capitalism. Vol. 21. March 4, 2019 (https://www.unicor.gov/publi- Cambridge, MA: MIT Press. cations/reports/FY2017_AnnualMgmtReport. Weiss, Robert P. 2001. “Repatriating Low-wage pdf). Work: The of Prison Labor UNICOR. 2019. “FDI General Overview.” Reprivatization in the Postindustrial United Retrieved January 18, 2019 (https://www States.” Criminology 39(2):253–91. .unicor.gov/FAQ_General.aspx#4/). Western, Bruce and Christopher Wildeman. 2009. U.S. Census Bureau. 2018. Retrieved November “The Black Family and Mass Incarceration.” 15, 2018 (https://www.census.gov/). Annals of the American Academy of Social Urban Justice Center. 2018. “The Prison Industrial and Political Science 621:221–42. Complex: Mapping Private Sector Player.” York, Richard and Julius Alexander McGee. Retrieved January 18, 2019 (https://static1.square- 2017. “Does Renewable Energy Development space.com/static/58e127cb1b10e31ed45b20f4/t/5 Decouple Economic Growth from CO2 ade0281f950b7ab293c86a6/1524499083424/ Emissions?” Socius 3:2378023116689098. The+Prison+Industrial+Complex+-+Mapping+Pr York, Richard, Eugene A. Rosa, and Thomas ivate+Sector+Players+%28April+2018%29.pdf). Dietz. 2003. “STIRPAT, IPAT and ImPACT: Wacquant, Loïc. 2009. Prisons of Poverty. Analytic Tools for Unpacking the Driving Minneapolis, MN: University of Minnesota Forces of Environmental Impacts.” Ecological Press. Economics 46(3):351–65.