Sage Journals: Discover world-class research

Abstract

This article explores the relationship between supervisory approaches to governance, punishment, and poverty among people with drug convictions. Tying government assistance to supervision could improve employment and economic outcomes. However, if experienced as punishment, recipients may forgo financial assistance and be more likely to experience poverty. Using information on policies that restrict access to welfare for people with drug felony convictions in the Temporary Assistance to Needy Families (TANF) and the Supplementary Nutrition Assistance Programs (SNAP), this paper makes two contributions. First, it documents state variation in the balance between supervision and punishment in these bans. Second, using data from NLSY97, it estimates how individuals’ likelihood of being in poverty is related to state SNAP drug ban policies. States have shifted away from overtly punitive policies denying access to welfare toward policies that increase supervisory requirements, especially for SNAP. This shows that punitiveness extends beyond work activation programs like TANF. Additionally, poverty among people with drug convictions is almost half in no ban states compared to those in full ban states. While poverty is lower in states that include supervisory requirements than in those for which a drug conviction fully blocks access to welfare, this difference was not statistically significant.

Keywords

poverty state variation supervision welfare

Introduction

Theorists contend that governance in the U.S. has shifted dramatically, with criminal justice and welfare systems increasingly working together to control the behavior of socially marginalized people through punishment and supervision (Gustafson, 2011; Wacquant, 2009). Punitive drug laws and other policies that have increased incarceration rates have been criticized as seeking to isolate, contain, and exclude marginalized segments of the population, especially African American men (Lynch, 2012). Welfare provision has also changed. Support has been redirected towards working families and away from the deeply poor (Shaefer et al., 2019). Beckett and Western (2001) argue for a resulting new penal-welfare regime, marked by increased incarceration rates and reduced welfare support for socially marginal groups.

In this view, punishment forms part of a wider system whose aim is to shape the behavior of poor people. Behavior modification includes setting expectations or goals; supervising people to make sure they are meeting those goals; and punishing those who fail to do so. As stated by Wacquant (2010: 199), both social welfare and penal policies are now “informed by the same behaviorist philosophy relying on deterrence, surveillance, stigma and graduated sanctions in order to modify conduct.” Importantly, this shift to supervision is often justified as a means to improve economic and other social outcomes of poor people. Rather than simply withdrawing state support, Lawrence Mead (1997) observes that this ‘new paternalism’ will increase the likelihood that the poor will constructively engage in programs that will help them overcome the personal barriers that keep them in poverty.

In the criminal justice system, this shift can be evidenced by the widespread use of community supervision, in which people are allowed to serve time in the community, as long as they meet the behavioral conditions imposed. Indeed, the majority of people under the supervision of the criminal justice system are in the community completing a period of probation or parole (Phelps, 2017). Supervision is also a key part of welfare provision. The focus of welfare services has changed from providing benefits with the aim of alleviating financial need to reducing welfare use, promoting marriage, as well as preparing and actively pushing recipients into work (Schram et al., 2010). Welfare case managers used to primarily determine financial eligibility for programs; they must now also match recipients with appropriate services to increase their employability, create job search requirements, monitor recipient compliance with these requirements, and decrease financial assistance (i.e. sanction) recipients who fail to do so (Raffass, 2017). Moreover, in both criminal justice and welfare systems, people who are supervised also become eligible for services beyond those available to other poor people, including job training and substance abuse treatment. Miller and Stuart (2017) have asserted that it is through their supervision that socially marginal groups become simultaneously eligible for both coercion and care.

Individuals may experience supervision as coercion, care, or a mix of both (Phelps, 2020). If experienced positively, one could argue that increased supervision may lead to improved outcomes, especially compared to punitive measures such as incarceration and the withdrawal of welfare support. However, both criminal justice and welfare scholars have shown that punitive supervision can produce negative emotional and financial consequences (Edin and Shaefer, 2015; McNeill, 2019; Sherman, 2013; Wright et al., 2020).

In this article, as an example of the new regime combining supervision and punishment, I use drug ban policies enacted by states in order to explore their effects on one financial outcome: poverty. These bans were enacted with the passage of Section 115 of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (P.L. 104–193), commonly called welfare reform. In this legislation, the U.S. federal government prohibits people with drug felony convictions from receiving financial support from two programs – Temporary Assistance to Needy Families (TANF) and the Supplementary Nutrition Assistance Program (SNAP). Although household members without drug felony convictions are still eligible to receive financial support and services, this provision has been widely criticized for its potential to increase poverty by limiting financial assistance to a group of people whose employment prospects are restricted both by their criminal record, as well as their need for substance abuse treatment (Allard, 2002; Godsoe, 1998; Hirsch, 1999).

Importantly, welfare reform also allows states to enact legislation to modify or opt out of the ban. Twenty years after the federal legislation was passed, there is considerable variation between states in their felony drug ban policies (Luallen et al., 2018; Martin and Shannon, 2020). Variation occurs because some states have opted out of the ban altogether, while others have enacted policies to allow people convicted of drug felonies to access benefits under specific conditions, including the imposition of increased supervisory requirements. Inherent differences between SNAP and TANF are an additional source of variation. While SNAP is a program aimed to help recipients seen as ‘deserving’ of government assistance because their financial need is due to factors outside their control, TANF is viewed as a program that primarily serves ‘undeserving’ mothers who are dependent on welfare and unwilling to work (Shaefer et al., 2019).

This paper makes two specific contributions. First, I document the nature of, and changes in, state-specific supervisory requirements in drug ban policies since the passage of welfare reform. Doing so addresses two shortcomings of existing attempts to describe variation in state drug ban policies. Despite the profound differences between SNAP and TANF, earlier studies simply focused on TANF bans or combined these two programs into a single policy choice (Luallen et al., 2018; Owens and Smith, 2012). These studies also pooled states that opted out and modified TANF bans into a single category, although recent research has shown that the state-level factors associated with these policy decisions are distinctive (Martin and Shannon, 2020). The present study addresses these issues. It does so by capturing state- and year-specific variation in drug ban policies separately for SNAP and TANF, classifying them into one of three mutually-exclusive categories: full bans, in which states adopted policies that ban people with drug felony convictions from receiving any financial assistance; partial bans, in which states only allow people to access assistance if they meet additional requirements; and no bans, in which states do not take drug convictions in account when determining eligibility. This three-level classification complements a recent contribution by Martin and Shannon (2020) on TANF, but extends coverage to SNAP, which has grown to become one of the most important programs in the social safety net (Parolin and Luigjes, 2019). I show that, as with TANF, this neglected middle category is of particular importance because partial bans are the policy choice that allows lawmakers to impose supervisory requirements. Importantly, I also show that there is substantial variation between state TANF and SNAP drug ban policies, which suggests that researchers need to extend the scope of their research beyond TANF to understand state approaches to welfare provision.

Armed with this improved understanding of state drug ban policies, the second contribution of this paper is to provide the first known estimates of how variation in SNAP drug ban policies may be related to poverty. While proponents of new paternalism assert that increased supervision will reduce poverty by encouraging recipients to engage in work and treatment, a wealth of welfare research on work activation programs has shown that arduous supervision prompts people to simply forego needed government support (Raffass, 2017; Wright et al., 2020). In the case of drug bans, it is possible that increased supervision will lead to poverty if it is strenuous enough to deter people from applying for or from remaining in the program. The focus on SNAP is important here because, unlike TANF, it imposes few behavioral requirements on recipients (Sugie, 2012). Relative to TANF, SNAP therefore ought to provide a clearer signal on the relationship of interest. Furthermore, SNAP caseloads overlap more with those individuals who are likely to be involved in the criminal justice system, namely men without children (Tuttle, 2019).

Overall, study findings suggest that supervisory requirements can be experienced as punitive even when part of relatively generous programs. More broadly, supervisory requirements may not lead to increased engagement in potentially beneficial programs, as envisioned by supporters of new paternalism. Instead, by showing the link between supervision and poverty, the study adds to the burgeoning literature on the potential punitiveness of coercive state supervision in both welfare and criminal justice systems.

Background

Drug addiction, punitiveness and supervision in the criminal justice system

Governmental responses to people with substance abuse problems are an interesting case example of how criminal justice and welfare systems blend supervisory with overtly punitive approaches. Substance abuse policies have oscillated between considering addiction as a medical condition to be treated and a moral failing to be punished (Wahler, 2015). In the criminal justice system, the substantial population incarcerated for drug-related offenses is evidence of a punitive response (Pearl, 2018). Meanwhile, alternatives to incarceration for people with substance abuse disorders have proliferated, including diversion and community supervision paired with drug treatment (Belenko et al., 2013).

Despite these developments, research finds that community drug treatment programs do not reach the majority of people in need of substance abuse treatment (Chandler et al., 2009), and there is little consensus about the types of treatment that are most effective (Hamilton et al., 2015). There is also considerable debate about whether the benefits derived from the services provided outweigh the potentially negative consequences associated with the close scrutiny that these services include (Zhang et al., 2013). McKim’s (2017) study of a residential substance abuse rehabilitation program for women showed that while the program studied was one of the few means through which poor women could access services and treatment, the program was more focused on disciplining behavior than providing resources. Women needed to reflect program staff’s view of them as ‘diseased and disordered,’ as well as stay sober, in order to succeed. Overall, McKim (2017) concluded that, while an alternative to incarceration, this program was also experienced as a form of punishment.

This confirms findings of other studies that community supervision can be another manifestation of punitiveness (McNeill, 2019; Natapoff, 2015; Phelps, 2017). In his study of community supervision in Romania, Durnescu (2011) identified ‘pains’ experienced by probationers, including financial costs and the deprivation of time. Community supervision also engendered ‘emotional pains’ among probationers deprivation of autonomy as they were consistently required to meet and provide various forms of evidence to the probation services; stigmatization by employers and neighbors resulting from their probation status; and the feeling of living under the threat of punishment. Along with punitiveness being experienced by the persistence of supervision, McNeill’s (2019) study in Scotland also revealed that probationers suffer due to being constructed as untrustworthy and unworthy of autonomy. Indeed, the idea of having to present oneself in a way that reflects the views of authorities in order to avoid punishment has also been documented among juvenile offenders (Cox, 2011), mothers serving community sentences (Haney, 2010), as well as incarcerated women (McCorket, 2013).

Drug addiction, punitiveness and supervision in the welfare system

Gustafson (2009) has documented a ‘criminalization’ of state policies relating to poverty in which the historical focus of welfare policies that have included the stigmatization, surveillance and regulation of the poor have become combined with two new elements from the criminal justice system. First, there has been increased collaboration between these systems, including the use of shared information systems (see also Gillom, 2001). Second, welfare policies are increasingly focused on addressing recipients’ ‘latent criminality.’ This focus is evident in efforts to control fraud among welfare recipients, as well as to limit access to aid for people with drug felony convictions (Gustafson, 2009).

Like most scholars studying the increased links between the criminal justice and welfare systems (Garland, 2001; Soss et al., 2011; Wacquant, 2009), Gustafson’s (2009, 2011) focus is on the TANF program, whose recipients have long been seen as undeserving of aid since their poverty is attributed to their own values and behaviors (Katz, 2013). Overall, TANF has become increasingly restrictive, as well as paternalistic for those left on the rolls (Handler and Hasenfeld, 2006). TANF recipients are told the proper ways to act, including engagement in the labor market, are closely supervised to make sure they are doing so (Perez-Munoz, 2017). Failure to meet these obligations results in financial penalties or sanctions. The process of establishing work requirements, monitoring compliance, and imposing sanctions translates into frequent meetings with case managers.

Although substance abuse issues are not widespread among TANF recipients and are less important than other employment barriers (Metsch and Pollack, 2005), the presumed link between substance abuse and receiving welfare has informed both explicitly punitive, as well as supervisory measures. Thirteen states now require some TANF applicants to undergo drug tests as part of establishing eligibility. If applicants test positive, their benefits could be terminated (CLASP, 2019). Some state TANF programs also include supervisory provisions blending coercion and care for people with substance abuse issues. Many states have adopted drug testing policies that require applicants who test positive to engage in treatment programs in order to receive assistance. To encourage participation, states can count the time spent in drug treatment as work-related activities. Rather than being exclusionary, these policies adopt a paternalistic stance and seek to increase supervision in order to shape the behavior of recipients to conform with societal expectations about proper behavior (Perez-Munoz, 2017).

Recipients’ experiences of supervision in TANF and other work activation programs closely mirror those revealed in studies of the criminal justice system. Specifically, recipients view supervision as punishment, rather than as a means to support them (McNeill, 2020). Mirroring the emotional pains of criminal justice supervision, welfare recipients report feelings of being continuously monitored (Jordan, 2018) and stigmatized (Seccombe et al., 1998). Recipients also describe pressure to act in ways that reflected how officials viewed them (Koch, 2018). Research on work activation policies has also consistently shown that financial penalties for noncompliance are linked to material hardship (Raffass, 2017) and are applied more often to recipients with multiple barriers to employment, including histories of substance use (Bauld et al., 2012). However, in contrast to nonvoluntary interactions with the criminal justice system, when welfare surveillance becomes too burdensome or stigmatizing, recipients simply withdraw from the system (Morash et al., 2017; Sherman, 2013). A study of recipients in the UK documented that in order to avoid interactions with the welfare system, recipients turned to other sources of income, such as borrowing money from friends, visiting food banks, and crime (Wright et al., 2020).

Although research on work activation policies has yielded insights about the consequences of supervision, it is important to note the welfare system is comprised of more than TANF. Over time, welfare provision has shifted from providing financial assistance to poor families to offering in-kind support to the working poor (Edin and Shaefer, 2011). One of the most significant welfare programs is SNAP, which provides poor households with vouchers that can be used to purchase food. As compared to TANF, SNAP serves a much broader population of recipients, including families, people with disabilities, and the elderly (Lauffer, 2017). In 2018, there were 2.3 million TANF recipients (U.S. Department of Health and Human Services, 2019) compared with 39.3 million recipients of SNAP (Cronquist, 2019).

In terms of behavioral requirements for SNAP recipients, ‘Able-Bodied Adults without Dependents’ must either search for work or work at least 20 hours per week, but other populations are exempt from these rules. If recipients do not meet work requirements, they are barred from receiving SNAP for at least a month and might become disqualified from the program. As another example of its relative generosity compared to TANF, over time, the federal government has encouraged states to enact rules that help potential applicants access SNAP (Bitler and Hoynes, 2016), while also restricting the ability of states to enact supervisory provisions, such as requiring drug testing as a condition of receiving aid (CLASP, 2019).

Overall, there is evidence that recipients in both the criminal justice and welfare systems experience increased supervisory requirements more as a means of coercion than as a way to access beneficial care. Given this literature, I expect that poverty among people with drug convictions will be higher in full ban states than in states with no bans. The more interesting comparison to consider is between full and partial ban states. Proponents of new paternalism have asserted that increased supervision, especially the requirement to engage in drug treatment, could help recipients overcome barriers to employment and thus reduce poverty. However, the review above has highlighted the negative emotional and economic consequences of punitive supervisory programs. I therefore expect that poverty will be lower in no ban states than in states with partial bans. One of the primary contributions of this paper is to extend the analysis beyond work activation programs like TANF to SNAP, which is a relatively generous program providing assistance to a population viewed as deserving of state support.

Methods

Empirical approach

This paper has two research aims. First, it documents changes in SNAP and TANF drug ban policies over time, including their emphasis on supervision versus punishment. I include both SNAP and TANF in the analysis to assess whether states enact comparable drug ban policies across these programs.

I also examine the association between SNAP drug ban policies (full, partial, and none) and poverty among people with drug convictions. At least two approaches could be used to addressing this aim, both of which exploit variation in state drug ban policies. First, one could look within individuals across time, relating changes in their poverty status to changes in their exposure to drug ban policies. Variation in drug ban policies can be obtained either from policy changes in the state where they reside, or because the individual moves between states with different policies. This approach would require accessing data that includes economic outcomes on people with drug convictions, which is not publicly accessible at a sufficient scale to permit inferential statistics. The second option, the one I pursue in this paper, is to compare outcomes of individuals in states with different state SNAP drug ban policies. To do so, using data drawn from NLSY97, I create a pooled cross-sectional dataset containing respondents with drug convictions. Respondents are matched to the states in which they reside, covering the period from 2000 until 2016.

Using these data, I estimate variants of the following baseline logistic regression model: $y_{ijt}^{*} = β_{1} {SNAP}_{j t} + {X'}_{ijt} + {Z'}_{j t} + μ_{t} + ϵ_{ijt}$ (1)where $y_{ijt}^{*}$ refers to log odds of poverty for individual i in state j at time t. SNAP captures each state’s drug ban policy (none, partial, full) in a given year; $X'$ is a vector of individual, time-varying characteristics, and $Z'$ is a vector of state-level features. Year fixed effects, $μ_{t}$ , are included in order to absorb bias from year-specific economy-wide features, such as recessions, that may influence the likelihood that people with drug convictions would experience poverty. Standard errors are clustered at the individual level to account for the fact that respondents with drug convictions can be included in multiple years of data. Individuals only appear in the sample in waves after their drug conviction.

Given the scarcity of people with drug convictions in NLSY97, there were years in which some states had no observations. Therefore, it was infeasible to use state fixed effects to control for stable, unmeasured but potentially relevant state characteristics.¹ It was also not possible to use individual fixed effects since there is not always multiple observations per person. These limitations of the data and approach means the current analysis permits description of the relationship of interest, but cannot be used to make confident causal claims.

To ease interpretation, results are presented as predicted probabilities or average marginal effects.

State-level data and variables

I used several data sources to create a dataset of state drug ban policies from 1996 until 2016. Given that states had to enact legislation to opt out of the federal ban, I started by using the online legal databases Lexis Library and Westlaw to find legislation or regulations outlining the treatment of people with drug felonies. Second, I compared the information gathered from existing research in this area (Luallen et al., 2018; Mauer and McCalmont, 2015; Mohan et al., 2017), as well as yearly government reports on SNAP (U.S. Department of Agriculture, 2019) and TANF (Urban Institute, 2020) policies. Third, if these sources yielded missing or contradictory information, I contacted local welfare administration departments.

For each state and year, I created three categories of state drug ban policies:

Full ban: Prohibits people with drug felony convictions from accessing aid under any circumstances.

Partial ban: Allows people with drug felony convictions to access aid, as long as they meet additional requirements. These include waiting a specific period of time after being convicted or released from incarceration, or not having over a specified number of convictions. Most of these bans also include supervisory requirements, including engagement in drug testing and treatment, as well as compliance with probation and parole requirements.

No ban: Does not consider drug felony convictions when determining eligibility.

In the descriptive analysis, I present data on state drug ban policy from 1997 until 2016 as it took time for states to enact their own welfare legislation in reaction to the new federal law (Owens and Smith, 2012).

In the analysis looking at the relationship between SNAP drug bans and poverty, I include other state-level variables that may be related to both a state’s decision to pass a drug ban policy, as well as the likelihood that an individual would experience poverty. Following Beckett and Western (2001), I include measures of: the percentage of the state’s labor force that is unemployed, and the percentage of the state’s population that is African American and Hispanic. In line with related work, I also include variables measuring the ideology of state residents and lawmakers (Martin and Shannon, 2020; Soss et al., 2011); and the number of people incarcerated/100,000 residents (Martin and Shannon, 2020; Owens and Smith, 2012). Although prior studies have included a measure of the state violent crime rate (Martin and Shannon, 2020; Owens and Smith, 2012; Soss et al., 2011), given this study’s focus on people with drug felony convictions, I instead include a control variable for the number of drug arrests/100,000 residents. Last, I control for other SNAP policies to ensure that I am measuring the relationship between drug ban policies and poverty instead of the overall generosity of the SNAP program: broad-based categorical eligibility, the average recertification period, and whether the state exempts the value of vehicles from asset limits on eligibility. Further details about the measurement and data sources for these variables are included in online Appendix Table A1.

Individual-level data and variables

Individual-level data on people with drug convictions come from the NLSY97. NLSY97 is an ongoing study designed to follow the experiences of young people as they transition from school into the workforce (Bureau of Labor Statistics, 2018). The initial survey was conducted in 1997 and was followed up with yearly interviews until 2011; since then interviews have been conducted bi-annually. This yields 17 waves of data, covering 1997 to 2016. Data for this study are restricted to the years between 2000 and 2016 since there were too few respondents with drug convictions in early waves of data collection. Along with detailed demographic and socioeconomic questions, at each wave of data collection, respondents are asked about their interactions with the criminal justice system since the last interview, including whether or not they were convicted. If the respondent was convicted, s/he is also asked to indicate the type of conviction.

NLSY97 is the only dataset with information on people with drug convictions covering all of the U.S. over multiple years.² It thus allows me to examine variation in state SNAP drug ban policies, as well as controlling for individual-level characteristics that are likely to influence both the state in which people reside and the likelihood that they will experience poverty. However, these data impose limitations. Most importantly, until wave 12, the survey included a question about whether or not respondents were convicted of a crime since the last interview, but did not distinguish between misdemeanor and felony convictions. To maximize the number of cases, as well as the number of years in the analysis, I include all people with drug convictions in the sample. This means that I am most likely underestimating the association between drug felony bans and poverty. Second, while the dataset is nationally representative, sampling is not designed to ensure a specific number of respondents from each state. This translates into a small number of people with drug convictions per state, and some years where there are no observations in some states. Given this fact, it was not possible to conduct a more causal analysis exploiting policy changes, such as a difference-in-difference approach. Third, although research has suggested important gender and ethnic differences in the way that people with drug addictions are treated in both criminal justice and welfare systems (McKim, 2017), the size of these subgroups was too small to investigate. Last, NLSY97 is a cohort study, which means that respondents have only been followed until their mid-thirties. Findings can therefore only apply to younger populations who came of age after the passage of welfare reform in 1996. The cohort’s relative youth also means that there were only 69 respondents with drug convictions that have children. I could not investigate the relationship between TANF drug bans and poverty since very few of these respondents would qualify for TANF.

Despite these limitations, NLSY97 remains the dataset most suited to address my research question. NLSY97 is the only dataset with respondents residing in states across the U.S. that collects information on criminal justice system involvement. This means that I can provide descriptive information about the relationship between different levels of SNAP drug bans and poverty rather than looking at the effects of changing the policy in one state. The detailed information collected by NLSY97, as well as its geographic coverage is evidenced by the fact that the data are widely used by researchers studying outcomes among criminal justice-involved people (Apel and Sweeten, 2010; Brame et al., 2014), as well as the effects of state policy on individual-level outcomes (Pacula et al., 2015).

From this dataset, I gather information on poverty, as well as other individual-level control variables. To assess poverty, I use a variable created by NLSY97, which compares a family’s total annual cash receipts before taxes from all sources to poverty thresholds that are based on the number of household members and the number of members under age 18. If the household’s income was below the federal poverty threshold (less than 1), the respondent was coded as living in poverty. I use federal poverty thresholds in this study as they are easily calculated using the dataset and are also linked to determining eligibility for welfare programs. However, it is important to note that this measure likely underestimates poverty in the sample as it does not reflect regional differences in the costs of living, include expenditures related to work, childcare, housing and medical expenses, or reflect the amount of financial assistance received from the government and other sources (Hutto et al., 2011).

I also include a series of control variables that are related to both the likelihood of being convicted of a drug felony and the likelihood of experiencing poverty. Thus, I measure whether respondents are female or male (gender), African American or non-African American (ethnicity), have at least a high school education (educational attainment), and live with at least one of their biological children (parental status). For all of these characteristics, I created dichotomous variables due to the small size of the sample. Last, to ensure that I was testing the association between supervisory requirements and poverty, I excluded the 38 respondents residing in Louisiana and North Dakota whose partial drug ban policies did not include increased supervision.

Results

State supervision in drug felony bans

Figures 1 and 2 show trends in state TANF and SNAP drug ban policies over time. States are categorized based on whether they have a full, partial or no drug ban in place. These figures show substantial changes in access to welfare over time, as well as important differences between SNAP and TANF policies. The first notable finding is that, between 1997 and 2016, states with full drug bans have decreased from a majority to a minority of states. In 1997, just over half of states had a full ban on benefits in place for TANF and SNAP programs. However, by 2016, only 10 states still had a full ban in place for TANF and 7 states had a full ban for SNAP. Second, for people with drug felony convictions, states have increased access to SNAP more than to TANF. Thus, more states still have full bans in place for TANF than for SNAP. Additionally, while 19 states have completely removed bans on SNAP for people convicted of drug felonies, only 14 states have done so for TANF.

Figure 1.

State SNAP policies, 1997–2016.

Figure 2.

State TANF policies, 1997–2016.

Table 1 shows the combination of SNAP and TANF drug ban policies enacted by each state in 2016. This table shows that, for around 70 percent of states, SNAP and TANF drug ban policies are consistent with one another. The most common combination of state policies is to have partial bans for both SNAP and TANF. Further demonstrating the stringent nature of TANF provision, there is only one state (Indiana) where the drug felony ban for TANF is less stringent than the ban for SNAP. Meanwhile, 11 states have SNAP bans that are less stringent than in their TANF program.

Table 1.

Distribution SNAP and TANF drug felony ban provisions among states and the District of Columbia, 2016.

		TANF policy
		None	Partial	Full	N
SNAP Policy	None	District of ColumbiaNew HampshireNew JerseyNew MexicoNew YorkOhioOklahomaOregonRhode IslandUtahVermontWashingtonWyoming	IllinoisIowaMaineMassachusetts	DelawareSouth Dakota	19
	Partial		AlabamaAlaskaCaliforniaColoradoConnecticutFloridaHawaiiIdahoKansasKentuckyLouisianaMarylandMichiganMinnesotaMontanaNevadaNorth CarolinaNorth DakotaPennsylvaniaTennesseeWisconsin	GeorgiaMissouriNebraskaTexasVirginia	26
	Full		Indiana	ArizonaArkansasMississippiSouth CarolinaWest Virginia	6
	N	13	26	12	51

Source: Author’s compilation of state legislative data.

While the replacement of full bans with partial bans has increased the ability of people with drug felony convictions to access assistance, most partial bans have made assistance contingent on increased supervision (information on each state’s policy is included in Table A2 in the online Appendix); 88 percent of states with partial bans for SNAP require supervision, while 81 percent of states with partial bans for TANF do so. The most common requirements among states with supervisory bans are participation or completion of a drug treatment program and meeting the conditions of their probation or parole. Additionally, far more states require drug testing as a condition of receiving assistance in their TANF program than in their SNAP program. The number of supervisory requirements also varies among states: while Alabama, California and Connecticut allow aid as long as people with drug felony convictions meet the conditions of probation and parole, in Missouri, to access SNAP, people with drug felony convictions must also engage in drug treatment and undergo drug testing.

Partial bans without supervisory requirements include other restrictions, including waiting periods and caps on numbers of convictions. For example, in Louisiana, for both TANF and SNAP, people with drug felony convictions are not allowed to obtain financial support until one year after they are convicted or released from incarceration.

Overall, this analysis shows that states have shifted from blocking people with drug convictions from accessing benefits to subjecting them to supervisory requirements. However, compared to TANF, state SNAP policies are more generous to people with drug felony convictions. Thus, states are more likely to opt out of drug bans in their SNAP program compared to TANF. Importantly, states are also more likely to enact partial bans that include supervisory requirements under their SNAP programs.

Drug felony bans, increased supervision and poverty

In this section, I explore how SNAP drug ban policies are related to the likelihood that people with drug convictions will experience poverty. Table 2 shows the characteristics of the 413 respondents with drug convictions included in the analysis, as well as the characteristics of the states in which they reside. Importantly, around one-third of people with drug convictions had income below the poverty level, which is high compared to the level of poverty in the overall NLSY97 sample (18 percent). It is also important to note the overrepresentation of men in the analytical sample; only 17 percent of the sample is female. Additionally, only 15 percent of the sample currently resides in a household with children, which means only a small portion of the study sample would meet the household eligibility requirements for TANF.

Table 2.

Descriptive statistics of individual and state characteristics.

Variable	Percent or mean
Individual characteristics
Poverty	33.13
Female	17.42
African American	25.76
High School diploma or more	68.25
Household with children	15.25
State characteristics
Percent unemployed	6.13 (0.13)
Citizen ideology	50.31 (1.04)
Government ideology	45.14 (1.32)
% residents African American	13.82 (0.82)
% residents Hispanic	13.92 (1.23)
Incarceration/100,000 residents	476.44 (14.47)
Drug arrests/100,000 residents	506.65 (13.96)
SNAP policies
Full drug ban	38.61
Partial drug ban with supervision	43.93
No drug ban	17.46
Broad-based categorical eligibility	47.21
Average recertification period	8.46 (0.26)
Vehicle exemption policy	57.59
N	413

Note: Standard deviations in parentheses for continuous data. Sample includes respondents with drug convictions between ages of 18 and 35. Estimates are weighted to adjust for complex sampling design.

Source: Individual-level data from rounds 4–17 NLSY97, collected between 2000 and 2016. State-level data compiled from multiple sources.

Figure 3 shows the predicted probability of poverty estimated from equation (1), generated from the sample of respondents with drug convictions residing in states with different SNAP drug ban policies. Full results are presented in Table 3.

Figure 3.

State SNAP drug ban policies and poverty among people with drug convictions.Note. *p<0.05.

Table 3.

SNAP drug felony bans and poverty among people with drug convictions: Average marginal effects.

	(1)	(2)	(3)
SNAP drug ban policy
No ban	−0.1997**	−0.1932**	−0.2082*
	(0.0634)	(0.0620)	(0.0842)
Partial ban with supervision	−0.1376*	−0.1193*	−0.1331
	(0.0594)	(0.0591)	(0.0706)
Individual controls
Female		0.0461	0.0409
		(0.0696)	(0.0676)
African American		0.1017*	0.1407**
		(0.0486)	(0.0494)
High school education or more		−0.1398**	−0.1410**
		(0.0538)	(0.0536)
Household with children		−0.0505	−0.0596
		(0.0639)	(0.0620)
State controls
Percent unemployed			0.0029
			(0.0220)
Citizen ideology			−0.0031
			(0.0032)
Government ideology			0.0023
			(0.0021)
% state residents: African American			−0.0082*
			(0.0038)
% state residents: Hispanic			−0.0002
			(0.0031)
Incarceration/100,000 residents			0.0001
			(0.0003)
Drug arrests/100,000 residents			0.0002
			(0.0002)
SNAP: Broad based eligibility			0.0578
			(0.0641)
SNAP: Recertification period			−0.0003
			(0.0096)
SNAP: Vehicle exemption			0.0006
			(0.0633)

Note: *p < 0.05, **p < 0.01, ***p < 0.001. SNAP policy reference group is Full ban. All models include year fixed effects. N = 414. In parentheses: standard errors clustered at the individual level.

Source: All data sources cover years 2000–2016. Individual-level data from rounds 4–17 NLSY97. State legislative data compiled by author. Other data sources for state-level data detailed in online Appendix A1.

The analysis reveals that predicted probabilities of poverty among people with drug convictions are almost double in states that have a full ban on SNAP benefits than in states with no drug ban, even after controlling for individual and state characteristics. In states that impose a full ban, the probability of living in poverty was 46 percent among people with drug convictions. The predicted probability of poverty among people with drug convictions was 33 percent, or 13 percentage points lower, in states whose ban included supervisory requirements. As seen in Table 3, living in a partial ban state, as opposed to a full ban state, was not associated with a statistically significant difference in expected poverty after controlling for state-level characteristics and policies. Predicted poverty among people with drug convictions in states that opted out of the drug ban was even lower at 25 percent, which is significantly lower than predicted poverty levels in full ban states even after controlling for other individual-level and state characteristics. Thus, we see that for SNAP, any modification of the drug felony ban is associated with decreased levels of poverty. However, the largest poverty reduction comes from states that have opted out of the ban completely.

Discussion

This paper examines changes in drug ban policies in SNAP and TANF programs over time. Prior research has merged TANF and SNAP into a single policy or created a category including both modified and no bans. Recognizing the important differences between SNAP and TANF, I describe trends in these policies separately. Supporting research showing the importance of supervision in both welfare and criminal justice systems, as well as research on drug felony bans in the TANF program, I find that states have increasingly adopted SNAP and TANF policies that allow people with drug felony convictions to access aid, as long as they comply with increased supervisory requirements. The most common requirements include engaging in drug treatment programs, consenting to drug testing, and complying with the conditions of probation and parole. For people in community supervision, some of these requirements might overlap with those imposed as a condition of their probation and parole. For these recipients, welfare supervision requirements will still create additional burden as they must provide proof of compliance to their caseworkers in both criminal justice and welfare systems. The additional requirements indicate that lawmakers have embraced the idea of new paternalism. Thus, along with providing additional financial resources, these policies seek to change the underlying behavior of poor people.

Despite the increased focus on supervision, important differences still exist between drug felony bans for TANF and SNAP. Lawmakers have enacted less stringent bans on SNAP than on TANF. These trends clearly reflect differences in the underlying history of these programs, with TANF serving a highly stigmatized population. At the same time, the U.S. welfare system has become more generous in programs like SNAP that provide in-kind assistance to recipients who are seen as deserving of aid (Shaefer et al., 2019). These differences are important, since most researchers interested in overlapping welfare and criminal justice systems have either solely focused on work activation programs like TANF or lumped all welfare programs together.

I then consider the association between the severity of SNAP bans and poverty among people with drug convictions. I find that poverty is lower among people with drug convictions in states that opted out of the drug ban, compared to full ban states. This makes sense, since people with drug felony convictions in these states are allowed to access additional financial support irrespective of these convictions. What is more surprising is the strength of the association: the predicted probability of poverty among people with drug convictions in no ban states was almost half of that in full ban states. Turning to contrasts between full and partial bans, results do not support the assertion greater paternalism will lead to greater reductions in poverty. Differences in poverty comparing partial and full ban states were not statistically significant. This could be due to the emotional pains of supervision documented in the criminal justice literature. While people involved in the criminal justice system must endure criminal justice supervision or face penalties like incarceration, people in the welfare system can always simply drop out if they are willing to face the economic pains of doing so. In fact, existing research indicates that SNAP participation rates are sensitive to changes in supervisory requirements; participation significantly decreased after some states enacted work requirements (Harris, 2019).

Study findings make contributions to several literatures: the effects of drug bans; the increasing links between welfare and criminal justice systems; and the consequences of mass supervision. First, this is the first known study that uses quantitative methods to show that poverty is lower among people with drug convictions in states that do not have SNAP drug bans. Instead of looking at outcomes, most of the existing literature has focused on either documenting the characteristics of states with specific drug ban policies (Martin and Shannon, 2020; Owens and Smith, 2012) or on critiquing the underlying assumptions of drug ban policies, especially the assumed link between substance abuse and welfare receipt among women (Eadler, 2011; Godsoe, 1998; Gustafson, 2011). A notable exception is a qualitative study following formerly incarcerated women in Pennsylvania showing that full drug ban policies further restricted the already limited financial support available to these women, given their poor labor market prospects and lack of family financial support (Hirsch, 1999).

Second, this paper uncovers broader links between criminal justice and welfare systems. Theorists have largely assumed that these two systems divide along gendered lines, with the criminal justice system used to control men and the welfare system used to control women (Wacquant, 2009). These assumptions, and the fact that most TANF recipients are single mothers with children, have led to a focus on TANF – and consequently on women (Soss et al., 2011). However, programs like SNAP “serve as important safety nets for needy families and demand relatively minimal requirements” (Sugie, 2012: 1423). Importantly, this paper shows that the stringency or leniency of a state’s welfare system is dependent on the program being considered. Scholars interested in punitiveness and mass supervision must look beyond TANF, which serves an increasingly small proportion of poor people. Along with extending the analysis to SNAP, I show that the negative economic consequences of drug ban policies are not concentrated among women; the sample in this analysis was primarily composed of men. Further, a close analysis of drug ban policies themselves show that they create explicit ties between criminal justice and welfare systems. For example, the requirements that people with drug felony convictions must comply with the conditions of probation and parole to receive financial assistance adds to Gustafson’s (2011) finding that welfare and criminal justice bureaucrats are increasingly working together to manage the same caseload.

Finally, this article adds to both criminal justice and welfare literatures, showing the potential negative consequences of mounting state supervision in the lives of economically vulnerable groups by explicitly considering a policy at the intersection of these two systems (Fletcher and Wright, 2018; Shaefer et al., 2019). The finding from this study that one’s likelihood of being poor is lower in a no ban state than in one with a partial ban calls into question the utility of supervision. It is important to note that supervision was associated with poverty even in SNAP, which is seen as a program that helps a population deserving of state support and that is relatively generous in terms of other behavioral requirements. Overall, policymakers must ensure that supervisory programs, in both criminal justice and welfare systems, are designed in a way that respects the needs and autonomy of recipients instead of simply seeking to coerce them into proper behavior.

Footnotes

Acknowledgements

I am grateful to the LSE International Inequalities Institute for providing funding for this research. I would also like to thank Johann Koehler,Lucinda Platt,Tom Kemeny,as well as the editor,and anonymous reviewers for their helpful comments on earlier drafts of this paper. AshLee Smith provided invaluable research assistance.

Amanda Sheely is an assistant professor in the Department of Social Policy at the London School of Economics and Political Science. Her research investigates the relationships between socially marginal groups and state institutions,with a primary focus on women. Her work has been published in Crime & Delinquency,Social Science & Medicine,Social Work,and the Policy Studies Journal .

ORCID iD

Amanda Sheely

Supplemental material

Supplemental material for this article is available online.

Notes

References

Allard

(2002) Life Sentences: Denying Welfare Benefits to Women Convicted of Drug Offenses. Washington, DC: Sentencing Project.

Apel

Sweeten

(2010) The impact of incarceration on employment during the transition to adulthood. Social Problems 57(3): 448–479.

Bauld

McKell

Carroll

, et al. (2012) Benefits and employment: How problem drug users experience welfare and routes into work. Journal of Social Policy 41(4): 751–768.

Beckett

Western

(2001) Governing social marginality: Welfare, incarceration, and the transformation of state policy. Punishment & Society 3(1): 43–59.

Belenko

Hiller

Hamilton

(2013) Treating substance use disorders in the criminal justice system. Current Psychiatry Reports 15(11): 414.

Bitler

Hoynes

(2016) The more things change, the more they stay the same? The safety net and poverty in the great recession. Journal of Labor Economics 34(S1): S403–S444.

Brame

Bushway

Paternoster

, et al. (2014) Demographic patterns of cumulative arrest prevalence by ages 18 and 23. Crime and Delinquency 60(3): 471–486.

Bryan

Jenkins

(2016) Multilevel modelling of country effects: A cautionary tale. European Sociological Review 32(1): 3–22.

Bureau of Labor Statistics (2018) National Longitudinal Survey of Youth 1997 Cohort, 1997–2016 (Rounds 1–17). Washington, DC: Bureau of Labor Statistics, U.S. Department of Labor.

10.

Chandler

Fletcher

Volkow

(2009) Treating drug abuse and addiction in the criminal justice system: Improving public health and safety. JAMA 301(2): 183–190.

11.

CLASP (2019) Drug Testing and Public Assistance. Washington, DC: CLASP.

12.

Cox

(2011) Doing the programme or doing me? The pains of youth imprisonment. Punishment & Society 13(5): 592–610.

13.

Cronquist

(2019) Characteristics of Supplemental Nutrition Assistance Program Households: Fiscal Year 2018. Alexandria: US Department of Agriculture, Food and Nutrition Service.

14.

Durnescu

(2011) Pains of probation: Effective practice and human rights. International Journal of Offender Therapy and Comparative Criminology 55(4): 530–545.

15.

Eadler

(2011) Purging the drug conviction ban on food stamps in California. Scholar 14: 117–164.

16.

Edin

Shaefer

(2015) $2.00 a Day: Living on Almost Nothing in America. New York: Houghton Mifflin Harcourt.

17.

Fletcher

Wright

(2018) A hand up or a slap down? Criminalising benefit claimants in Britain via strategies of surveillance, sanctions and deterrence. Critical Social Policy 38(2): 323–344.

18.

Garland

(2001) The Culture of Control: Crime and Social Order in Contemporary Society. Chicago: University of Chicago Press.

19.

Gillom

(2001) Overseers of the Poor: Surveillance, Resistance, and the Limits of Privacy. Chicago: University of Chicago Press.

20.

Godsoe

(1998) The ban on welfare for felony drug offenders: Giving a new meaning to ‘life sentence. Berkeley Women’s Law Journal 13: 257–267.

21.

Gustafson

(2009) The criminalization of poverty. Journal of Criminal Law and Criminology 99(3): 643–716.

22.

Gustafson

(2011) Cheating Welfare: Public Assistance and the Criminalization of Poverty. New York: NYU Press.

23.

Hamilton

Koehler

Lösel

(2015) Treatment programmes for substance abusing offenders in Europe: A survey of routine practice. European Journal on Criminal Policy and Research 21(3): 371–384.

24.

Handler

Hasenfeld

(2006) Blame Welfare, Ignore Poverty and Inequality. London: Cambridge University Press.

25.

Haney

(2010) Offending Women: Power, Punishment, and the Regulation of Desire. Berkeley: University of California Press.

26.

Harris

(2019) Do SNAP Requirements Work? Kalamazoo: W.E. Upjohn Institute for Employment Research.

27.

Hirsch

(1999) Welfare reform and women with felony drug convictions: Research results and policy recommendations. Clearinghouse Review 33: 587.

28.

Hutto

Waldfogel

Kaushal

, et al. (2011) Improving the measurement of poverty. The Social Service Review 85(1): 39–74.

29.

Jordan

(2018) Welfare grunters and workfare monsters? An empirical review of the operation of two U.K. ‘work programme’ centres. Journal of Social Policy 47(3): 583–601.

30.

Katz

(2013) The Underserving Poor: America’s Enduring Confrontation with Poverty. 2nd ed. New York: Oxford University Press.

31.

Koch

(2018) Personalizing the State: An Anthropology of Law, Politics, and Welfare in Austerity Britain. Oxford: Oxford University Press.

32.

Lauffer

(2017) Characteristics of Supplemental Nutrition Assistance Program Households: Fiscal Year 2016. Washington, DC: U.S. Department of Agriculture, Food and Nutrition Service, Office of Policy Support.

33.

Luallen

Edgerton

Rabideau

(2018) A quasi-experimental evaluation of the impact of public assistance on prisoner recidivism. Journal of Quantitative Criminology 34(3): 741–773.

34.

Lynch

(2012) Theorizing the role of the ‘war on drugs’ in US punishment. Theoretical Criminology 16(2): 175–199.

35.

Martin BT and Shannon SK (2020) State variation in the drug felony lifetime ban on Temporary Assistance to Needy Families: Why the modified ban matters. Punishment & Society 22(4): 439–460.

36.

Mauer

McCalmont

(2015) A Lifetime of Punishment: The Impact of the Felony Drug Ban on Welfare Benefits. Washington, DC: Sentencing Project.

37.

McCorkel

(2013) Breaking Women: Gender, Race and the New Politics of Imprisonment. New York: New York University Press.

38.

McKim

(2017) Addicted to Rehab: Race, Gender, and Drugs in the Era of Mass Incarceration. Rutgers: Rutgers University Press.

39.

McNeill

(2019) Mass supervision, misrecognition and the ‘malopticon’. Punishment & Society 21(2): 207–230.

40.

McNeill

(2020) Penal and welfare conditionality: Discipline or degradation? Social Policy & Administration 54(2): 295–310.

41.

Mead

(1997) The New Paternalism: Supervisory Approaches to Poverty. New York: Brookings Institute.

42.

Metsch

Pollack

(2005) Welfare reform and substance abuse. The Milbank Quarterly 83(1): 65–99.

43.

Miller

Stuart

(2017) Carceral citizenship: Race, rights and responsibility in the age of mass supervision. Theoretical Criminology 21(4): 532–548.

44.

Mohan

Palacio

Lower-Basch

(2017) No More Double Punishments: Lifting the Lifetime Ban on Basic Human Needs Help for People with a Prior Drug Felony Conviction. Washington, DC: CLASP.

45.

Morash

Kashy

Bohmert

, et al. (2017) Women at the nexus of correctional and social policies: Implications for recidivism risk. The British Journal of Criminology 57(2): 441–462.

46.

Natapoff

(2015) Misdemeanor decriminalization. Vanderbilt Law Review 68: 1055–1116.

47.

Owens

Smith

(2012) Deviants” and democracy: Punitive policy designs and the social rights of felons as citizens. American Politics Research 40(3): 531–567.

48.

Pacula

Powell

Heaton

, et al. (2015) Assessing the effects of medical marijuana laws on marijuana use: The devil is in the details. Journal of Policy Analysis and Management 34(1): 7–31.

49.

Parolin

Luigjes

(2019) Incentive to retrench? Investigating the interactions of sate and federal social assistance programs after welfare reform. Social Service Review 93(2): 305–339.

50.

Pearl

(2018) Ending the War on Drugs: By the Numbers. Washington, DC: Center for American Progress.

51.

Perez-Munoz

(2017) What is wrong with testing welfare recipients for drug use? Political Studies 65(4): 912–929.

52.

Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (1996) P.L. 104-193, 110 Stat. 2105.

53.

Phelps

(2017) Mass probation: Toward a more robust theory of state variation in punishment. Punishment & Society 19(1): 53–73.

54.

Phelps

(2020) Mass probation from micro to macro: Tracing the expansion and consequences of community supervision. Annual Review of Criminology 3(1): 261–279.

55.

Raffass

(2017) Demanding activation. Journal of Social Policy 46(2): 349–365.

56.

Schram

Soss

Houser

, et al. (2010) The third level of U.S. Welfare reform: Governmentality under neoliberal paternalism. Citizenship Studies 14(6): 739–754.

57.

Seccombe

James

Walters

(1998) “ They think you ain’t much of nothing”: The social construction of the welfare mother. Journal of Marriage and the Family 60(4): 849–865.

58.

Shaefer

Edin

Fusaro

, et al. (2019) The decline of cash assistance and the well-being of poor households with children. Social Forces

59.

Sherman

(2013) Surviving the great recession: Growing need and the stigmatized safety net. Social Problems 60(4): 409–432.

60.

Soss

Fording

Schram

(2011) Disciplining the Poor: Neoliberal Paternalism and the Persistent Power of Race. Chicago: University of Chicago Press.

61.

Sugie

(2012) Punishment and welfare: Paternal incarceration and families’ receipt of public assistance. Social Forces 90(4): 1403–1427.

62.

Tuttle

(2019) Snapping back: Food stamp bans and criminal recidivism. American Economic Journal: Economic Policy 11(2): 301–327.

63.

U.S. Department of Agriculture (2019) SNAP State Options Report. Washington, DC: U.S. Department of Agriculture.

64.

U.S. Department of Health and Human Services (2019) Characteristics and Financial Circumstances of TANF Recipients, Fiscal Year 2018. Washington, DC: U.S. Department of Health and Human Services.

65.

Urban Institute (2020) Welfare Rules Database. Available at: https://wrd.urban.org/wrd/WRDWelcome.cfm

66.

Wacquant

(2009) Punishing the Poor: The Neoliberal Government of Social Insecurity. Duke: Duke University Press.

67.

Wacquant

(2010) Crafting the neoliberal state: Workfare, prisonfare, and social insecurity. Sociological Forum 25(2): 197–220.

68.

Wahler

(2015) Retribution or rehabilitation? Conflicting goals of U.S. policies pertaining to drug felonies and their impact on women. Journal of Women, Politics and Policy 36(1): 95–106.

69.

Wright

Fletcher

Stewart

ABR

(2020) Punitive benefit sanctions, welfare conditionality and the social abuse of unemployed people in Britain: Transforming claimants into offenders? Social Policy & Administration 54(2): 278–294.

70.

Zhang

Roberts

REL

Lansing

(2013) Treatment or else: Coerced treatment for drug-involved California parolees. International Journal of Offender Therapy and Comparative Criminology 57(7): 766–791.