Key Points

• Evidence shows that the Supplemental Nutrition Assistance Program (SNAP), like other means-tested programs, reduces employment levels among those eligible.
• Work requirements in SNAP are one tool to address the program’s work disincentives.
• Existing evidence on the effectiveness of SNAP’s work requirements provides important insights but requires careful interpretation and understanding of data and methodological challenges.

Introduction

Work has long been viewed as a critical pathway out of poverty. Therefore, it is important to understand the effects of federal safety-net programs—designed to alleviate poverty and promote economic mobility— on work. One commonly used policy tool to mitigate work disincentives in federal safety-net programs is work requirements. These typically mandate that indi­viduals work or engage in work-related activities for a set number of hours to maintain their eligibility for benefits.

When evaluating work requirements for the Sup­plemental Nutrition Assistance Program (SNAP)— previously the food stamp program—policymakers face two key questions:

1. Does receiving SNAP discourage work?
2. Are SNAP work requirements effective?

A recent House Committee on Agriculture hearing underscored the continued debate over these questions.¹ While existing research provides valuable insights, the answers are complex and require careful interpretation. Ultimately, the answers to these questions depend on context and the specific populations served.

Broadly, the evidence indicates that SNAP can dis­incentivize work and that work requirements may help address that problem in certain contexts and for spe­cific populations. However, existing data limitations and methodological challenges contribute to mixed find­ings when assessing the effectiveness of work require­ments at increasing employment levels. What is more consistently observed is a decline in SNAP participation following the implementation of work requirements. In this report, I review the evidence surrounding each question and underscore the complexities involved in reaching broad, conclusive judgments.

Does Receiving SNAP Discourage Work?

Economic theory suggests that access to SNAP ben­efits will reduce employment. In simple terms, when an individual receives SNAP benefits, it reduces their

preference for work in the classic labor-leisure trade-off. Hilary W. Hoynes and Diane Whitmore Schanzenbach explained this “static labor supply model” in their 2015 paper, “U.S. Food and Nutrition Programs.”² Their description serves as the basis for understanding how SNAP affects employment decisions among recipients.

In explaining the application of this theory to the food stamp program, Hoynes and Schanzenbach, in a 2012 paper, reviewed the literature on the labor-supply effects of Aid to Families with Dependent Children (AFDC), as well as other literature.³ On the basis of the AFDC literature, Hoynes and Schanzenbach argued that we should expect negative employment effects related to the food stamp program. They wrote:

The general findings from the literature on the AFDC program (which provided cash income support) are that AFDC reduces labor supply among program participants by 10 to 50%. In particular, the influential paper by Moffitt (1983) finds that annual hours worked by AFDC recipients are 546 lower per year because of the program. This translates into a reduction of 208 h/year among female-headed households as a whole.⁴

It remains important to consider the broader evidence when assessing whether SNAP disincentivizes employment. As Hoynes and Schanzenbach argue, a large body of literature provides evidence from AFDC confirming the economic theory that income support programs such as SNAP have work disincentives.

A few studies have examined how this theory applies to SNAP in practice. In the same 2012 study (published in the Journal of Public Economics and titled “Work Incentives and the Food Stamp Program”), Hoynes and Schanzenbach noted, “Consistent with theory, we find reductions in employment and hours worked when food stamps are introduced.”⁵ They found that when the food stamp program expanded county by county in the 1960s and 1970s, it led to lower rates of employment and hours worked. Furthermore, they found the largest and most robust estimates among female-headed households. They also showed how their estimates align with the evidence on AFDC. While the labor market today is very different from that of the 1960s and 1970s, there is little evidence that individuals currently receiving SNAP benefits would respond differently.

In fact, other studies have confirmed these findings in a more contemporary setting. Chloe N. East, publishing in Labour Economics in 2018, exploited changes to immigrant eligibility in 2002 to study the effects on the labor supply among noncitizens.⁶ She wrote:

I find strong evidence of labor supply disincentives [in the food stamp program], and the magnitude and margin of this response varies across demographic groups. Access to the [food stamp] program reduces the employment rates of single women by about 6%, whereas married men continue to work but reduce their hours of work by 5%. These findings confirm the predictions of traditional labor supply theory regarding the response to a means-tested program.⁷

Therefore, both economic theory and empirical research find that SNAP benefits—or means-tested transfers more broadly—disincentivize work. Importantly, beginning in 1970 and especially in 1996, SNAP began to use work requirements to counteract these labor-supply disincentives. However, both the 2012 study by Hoynes and Schanzenbach and the 2018 study by East rely on data collected before the work requirements were in effect or before they were strongly enforced. Therefore, work requirements likely did not affect the results of those studies.

Although SNAP’s general work requirement existed beginning in 1970 (during the time period covered by Hoynes and Schanzenbach), it is unclear how strictly state and local food stamp offices enforced the requirement. Furthermore, the requirement was relatively weak—only stipulating that nonexempt individuals reg­ister for work, not quit a job without good cause, and take a suitable job when offered.

Beginning in 1996, able-bodied food stamp recipients age 18–49 without dependents could obtain benefits for only three months in a three-year period unless they worked or participated in a work-like activity for 80 hours per month. This rule is commonly referred to as the able-bodied adult without dependents (ABAWD) work requirement. Congress passed the ABAWD work requirement after the time period covered by Hoynes and Schanzenbach’s 2012 study—which relied on data from the 1960s and 1970s. Therefore, the food stamp recipients covered by the study did not face the ABAWD work requirement.

East’s 2018 study uses data from 2002, suggesting that food stamp recipients covered by the study may have faced the ABAWD work requirement. However, the Food and Nutrition Service (FNS) did not finalize its rules until June 19, 2002, so states were likely not enforcing the ABAWD work requirement at the time of East’s study. In fact, data suggest that very little enforcement of the ABAWD work requirement occurred in the early 2000s, and until 2007, states were allowed to carry forward their caseload exemption and thereby avoid implementing the ABAWD work requirement altogether.⁸ This makes it likely that immigrants included in East’s study did not face the ABAWD work requirement and were subject only to the relatively weak general work requirement.

These points are important because work requirements were either nonexistent, relatively weak, or weakly enforced during the periods covered by these two studies. Hence, the labor-supply disincentives found by Hoynes and Schanzenbach in 2012 and by East in 2018 occurred without modern-day SNAP work requirements. In other words, the evidence shows that receipt of SNAP discouraged work, and any work requirements nominally applied were beside the point.

In contrast to those two studies, work published by Jason B. Cook and East in 2024 studied the labor-supply effects of SNAP from 2012 to 2016 in one state where work requirements were in place and enforced.⁹ They found that access to SNAP did not affect employment levels and had only small negative effects on earnings. The authors wrote:

Taken together, this analysis shows no evidence of meaningful effects of SNAP on the extensive margin of work. The analysis on earnings is more mixed but indicates at most a modest decline in earnings that is short-lived. This short-run change in earnings is in part driven by a significant increase in part-time work (quarterly earnings of $1–1,999) though this effect is also quantitatively small.¹⁰

Importantly, Cook and East’s study involved one state—identified as a mountain state for anonymity—and analyzed application data from 2012 to 2016. Cook and East acknowledge that the ABAWD work requirement existed in this mountain state, affect­ing 4 percent of all recipients. (A relevant metric—the share of working-age adult recipients affected— is unclear.) Furthermore, the authors acknowledge that 16 percent of all recipients were subject to the general work requirement. (Again, the share of working-age adults subject to the general work requirement is left unclear.)

The authors do not mention any further work requirements, such as a mandatory SNAP employment and training (E&T) requirement. However, in another study also published in 2024 by the same authors, which likely included the same “mountain state,” Cook and East describe the existence of a mandatory SNAP E&T program in the state.¹¹ In that study, the mountain state had a mandatory SNAP E&T program for SNAP participants with a youngest child age six or older.¹² While the existence of work requirements in this mountain state remain unclear because Cook and East did not name the state, the ABAWD work requirement, the general work requirement, and a mandatory SNAP E&T requirement likely existed during the period examined by Cook and East in their other study.

This information is important when interpreting the findings on whether SNAP affects work, because all work-capable SNAP recipients in the mountain state were likely subject to a work requirement during the study period. Therefore, the labor-supply incentives created by the work requirement (or mandatory E&T) likely offset any labor-supply disincentives of SNAP receipt. In this context, the results from Cook and East’s work are consistent with the theory that SNAP disincentivizes employment and that work requirements would counteract those work disincentives.

Are SNAP Work Requirements Effective?

Given the evidence showing that SNAP disincentivizes employment, the next question is whether the program’s work requirements are effective. Although there are different ways to define “effective” (which I address later), the literature primarily focuses on two outcomes: how work requirements affect SNAP caseloads and how work requirements affect labor supply.

Here the large literature on AFDC and welfare-to-work experiments also offers important insights. Generally, the welfare-to-work experiments from the 1990s that imposed work requirements on AFDC

recipients increased employment levels.¹³ Evidence also points to employment gains from replacing AFDC with Temporary Assistance for Needy Families (TANF), although TANF involved policy changes beyond work requirements.¹⁴

As Hoynes and Schanzenbach argue, the AFDC and welfare-to-work literature implies a similar positive effect of work requirements on employment in the SNAP context.¹⁵ However, the evidence is mixed and less straightforward than the AFDC literature implies. Although most studies of SNAP evaluate the ABAWD work requirement, one study evaluates the program’s general work requirement combined with a mandatory SNAP E&T referral.

The evidence is challenging to interpret given the data limitations and methodological constraints. All the existing research on SNAP’s work requirements relies on quasi-experimental methods rather than the more rigorous random assignment methods. Quasi-experimental methods are useful and contribute importantly to our understanding of how work requirements affect behavior. However, when uncertainty exists, random assignment methods are considered more reliable. The welfare-to-work experiments in the 1990s, for example, used random assignment methods, making the results clearer. Moreover, the absence of comprehensive data on SNAP receipt and employment outcomes further complicates the interpretation of existing research on the effectiveness of SNAP’s work requirements.

Table 1 presents a selection of studies that highlight the complexities involved in interpreting this research and helps explain why the findings are often mixed. Although the literature review in Table 1 is not intended to be comprehensive, many of the other studies in this literature have similar data or methodological constraints.¹⁶

The initial set of studies in Table 1 rely on administrative data to explore SNAP participation and employment. Administrative data are generally considered more reliable than survey data, which tend to underreport SNAP receipt. However, administrative data also have limitations—particularly in measuring employment.

These studies typically track employment through wages covered by unemployment insurance (UI), which do not account for self-employment or informal “off-the-books” work. Estimates vary, but surveys suggest that between 20 and 50 percent of prime-age adults engage in some form of informal work.¹⁷ One survey found that 26 percent of individuals classified as “not in the labor force” were participating in informal employment.¹⁸ The studies featured in Table 1 that use UI-covered wages to measure labor market outcomes miss this informal employment. As a result, studies that find no employment effects of work requirements using UI-covered employment may overlook potentially significant impacts on informal work.

Moreover, studies relying on administrative data often focus on individual states, thereby making their findings context specific. The study by Cook and East is a notable example.¹⁹ They examined the effects of a mandatory SNAP E&T program in one mountain state at the time of recertification on SNAP participation and UI-covered employment, using age-of-youngest-child criteria: Households with a youngest child under age six were exempt, while those with a child over age six were subject to the requirement. They found that parents newly exposed to the general work requirement were more likely to disenroll from SNAP but not more likely to work.

The policy context surrounding these findings is crucial. A review of Utah’s SNAP E&T plan suggests that the mountain state is likely Utah.²⁰ Utah’s mandatory SNAP E&T program is relatively low intensity, at least initially, with most compliance issues stemming from missed virtual workshops. Sanctions apply only after repeated noncompliance, and even minimal earnings—as little as one dollar—could exempt participants from the requirement. Additionally, the study only included UI-covered employment for the individual, not informal employment or employment in the household among adults not on the SNAP case. These factors must be considered when interpreting the study’s results.

The reasons that individuals disenroll at the time of recertification when faced with a minimal earnings requirement (i.e., any earnings) and low-intensity requirements remain unclear. Possibly this phenomenon reflects a preference on the part of the recipient or perhaps an information gap, and program administrators should investigate the reasons. Furthermore, the inability to fully measure earnings and household income using UI-covered employment might lead to an underestimation of the impact of the work requirements on employment or household income.

Beyond that study by Cook and East, many other studies exploit the age-based cutoff in SNAP’s ABAWD work requirement. Before a revision in 2023, the policy applied only to ABAWDs age 18–49. This age threshold provides a natural comparison group, enabling researchers to assess the impact of the work requirement by comparing employment outcomes for SNAP recipients just below age 50 (who are subject to the requirement) with those just above age 50 (who are exempt).

Using these comparison groups is methodologically useful, but it limits the generalizability of the findings. The results may not apply to younger ABAWDs who are still many years away from being exempt, and those individuals may respond differently to work requirements from those nearing the exemption threshold.

The interpretation of these studies is further complicated by differences in data sources. Some rely on administrative data, others on survey data, and others on a combination of both—each method has its own strengths and limitations. Administrative data provide more accurate measures of SNAP participation and UI-covered employment, but they typically exclude informal employment and are often limited to specific states, which can constrain the findings’ generalizability. In contrast, survey-based studies face challenges in accurately identifying SNAP participants, often relying on proxy measures that reflect a likely eligible popu­lation rather than actual recipients. While survey data may better capture informal employment, they are also subject to reporting biases and may understate actual employment because of their reliance on self-reports.

Taken together, these complexities make it difficult to draw definitive conclusions about whether SNAP’s work requirements—particularly the ABAWD requirement—lead to increased employment. Findings on SNAP participation are more reliable when based on administrative data, but employment outcomes require thoughtful interpretation given the data and methodological limitations. The evidence suggests employment impacts are primarily isolated to groups just above or below age 50 and involve measures of UI-covered employment at the exclusion of informal or self-reported employment.

Overall, the most balanced conclusion from the literature is that the impact of work requirements on employment likely varies by context and population and ranges from no impact to a small positive impact on employment. In some settings and for certain groups, the requirements appear to have little or no effect, while in others and for certain groups, they may lead to increased employment or earnings. Research suggests that the most positive effects are seen among individuals without a high school diploma and in states with active SNAP E&T programs (Table 1). Conversely, often no employment effects are observed among individuals nearing age 50 and parents subject to minimal requirements with only marginal benefit reductions.

Other Outcomes

Much of the literature on SNAP’s work requirements centers on their impact on caseloads and labor supply. While these are important metrics, they do not capture the full scope of considerations in evaluating the appropriateness of work requirements.

Other intended goals include fostering self-sufficiency, encouraging personal responsibility, enhancing program integrity, and directing government assistance to the most vulnerable. Empirically measuring the effectiveness of work requirements in achieving these broader objectives is even more complex than estimating their effects on caseloads and labor supply. Nevertheless, these goals remain central to ongoing policy discussions.

Conclusion

Policymakers must deliberate the purpose and expected outcomes of SNAP work requirements. Consistent with economic theory, the evidence remains strong that SNAP, as well as other means-tested aid programs, reduces employment. Are work requirements an effective way to counteract those disincentives? As part of this deliberation, it is important to consider a broad range of objectives.

As to whether work requirements improv employment outcomes, evidence from welfare reform suggests they will. However, in the context of SNAP, the evidence is mixed. Research on the effects of SNAP work requirements on caseloads and labor supply remains challenging to interpret due to data and methodological limitations. Nonetheless, existing evidence indicates that the impact likely varies by context and target population.

Effects on SNAP caseloads reliably measured using administrative data consistently show reductions in participation when work requirements are in place. In contrast, estimating effects on labor supply is more complex due to difficulties in identifying appropriate comparison groups and accurately measuring employment and earnings. Despite these challenges, the evidence generally ranges from demonstrating no effect to modest changes in employment among groups subject to work requirements, depending on the population studied and the data and methods used.

Even though the findings are inconclusive, there remain clear theoretical and empirical reasons to think that SNAP and other means-tested programs disincentivize work and that work requirements can counteract those disincentives. State experimentation and rigorous research are needed to further inform these important policy questions.