Why SNAP Work Requirements Aren’t Working A Case for Waiver Reform

The Supplemental Nutrition Assistance Program (SNAP), formerly known as food stamps, is the leading antihunger program in the United States. Once just a temporary program created to combat rampant food insecurity at the height of the Great Depression, SNAP has steadily grown into an important source of nutrition assistance to millions of low-income Americans.[1] Like many means-tested programs, SNAP disincentivizes work by partially offsetting increases in recipients’ income with decreases in benefit levels.[2] To counter this effect, SNAP requires that all able-bodied adults without dependents (ABAWDs), or adults aged 18–54 without children or a disability, work at least 80 hours each month to retain their benefits. Although this ABAWD work requirement’s effectiveness is difficult to quantify, multiple high-quality studies recently found that it produces moderately better work outcomes, albeit at the cost of slightly reduced SNAP enrollment by disqualifying some low-income ABAWDs who are unable to find employment.[3]

To avoid punishing SNAP recipients who fail to find sufficient employment due to a bad economy rather than an unwillingness to work, states can apply for temporary waivers to the ABAWD work requirement for areas where jobs are scarce. While sound in theory, bad policy design has enabled states to widely misuse the waivers and either exempt as many SNAP recipients from the work requirement as possible or leave the work requirement in place even in high-unemployment areas. As a result, it is likely that thousands of eligible SNAP recipients have lost their benefits through no fault of their own, and millions more have faced steep disincentives to work that will hold them back from escaping poverty in the long run.

The problems with current waiver requirements will only continue to worsen over time. SNAP participation has steadily grown in recent years—more than doubling, from 17.1 million individuals in 2000 to 41.2 million in 2022.[4] Meanwhile, Congress has expanded the ABAWD work requirement age range three times since 2023, most recently raising the age limit to 54 in October 2024.[5] In light of this rapidly growing affected population, it is essential that Congress act quickly to fix the eligibility process for ABAWD work requirement waivers to ensure that SNAP participants are encouraged to work while also guaranteeing that those unable to do so are not penalized.

SNAP is a simple transfer program in which eligible households receive a monthly stipend that can be spent only on foods preapproved by the Department of Agriculture (USDA).[6] It is a means-tested program, meaning that only individuals who fall below certain income and wealth limits qualify for SNAP benefits. Specifically, the program is restricted in most states to households with an after-deductions monthly income below the federal poverty line (for a household of two, this is $1,644 per month, or about $19,700 per year) and with assets, such as money in the bank, not exceeding $2,750.[7] Full-time college students and noncitizens are typically not eligible for SNAP.[8] These eligibility limits mean that SNAP benefits are available only to Americans most in need. The academic literature shows that SNAP is effective at improving the health and nutrition outcomes of those it serves.[9]

Despite these benefits, SNAP, like most means-tested welfare programs, disincentivizes work. The amount of SNAP benefits that a household receives is a function of its income; a SNAP household is allotted a base benefit level (varying by household size) minus 30% of its monthly income.[10] In effect, this amounts to a 30% income tax on top of whatever federal and state taxes the household is already paying, because each additional dollar of income is offset by a 30-cent loss of benefits just from SNAP. The Department of Health and Human Services (HHS) recently estimated, however, that in 2019, more than 80% of SNAP recipients received benefits from at least one other welfare program, and just over 60% collected benefits from at least two others.[11] Since many other safety-net programs come with similar benefit phaseouts as household income increases, their use results in additional effective taxes on earned income on top of the 30% from SNAP. A recent study found that, largely because of these overlapping benefits phaseouts, households in the lowest income quintile face total effective income-tax rates exceeding 40%, on average, over the course of their lifetime, and 10% of low-income households are subject to effective income-tax rates of over 70%.[12]

It is no wonder that some studies find that receiving SNAP benefits is followed by a considerable average decrease in work effort.[13] Given that SNAP households already have low pretax incomes, these steep marginal effective tax rates nearly eliminate any earnings resulting from additional work, leaving little reason to work in the short term. Yet consistent work effort fuels the wage increases or promotions that serve as most low-income individuals’ only opportunity to escape poverty and contribute to economic growth in the long run. Therefore, it is to everyone’s benefit, especially SNAP participants, to combat the program’s inherent disincentive to work.

SNAP has two work requirements: a looser, general work requirement that applies to most nondisabled SNAP recipients; and the stricter able-bodied adults without dependents (ABAWD) work requirement, which affects adults aged 18–54 without children or a work-inhibiting disability.[14] States can request temporary waivers to the ABAWD work requirement for high-unemployment areas, but these waivers do not apply to the general work requirement. This carve-out was designed to help SNAP recipients who might be unemployed because of a weak labor market retain their benefits while maintaining a strong motivation to work for those who can.[15]

Unfortunately, the criteria used to determine whether an area is eligible for a waiver are vague and, at times, counterintuitive. In practice, the process leaves states enough leeway to enact or eliminate waivers at will, with little consideration for job availability. Consequently, many SNAP recipients in Democratic-controlled states face no short-term incentive to work, while those in high-unemployment areas of Republican-controlled states who genuinely lack work opportunities often risk losing their SNAP benefits. With national SNAP enrollment at an all-time high and in light of recent expansions to the ABAWD work requirement, the need to remedy these major flaws in the policy’s implementation has never been stronger.[16] Congress must reform the ABAWD work requirement now to ensure that SNAP recipients see real benefits for their work while also guaranteeing that those who live in areas with weak labor markets are not punished for failing to find employment, through no fault of their own.

There are a few key distinctions between each of SNAP’s work requirements. The general work requirement is based on earnings: SNAP recipients must earn the equivalent of 30 times the federal minimum wage each week.[17] For those making the federal minimum wage, this amounts to a 30-hour-per-week work requirement, but individuals who earn more per hour do not have to work as much. An individual making the New York minimum wage of $15 per hour, for example, would need to work only about 14 hours per week to satisfy this requirement. Additionally, merely registering for work is sufficient to meet the general work requirement if SNAP caseworkers are unable to find suitable work for an individual.[18] This is a notable carve-out, as it lends significant leniency to individuals with sympathetic caseworkers and to those who might otherwise lose their benefits because of local job-market circumstances.

The ABAWD work requirement, meanwhile, mandates 80 hours per month of some combination of work, volunteer service, or job-training program participation to remain eligible for SNAP, regardless of how much the individual is paid for that work.[19] Individuals who fail to meet this requirement for more than three months in any 36-month period will lose their SNAP benefits until the next month in which they work at least 80 hours.[20] While there are no exceptions to the ABAWD work requirement for individuals unable to find work, states can apply to temporarily waive the ABAWD work requirement in high-unemployment areas.[21] States must submit data to the Food and Nutrition Service (FNS) demonstrating that each requested area recently experienced sustained high unemployment. FNS will approve or deny the request, usually for a year at a time, based on those data. The federal government can also waive the policy nationally, but this has occurred only during periods of unusually high unemployment, such as from 2009 to part of 2010, at the peak of the Great Recession; and the earlier half of 2020, when Covid-19-related shutdowns made many jobs effectively illegal.[22]

The ABAWD work requirement is a difficult policy to study for two reasons. First, the waivers, which serve as the primary source of variation in its implementation, are not randomly distributed because their approval is largely driven by economic circumstances. Because it is impossible to fully control for economic health in empirical models, even the best studies cannot completely separate out the effects of the waivers on any outcome that we may be interested in studying. Second, local-level data on states’ use of the work requirement became available to the public only after the Fiscal Responsibility Act of 2023 required its release.[23]

Because of these two factors, quality academic literature on the ABAWD work requirement is sparse. A handful of studies exist, but most fail even to attempt to account for the endogeneity between economic health and the work requirement’s implementation;[24] and many of the studies that do account for the endogeneity have other major quality concerns.

For example, most studies on the work requirement’s effects on SNAP enrollment compare self-identified SNAP enrollment rates from census surveys in counties that have a waiver with enrollment rates in counties that do not have a waiver.[25] However, recent studies have shown that these surveys tend to massively (and nonrandomly) understate true SNAP enrollment rates compared with actual government administrative records, which makes the findings of studies using these self-identified SNAP enrollment rates as the measured outcome extremely unreliable.[26] Additionally, many papers examining the policy’s effects on work outcomes consider only the work requirement’s relationship with overall employment.[27] Given that the general work requirement, which applies to all ABAWDs even when the ABAWD work requirement is waived, also requires consistent employment, it makes little sense to expect that adding the ABAWD requirement on top of it would cause a significant change in overall employment rates. A better metric of the policy’s work effects would be to examine whether it induces ABAWDs to work more hours, on average, than while it was waived.

The few studies that are well designed to account for these issues tend to find that the ABAWD work requirement is associated with modest improvements in work outcomes[28] but at the cost of slightly lower SNAP enrollment and, as a result, marginally worse in health and food insecurity outcomes among low-income ABAWDs.[29] These relatively small average differences are not surprising. As the ABAWD work requirement requires an average of only 20 hours of work per week, and the general work requirement demands 30 hours’ worth of earnings at the federal minimum wage each week, the ABAWD work requirement does not affect many of the SNAP recipients who are subject to it.[30] The only SNAP recipients required by the ABAWD work requirement to work more than the general work requirement demands are ABAWDs earning more than $10.90 per hour and those registered for work but unable to find it. Still, observable average changes among the entire SNAP population following its implementation show that the ABAWD work requirement can make a big difference for those whom it does affect. It is crucial, then, that policymakers ensure that the work requirement is in place in areas where work is available and that it is waived in areas without available employment opportunities.

States have the option of waiving the ABAWD work requirement for any area that has: (1) an unemployment rate averaging 10% or higher for a continuous 12-month period; or (2) a DOL designation as a “labor surplus area,” which is defined as any area with an unemployment rate at least 20% above the national unemployment rate for a continuous 24-month period.[31] States must submit to FNS unemployment data demonstrating that the areas under consideration recently met one of these criteria.

The first criterion ensures that there is a baseline unemployment rate to qualify counties for a waiver in the extremely rare event that the national unemployment rate climbs so high (above 8.3%) that 20% above the national rate is greater than 10%. States wishing to waive the work requirement, therefore, almost always use the second criterion.

Vague and overly permissive, these criteria’s poor design has been widely used by states to further their political goals, to the detriment of their SNAP recipients.

The top map in Figure 1 displays states’ use of waivers in 2019 against each county’s unemployment rates. The counties in blue had acquired a waiver that year, despite having an unemployment rate below that year’s qualifying labor surplus rate of 4.5% (already a very low unemployment rate, due to a strong national labor market). Counties in red kept the work requirement in place, despite having a relatively high unemployment rate of over 6%. The bottom map shows each state’s governor’s political affiliation. Together, these maps demonstrate that overuse of the waivers (in which the work requirement was waived, despite too low an unemployment rate to individually qualify for a waiver) was concentrated—with a few exceptions, such as Georgia and Idaho—in Democratic-controlled states, while underuse of waivers (in which the work requirement is kept in place, despite a relatively high unemployment rate) almost solely happened in states with a Republican governor.

As a result, in 2019, 463 counties accounting for more than 10 million SNAP recipients (over a quarter of the program’s total enrollees) were exempted from the work requirement, despite not individually qualifying; another 44 counties with more than 550,000 SNAP recipients retained the ABAWD work requirement even as they experienced high unemployment.[32] Consequently, thousands of ABAWDs facing a lack of employment opportunities appear to have lost their benefits through no fault of their own, and many more in better job markets saw only minimal increases in income for their work. In addition to undermining the intent of work requirements, therefore, these weak federal waiver guidelines exacerbated the economic vulnerability of SNAP recipients, underscoring the pressing need for waiver reform.

There are four key ways in which the two waiver criteria in their current form allow states to enact or reject waivers, with minimal regard for job availability. First, there is no limit on the size of an area for which states can request a waiver, enabling states to group high-unemployment areas with low-unemployment areas in the same waiver request and eliminate the work requirement in both. Second, there is no specific rule about how recent the unemployment data used for waiver requests must be, so states can use relatively old data to qualify an area for a waiver when more recent data would not support the request. Third, the minimum waiver-qualifying unemployment rate is a percentage of the national rate—paradoxically, making it easier for states to acquire a waiver when the economy is performing well and harder to get one when the national economy is suffering. Fourth, requesting waivers is entirely optional and at the discretion of the state’s SNAP department. This means that states can choose not to request waivers, even for areas with extremely high unemployment rates. Fortunately, a few relatively simple changes to the waiver application and approval process would go a long way toward fixing these problems.

1. Modify the Waiver Area Request Size

Currently, states have complete freedom to define what is considered an “area” for each waiver request, likely in order to recognize that geographic areas smaller than counties (such as cities or Native American reservations) might experience unique economic circumstances; and it might be intended to afford states the necessary flexibility to request waivers for these smaller areas without including the counties surrounding them. Unfortunately, this freedom has also given states the ability to lump counties with unemployment rates too low to qualify for a waiver together with counties with unemployment rates above one of the unemployment criteria that year and then request a waiver for the entire group. One county with a high unemployment rate raises the group’s average unemployment rate, and, as a result, all of them qualify for a waiver. In this way, states can gerrymander groups of counties to ensure that the maximum number of low-income ABAWDs qualify for a waiver, regardless of how few of them face job scarcity. In some cases, states have used this loophole to entirely abandon the work requirement statewide when only a few counties individually qualify for a waiver.[33]

This practice of allowing states to request a single waiver for several counties is unnecessary and leaves too much room for states to take advantage of high unemployment rates in a handful of counties to qualify low-unemployment counties for waivers, too. Policymakers should therefore require states to justify waiver requests for individual labor markets separately. There are several options for how a labor market might be defined; for example, waiver requests could be limited to individual metropolitan areas, which the census recognizes as having cohesive employment opportunities and uniform economic circumstances.[34] Congress could also continue to allow states to request waivers for counties but require that each be justified separately. In either case, the goal should be to grant waivers only to areas with a genuine labor surplus in which most unemployed or underemployed SNAP recipients would struggle to find more work.

2. Require Recent Data to Justify Waiver Requests

Currently, states can rely on outdated labor-market data to request waivers for counties whose labor markets have since improved. States are directed to provide “recent” evidence in their waiver requests but also must justify their requests with 24 months of unemployment data.[35] The lack of specific guidance on what qualifies as recent evidence and the requirement for two full years of data have led FNS to grant waiver requests based on unemployment data going back as far as five years, when more recent data would not have justified the request. This is especially problematic in the years following a recession, when unemployment rates are steadily dropping; old unemployment data can be several times higher than the most recent data and used to acquire waivers for areas with previously bad labor markets that have since recovered. Recently, for example, 11 states used unemployment data from 2020’s Covid-19-related shutdowns to justify statewide waivers lasting through 2024, even though more current data at the time the request was submitted showed that their economies had already bounced back.[36]

Policymakers should amend the evidence requirements for waiver requests to accept only the most recent year of available unemployment data. To a small extent, relying on old data is inevitable because it takes the Bureau of Labor Statistics (BLS) a few weeks to calculate each month’s local area unemployment rates.[37] But it is senseless to require two full years of data, and then to accept anything other than the most recently available two years. Nor is it sensible to accept data going back five years. One year of the most current data showing that an area has, on average, exceeded whatever waiver-qualifying unemployment rate is in place is sufficient evidence to demonstrate that its unemployment is unusually high.

In recent years, states have used sudden recessions from which unemployment took years to recover to make current average unemployment rates appear higher. But it is also easy to imagine the states in a recession using old data to justify not requesting waivers for areas currently experiencing heightened joblessness. Policymakers interested in maximizing the work requirement’s effectiveness while minimizing its harms to the cyclically unemployed should therefore reduce the amount of required unemployment data to a single year and limit the time frame of data that can justify waiver requests to the most recently available year.

3. Recognize Linking Local Waiver Use to National Unemployment is Counterproductive

Even without these loopholes, the second waiver-qualifying criterion, which sets the unemployment rate that qualifies an area for an ABAWD waiver at 120% of the national unemployment rate, is heavily counterintuitive for several reasons. First, it makes little sense for what is considered high unemployment to vary from year to year, as it is equally difficult to find work in a labor market with a given unemployment rate, regardless of what year it is or what is happening in the rest of the country. More concerning is that by setting the qualifying unemployment rate as a percentage of the national rate, it becomes easier to acquire a waiver when the economy is performing well and harder to get one when it is not.

For example, in 2012, when the national unemployment rate was 8.2%, areas needed an unemployment rate of 9.8% to qualify for waivers, which was 1.6 percentage points higher than the national rate. By 2019, with national unemployment at 3.3%, the waiver-qualifying threshold dropped to 3.9%, a difference of only 0.6 percentage points. In other words, the qualifying local unemployment rate for a waiver is not only higher during economic downturns but also exceeds the national rate by a larger amount than when the economy is thriving. As a result, it is more difficult to provide adequate support for the jobless or underemployed during periods of economic stagnation, which runs completely contrary to the waivers’ entire purpose of protecting SNAP recipients who are unable to work.

Setting absolute unemployment rates for waiver eligibility (discussed next) would help solve this problem because that would make it much easier for areas with unfavorable labor markets to acquire much-needed waivers during recessions because they would qualify even if many other areas in the country were suffering at the same time.

4. Set Required, Absolute Unemployment Rates for Waiver Eligibility

The 3.9% benchmark in 2019 highlights another major issue with the 20% criterion: areas with an unemployment rate only slightly above the national average will always be eligible for a waiver, no matter how low the national rate might be. A 3.9% unemployment rate is historically low and suggests an abundance of available jobs. The fact that some other parts of the country happen to have even better labor markets does not change this.[38] No matter what might be going on in the rest of the country, it does not make sense for areas experiencing such low unemployment to qualify for a waiver. This problem, along with the ability to group many counties into a single waiver request, is how Illinois in 2019 was able to waive the work requirement in all but one county (Figure 2). This is despite almost every county in the state maintaining an unemployment rate below 5% (which is widely viewed as a very healthy unemployment rate)[39] and not exceeding 7% in any.[40]

The 20% criterion leaves states that wish to retain the work requirement without any regard for job availability the complete freedom to do so. The rule limits states’ ability to overreach in their use of waivers but leaves the entire waiver request process fully optional, so states have no obligation to request any waivers, no matter how high unemployment climbs.[41] This is how Mississippi in 2019 avoided waiving the work requirement in any county, even though all but six would have qualified. In some counties, unemployment was severe that year; as Figure 3 shows, about half experienced relatively high unemployment of at least 6%, a few exceeded 8%, and one even climbed as high as 12%.[42]

Although the ABAWD work requirement is probably helpful when opportunities to work are abundant, retaining it where jobs are scarce can only serve to deny food assistance to those who, through no fault of their own, are unable to find work.[43] Worse, this loss of benefits occurs when these would-be SNAP recipients’ income is already very low or nonexistent because of insufficient employment. This is not a problem caused by the recipients’ laziness; it is a problem caused by a poor labor market. Congress created the waivers for high-unemployment areas with the hope of avoiding this outcome, which is why they must be used wherever needed.

Congress should abandon the 20% criterion and replace it with a minimum unemployment rate below which an area (as modified above) is ineligible for a waiver and a maximum rate above which an area is compelled to issue a waiver, irrespective of the state, year, or national unemployment rate. States can be left with the freedom to choose how to implement waivers for areas with unemployment rates falling between these rates. This would prevent states from doing away with the work requirement in areas with ample job openings just because the national labor market happens to be slightly stronger, or failing to use waivers even in areas where jobs are scarce. As discussed in the previous section, it would also make it much easier for areas with unfavorable labor markets to acquire a waiver during recessions because they would qualify even if many other areas in the country were experiencing even worse unemployment at the same time.

The administrative burden would also likely be lighter than under the current process, which needlessly requires state SNAP offices to get involved with the submission of federal unemployment data back to a different federal department.[44] FNS could collect local unemployment data from BLS and use these data to inform states which areas must waive the work requirement, which areas must retain it, and for which areas states have the freedom to decide work requirement status.

Bringing some national uniformity to the implementation of waivers would additionally help level the playing field across states. Under the current system, SNAP recipients in Mississippi in 2019 had to find work in order to continue receiving benefits, even in counties with an unemployment rate above 10%; but those living in counties in Illinois with unemployment rates as low as 2% were exempt from the ABAWD work requirement that same year.[45] By administering the work requirement using nationally consistent unemployment criteria, the federal government could ensure that all ABAWDs are held to roughly the same expectations and protected from being penalized for living in an area without available jobs, regardless of the state where they reside.

A maximum threshold for waiver eligibility already exists in the 10% criterion, so policymakers could retain that rate or choose a different one, above which they decide it unreasonable to maintain the work requirement and require that it be waived in these areas. The minimum threshold, meanwhile, should be set somewhere around the natural rate of unemployment, which is the unemployment rate at which a job market is in equilibrium. This rate is not directly observable, but most economists agree that the national natural rate of unemployment is 4%–6%.[46] Although individual labor markets’ natural rates of unemployment vary considerably more than the national rate because of differences in average education levels, mobility, demographics, and other factors, this range of unemployment rates almost universally indicates a very healthy job market with abundant opportunities to work. The ideal minimum qualifying rate for an ABAWD waiver, therefore, would be in this range.

These changes might be objectionable to some who favor greater state and local involvement in policymaking, but SNAP is already federally funded and federally implemented,[47] and states already must gain FNS permission to acquire waivers. This reform would therefore not significantly increase federal oversight of the program.

Furthermore, the average distribution of unemployment rates across counties tends to be rather small (unemployment tends not to vary too significantly across counties, other than a few outliers, as shown in the Illinois and Mississippi examples above), so delegating states’ control over waiver use over even a relatively narrow window of unemployment rates would leave them with significant power over the work requirement’s implementation. In 2014, for example, nearly half of all SNAP recipients lived in counties with unemployment rates of 6%–8%.[48] Whatever rates policymakers settle on, additional federal action is needed to guarantee that waiver use is driven by a genuine unavailability of jobs, and not politics or arbitrary fluctuations in national economic health.

A combination of loopholes and bad policy design has allowed for widespread overuse and underuse of ABAWD waivers relative to job availability and federal guidelines, at the cost of millions of SNAP recipients’ economic well-being. Policymakers can put an end to this by:

• Restricting waiver requests to individual labor markets
• Requiring more up-to-date unemployment data to support waiver requests
• Guaranteeing waivers in exceptionally high unemployment areas
• Preventing waiver use in very low unemployment labor markets

As SNAP enrollment continues to grow and the work requirement expands to apply to a greater proportion of SNAP recipients, Congress should enact these changes now to ensure that the program better fulfills its mandate of providing assistance to those who need it and promoting economic mobility.

Please see Endnotes in PDF

Photo: jetcityimage / iStock Editorial / Getty Images Plus