What AEI Gets Wrong about SNAP Work Requirements
They cited my research, so let me respond!
The Congressional Budget Office estimated that President Trump and House Republicans’ “Big Beautiful Bill” is regressive: the most disadvantaged households stand to lose resources while the most advantaged stand to gain. One policy that contributes to already disadvantaged households being worse off are proposed expansions to work requirements in the Supplemental Nutrition Assistance Program (SNAP), which provides vital food assistance to low-income households.
Researchers, including myself, have studied the effectiveness of these types of work requirements for many years. While for some policy questions it can be hard to reach a consensus, this is not one of those cases. The body of evidence is crystal clear: work requirements do not increase work. On top of that, work requirements cause large drops in program participation, meaning people can go hungry.
Despite this, the American Enterprise Institute (AEI) recently claimed that work requirements do improve employment outcomes. Every so often you might read about competing claims from researchers about what the evidence actually says. How do you sort out these rival claims if you are not a specialist?
Now, imagine you are a researcher on one side of the argument, and your own research is being cited by people on the other side! This is the position I find myself in, where AEI cites three of my research papers to reach a conclusion that is not supported by the evidence. In this piece, I weigh in.
I have studied the safety net in the U.S. for over a decade, with a particular focus on SNAP, formerly known as Food Stamps. When someone makes bad faith arguments about one’s own work, it can be hard not to get petty. But, in this case, the evidence clearly and plainly speaks for itself, so let’s get to it!
What are SNAP work requirements and why are policymakers advocating to increase them?
In simple terms, work requirements are policies added to existing programs that make benefit eligibility not just dependent on income, but also on working or engaging in work-related activities.
What makes it more complicated is that there is not a single work requirement. In fact, there are three work requirements in SNAP right now, described in much more detail in this Hamilton Project piece I co-authored. The first is the General Work Requirement, which applies to many participants ages 16-59 and requires them to keep their job if they are currently employed, or search for work if they are not. The second is the so-called “ABAWD” Work Requirement where ABAWD is short for able-bodied adults without dependents. Complying with the ABAWD requirement means working at least 80 hours per month or participating in a job training program for the equivalent amount of time, though there are very few slots available in these programs. (Volunteer work for 80 hours per month can also count, but is relatively rare). States and sub-state areas can be granted waivers for the ABAWD Work Requirements (but not the other requirements) if they have poor labor market conditions making it hard for people to find work. Third, the Employment and Training program (E&T) is a nominal job search and training assistance program that those subject to the first two requirements can be placed in.
There are several proposed changes to SNAP Work Requirements in the budget reconciliation bill.
First, single parents and cohabitating parents whose youngest child is between the ages of 7 and 17 will now need to meet the ABAWD work requirements.
At least one parent in married cohabiting couples, with children ages 7 and above, will have to meet these requirements.
Additionally, the bill expands work requirements to older adults (going from 18-54 to 18-64), and
Limits the ability of states to apply for waivers from the ABAWD Work Requirements when they are in an economic downturn and reduces the hardship exemptions from eight percent of the ABAWD caseload to one percent.
Advocates of work requirements argue that they help to counteract the disincentives to work created by the SNAP program and set people on a path to “self-sufficiency” and “personal responsibility.” This rhetoric puts the blame on safety net recipients themselves for being low-income, instead of acknowledging structural factors that can cause people to be low-income, such as lack of good-paying jobs and affordable childcare.
If we take proponents of work requirements at their word on their motivations, this raises two important questions that the existing research answers:
Does SNAP create enough disincentives to work that justify work requirements?
Do studies of the existing work requirements in SNAP show evidence that they increase employment?
The answers to these questions are “No” and “No”, and I’ll dig into these answers next.
Does SNAP create large enough disincentives to work that justify adding work requirements?
For a long time, it was hard for researchers to study the impacts of SNAP on labor market decisions, so the research evidence was largely limited to one study looking at the program in the 1960s-70s and one study, written by me, studying immigrants in the 1990s-2000s. Both sets of results find null to small negative impacts on employment and earnings when individuals have access to SNAP. Much has changed in the intervening years making it hard to use these studies to predict impacts today.
For example, not only have mothers’ employment rates climbed substantially since the 1960s, but the other support programs available to them in the absence of work are very different. Thus, the impact of SNAP on employment in today’s context could be very different, contrary to what AEI states. Indeed, one of the authors of the study on the 1960-70s made this point recently in a Congressional hearing that AEI was in attendance at.
Fortunately, we have more recent evidence. In new work of mine with my coauthor, Jason Cook, we use recent, high-quality data to study whether receipt of SNAP (in this century!) changes whether and how much people work among a large, representative group of working-aged people. We find very clear evidence that SNAP does not lead to large decreases in employment or earnings overall.
The only decline in earnings we document is on a subgroup (people who had some work history), and it is short-lived and followed quickly by a rebound and longer-run positive effects on earnings. My read of this evidence is that participating in SNAP is actually a helpful policy for increasing work among some recipients.
AEI mischaracterizes our central finding by claiming that our result is in fact evidence that work requirements are effective because they counteract disincentive effects, thus the net effect is zero. Why is this wrong?
1) Most people in our analysis are not subject to the work requirements. While we did not report this explicitly in the paper, we have calculated that only 23% were subject to a work requirement (13% are subject to General Work Requirements, 2% to ABAWD, 17% to E&T, and some are subject to multiple).
2) The research evidence on the imposition of work requirements themselves would need to show a positive employment effect, which it does not, as I discuss next.
Finally, if AEI views this as evidence of work requirement efficacy, this means that effective policy has no impact on whether people work but removes a large number of people from the program. Perhaps they inadvertently said the quiet part out loud?
What is the evidence on whether SNAP Work Requirements impact work?
Lauren Bauer and I summarize the evidence here. We focus on the papers that study SNAP recently, use the highest-quality data, and employ the most reliable methods.
As shown in Figure 1, across all papers that meet these criteria, the findings are consistent: there is no large impact on whether someone works at all or on their earnings. And there is a large decrease in receipt of SNAP.
AEI makes several arguments about why this evidence should not be trusted. Some of this gets pretty into the weeds on methodology and data issues, but, for full transparency, I will explain how the analysis in two of these papers Gray et al. (2023) and Cook and East (2024) address these issues.
Issue Raised: The use of administrative earnings data sometimes exclude informal or self-employment.
Researcher Solution: Administrative earnings data do indeed come with pros and cons (as is true of every data set). Both papers, however, confirm the results are the same using both administrative and a more inclusive measure of self-reported earnings.
Issue Raised: The papers only study groups of people in a certain age range so are not generalizable.
Researcher Solution: It is true that the research designs used in these papers provide estimates that are specific to people right around the age 50 eligibility cutoff, or with children right around the age 6 cutoff. However, both papers make sure the results can be generalized by showing the characteristics of those around the age cutoffs are similar to those farther from it. And, in my paper, the fact that we use child age and find the same effect as papers that look at adult age further suggests the results are generalizable.
Issue Raised: Not using a randomized control trial approach.
Researcher Solution: The economics community has long accepted that quasi-experimental methods carry the same causal force as randomized control trials when carried out effectively. This is especially true of papers using a regression discontinuity design, as many of the papers in this literature do. Additionally, it is hard not to notice that this same criticism is not leveled at the research that supports AEI’s hypothesis and uses similar methods. Finally, AEI cites evaluations of older, welfare-to-work programs, but this evidence is not as strong as AEI claims, and looks at very different programs than those currently being proposed.
Why do people have a hard time complying with work requirements?
First, there are many administrative burdens imposed with work requirements. For those who are subject to work requirements, they must learn about the program rules, figure out how to comply, and try to comply with them. This adds extra burdens such as getting a letter from one’s employer to submit to the SNAP office to prove employment.
This red tape can keep many SNAP recipients from meeting work requirements. Figure 2 shows an example form for E&T participants. AEI calls on the USDA to study how much these administrative burdens matter, however, the USDA already commissioned investigations as to why individuals disenroll from SNAP and found these burdens do cause disenrollment. This is consistent with a much wider body of causal research showing that administrative burdens including in work requirements reduce SNAP participation.
Second, not everyone who wants to work will be able to immediately find a job. In forthcoming work, Lauren Bauer and Diane Schanzenbach find that unemployment is pervasive among those subject to work requirements. So, these individuals are trying to find work but having a hard time doing so. This is even more difficult for individuals who live in areas with a strained economy. Others have childcare responsibilities or transportation issues preventing them from increasing work. Additionally, research shows that SNAP recipients work in jobs that are much more likely to lay them off through no fault of their own.
Third, not everyone who is working will be able to control their schedule to satisfy the strict work requirements. The low-income job market SNAP recipients typically work in offers very little flexibility. For example, a recent analysis found that 77% of low-income parents working in the service sector are forced to work irregular schedules and 66% of them would like to work more than they do. So, individuals might not be able to simply pick up another shift to meet work requirements, even if they want to.
What does the research say about work requirements reducing SNAP participation?
For the reasons mentioned, individuals have a hard time complying with work requirements even if they want to. The research, as summarized in Figure 1 above, shows that work requirements lead to large drops in program receipt, from about 20% to 50%.
Proponents of work requirements say that this disenrollment effect is just removing people from the program who do not “deserve” benefits. The research points against this narrative and finds that those removed from SNAP because of work requirements are more disadvantaged.
Additionally, my recent work shows that even though only parents are subject to the requirements, work requirements reduce benefits available to children too, which can cause long-run harm for these children. This is especially of concern given the proposals to greatly expand work requirements for parents.
The conclusion is clear: work requirements will not increase work and will leave many needy households without important food support. Congress would do well to give up on this ineffective policy.
Chloe N. East is an Associate Professor of Economics at the University of Colorado Denver, a Research Associate at the National Bureau of Economic Research (NBER), a Non-Resident Fellow at the Brookings Institution, a Research Fellow at the Institute for the Study of Labor (IZA), and an affiliate of the Institute for Research on Poverty (IRP) at the University of Wisconsin. @chloeneast.bsky.social
People either "deserve" to eat because this is a rich country or they do not because some other people are greedy and parsimonious. The entire surrounding apparatus is around this question which we are forbidden to discuss.