Overview

The final judgment on SNAP’s ABAWD work requirement is conditional but direct. In the 2026 paper’s preferred design, reinstatement in selected settings reduces participation by 4.4 percentage points, about 7 percent, without a detectable employment or earnings gain, while the remaining caseload shifts away from lower-income recipients. Earlier CBO evidence leaves room for modest employment gains in some states and periods. The rule can therefore screen people out more reliably than it creates work in the studied setting, but its effects are not universal. Policy success must be judged by access, employment, income, targeting, burden, fiscal effects, and local conditions.

Learning outcomes

In the preferred 2026 study design, SNAP ABAWD reinstatement clearly reduces participation but does not produce a detectable positive labor-market response.
Lower-income recipients exit at higher rates in the preferred setting, making targeting a central outcome rather than a caseload footnote.
A work requirement may create employment in some contexts, but success requires measuring work and income rather than inferring them from benefit loss.

Keywords

SNAPABAWDScreeningEmploymentTargetingWelfare

The Policy Test: Can a Work Requirement Create Work Without Losing the Neediest?

The final question for SNAP’s ABAWD work requirement is not whether it can reduce the number of people receiving benefits. The evidence shows that it can. The harder question is whether it creates enough work, income, and improved targeting to justify the access it removes. Bringing the participation estimates, labor-market results, composition changes, welfare model, and policy alternatives together produces a clear but bounded answer: in the reinstatement settings studied by the 2026 paper, screening is a strong observed effect; work creation is not established by the preferred estimates, and the people screened out appear to be disproportionately lower-income.

That conclusion is narrower than saying work requirements never work. Earlier evidence summarized by CBO leaves room for modest employment gains in some states and periods. It is also stronger than saying the policy simply failed because a single employment estimate was not positive. The policy changes several outcomes at once. A fair judgment must ask which outcome moved, for whom, through what mechanism, and at what social and fiscal cost.

What the evidence can and cannot say

The most reliable empirical result is a change in SNAP access. In the preferred triple-difference estimate, reinstatement reduces participation by 4.4 percentage points, about 7 percent relative to the pre-reinstatement mean. The decline appears after roughly three months, consistent with the rule’s grace period before the time limit can produce benefit losses, and it persists through the 18-month follow-up. Average benefits including zeros fall by about $8 per person in the same estimate. The alternative DiD estimates show larger participation declines, but the preferred estimate is the safer anchor because it uses the age, geography, and timing contrasts that address the comparison problem more directly.

The evidence does not show a corresponding positive labor-market response in that preferred design. Employment falls by a small, statistically insignificant point estimate of about 0.4 percentage points, and the reported confidence interval rules out a meaningful positive increase within the measured outcome and sample. Quarterly earnings, available for only four states, show an insignificant reduction of about $15. This is evidence against a large aggregate employment response in the studied setting. It is not evidence that no person worked more, nor does it rule out a response in a different labor market or under a different implementation system.

The CBO comparison keeps that statement properly qualified. Its 2022 synthesis reports earlier studies near the age cutoff that found employment increases of roughly 1 to 4 percentage points in some states. CBO therefore judged that SNAP work requirements probably increased employment for some adults without dependents, while reducing average income overall because benefit losses exceeded the earnings gains for the broader group. Those studies use different states, periods, enforcement environments, and comparison assumptions. The newer preferred DDD result does not erase them, but it does make it unsafe to treat a positive estimate from one setting as the default effect of the rule.

The most defensible synthesis is therefore not a compromise average between zero and four percentage points. It is a conditional statement about evidence quality. Participation loss is the clearest recurring result in the Census study and much of the CBO evidence, although CBO reports variation across states. Employment gains are smaller, more variable, and sensitive to design and setting. A positive labor response may occur where jobs are available, supports are credible, and the comparison is valid. But a lower caseload is not a measure of that response. It is an access outcome that requires separate labor-market evidence.

The composition result gives the access outcome a distributional direction. Among current participants, average pre-reinstatement gross income rises by $23.58, average SNAP benefits fall by $1.77, and the share receiving the maximum benefit falls by 2.2 percentage points in the preferred estimate. These are not direct income gains for the people who stay.

Published by DeepMox

FAQ

What would count as success for a SNAP ABAWD work requirement?

Success would require evidence that affected people work more or earn more, while access and income losses, targeting, burden, and fiscal effects are acceptable for the stated objective. A smaller caseload alone is insufficient.

Why is the evidence about SNAP ABAWD effects conditional?

The evidence covers selected states and counties and childless adults aged 45–55, uses a proxy for ABAWD status and UI-covered employment, and draws on CBO studies with different settings. Effects may vary with jobs, waivers, supports, and implementation.

What should policymakers measure before expanding a SNAP ABAWD requirement?

They should measure access, credible employment and earnings, household income and expenses, composition, compliance burden, fiscal effects, and variation by local conditions and supports.

Previous9 minSNAP ABAWD Policy Options: Waivers and Supports

All chaptersBack to Topic

SNAP ABAWD Policy Test: Work or Screening?

The Policy Test: Can a Work Requirement Create Work Without Losing the Neediest?

What the evidence can and cannot say

FAQ