Under US law, “able-bodied adults without dependents” (ABAWDs) on the Supplemental Nutrition Assistance Program (SNAP) must work, volunteer, or take part in training or workfare for at least 80 hours a month. Those who do not can receive benefits for only three months in any 36-month period. These requirements were suspended nationwide during the Great Recession, and waivers remained widespread until about 2015. They were then reinstated county by county.
Cai, Kim, and Leung (2026) study this reinstatement, many of which took place in 2016. The paper uses administrative SNAP records linked to the Census Bureau’s Longitudinal Employer-Household Dynamics (LEHD) employment and earnings data, for childless adults aged 45 to 55 who received SNAP in the 18 months before reinstatement. The paper uses a triple difference in five states that reinstated the requirement only in some areas: Colorado, Connecticut, Maryland, Massachusetts, and New Jersey. It compares 45-to-49-year-olds, who are subject to the rule, with exempt 50-to-55-year-olds, in reinstated counties versus counties that kept their waivers.
Reinstatement reduced SNAP participation by 4.4 percentage points, or 7 percent, with no statistically significant effect on employment or earnings. The people who left had lower incomes than those who stayed: among remaining participants, average pre-reinstatement gross income rose by $23.58 a month, and the share receiving the maximum benefit fell by 2.2 percentage points.
The paper uses these estimates to estimate the MVPF of reinstating the SNAP work requirements.
MVPF = 2.2
The effect on the government budget of reinstating the work requirement is the SNAP benefits paid to the people who left the program as a result of the requirements. Following Finkelstein and Notowidigdo (2019), the paper splits participants into two types:
The requirement reduces participation by 6.5 percentage points among the lower-income type and 2.2 percentage points among the higher-income type. The impact on the government budget is therefore:
(0.49 * 0.065 * \$192) + (0.51 * 0.022 * \$93) = \$6.12 + \$1.04 = \$7.16
per person per month.
This estimate does not include any change in tax revenue. The paper estimates the requirement’s effects on employment (-0.4 percentage points) and on quarterly earnings (-$15.49). Neither is statistically significant.
Willingness to pay to avoid the requirement has two components.
The first is the compliance cost for participants who would stay on SNAP under the requirement. The paper assumes the requirement costs them two hours a month, valued at the federal minimum wage of $7.25, or $14.50. Participation under the requirement is 59 percent for the lower-income type and 66 percent for the higher-income type, so this component is $14.50 × (0.49 × 0.59 + 0.51 × 0.66) = $9.07.
The second is the value to those who would leave SNAP because of the requirement. In the standard model of an administrative hurdle, people who leave are nearly indifferent between participating and not, so their loss is negligible. The paper instead assumes that complying is infeasible for everyone who leaves: they cannot secure enough documented work hours, a workfare slot, or an exemption. They therefore value the benefits those individuals would lose in full, less the participation costs they would again incur. Those costs are $12.50 a month: the $75 application time cost estimated by Finkelstein and Notowidigdo (2019), spread over a six-month recertification period. This component is:
(0.49 * 0.065 * (\$192 – \$12.50) )+ (0.51 * 0.022 * (\$93 – \$12.50)) = \$5.72 + \$0.90 = \$6.62
Total willingness to pay is then $9.07 + $6.62 = $15.69 per person per month.
Dividing willingness to pay by net cost yields an estimated MVPF of $15.69 / $7.16 = 2.19: reinstating the requirement costs participants $2.19 for every dollar of budget savings.
If the losses of those who exit are assumed not to enter the willingness to pay, the MVPF drops to 1.27. The paper also reports MVPFs under alternative assumptions about compliance costs: Assuming one, three, or four hours a month instead of two gives MVPFs of 1.56, 2.83, and 3.46.
Cai, Lexin, Hyewon Kim, and Pauline Leung (2026). “Screening Out the Needy: The Effects of SNAP Work Requirements.” Working Paper, arXiv:2609.19660. https://arxiv.org/abs/2609.19660
Finkelstein, Amy, and Matthew J. Notowidigdo (2019). “Take-Up and Targeting: Experimental Evidence from SNAP.” The Quarterly Journal of Economics, 134(3): 1505–1556. https://doi.org/10.1093/qje/qjz013