How Has the Composition of TANF Recipients Changed Under Work Requirements?

Richard and Bart (2026) examine how TANF participation changed over time and how recipients differed according to exposure to work requirements and sanctions. They analyze Michigan administrative records covering all TANF participants from 2010 through 2018, linked to SNAP, Medicaid, and unemployment-insurance earnings data. Michigan’s monthly TANF caseload fell from about 90,000 cases in 2010 to 20,000 by 2018. The share subject to work requirements declined from roughly 60% to 30%, while child-only cases increased from 25% to more than 50%. Recipients who violated work requirements had younger children and, when employed, earned substantially less than compliant recipients.

Why This Article Was Selected for The Policy Scientist

The article addresses an important question about how work requirements shape participation in the social safety net and which families remain connected to public assistance as eligibility rules become more restrictive. This issue is timely because work requirements continue to feature prominently in debates over TANF and other means-tested programs. Published in AEA Papers and Proceedings, a respected outlet of the American Economic Association, the study contributes valuable descriptive evidence to a longstanding literature on welfare reform. Its linked administrative data are unusually strong, although reliance on Michigan limits direct generalization to other states. The analysis is primarily descriptive; future research using stronger causal-inference designs would better identify the effects of work-requirement policies.

Full Citation and Link to Article

Richard, K., & Bart, L. (2026). Who participates in TANF? Variation by exposure to work requirements. AEA Papers and Proceedings, 116, 314–319. https://www.aeaweb.org/articles?id=10.1257/pandp.20261087

Central Research Question

The study examines how the composition of the Temporary Assistance for Needy Families (TANF) caseload changed as work requirements and associated sanctions became more stringent. The authors focus on Michigan between 2010 and 2018, a period that encompassed both recovery from the Great Recession and major changes in the state’s TANF rules. The central question is not simply whether TANF participation declined, but who remained in the program as the caseload contracted and whether participants differed systematically according to their exposure to work requirements.

The analysis distinguishes among several groups: TANF cases not subject to work requirements, cases subject to work requirements, and cases in which participants were sanctioned for violating those requirements. This distinction allows the authors to examine whether the characteristics of participants changed differently across these groups. Of particular interest are employment, earnings, family structure, the age of children, race and ethnicity, receipt of other safety-net benefits, and the prevalence of sanctions. The broader objective is to establish who is affected by work-requirement policies and whether those who fail to satisfy the requirements differ meaningfully from those who comply.

Previous Literature

The study builds on a long literature examining TANF and the transformation of cash assistance following the 1996 replacement of Aid to Families with Dependent Children with TANF. Moffitt (2003) provides an important foundation by documenting the structure and evolution of TANF and the broader policy shift toward employment expectations, time limits, sanctions, and state discretion. This literature emphasizes that reductions in welfare caseloads can result from both changes in economic conditions and changes in program rules.

The authors also draw on the Welfare Rules Databook series. Kassabian et al. (2011) document state TANF policies as of 2010, providing information on income eligibility, benefit levels, earnings disregards, and other program characteristics. Kassabian et al. (2013) document subsequent changes in state welfare policies, including variation in lifetime limits and penalties for failure to comply with work requirements. These sources place Michigan within the larger national policy environment and show that its reforms reflected a broader pattern of increasingly restrictive TANF administration in some states.

Michigan-specific policy changes are described using Best and Koorstra (2011), who summarize legislative revisions affecting the state’s Family Independence Program, and Carley (2011), who discusses Michigan’s 48- and 60-month time limits and the litigation surrounding their implementation. These sources are particularly important because the study period includes the introduction of substantially stronger sanctions and time limits.

The article also references Al-Chanati and Husted (2021), who examine the effects of losing cash welfare support, and Richard and Bart (2025), a companion study examining the economic consequences of TANF work-requirement enforcement. Together, this literature shifts attention from aggregate caseload decline toward the experiences and subsequent outcomes of families removed from or sanctioned within TANF.

Data

The analysis uses unusually detailed administrative records from the Michigan Department of Health and Human Services. The underlying database covers every individual enrolled in Michigan TANF for at least one month between 2008 and 2020, although the primary analysis concentrates on 2010 through 2018. This provides a population-level administrative dataset rather than a sample survey, substantially reducing concerns about sampling error and inaccurate self-reported program participation.

For each month of TANF enrollment, the records identify benefit amounts, cumulative program utilization, whether participants were subject to work requirements, and demographic characteristics including age, gender, race, and ethnicity. The data also provide information on family composition and whether a participant received a sanction for violating a work requirement.

The TANF records are linked to administrative information on Supplemental Nutrition Assistance Program and Medicaid enrollment. They are also linked to quarterly earnings records from Michigan’s Unemployment Insurance system. These linked records allow the authors to measure formal-sector employment and earnings while individuals are receiving TANF.

The scale of the data is substantial. In 2010, the analysis includes approximately 101,940 unique cases subject to work requirements and 36,067 cases not required to work. By 2018, these figures had fallen to 20,267 and 17,704, respectively. The data therefore permit detailed comparisons across groups and over time during a period of exceptionally large changes in TANF participation.

Methods

The study is primarily descriptive rather than causal. The authors track the number and composition of TANF cases over time and compare participant characteristics according to exposure to work requirements. Cases are classified as subject to work requirements if a household member was subject to those requirements at any point during the calendar year.

Among work-required cases, the authors further distinguish between households that received a sanction for violating a work requirement and households that did not. Characteristics are compared for 2010, 2014, and 2018. The authors calculate changes between 2010 and 2014, which spans Michigan’s major policy reform, and between 2010 and 2018, which captures longer-term changes.

Statistical tests determine whether changes over time and differences across groups are significant at the 95 percent confidence level. Employment is measured as the fraction of quarters during TANF receipt in which the participant has positive earnings in the unemployment-insurance records. Conditional earnings are calculated for quarters in which participants both receive TANF and have positive reported earnings.

These methods provide a strong descriptive account of caseload composition but do not isolate the causal effect of work requirements or sanctions. Changes in the economy, program rules, participant composition, and other factors occurred simultaneously. The study therefore appropriately treats its results as evidence about patterns and differences rather than causal estimates of policy effects.

Findings/Size Effects

Michigan’s TANF caseload contracted dramatically over the study period. Approximately 90,000 cases were enrolled each month in 2010, compared with roughly 20,000 by the end of the study period, a decline of approximately 78 percent. At the same time, the composition of the caseload changed substantially.

In 2010, approximately 60 percent of active TANF cases were subject to work requirements, while about 25 percent were child-only cases and approximately 15 percent received hardship exemptions. By 2018, only about 30 percent of cases were subject to work requirements, while more than half were child-only cases. The hardship exemption had been eliminated.

The 2011 reforms produced particularly large immediate changes. Nearly 20,000 cases exited TANF around implementation of the reforms, which strengthened penalties for work-requirement violations, eliminated the hardship exemption, and imposed a 48-month time limit. Additional declines followed after litigation temporarily delayed implementation of some provisions.

The demographic composition of TANF also shifted. Among participants not subject to work requirements, the proportion identified as non-Hispanic Black increased from 36.5 percent in 2010 to 51.5 percent in 2018, a 15-percentage-point increase. Among those subject to work requirements, the corresponding proportion increased from 52.3 percent to 61.9 percent, an increase of 9.5 percentage points.

Marriage or partnership became less common. Among work-required beneficiaries, the share married or partnered declined from 10.9 percent to 7.0 percent. Among beneficiaries not required to work, it fell from 11.5 percent to 5.1 percent.

Employment increased substantially during the economic recovery. Among participants subject to work requirements, average quarterly employment increased from 54.5 percent in 2010 to 63.3 percent in 2018, an 8.9-percentage-point increase. Among participants not subject to work requirements, employment increased from 41.4 percent to 56.7 percent, a 15.3-percentage-point increase.

Differences between participants who violated work requirements and those who complied are particularly important. In 2018, sanctioned participants had quarterly employment rates of 58.4 percent, compared with 64.6 percent among participants who did not violate work requirements. Thus, the employment gap was only about 6 percentage points.

Differences in earnings were larger. In 2018, sanctioned participants who were employed earned an average of approximately $2,045 per quarter, compared with $2,278 among compliant participants. Sanctioned households also had younger children. The youngest child averaged 3.19 years of age among sanctioned households compared with 4.07 years among nonsanctioned work-required households.

Conclusion

The study demonstrates that TANF caseload decline involves substantial changes in who remains enrolled. Michigan’s TANF program became dramatically smaller between 2010 and 2018, increasingly concentrated among child-only cases, and less likely to include adults subject to work requirements.

The comparisons between sanctioned and nonsanctioned participants are especially informative. People who violated work requirements were somewhat less likely to be employed and earned less when working, but the differences in employment were relatively modest. Many sanctioned participants remained connected to formal employment, suggesting that failure to satisfy TANF work requirements does not necessarily indicate an absence of labor-force participation. Younger children and lower earnings may instead indicate greater difficulty maintaining sufficient hours, documenting qualifying activities, or navigating unstable employment.

The study therefore provides an important empirical foundation for evaluating work requirements. Its principal contribution is to document which households are exposed to sanctions and how those households differ from other TANF participants. Because the analysis is descriptive, it does not establish the causal consequences of work requirements. It does, however, identify precisely the compositional changes that future causal research must address when estimating the effects of increasingly stringent work-requirement policies.