In the 1990s, employment among single mothers rose dramatically and welfare caseloads fell sharply. A large and influential body of research has credited the 1993 expansion of the Earned Income Tax Credit (EITC) for much of that shift—the maximum credit nearly doubled for families with two or more children.
If the 1990s employment gains were driven mainly by the EITC, the story is one of “carrots”: increasing the payoff to work through an earnings subsidy. If those gains were driven mainly by welfare reform, the story is more about “sticks”: pushing people off cash assistance and into the labor market. The distinction matters for what lessons we should draw for today’s debates about child-related tax credits, cash assistance, and work requirements.
The prevailing interpretation credits the EITC for employment gains of the ‘90s. That view became a cornerstone of how scholars and policymakers understand the direction of U.S. family policy. Since the 1990s, essentially all increases in social spending have been directed towards working families and tied to earnings.
In new research, I argue that the canonical story is unsupported by the evidence. A re-examination of the same data and research designs that built the conventional view shows that there is no evidence that the 1993 EITC expansion caused the employment boom among single mothers. Instead, the results point to changes in welfare policies, otherwise known as “welfare reform.”
The missing piece: Welfare reform hit some families harder than others
The paper’s central argument is that the main strategies used to attribute the 1990s shift to the EITC overlook a critical feature of welfare reform: Its effects were largest for families most entangled with the welfare system—those more likely to apply, recertify, interact with caseworkers, face sanctions, or confront new administrative hurdles. Ethnographic accounts and implementation histories of the period document exactly these channels: Local offices reoriented around caseload reduction and procedural changes became a powerful force shaping whether families stayed on welfare or instead had to find work.
Most studies account for welfare reform by controlling for policies or economic conditions in a state, assuming they apply equally to all residents. But those controls are insufficient because the relevant variation is not only which state enacted which policies but which families were most exposed to those reforms—-and most affected when rules tightened.
Welfare exposure, not the EITC, predicts the employment boom
Mothers with two or more children saw far larger employment gains in the 1990s than mothers with one child—17 percentage points compared to 7. But those same mothers also had much higher rates of welfare participation: Before welfare reform began, about 26% of mothers with two or more children were on welfare and not working, compared to 13% of mothers with one child. Given the strong relationship between baseline welfare exposure and subsequent employment gains, this difference alone predicts the actual changes almost exactly—with nothing left over for the EITC to explain.
This pattern holds up in formal replication. The paper replicates four influential studies that share a core approach and attribute the post-1993 employment divergence to the EITC. In each case, once the pre-reform welfare participation is properly accounted for, the estimated employment effect of the EITC disappears. In other words, the divergence in employment found by these studies is explained by welfare reform, not the tax credit.
Welfare reform pushed mothers into work
If the 1990s employment gains resulted primarily from tightened welfare access rather than an earnings subsidy, the standard “success story” becomes more complicated. Rather than carrots pulling people into better-paying work, the story is one of sticks: Many were pushed into work because cash assistance became harder to get, harder to keep, or more costly to rely on.
That distinction matters most for families most exposed to the welfare system—often those with the lowest earnings capacity and greatest caregiving burdens. If sticks were doing more of the work than carrots, the average rise in employment likely came with more hardship than typically assumed, particularly for mothers with young children facing the sharpest constraints on time and childcare. To be sure, total after-tax income from earnings and the EITC did rise substantially over the 1990s. But that increase came alongside the loss of a cash safety net—and for many families, the tradeoff meant more work in families juggling care for young children.
The EITC’s perceived success shaped decades of policy—and may have been wrong
The perceived success of the 1993 EITC expansion has echoed through decades of policymaking. Because the 1990s are widely viewed as proof that work-conditioned support raises employment and improves child outcomes, the safety net increasingly emphasizes benefits linked to earnings. That history is directly relevant to modern debates about the Child Tax Credit, cash assistance, and work requirements.
If the 1990s evidence has been misread, policymakers may be overconfident about how much employment gains can be attributed to earnings subsidies alone. If welfare reform drove much of the change, the “policy lever” was not simply making work pay—it was making nonwork support harder to access, with the greatest consequences for families who cannot work consistently due to caregiving, disability, or unstable employment.
That matters today. Proposals to add work requirements to the Child Tax Credit or Medicaid often draw on the 1990s as evidence that conditioning benefits on earnings comes at little cost to families. This paper’s findings suggest that assumption is built on a misreading of the evidence. The 1990s showed that policy can move employment, but the families pushed into work were often those least equipped to sustain it, and doing so through withdrawal of unconditional support may have created larger costs than the EITC-centered narrative suggests.
Lessons for today’s safety net debates
The central message is not that the EITC is unimportant for families’ take-home pay or that earnings subsidies do not affect work. Rather, the best-known evidence that the 1993 EITC expansion caused the 1990s employment surge among single mothers does not hold up once welfare reform is properly accounted for. The employment gains appear driven by caseload-reduction pressures, not the EITC. The tradeoff between promoting employment and protecting family wellbeing may be sharper than the standard story implies—and debates over cash support for non-working families cannot be resolved by pointing to the 1990s as clean evidence of the virtues of work-conditioned aid.
Related Content
Author
The Brookings Institution is committed to quality, independence, and impact.
We are supported by a diverse array of funders. In line with our values and policies, each Brookings publication represents the sole views of its author(s).
