The Earned Income Tax Credit’s Hidden Tradeoffs

How a popular anti-poverty program distorts work, education, and long-term growth

In a Tax and Budget Bulletin that Chris Edwards and I wrote back in 2015, we warned that the Earned Income Tax Credit (EITC) delivers small benefits at large costs. Although the program enjoys broad bipartisan support (it reduces poverty and is a more effective approach than raising the minimum wage or other welfare programs), it functions far more like a spending program than a tax cut—roughly 88 percent of its benefits are direct cash transfers rather than tax relief. One of the problems with the EITC, as we explained, is that while it was designed to encourage work, in practice it often discourages additional earnings. For most recipients, the phase-out of benefits imposes steep effective marginal tax rates, creating a strong disincentive to work more.

Many also like the program because it was shown to raise single mother workforce participation. Yet, empirical studies have found this claim to be very fragile—most of the increase in single mother labor force participation in the 1990’s was due to other confounding effects from welfare reform and a booming economy. Other studies also find that EITC reduced the participation of women with husbands with low earnings by 10-12% when EITC was expanded in the 1990s. Eligible women can also expect to lose about half their pre-marriage benefits following marriage, making them 2.5 percentage points less likely to marry their partner than single mothers who expect no change in benefits.

The program’s costs go well beyond its economic distortions (over $800 billion over 10 years). It has one of the highest error and fraud rates of any federal benefit, roughly 33.5% ($21.9 billion in 2023. Its complexity forces two-thirds of claimants to rely on paid preparers, and the taxes needed to fund it impose further deadweight losses on the economy. For these reasons, Chris and I argued that the EITC should not be expanded and that policymakers should instead focus on reforms that directly raise market wages—such as lowering taxes on business investment and employment.

That argument has held up reasonably well, but a new paper in The Review of Economic Studies by Julien Albertini, Arthur Poirier, and Anthony Terriau (HT to Tyler Cowen at Marginal Revolution) complicates the story in a fascinating way. Using state-level EITC variation and border discontinuities, the authors find that more generous state EITCs are associated with higher high school dropout rates. Specifically, a one-percentage-point increase in the state EITC rate reduces high-school completion by roughly 0.07 percentage points. Their interpretation is straightforward: wage subsidies lower the relative returns to schooling, leading some young people to enter the workforce earlier and forgo education.

At first glance, the main empirical strategy appears straightforward—a classic border discontinuity design. Still, it raises tricky questions about what we actually want the EITC to do. The estimated effect is small but robust, and it’s striking that the authors pick up a difference at all.

But the policy implications are not so clear-cut. As my friend, economist Scott Winship, asked over email after I sent him this study, “Do we want more high-school graduation but less employment?” That’s a fair question. Winship is right that the marginal students induced to stay in school by a smaller EITC may not acquire much additional skill; for them, an extra credential might have symbolic value but limited productivity impact. On the other hand, if the EITC draws young people prematurely into low-skill jobs, it could depress their long-term earnings potential.

For now, though, the study reinforces one aspect of our earlier critique: the EITC, like many well-intentioned programs, distorts behavior in multiple directions simultaneously. It may increase participation in the short run while reducing incentives to acquire skills in the long run. That’s a poor trade-off if what we really want is sustainable wage growth through higher productivity rather than larger subsidies.

Still, one question lingers in my mind: how much should we care about traditional credentials in an age when artificial intelligence may soon make them less valuable? The policy literature tends to assume that more schooling automatically yields more skill and higher productivity. But that relationship is becoming less certain. As AI diffuses, it’s starting to flatten hierarchies and make complex tasks accessible to non-credentialed workers. If a high-school diploma—or even a college degree—signals less about real ability than it once did, perhaps the old trade-off between “more employment” and “more schooling” looks different.

That doesn’t make the EITC good policy—it still distorts wages and shifts billions through an inefficient, fraud-prone channel. But it does remind us that the world we’re measuring is changing faster than our models. Education, work, and human capital are not static categories. And the value of a credential may decline long before policymakers notice.

So yes, the EITC may still be overrated. But I’ll keep thinking about it.