The EITC vs. the NIT

by Reihan Salam

Recently, David Cay Johnston published a column identifying potential pitfalls of the earned-income tax credit, one of the more popular anti-poverty measures in the United States. EITC-like programs have been adopted in a number of other countries, perhaps most prominently in Britain. Johnston’s provocative observations comes early in the column:

The tax credit, the largest U.S. program to alleviate poverty, is meant to be an incentive to work, but it may also contribute to poverty, effectively holding down wages of all low-skilled workers. Now how is that?

Imagine that more people whose skills limit them to low-paid work decide to work more hours so they can get the credit. Add in people who have the skills for better-paid work but cannot find it and so take lower-paid jobs because of the implicit supplement to their wages provided by the earned income tax credit.

The result is more workers seeking more work, allowing employers to hold back wage increases or even reduce wages because of the enlarged supply of labor. The employers benefit. The American median wage, in 2011 dollars, has hovered at just above $500 per week for more than a decade.

Add in the elimination of America’s largest welfare program 15 years ago, Aid to Families with Dependent Children, and the labor market is flooded with single mothers with few job skills. This not only holds down their pay, it also tends to depress the wages of all low-skilled workers.

Johnston goes on to make a number of interesting observations, drawing on the work of John Schmitt, e.g.,

Schmitt’s work shows that the share of American workers earning less than two-thirds of the median wage has been slowly increasing since 1980, a trend that also goes to the decline of unions due to anti-worker laws enacted by Congress.

To be sure, one could also attribute some of this shift to the changing demographic composition of the workforce. Johnston also cites Bruce Meyer of the University of Chicago’s Harris School:

Research by Professor Bruce Meyer of the University of Chicago Harris School of Public Policies shows that the largest increase in low-wage work was among single mothers with three or more children. In place of AFDC, as it was known, Congress in 1996 adopted Temporary Assistance to Needy Families with a maximum of 60 months of assistance.

Providing day care for children or poor working mothers has become a growing subsidy expense borne heavily by federal and state governments. In many cases the cost of child care, especially when a single mother has two or more children, exceeds what she can earn even working fulltime.

This raises an interesting question. Might we have a good reason to encourage participation in the mainstream — as opposed to the off-the-books — labor force, even if the subsidy expense exceeds the earning potential of the less-skilled worker in question? One is reminded of our discussion of Chen and Chevalier, and specifically their finding that the high upfront cost of a medical education actually outweighs the earning potential of the average female physician due to time devoted to child-rearing, etc. People nevertheless freely chose to become physicians in large numbers for a number of reasons, among them the desire to pursue meaningful. intellectually stimulating work.  

Now, this is obviously not the set of circumstances facing poor mothers. Rather, there is a paternalistic motivation, i.e., the implicit and sometimes explicit idea that it is good and useful for children raised in single-parent households to have some exposure to the world of paid work through an employed parent. The idea is that having some exposure to the norms and expectations of the working world is sufficiently valuable that it would be preferable to unconditional transfers even if unconditional transfers were, in the short term, much cheaper, as the long-term costs of having children grow up in households disengaged from the norms and expectations of the working world might actually prove quite high. 

Johnston concludes his column on the following note:

Milton Friedman, the Nobel prize-winning Chicago School economist, proposed what became the EITC, a form of negative income tax, to encourage people to work. He noted that many people on welfare faced marginal tax rates of more than 100 percent if they left the dole for low-wage jobs. President Ronald Reagan championed the EITC because it required people to work to get benefits.

The credit provides its greatest benefits to people making from about $10,000 to $14,000. Earn more and the credit falls off. Work 1,300 hours at $10 an hour and you are in the sweet spot to get the biggest tax credit. Work an extra week and benefits slip.

If the earned income tax credit, combined with the end of welfare as we knew it, hold down wages for low-skill workers then it is time to find smarter ways than Chicago School theories to reduce poverty for those who work. [Emphasis added]

To my great surprise, Johnston doesn’t reference the work of Berkeley economist Jesse Rothstein, who is much-admired, particularly in left-of-center circles, for his careful empirical work. Rothstein argues [subscriber-link; NBER has a subscriber-link to an earlier version] that the EITC is actually quite different from the traditional negative income tax, and that this is precisely the problem with the EITC.  

The confusion is understandable. Milton Friedman championed a negative income tax (NIT) and many people believe that the EITC and the NIT are essentially the same. For example, consider Robert H. Frank’s take in 2006. After briefly describing Friedman’s NIT proposal, he writes:

Mr. Friedman’s policy prescriptions were shaped by his desire to minimize adverse economic incentives, a feature that architects of earlier welfare programs had largely ignored. Those programs, each administered by a separate bureaucracy, typically reduced a family’s benefits by some fraction of each increment in earned income. Rates of 50 percent were common, so a family participating in four separate programs might see its total benefits fall by $2 for each extra dollar it earned. Under the circumstances, no formal training in economics was necessary to see that working didn’t pay. In contrast, someone who worked additional hours under Mr. Friedman’s plan would always take home additional after-tax income.

The negative income tax was never adopted in the end, because of concern that a payment large enough to support an urban family of four might induce many to go on the dole. With a payment of $6,000 per person, for example, rural communes of 30 would have a pooled annual payment of $180,000, which they could supplement by growing vegetables and raising animals. Because these groups could live quite comfortably at taxpayer expense, there would be an eager audience for accounts of their doings on the nightly news. Political support for such a program would be difficult to sustain.

Instead, Congress adopted the earned-income tax credit, essentially the same program except that only people who were employed received benefits. One of the few American welfare programs widely adopted in other countries, the earned-income tax credit has proved far more efficient than conventional programs, just as Mr. Friedman predicted. Yet because it covers only those who work, it cannot be the sole weapon in society’s antipoverty arsenal. [Emphasis added]

The fact that “only people who were employed received benefits” is obviously a very significant difference between Friedman’s traditional NIT and the EITC. The Chicago School theory in question that Johnston seems to suggest is not smart enough counsels a strategy quite different from the EITC. 

Let’s return to Jesse Rothstein’s work. The abstract of “Is the EITC as Good as an NIT? Conditional Cash Transfers and Tax Incidence” reads as follows:

The EITC is intended to encourage work. But EITC-induced increases in labor supply may drive wages down. I simulate the economic incidence of the EITC. In each scenario that I consider, a large portion of low-income single mothers’ EITC payments is captured by employers through reduced wages. Workers who are EITC ineligible also see wage declines. By contrast, a traditional Negative Income Tax (NIT ) discourages work, and so induces large transfers from employers to their workers. With my preferred parameters, $1 in EITC spending increases after-tax incomes by $0.73, while $1 spent on the NIT yields $1.39.

So what exactly is going on here?

A basic result in the economics of taxation is that the economic incidence of taxes depends on the elasticities of supply and demand for the good being taxed and not on their statutory incidence. If demand is less than perfectly elastic, supply-side taxes are partially passed through to the demand side via changes in the equilibrium price. Effects on prices are of the opposite sign as those on supply, so any program that increases labor supply will lead to reduced pre-tax wages. This implies that employers of low-skill labor capture a portion of the intended EITC transfer. Moreover, because EITC recipients (primarily single mothers) compete in the same labor markets as others who are ineligible for the credit, wage declines extend to many workers who do not receive offsetting EITC payments. These unintended transfers limit the EITC’s value as a tool for income redistribution. Recognizing the endogeneity of wages thus reduces the attractiveness of work-encouraging transfers like the EITC. But the practical importance of incidence effects is unclear.

In this paper, I show that incidence effects are extremely important to the evaluation of the EITC. With plausible labor supply and demand elasticities, the unintended consequences of the EITC operating through the pre-tax wage are large relative to the direct, intended transfers. Neglecting these wage effects leads to misleading assessments of the impact of a hypothetical EITC expansion on labor supply, incomes, and welfare.

At the end of the paper, Rothstein acknowledges the limitations of his analysis:

First, I have assumed that labor supply elasticities are constant across female workers of different types. It would be straightforward to extend the formulas in Section I to allow for heterogeneity in labor supply behavior. Eissa and Hoynes (2004) estimate a wage elasticity of participation for married women that is much smaller than those typically obtained for single women. With uniform elasticities, the EITC’s negative effect on married women’s labor supply partially offsets its positive effect on that of single mothers. If married women are less responsive than are single women, this offsetting effect is overstated, and the EITC’s net wage effects are even more negative than those presented above.

Second, I have ignored the interaction between my proposed EITC and NIT policies and other preexisting distortions to the low-skill labor market. These would affect the welfare results. By treating my proposed policies as the only taxes, I have been able to ignore deadweight losses as second-order, where the EITC might yield first-order reductions in deadweight loss produced by other work-discouraging programs. Interactions between the EITC and other programs might also have firstorder effects on the government budget. But my results on after-tax incomes would not be affected by the inclusion of other programs in the simulation.

Finally, it would be interesting to examine the impact of incidence effects on the design of optimal transfers. The results here indicate that labor-supply-promoting schedules are less desirable than one might otherwise expect. A plausible consequence is that the optimal tax should have higher (less negative) tax rates at low incomes. This would be a fruitful topic for future research.

Yet none of Rothstein’s caveats challenge the basic conclusion that an NIT would have a somewhat more positive impact on after-tax incomes for those who choose to work. We might as a society decide that this isn’t our goal — rather, that our goal is to encourage labor force participation, particularly for parents. The Chicago School theory behind the traditional NIT turns out to be a pretty smart way “to reduce poverty for those who work.” 

Again, it seems odd to completely ignore the fate of those who in the absence of the EITC would choose not to work. How we feel about this clearly depends on how we feel about the value of labor force participation as a means of being included in the wider society, etc. (See Judith Shklar.) But we can bracket that question for now. 

For an unconventional variation on the NIT concept, you might want to check out Charles Murray’s call for replacing all government transfers with an unconditional basic income grant