Yesterday at the Tax Policy Colloquium, Henrik Kleven presented his paper, The EITC and the Extensive Margin: A Reappraisal.
This paper challenges the previously prevailing view that the earned income tax credit (EITC), at least when it was significantly expanded in 1993, had significant positive labor supply effects at the extensive margin (pertaining to whether or not one worked in a market job, as distinct from how many hours one worked). It notes that all EITC changes other than that in 1993 seem not to have affected employment levels among low-wage workers, despite the positive substitution effects that it should have in the phase-in range (and indeed, at the extensive margin, until full phase-out).
As to the 1993 change, it finds that the previously claimed positive employment effects appear better attributable to two confounding (from a statistical measurement standpoint) sources of employment gain from the same era. The first is the enactment of welfare reform. The second is the era’s booming economy. The paper finds that the era’s employment gains from that era better fit welfare reform and the booming economy, and in any event cannot with any confidence be attributed to the EITC change, especially given evidence from all the other EITC changes.
Why would the prior literature have gotten this wrong? Disentanglement is indeed extremely difficult, and empirical techniques have improved since some of the earlier papers were written. Plus, economists may naturally expect that people will respond to incentives, even if, with respect to real labor supply, it is not so clear that we should expect to see this at any significant level. There also may be publication bias against papers that fail to find a significant response of a given X to a given Y.
However, the topic of this particular paper makes it no-response finding highly controversial. Many people on both the left and the right would like there to be a positive response. On the left, the fact that the EITC has greater potential political support than, say, unconditional cash grants – because it only goes to the working poor, who are less easily slandered than the rest as “undeserving” – can make attacks on its efficacy at the extensive margin quite unwelcome. On the right, while scoundrels may want to deny all aid to the poor no matter what, those who are acting in good faith may be glad that there is a poverty relief program they can point to, and that (previously at least) was thought to have effects that they may not just welcome but consider vital.
The Wall Street Journal added gas to the flames surrounding this paper by publishing an editorial that said: Given the paper’s findings, we no longer support the EITC. This was transparently disingenuous on the WSJ’s part, given the utter impossibility that papers showing low responsiveness to taxes’ substitution effects would ever affect its views of taxing the rich. But the WSJ’s eagerness to advance its top-down class warfare by citing the paper certainly helped to show the temperature-raising stakes here.
I am not an empiricist, but I found the paper’s empirical analysis very impressive. Empirical debate should and will go on, but this is a major contribution that appears to me to move the needle a bit. But that said, how should it affect one’s views of the EITC? These should not, after all, turn just on the EITC’s labor supply effects at the extensive margin. So here are some thoughts on the surrounding issues.
1) Why seek to increase low-wage labor supply, and how much does it matter to the EITC’s merits? From a standard neoclassical standpoint, labor supply is merely a commodity choice between (a) the market goods that one can buy with wages, and (b) “leisure” in the economist’ sense. (I add the scare quotes because, say, raising a couple of young children in lieu of taking a market job is not exactly “leisure” in the layperson’s sense of the word.) So why tilt the commodity choice towards working more?
One reason for doing this might be that low-wage workers face high implicit marginal tax rates in other respects, such as by reason of the phaseout of income-conditioned benefits. So here the point would not be more work for its own sake, but making the overall labor supply choice in this range more neutral.
A second reason would be the view that positive internalities and externalities strongly support increasing low-wage labor supply. Hence, according to this view, one might not only want to subsidize work, creating substitution in its favor, but also induce it via income effects (by not giving able-bodied people enough to live on unless they work).
While that view may have some merit, at least in some cases, I think we should also keep in mind that low-paying jobs (a) are often quite miserable, and are not always good stepping stones to something better, and (b) may compete with other valuable uses of one’s time, such as spending time with one’s own young children, e.g., in a single-parent household.
If one would favor unconditional cash grants – the position towards which I lean – then a lack of positive labor supply response to the EITC is nowhere near fatal. So long as the EITC can’t be traded in as a political matter for more cash grants – i.e., so long as the legislative counterfactual is simply less aid to the poor – the program still has merit.
That said, one should keep in mind the EITC’s second-bestness insofar as one favors more aid to the poor without conditioning it on labor supply. From a behind-the-veil standpoint, the EITC is social insurance against the risk of being a person who can only earn low, rather than high wages. But it is anti-insurance as between poor people who succeed in finding jobs vs. those who fail, rewarding the winners and offering nothing to the losers.
2) Broader lessons – Why doesn’t substitution seem to matter here, assuming that one accepts the paper’s empirical findings? The two main suspects here are (a) generally low real labor supply elasticity, and (b) the EITC’s complexity and lack of salience. Based on (a), does it offer indirect support for taxing the rich at high levels, even though that is admittedly a different population? (And one with greater access to formal tax planning responses.) Based on (b), should the EITC be revised or else replaced with means of offering cash support to low-wage workers that are more transparent and salient?