My recent radio silence here reflects multiple demands on my time these days, including health issues for a more senior close relative. These things are never fun, but everyone goes through them at some point, unless something even worse happens.
But anyway, back here at NYU, after a week off (due to calendar complexities), the Tax Policy Colloquium resumed yesterday, with Jacob Goldin presenting this paper on increasing EITC take-up by individuals who are eligible for it. It draws on related empirical work (by Goldin and Taylor Cranor) suggesting that, when jurisdictions require employers to provide EITC information to their employees, this has “precise null effects on EITC take-up.” That is, Cranor and Goldin conclude in the empirical paper, not that they can’t find any effect on take-up, but that they have found, with high confidence, a lack of any effect.
In the paper that we discussed yesterday at the colloquium, Goldin assessess how EITC take-up might best be increased. He concludes that by far the most efficacious method would be to increase income tax filing by eligible individuals, in particular those who would get a net refund from filing but don’t currently do so.
A key point on which this argument relies is the near-ubiquity of assisted preparation methods (APMs) among filers who qualify for the EITC. An APM can either involve electronic software, a la TurboTax, or the aid of a tax professional, including through free VITA assistance. (Or both, as the tax professional is likely to be using electronic software.)
With APMs that use electronic software, the paper argues that the issue of computational complexity for the EITC pretty much disappears. Enter the pertinent information, and presumably the net liability or refund, including by reason of the EITC, will come out on the other end.
There is still the problem of informational complexity, given the multiple factual prerequisites for claiming the EITC. (E.g., where a child is spending time in multiple households &/or is supported in part by multiple adults, one gets the question of who is entitled to claim the EITC.) But the paper argues that the EITC has almost no marginal effect on informational complexity for a given taxpayer, since nearly all of the facts that are needed to determine EITC eligibility and its amount are also needed on the tax return for other purposes.
Here are a few other interesting issues that the paper invites one to think about:
1) Is there any way to loosen up a bit on who gets to claim the EITC? In principle, ought it to be allowed to anyone who is involved in the child’s support? Obviously the problem here is coordinating between potential claimants, so that only one claims it, and preferably it’s whomever they agreed to. Probably not realistically feasible to address this without much greater changes to how tax filing takes place, and/or how the income tax is generally administered. But there may be a decent number of cases in which, seemingly, there is both under-claiming by the person who’s supposed to claim it, and an “incorrect” claim by someone else, in circumstances where it makes little or no difference. This can lead to misleading statistics regarding both under-claiming and EITC malfeasance.
2) Is stigma a significant issue in discouraging EITC take-up? I am inclined to doubt it. After all, one gets it via one’s tax return, a refund is cash, and it’s associated with work that our society valorizes (and perhaps even over-valorizes). But on the other hand, Vanessa Williamson’s paper from week 5 suggested that people value being called ‘taxpayers.”
The main reasons for underclaiming may be (a) under-awareness of it, leading to non-filing where one would have gotten a net refund, and (b) rationally choosing not to file, e.g., because the refund would be smaller than the cost of the APM one would use. But rational non-filing is less benign when what makes it personally rational is that the refund would be garnished and used, say, to pay back arrears in child support.
3) The EITC, of course, is far more generous for taxpayers with children than those without. This seems misguided in one sense, but not another.
It makes a whole lot of sense for the fiscal system to be more generous, especially at the low end of the distribution, to households with children (or more children) than to those without them. The welfare of the children is the main reason for this, but their presence may also greatly increase the marginal utility of a dollar to their parents or other caregivers.
For reasons I’ll discuss shortly, it also may make sense to have an EITC – even taking as given that the overall amount of aid given to poorer individuals will be constant with or without it, so that the only question at issue is whether some of the aid is positively tied to wages, EITC-style.
But the case for basing such aid on wages seems to me to be primarily independent of the case for giving more aid to poor households with children than to those without.
Why would one tie the relative benefit that poor households with children receive, relative to those without children, to whether the caregivers work? One reason is that working raises the cost of childcare. But that in principle ought to be handled at a more general level, without tying it to wages earned in the way that the EITC does. Perhaps a second reason is that the reasons for encouraging low-wage work, where the alternative is no work rather than higher-wage work, might include modeling working to the kids. But the latter is a bit of a thin reed here, e.g., since why punish children in households where the caregivers don’t have market jobs.
4) Why favor increased EITC take-up? I agree with the paper that we should favor it (other than perhaps in cases where it’s rationally not claimed and this doesn’t reflect the garnishment issue. But the analysis of why we should favor it has several complications that are worth noting.
The main reason for favoring increased take-up, of course, is to increase the welfare of people in the claimants’ households. But also, the EITC can have positive labor supply effects that are part of the EITC’s underlying rationale.
Labor supply effects may arise either on the extensive margin (whether or not one works) or on the intensive margin (how much one works). I believe the EITC empirical literature suggests that the extensive effects tend to be more significant empirically. This is a good thing, given that the EITC’s extensive effects ought to be purely positive. One never has a negative EITC – it’s either positive (rising via an initial 40% wage subsidy) or zero (reflecting its being phased out at a 21% rate). The intensive effects, by contrast, can be either positive or negative. Wage-earning is encouraged at the margin during the 40% phase-in stage, and discouraged at the margin during the 21% phase-out stage.
Why encourage work, via a wage subsidy? In a standard optimal tax model, it usually turns out that the tax rate on earnings should be somewhere between 0% and 100%, rather than being negative 40% via the payment of a wage subsidy.
Note, by the way, that the wage subsidy causes the EITC to be anti-insurance, as between successful and unsuccessful job-seekers. The former are already doing better than the latter by reason of having succeeded in the mutual job quest, then they’re further rewarded by the EITC, while the latter get nothing.
The reasons for nonetheless possibly favoring a wage subsidy via the EITC include:
(a) claims about positive externalities, if working today increases the likelihood of working in the future & this benefits others, such as one’s children or for that matter other taxpayers,
(b) parallel claims about positive internalities, and
(c) concern about the overall marginal tax rate. At low income levels, even if one gets a 40% wage subsidy via the EITC, one may still have a positive MTR due to other features of the fiscal system, such as the phase-out of means-tested benefits.
But now let’s bring this back to the question of increasing take-up. Suppose people get the EITC despite not knowing it’s there. Or more precisely, suppose they know too little about it to have any well-developed understanding of how it is affecting their after-tax-and-transfer bottom line. This wouldn’t negate favoring increased take-up due to the welfare effects in claimants’ households – but does it negate reliance on the work-encouraging rationales for the EITC?
Not necessarily. Many years ago at the colloquium, we had an econometrics paper (I believe, by Nada Eissa and Hilary Hoynes) that showed apparent behavioral responses to fine points of EITC design. An attendee scoffed at the findings, saying they couldn’t be true because surely none of the people whose actions the data reflected had more than the crudest understanding of how the rules actually worked. The author’s response was: That may well be, but nonetheless our empirical finding is what it is.
My thought (and that of others in the room) at the time was that this wasn’t actually implausible. Suppose there was a work subsidy, always positive at the extensive margin whether or not at the intensive margin, that no-one knew was there (although workers knew how much $$ they ended up with, reflecting payment of the subsidy into their bank accounts). It might still increase labor supply at the extensive margin, through the following mechanism. It would cause people who worked to do better than they would have otherwise. People might have a sense of how they were doing overall when they worked vs. didn’t work, and might also have this sense as to their neighbors. The invisible wage subsidy might thus encourage work indirectly, by causing people to associate working with a better end state of affairs.