Recently, Ken Jacobs, Ian Perry and Jenifer MacGillvary from UC Berkeley Center for Labor Research and Education released a detailed report showing that the majority of public assistance payments—such as EITC, food stamps (SNAP), child care subsidies, TANF, Medicaid–go to working families.  Working, per se, turns out not to be a guarantee against reliance on such assistance. When wages are low (or workers have insufficient hours) families often qualify for tax-payer funded safety net programs. Across states, the working family share of pubic assistance ranged from 43% in Alaska to 66% in Texas.

So is this a problem? That depends on who you ask. Moreover, there are different reasons for considering this to be a problem.  Here I lay out three such reasons, and assess their relevance.

While I am not certain, I would guess that most Americans find it troubling that if someone is working hard, they still qualify for food stamps. This is a normative judgment about who should pick up the tab to ensure a basic quality of life: employers or taxpayers.  Now, it’s not quite that simple, since cost of higher wages may also be borne by customers of those products and services. But fundamentally, most people who think this is a problem have a fairness argument like this in mind. From a purely economic framework, what matters is the ultimate incidence on various groups.  If we fund redistribution via taxes and transfers, and it’s paid for by wealthy taxpayers, that’s no worse than funding it via reduced profits going to wealthy investors. Moreover, the redistribution could be made more progressive if done through the tax code than via a mix of lower profits and higher product prices. But people often have procedural concerns that go beyond economic incidence. The idea that work should pay sufficiently to attain a standard of living (regardless of who ultimately bears the incidence of that pay) is an appealing idea to many people.

Besides this fairness rationale, there is also an efficiency argument for paying people via wages rather than public assistance. As Nathan Hendren from Harvard points out in his paper on the “inequality deflator,” existing research suggests it may be more costly to transfer money from the rich to the poor using tax policies than it is the other way around.  He argues that $1 of income falling to the bottom of the distribution can be turned into $1.15 of surplus to everyone. In contrast, transferring from the top to the bottom is more costly: $1 of income falling to the top can be turned into $0.60 of surplus for others. Why? Because taxable income of the wealthy is more sensitive to tax rates, perhaps due to tax avoidance possibilities.  As a result, a $1 of pre-tax income is more valuable socially when it accrues to the bottom of the income distribution. In turn, this means that wage inequality is costly because it requires compensation through costly tax and transfer mechanisms.  These calculations assume that the redistribution occurs through the tax code (like EITC), which is relatively more efficient. Redistribution through other transfers may be more distortionary: for example, he argues that food stamps are relatively more costly for that reason.  Another practical concern is that it maybe hard to raise taxes politically, so the shadow cost of a public dollar may be even higher than just from economic distortions. The bottom line: the efficiency costs associated with means tested programs means that—all else equal—we should prefer someone getting paid a $1 more in wages than through public assistance. And when all else is not equal, we need to trade off the distortions from raising wages with the shadow cost of providing public assistance.

A final line of argument is that these public assistance programs have become de-facto subsidies for low-wage employers. For a program to be a subsidy for an employer, it needs to lower wages. Is this plausible for the public assistance programs considered? I think it is for the EITC, but not for other programs. Depending on where one is on the EITC schedule, that policy can increase work incentives. And there is a lot of empirical evidence showing EITC encourages labor force participation. An unintended consequence of that labor supply response, however, is that employers capture some of the tax subsidies. This can happen in a simple supply and demand framework, where an increased labor supply to the market drive wages down. This can also happen in a bargaining context where the size of the bilateral surplus expands from lower taxes, and employers capture some of this increased surplus. Work by UC Berkeley’s Jesse Rothstein suggests that for every $1 of transfer to workers using the EITC, post-tax income rises only by $0.73 because of employer capture.

But what about other programs like food stamps or housing assistance? These means tested public assistance programs are not tied to work, and we should not expect them to lower wages. Let’s take food stamps, which are available to eligible families whether or not a family member works or not. Indeed, when people are not working, they are more likely to be eligible for food stamps since their family incomes will be lower. Therefore, SNAP is likely to raise, and not lower a worker’s reservation wages—the fallback position if she loses her job.  This will tend to contract labor supply (or improve a worker’s bargaining position), putting an upward pressure on the wage. Whether or not wages are increased is an empirical matter: there is evidence that the initial roll-out of the food stamps program across counties in the 1970s lowered work hours, consistent with an increase in the reservation wage. The key point is that it is difficult to imagine how food stamps would lower wages.  And if they don’t lower wages, they can’t be thought of as subsidies to low wage employers. The same logic applies to other means tested programs like energy or housing assistance. Moreover, these conclusions hold in a wide array of models of the labor market, including ones that emphasize bargaining or efficiency wage concerns.