If you’re working on your taxes this week, you may find yourself buried in various forms and schedules. Why is it all so complicated? One reason is that companies that benefit from the annoyance of tax season, like TurboTax, have put a lot of energy into making sure tax forms stay complicated. But another is that the U.S. government uses tax breaks to accomplish a lot of things that other countries do through direct payments to citizens. For example, instead of a child allowance sent to all parents, we have the child tax credit.

Why do we do things this way? In 2014, political scientists Christopher Faricy and Christopher Ellis investigated.

Faricy and Ellis note that direct spending on things like Social Security, Medicare, and cash benefits to low-income families is highly visible. On the other hand, tax breaks often go unnoticed. Many people don’t think of the mortgage interest tax deduction as government spending, even though it has the exact same relationship to the federal budget as a direct payment. Other federal spending in that category includes the tax breaks that help employers and employees pay for health insurance and 401(k) accounts.

To examine public opinion, the authors created a survey in which they described versions of three existing programs—the mortgage interest deduction, the retirement savings tax deduction, and food stamps (the Supplemental Nutrition Assistance Program, or SNAP). They varied the descriptions in two ways. In some versions of the survey, they said the program was a tax break, while in others they called it direct spending. Some descriptions also included information about who would benefit most from the program—for example, that the mortgage assistance helped people with expensive homes the most—while others left that out.

Faricy and Ellis recruited a politically diverse sample of 373 undergraduate students to take the survey. They found that, in general, respondents supported food stamps the most of the three programs, and mortgage interest subsidies the least. Hearing about the upward redistributional effects of the mortgage and retirement programs dampened support for them. So did having these programs presented as direct spending rather than tax breaks. On the other hand, neither tweak to the description had much effect on support for food stamps.

Democrats and Republicans responded to the different descriptions of the mortgage and retirement programs in dramatically different ways. Hearing that programs were particularly helpful to well-off people made Democrats much less likely to support them. For Republicans, these redistributional effects didn’t make much difference. But Republicans were much more likely to support these measures if they operated as tax benefits rather than direct spending. Payment methods didn’t matter much to Democrats.

Faricy and Ellis’s results suggest that the reason for packaging government spending as tax breaks is clear—at least when it comes to benefits that flow toward richer people, it seems to make it more palatable to a lot of voters.