What’s the difference between a universal and means-tested child allowance?

In the discourse, mean-tested benefits are often referred to as more “progressive” than universal alternatives. In a piece defending the EITC, Max B. Sawicky remarked:

A means-tested children’s allowance is preferable to a universal one because it can provide more aid to families in greater need.

The thing that’s important to understand is that means-testing is essentially a tax implemented through the welfare state. So the claim really is that if we tax the rich a bit more, we can provide more generous benefits. As a fan of taxing the rich, this sounds intuitively good. But is means-testing a good way of going about this? Let’s consider two hypothetical Child Allowance program designs.

• Phased-out Child Allowance (PCA): every parent gets $3600/year, per child. It phases-out at $100,000, at a rate of 5% (that is to say, for every dollar a parent makes above $100,000, they lose 5 cents of the child benefit). At $172,000, parents get no child benefit.
• Universal Child Allowance (UCA): every parent gets $3600/year, per child. There is a 2% tax on all income above $100,000, including both parents and non parents.

These two programs have identical costs. The 5% phase-out and the 2% income tax raise the same amount. The remaining cost could be financed however you’d like, but since it’s identical it isn’t relevant to the analysis. So what’s the difference between these proposals? First lets look at how much income above $100k is taxed by each proposal (this excludes all other taxation). As you can see below, the PCA exempts non-parents from any implicit taxation, and it also has a regressive structure, because once the allowance has fully phased-out, it essentially turns into a head tax of $3600. That means that a parent earning $1 million pays a .4% effective tax rate on their income above $100,000 as a result of the phase-out. The UCA, meanwhile, taxes all income above $100,000 at 2%, no matter if you are a parent or not.

The UCA is doing what some people might call “broadening the base, lowering the rates.”

Here’s what it looks like in terms of total dollar amounts, taking into account the tax/phase-out:

What you see here is that not only is the UCA is more vertically equitable (as in, the richer you are, the more you pay), it’s also more horizontally equitable. At every point in the income distribution, if you have greater expenses due to having children, you have more resources than someone with the same market income, but no children. The PCA, on the other hand, is no longer horizontally equitable by the time it fully phases out. Horizontal equity is desirable because raising children is important societal work, and parents face considerable additional costs. At any given income level, parents should have more material resources than non-parents.

What if, you might ask, we do both a phase-out and a tax above $100k? The issue is that if we’re trying to maximize tax rates, the phase-out doesn’t achieve anything that a tax could do better. Say we want to have a 70% marginal tax rate on income over $100k (as Sweden does), and we suppose that this is the revenue maximizing rate, so we don’t want to go above this. Because the UCA has broadened the base and lowered the rates, it leaves us with room for an additional 3% tax on income over $100k.

All in all, it seems clear that phase-outs are just an inferior way of taxing the rich. They may not be the end of the world, especially at higher incomes, and so a bill like the American Family Act (similar to the hypothetical PCA) is still well worth passing (especially given that it eliminates the much worse phase-in that excludes the poor). But if we’re trying to determine what optimal policy is, just make it universal and save the taxes for the tax code.