Andrew Sullivan is smitten

A simplified tax code, consisting of a two-bracket income tax with a large standard deduction and a business consumption tax, would pay for a means-tested safety net, and a system of tax credits, risk pools and low-income subsidies would underwrite a free (or, well, somewhat freer) market in health care. In other words, Ryan would balance our books by shifting away from programs that shuffle money around within the middle and upper-middle classes — taking tax dollars with one hand and giving health-insurance deductions, college-tuition credits, home-mortgage deductions, Social Security checks and so forth with the other — and toward programs that tax the majority of Americans to fund means-tested support for the old, the sick, and the poor.

Paul Ryan is my new hero.

There is a lot about Ryan’s proposals to like: incentive based, progressive and in the long run, cost containing. Yglesias and Klein have focused on some of the downsides from a liberal perspective.

There is a potentially large libertarian downside as well: the  creation of extremely high marginal tax rates at low income levels. Ryan suggests that we means test everything. From the point of view of a household though, means testing is no different from marginal taxation. Every extra dollar you earn reduces the amount of government benefits you receive and so the return from working or investing is muted.

Means-testing can be particularly sinister because the tax rate is not obvious. The result is that effective marginal tax rates could climb to 100% or more. I tend to think that the elasticity of labor supply is pretty low and so the effects of marginal taxation can be easily exaggerated. Still, marginal rates approaching 100% are going to have serious incentive effects.

In addition, even more reasonable but very high marginal tax rates can have strong disincentive effects for the poor. High enough effective marginal tax rates and entering the underground economy becomes profitable, investing in education becomes unprofitable, etc.

Mechanically, means-testing lots of credits is little different than giving everyone the credit and then paying for it with high income tax rates. In some ways explicitly high tax rates are preferable because they are transparent.

Further still, there is not much difference between a tax credit and a voucher system. There can be technical details between means tested credits, universal credits funded by marginal taxation, and universal vouchers funded by marginal taxation but the basic incentive effects are all the same.

Under a Ryanesque approach the accounting will show that government spending is smaller and taxes are smaller. Under a universal voucher approach the accounting will show that government spending is huge and taxes are huge. But the basic economic effects are exactly the same. We should not get lost in the accounting and think that we have done something special because we get different numbers for “taxes” and “spending.”  What matters is the effect government has on the economy.