The Agenda

Yes, Virginia, There Is Such a Thing As a Tax Expenditure

Cato’s Michael Cannon takes issue with the concept of “tax expenditures,” a term referring to government revenues foregone due to credits and deductions in the tax code. For example, allowing a deduction for home mortgage interest reduces federal tax collections by approximately $120 billion per year. Various exemptions, credits and deductions for health care cut revenues by $313 billion.

Cannon objects that these aren’t really expenditures—this money was never the government’s to begin with, and using the term “expenditure” implies that the money belonged to the government which then benevolently decided to let the taxpayer keep it. I disagree with Cannon—the tax expenditure concept is important for thinking about the federal budget, and saying that tax expenditures are “not expenditures” leads you to some perverse policy conclusions.

Consider the Earned Income Tax Credit. This is a “refundable” credit that goes to people with no income tax liability, and is therefore counted as a true government expenditure rather than a tax expenditure. EITC recipients get actual checks from the government and their amounts are reflected on the spending side of the federal budget.

But most EITC recipients pay payroll tax. Theoretically, we could repeal part or all of the EITC and exempt part of former EITC participants’ income from payroll tax. This program could be designed to be economically identical to the current program, so all the former EITC participants receive the same benefit from the government net of taxes paid, and all face the same effective marginal tax rate.

Indeed, you could do this with almost any government program, especially transfer programs. Cash for Clunkers was a spending program, but it could just as easily have been structured as a tax credit. Such a change would have reduced government tax collections and on-budget spending and therefore have been a “tax cut.” But, again, either way all Americans would have faced the same marginal tax rates and the same net government burden (taxes paid less benefits received).

If you hold that tax expenditures aren’t expenditures, how you structure these programs matter a lot when measuring the government: “government spending” is lower if you turn the EITC into a payroll tax exclusion, or Cash for Clunkers into a tax credit. But if two policies have identical economic effects, shouldn’t we try to measure them in the same way? This is what the “tax expenditure” concept does: it allows us to measure economically identical government programs in comparable ways, whether they operate through the tax code or not.

I had a rather bizarre exchange about a year ago about tax expenditures with Ryan Ellis from Americans for Tax Reform. Ryan really hates the tax expenditure concept; he says it was invented by the Carter Administration as an excuse to raise taxes. What’s bizarre is that Ellis claims that tax expenditures are actually better for economic output than identically-designed direct government spending. Cash for Clunkers might be the same to an individual car buyer whether it’s a voucher or a credit, he says, but a credit structure would reduce taxes as a share of GDP and therefore increase economic growth.

This, of course, is nonsense. There is a correlation between the tax share of GDP and economic growth, but that does not mean that any policy that lowers the tax share of GDP necessarily increases economic growth. The way taxes harm economic growth is by driving a wedge between the private benefit and total benefit of economic activity: if you only get to keep 75 cents of the next dollar you earn, you’ll only provide a dollar’s worth of output if you can do so at a cost to you of less than 75 cents. When your marginal tax rate goes up, the tax wedge gets wider, and more economic activity is lost.

Tax share of GDP is correlated with GDP growth because achieving a higher tax share of GDP often involves applying higher effective marginal tax rates. However, this is not always the case. As discussed above, converting Cash for Clunkers from a voucher to a credit would reduce the tax share of GDP but would not change effective marginal rates and therefore would not affect economic growth. The logical error here is assuming that a correlation between two variables translates to an ironclad relationship.

There is also a simpler reason why substituting a tax credit for an on-budget expenditure cannot affect the size of the economy: a policy with no micro-level effects by definition has no macro-level effects. GDP is merely the sum of economic activities undertaken by individuals. If a policy change does not affect endowments or incentives for any individual, it cannot affect aggregate GDP.

This is the problem with denying the tax expenditure concept. A key focus of the conservative project in Washington is to reduce the tax and spending shares of GDP. Using an under-inclusive definition of “spending” encourages policymakers to use tax expenditures instead of standard expenditures. As such, we may achieve an appealingly low measure of taxes as a share of GDP, but with high marginal tax rates that are a drag on the economy.

If you had taxes at 19 percent of GDP and few tax expenditures, you could have much lower marginal tax rates than if you had taxes at 19 percent of GDP and many tax expenditures. The former case, of course, would much better for the economy. But if you deny the tax expenditure concept, the distinction gets lost—and tax share of GDP will tell you little about the tax code’s effects on the economic growth.

In his post, Cannon acknowledges that tax expenditures “give power to politicians, inhibit freedom, reduce economic output, unjustly enrich special-interest groups, et cetera.” But the he also says “When you hear a politician use the terms tax expenditure, tax subsidy, or backdoor spending in the tax code, beware.  He’s about to raise your taxes.”

So, I suppose I have two and a half questions for Mike: (1) What term would you like to use for the preferences formerly known as tax expenditures? (2) If the government converts a tax expenditure into an economically-identical on-budget expenditure, has any person suffered from a negative economic effect or a loss of freedom? And (2a): if “no”, then shouldn’t we be indifferent or even favorable to certain kinds of tax increases?

My argument for including tax expenditures in the overall expenditure discussion is they have similar or identical economic effects to direct government spending, reduce economic output, are often opaque, and are not subject to annual reauthorization and so less likely to face appropriate legislative scrutiny. Mike’s argument for separating them out from expenditures is that the term “tax expenditure” creates the rhetorical impression that individuals’ money belongs to the government first.

I tend to think the substantive concerns I list here are more important than this issue of rhetoric. But perhaps we could both be satisfied by developing a better term.

Josh Barro — Mr. Barro is the Walter B. Wriston fellow at the Manhattan Institute. His research is focused on state and local fiscal policy.


The Latest