## Candidates, Marginal Tax Rates, and Economic Welfare

Typically, TPC measures the impact of tax laws and proposals in terms of average tax rates. This gives a good measure of how taxes affect our pocketbooks, but economists also like to examine how taxes affect economic incentives. For that purpose, the effective marginal tax rate is most apt. That is, how much tax do we pay on an additional dollar of income. This is important because marginal rates affect the incentive to work that extra hour or to engage in tax avoidance.

My colleagues, Katie Lim and Jeff Rohaly, just released a fascinating analysis of how the presidential candidates' tax plans would affect marginal effective income tax rates on earnings. Katie and Jeff found that, on average, Obama's plan would leave marginal tax rates pretty much where they are now while McCain would cut them modestly. Although Obama would cut effective marginal rates for many more tax units than McCain (61 percent in 2009 vs. 20 percent), he'd also raise them on more taxpayers (15 percent vs. just 1 percent). Not surprisingly, the most dramatic difference is among those in the top brackets. On average, in 2009, Obama would raise effective tax rates on millionaires by almost 6 percentage points while McCain would leave them unchanged.

[Warning: The following gets a little technical. For non-economists, the upshot is that if we calculate marginal rates a different way, as suggested by Greg Mankiw, a Harvard prof who used to be President Bush's top economist, the comparison is more favorable to McCain. Caveats follow.]

When Greg Mankiw saw our study, he observed on his blog that average effective tax rates might not be the best way to measure the impact of taxes on economic welfare:

The deadweight loss of taxation rises roughly with the square of the tax rate. As a result, if one person sees the marginal tax rate fall from 20 to 15, while another sees it rise from 30 to 35, the average marginal tax rate is unchanged, but the deadweight loss increases. Perhaps the relevant average for thinking about deadweight loss is not the mean as conventionally computed but is, instead, the square root of the average squared tax rate.
Katie calculated this interesting “Mankiw effective tax rate” in the following table.

The Mankiw measure shows about a 1-percentage point increase in rates in 2009 under the Obama proposal and a 1-percentage point decline under McCain. As Greg conjectured, the difference between the two plans is somewhat larger under this metric than the traditional measure (2.4 percent vs.1.5 percent).

There's a question about how to treat negative effective tax rates—as apply to low-income households eligible for refundable tax credits that phase in with earnings (such as the EITC, the child tax credit, or Senator Obama's proposed working families tax credit). The negative rates are subsidies, not taxes, but a squared -10 percent rate would look just like a squared +10 percent rate. This is the right answer if the concern is the distortion in the choice between labor and leisure. However, if there are good policy reasons to favor work over leisure, it might be misleading. For that reason, we also show the average rates setting negative marginal tax rates to zero. This modification doesn't change the qualitative conclusions, but both McCain's and Obama's effective rates are slightly smaller by this measure.

Here's the caveat. McCain's and Obama's plans raise different amounts of revenue. The long-term economic effect of the two plans depends on their impacts on the deficits (and, implicitly, future tax rates). For example, if you believe that Senator McCain will do a better job at controlling spending than Senator Obama, then the lower tax rates under his plan are likely to be more durable. If, as the Obama campaign claims, McCain would spend more because of his policies in Iraq and elsewhere, then Obama's plan might be better for the economy over the long run.

Thus, a calculation of the long-term economic effects depends critically on what happens to spending. The Tax Policy Center will be examining this very question in a forum on Friday at the Urban Institute. Click this link to register.