A revenue-neutral carbon tax could recycle the revenues by: (A) reducing taxes on capital income (for corporations and individuals), (B) reducing taxes on labor income (including personal income taxes and payroll taxes), (C) reducing consumption taxes (like sales taxes), or (D) by redistributing the revenues to all households in a lump-sum payment. There are many variations, but that is a good set of fundamental policy types.

There is something of a prevailing political wisdom that among these four options, going from option A to D is going from the most efficient to the least efficient in terms of effect of GDP. This is because most economists believe that capital taxes are the most distorting, consumption taxes the least, and that lump-sum distributions distort both household consumption and production decisions in the most costly manner of all. The same prevailing political wisdom is that going from option D to option A is going from the most progressive to the least progressive, meaning it goes from the most politically palatable to the least politically palatable. Who besides Mitt Romney wants to stand up and give the One Percent a tax break, and leave the burden with the 99 percent?

But it turns out that it’s a little more complicated. The Initial Incidence of a Carbon Tax Across US States, by Williams et al, a discussion paper that follows on their earlier paper on the impacts across income groups,  analyzes these options. What they find is not particularly surprising upon reflection, but since few have done much reflection about the incidence of a carbon tax (why bother with something that is politically dead?), they are startling.

First, some not-too-startling findings. The most variation in carbon tax incidence stems from variation in electricity prices and electricity sources (coal, gas, hydro). The greatest incidence — the greatest amount of pain — stems from gasoline usage, but there is much less variation across states for that.

The first startling finding is that a carbon tax paired with a reduction in capital taxes benefits the New England and North Atlantic states the most — the deepest blue states. The states that would be hurt the most by a carbon-for-capital tax swap are the Southeastern states — the reddest states. The reason for that is that a reduction in capital taxes benefits the richest, and the highest income states are in those two regions. Parenthetically, Florida and Wyoming actually come out ahead, because of the extremely high proportion of incomes in those states deriving from capital income. That flips what you would think are the politics of revenue recycling — the Democratic strongholds should be pushing hardest for a reduction in capital taxes, and the Southerners should be screaming bloody murder.

Second, labor tax reductions produce the most even burden, and smallest variation in burden. The richest quintile and the poorest quintile bear more of the burden than the middle three quintiles.

Third, the lump-sum household rebate — by far the most progressive of the options — imposes the most pain on Connecticut, Wyoming, Virginia and Wyoming. The states that actually come out ahead are West Virginia, Kentucky, and Mississippi. Although these are coal-loving states, the incomes in these states are so low that the lump-sum rebate swamps the negative welfare effects of high electricity and lost coal production. This seems to underscore the theme of this paper — that in determining the incidence of a revenue-recycled carbon tax, income effects play an outsized role in determining who wins and who loses.

Lastly, the authors note that District of Columbia usually comes out pretty well in these models — only the lump-sum rebate puts D.C. in the losers camp. In fact, D.C. is a representation of how dense cities fare under the different carbon tax revenue recycling approaches. So if you want to sock it to those liberal urban snobs, support the lump sum distribution option.