I appreciate Scott Sumner’s post, "A Kindred Spirit," and Karl Smith’s post "Welcome Miles Kimball, Now Let’s Set the Record Straight on Taxes." Karl Smith raises the issue of whether taxes can raise GDP and rightly points out that a substantial income effect means that taxes can raise GDP. Indeed, it is a standard result in Real Business Cycle Theory that once prices adjust, extra lump-sum taxes to finance, say, an expensive war, will raise GDP by making everyone feel poorer so that they work harder. A simple lump-sum tax is a tax that falls equally on everyone in dollar terms: say every adult needs to pay $5000 per year.
Continuing to think about very simple models and government spending on something like an expensive war, what about a consumption tax that is a certain percentage of everyone’s consumption? On the one hand this makes people feel poorer so they want to work more, but on the other hand, what someone can buy with an extra hour of work is less, so they want to work less. The standard view is that these two effects will roughly cancel each other out. So the amount people want to work—and thus GDP in this simple model—will stay about the same. Matthew Shapiro and I take exactly this standard view in our paper "Labor Supply: Are the Income and Substitution Effects Both Large or Both Small?" The basic argument for the standard view is that to households, a consumption tax looks a lot like a wage cut. And we have a lot of information about what higher or lower wages do to desired work hours. Among people who have to live on their own wages, there is surprisingly little difference in how many hours people want to work based on whether they have high wages overall or low wages overall over the course of their lives.
Scott Sumner replies to Karl Smith, saying among other things:
It’s true that measured GDP might not fall if the money is wasted (as poorer people tend to work harder.) But that’s not much of an argument for big government.
Scott Sumner is absolutely right that any unnecessary government expenditure is to be lamented. Robbing people of their leisure time as well as of the wherewithal to spend on household wants and needs for no good purpose is a crime.
But suppose the war really is necessary—or that the President has decided on the war and that as his economic advisor the only thing you get to decide is how to pay for the war. What would be the best way to pay for the war if everyone in the economy had the same income level so that poor vs. rich weren’t an issue?
With everyone at the same income level, a lump-sum tax is fair. It raises GDP. The lump-sum tax leads everyone to take some of the hit from having to finance the war in lower consumption and some of the hit from having to finance the war in less leisure time. Less time away from work leads to higher GDP as more hours at work make it possible to produce more. Guns are counted in GDP as well as butter, and gun production goes up more than production of consumption goods such as butter goes down.
By contrast, the consumption tax with its tax distortion pushes everyone to take all of the hit from having to finance the war in lower consumption, and none of the hit in less leisure time. Production of consumption goods such as butter goes down so much that GDP stays the same even though production of guns goes up. This is inefficient. The production of guns is the same in both cases. But it is less painful to cut back some on leisure and some on consumption (what happens with the lump-sum tax) than to take the whole hit in consumption (what happens with the consumption tax).
So the bottom line is that, other than a concern with redistribution, a distortionary consumption tax to finance a war that leaves GDP the same is less efficient than a non-distortionary lump-sum tax to finance a war that has GDP go up to help make the war effort possible without sacrificing as much consumption.
The broader message is that taxes can be bad even if they don’t lower GDP.