Kevin Drum has been reading a survey of the literature on the elasticity of taxable income. He draws conclusions that Karl Smith suggests are basically right. Megan McArdle draws different conclusions, which I endorse.
Drum believes the following:
[W]hen taxes go up on the rich, they do report lower incomes. But that’s mostly because they’re fiddling with the tax code to report lower incomes, not because they’re actually earning any less. If that’s the case, we can draw a few conclusions:
* We should reduce high-end tax loopholes so that the rich have fewer options for moving income around solely to optimize their taxes.
* If we do that, modest increases in marginal rates on the rich will have very little impact on their taxable income.
* And even if we don’t, this sort of tax avoidance presents us with nothing worse than a mechanical issue of properly estimating tax receipts. Aside from the small inefficiency of paying tax accountants for lots of useless work, raising tax rates doesn’t have a negative effect on the economy and has little or no effect on the actual incomes of the rich.
This was very interesting to read.
Megan replies:
This is not at all what the cited literature says. It doesn’t tell us that marginal tax rates don’t matter, but that we haven’t been very successful at raising marginal tax rates. It would in fact be much more difficult than Kevin makes out to really ramp up marginal tax breaks on top brackets–it would involve things like ending the tax-deductibility of muni bonds, getting rid of most charitable deductions, and really taxing the hell out of capital income. If we did that, marginal rates would be higher. But growth would probably be lower.
The view that higher tax rates have very little effect on the economy will continue to persist, regardless of the arguments and evidence we bring to bear. I do recommend reading Arpit Gupta’s post on reconciling Edward Prescott’s macro estimates with the work of Raj Chetty et al. on Danish tax records:
Chetty points out that the micro estimates rely on instantaneous adjustment to higher tax rates, and typically focus on short durations after law changes. However, a variety of factors may combine to make the behavior responses to tax cuts a more long-run effect. People face costs in switching jobs or entering the job force. They may simply be unaware of tax changes or lazy. Any of these plausible frictions are compatible with large long-term effects of tax cuts that are difficult to capture in micro data.
This distinction is important, because policymakers are generally interested in the economy-wide and durable impacts of tax increases, rather than their short-term impacts. Macro estimates, which use economy-wide data, may be better suited to answer this question.
In a separate paper, Chetty and coauthors develop new techniques to capture broader responses to taxation while looking at firm-level data in Denmark. They are able to obtain a set of estimates that suggest that the work disincentives of taxes, properly computed, are closer to the higher macro estimates (though lower than some estimates Prescott prefers). These figures would suggest that a sizable portion of the Europe-America income difference can be accounted for by differences in marginal tax rates.
In another post, Arpit drew on the work of Harald Uhlig and Mathias Trabandt:
In particular, they finds that 32% of a labor tax cut, and 51% of a capital tax cut are self-financing, in the sense that lowering those taxes raises economy activity, which itself generates additional taxes, and partially offsets lower tax revenue.
By the same token, higher taxes – particularly higher capital gains taxes – will reduce economic activity, especially in the long run. This will result in a substantial amount of foregone income, as a result of the “deadweight loss” incurred through taxation. As the tax rate approaches the top of the Laffer curve, this loss grows higher and higher.
In other words, future tax hikes, which are necessary to pay for the projected path of spending, will come at a high cost. Even if they are sufficient to balance the budget and eliminate the deficit, and even if higher tax rates still result in more revenue, high taxes will still result in less output for all Americans.
Basically, I think the evidence that MTRs have efficiency costs is pretty clear. The question is whether or not these efficiency costs are worth bearing. If raising revenue is our goal, there are far less costly ways to do it, e.g., eliminating tax expenditures.
Part of the issue here, incidentally, could be a disagreement regarding what is and is not “bad for the economy.” I don’t think that small increases in MTRs are in themselves “a disaster,” and I’ve never suggested otherwise. Rather, I think that high MTRs create work disincentives that dampen labor supply and reduce the overall level of economic activity. If these revenue gains can be achieved through less costly measures, which is to say measures that create less in the way of deadweight loss, I’d much prefer that we pursue those measures. And I’d also argue that the public sector is sufficiently inefficient that I’d much rather see structural reform designed to increase public sector productivity than a bias towards raising revenue to pay for the existing public sector. There are many assumptions embedded here, and I can’t imagine everyone will embrace them.
One important thing to keep in mind is that tax-sensitivity varies across individuals — it varies by age, gender, ethnicity, geography, and much else. This is one reason I favor age-dependent taxes. Young workers relatively free of obligations seem to have a more elastic labor supply than prime-age workers with lots of obligations. Secondary earners seem to have a more elastic labor supply than primary earners, etc. In an economy like ours, we’re always trying to solve coordination problems across individuals with different preferences, e.g., some people prefer leisure to disposable income.