What Is Wrong With Incomplete Tax Policy Analysis?

-- Posted by Neil H. Buchanan

My new Verdict column discusses some recent positive changes in the public debate about taxes. I have always been struck by the weakness of the empirical evidence regarding the supposed harms and dangers of taxes, yet even the more liberal big-name economists have always been rather passive on the issue. It is not surprising, I suppose, when Paul Krugman gets on the case; but I was pleased to see that Christina Romer and Uwe Reinhardt are also now beginning to push back against some of the conventional wisdom. They are both reliably center-liberal, of course, but at least we are finally seeing some fight from the mainstream non-conservative economists on taxes.

In part, my column is a continuation of my Verdict column last month, in which I discussed how unsurprising it is that economists are largely incapable of contributing usefully to the public debate. In today's column, I recount an incident from a few years ago, in which a highly prominent tax economist told me bluntly that he simply refused to believe the overwhelming weight of the evidence about the estate tax -- evidence that generally shows no effect of estate taxes on behavior. The economist's response -- essentially, "It does not fit my theory, so the evidence cannot be believed" -- captures the acquired mindset of far too many mainstream economists.

Although it is good to see more voices -- highly respected voices -- coming out of the shadows to talk about the weak empirical case against taxes, it is the "taxes are presumptively bad" mindset that always drives the narrative. As I discuss in the column, that mindset fails to confront the most basic question that should be front and center in all policy debates: Compared to what? In light of the "baseline problem" that I think trumps everything else in this debate (see, most recently, here), it is simply amazing how many economists are willing to limit themselves to simply trying to determine "how bad" an individual tax might be, without comparing those effects to anything other than an imaginary no-tax baseline.

For an academic field that spends so much time studying "general equilibrium effects" -- that is, the full interactions of markets, as price changes in one market (wheat) affect buyers and sellers in other markets (corn) -- the failure to compare a given tax policy to its alternatives is a major failure. That is not to say that it is impossible to find economists offering comparisons to alternatives, but only that those instances are shockingly rare. The narrow, one-tax-only analysis should be the exception (appropriate in limited circumstances), not the overwhelming rule.

A good, compared-to-the-alternatives tax analysis does occasionally show up in public debate among economists, generally in two contexts. First, one will sometimes see a statement along the lines of this: "If we do not collect $x from this tax, we will have to collect it from some other, even worse tax." This is most commonly offered as a defense of the estate tax, because the empirical evidence shows (as noted above) that the estate tax leaves behavior essentially unchanged. If we reduce or eliminate the estate tax, then we might find ourselves raising money from other, more "distortionary" taxes. (I use scare quotes to highlight how the usual language among tax economists is so rhetorically loaded.)

The second context, related to the first, relates to budget deficits. If we do not collect $x of revenue from Tax Type I, and we refuse to collect $x of revenue from Tax Types II, III, and so on, then we should try to determine the consequences of increasing the deficit. The CBO often provides an analysis along these lines. As I point out at the end of today's column, however, there is yet another alternative: not to spend $x at all, which means that the analysis is incomplete unless it includes a comparison between raising $x from Tax Type I and not spending $x on Spending Type I.

One can reasonably object that this is analytically difficult. And it is. It is especially difficult because it requires making a reasonable guess about how Congress will respond to the loss of revenue from reducing (or failing to increase) a given type of tax. That difficulty, however, is no excuse for acting as if one-sided analyses are meaningful. "We can't raise Tax Type I, because it is inefficient," is the most common general form taken by tax policy arguments from economists and like-minded analysts. It is good that Romer, et al. are now at least saying, "Yes, but the tax is not all that inefficient," but that still cedes far too much ground.

This is especially unfortunate, because so much of the policy world takes economists to be all-knowing. Among tax law scholars, economics (especially efficiency analyses, in the very traditional mode of neoclassical economics) is often the trump card that shuts down all debate. This is true across the ideological spectrum, although tax scholars in the center and on the left are very good about emphasizing equity concerns, even as they generally concede the efficiency argument. (It is also true among legal scholars outside of tax law, where economists are far too often treated as if they have access to unique truths that the mathematically unsophisticated cannot understand.)

Among tax scholars who are not economists, evidence-free presumptions permeate policy discussions, based on standard neoclassical economic reasoning that too often goes unchallenged. One will hear or read statements like, "We all know that taxes are inefficient," and the souped-up version, "It is established by economists that the inefficiency of a tax goes up by the square of the tax rate." This is all based on an acceptance of theory, not an assessment of evidence. (The latter statement, indeed, is supposed to substitute for evidence, because it purports to show that any level of badness just gets worse and worse as tax rates rise.) And it replicates the error of not offering a comparison to any meaningful alternative.

My current crusade about the failures of economists, therefore, is based on the impact that they have on public policy, not only directly (through the interventions of economists themselves), but indirectly through the agenda-setting and creation of shared presumptions for which economists are uniquely responsible. Once those presumptions become widely accepted, policy becomes skewed in predictable ways. For some economists, this is very much on purpose. For others, however, I suspect that it is simply a matter of doing things as they have always been done.

If economists cannot change these basic shortcomings in their approach to analyzing policy (tax and otherwise), other analysts should at least be willing to stop taking standard-issue economic analyses so seriously.