Friday, April 01, 2011

March 31 NYU Tax Policy Colloquium (with Len Burman), part 1: What is "Spending?"

Yesterday Len Burman presented his paper draft (co-authored with Marvin Phaup), Tax Expenditures, the Size and Efficiency of Government, and Implications for Budget Reform. The central is that mislabeling tax expenditures (TEs as tax cuts rather than disguised spending both leads to bad program choices and is a major contributor to the long-term U.S. fiscal gap. In the later part of the paper that is particularly still in progress, it suggests using revised accounting for TEs both to report government operations differently and in designing budget rules that might in principle have some bite.

Obviously I tend to work similar territory part of the time (e.g., when not doing international tax or some such thing), though with less of a focus on budget rules, which I have discussed in a couple of my books (such as this one and that one) but tend to regard both somewhat skeptically and as outside my core expertise.

I guess I need to start by repeating what has become my Standard Tax Expenditures Gripe. (Indeed, I should probably adopt a shorthand and call it, from now on, the STEG.) My personal STEG starts from the following observation: Tax expenditure analysis typically purports to distinguish between the "normal income tax structure" and "spending through the tax code." The idea is fine - leaving aside more fundamental improvements to our fiscal language - but the expression is utterly incoherent. Decades of TE literature have focused on the difficulty of both defining, and motivating reliance on, the concept of the "normal income tax structure." But the other part, as almost no commentators seem to appreciate (David Bradford was of course an early exception) is at least equally problematic - as in, lacking any fundamental meaning.

The STEG consists of my having pointed this out about 8 years ago and proposed a substitute that is conceptually vastly superior. Various individuals have told me they agree, but I am pretty sure that I have never once in print seen anyone shift to using it. Instead (while a cite to my article usually appears, around footnote 83 or so), they keep on writing articles that trot out the old, incoherent structure or else what I consider confusing and jerry-built modifications to it that don't quite do the job either.

So I feel as if I'd invented the washing machine and people kept right on washing dirty underwear by hand in the sink. OK, that's a bit grandiose and overstated. But if my assessment is at all correct, you can appreciate that it would be personally frustrating.

This is obviously turning into a long post, so I will break it into two parts. I will close this one with some content from a little handout that I prepared for the colloquium session, addressing the question of whether "spending" is a meaningful term. In particular, it addresses the muddy line between ostensibly "normal" income tax features and identical features that could be placed outside the income tax system. It shows, no less than David Bradford's famous "weapons supplier tax credit" hypothetical, that we cannot accept form in making the distinction that commonly is called taxes versus spending, unless we are willing to turn the entire exercise into one of pure labeling that lacks broader meaning.


Example 1

Suppose that marginal tax rates are as follows:
0%, 0 - $20,000
20%, $20,000 - $100,000
35%, > $100,000

Astolphe earns $100,000 and pays $16,000 of tax. If not for the two lower rate brackets, however, he would have paid $35,000. Does this mean he got $19,000 worth of "tax expenditures," in the form of having 0% and 20% brackets? No, say TE proponents, because non-linear marginal rates are a "normal" feature of the income tax. I accept the conclusion but not the analysis. A point to note in passing: "normal" as used here has nothing to do with what the U.S. income tax code has ever looked like in its history. For example, a homeowner's imputed rent has never been taxed, yet there is no good argument of income measurement for regarding home mortgage interest deductions as tax expenditures OTHER than as indirect partial denial of the imputed rent exclusion. But there's worse to come regarding reliance on the "normal income tax structure" idea to explain why lower rate brackets aren't tax expenditures.

Example 2

Suppose we repeal the zero bracket. Marginal tax rates are therefore as follows:
20%, 0 - $100,000
35%, > $100,000

In addition to repealing the zero bracket, however, we made a strikingly related change outside the income tax system. Specifically, we have enacted an “earned and investment income transfer charge” (EIITC), equal to 20% of one’s first $20,000 of income, and administered separately from the income tax. Or alternatively, we can administer it through the income tax but call it a "tax expenditure" (like the actual earned income tax credit, or EITC).

The combined repeal of the zero bracket and enactment of the EIITC leaves everyone in the society in exactly the same position as before. For example, Astolphe, instead of paying $16,000 of tax, now pays $20,000 of "tax" and benefits from $4,000 of "spending."

So in substance Example 2 is identical to Example 1. Yet now, proponents of TE analysis as commonly formulated would be forced to say (and indeed might vociferously insist) that both taxes and spending have gone up by $4,000. You're not going to fool them by putting the EIITC into the income tax instead of leaving it to one side! So matters of pure form determine whether they regard lower rate brackets in substance as part of the "normal income tax system" after all.

Better still, we could enact a flat rate 35% tax and expand the EIITC to apply at a 15% rate to income between $20,000 and $100,000. Now Astolphe pays $35,000 tax and gets $19,000 of spending, all within the “normal” structure.

Example 3

Same as Example 1, but now we add a welfare benefit equaling 75% of the amount (if any) by which one’s income falls short of $20,000. Thus, for example, if you earn zero you get $15,000; if you earn $10,000 you get $7,500; if you earn $20,000 or more you get zero.

As all standard TE proponents think they know, this is a "spending" program from which Astolphe, with his $100,000 income, gets zero. Beulah, who earns $12,000, gets $6,000 in government spending.

Example 4

Same as Example 1, except now, in lieu of what we did in Example 3, we add a universal $15,000 demogrant and raise the tax rate on one’s first $20,000 of income from 0 to 75%.

All individuals end up in exactly the same positions as in Example 3, but we have increased both taxes and spending. For example, Astolphe now gets a $15,000 check and pays tax of $31,000, while Beulah gets a $15,000 check and pays tax of $9,000.

Now, you can't fool standard TE proponents. They know their "normal income tax structure," and they know "spending" when they see it. So, even though Examples 3 and 4 are identical, they are forced to agree - and they may indeed vociferously assert - that (considering only Astolphe and Beulah) in Example 3 taxes were $16,000 and spending was $6,000, whereas the completely non-substantive reformulation in Example 4 has raised taxes to $40,000 and spending to $30,000.

What are all these people getting wrong? (Even though, ahem, I thought I had explained all this. albeit evidently not well enough.) Let's leave that for the next post.

No comments: