Saturday, January 15, 2011

Tax expenditures conference in Loyola (LA)

Yesterday’s conference was reasonably lively. Four panels over 7-1/2 hours went pretty fast. There was widespread consensus among the attendees concerning:

(1) The basic validity or at least value of the tax expenditure concept, so long as one is not overly reductive and broad-brush about it. Ed Kleinbard noted that this is a significant change from where the debate stood 10 or 15 years ago.

(2) The point that, for many though not all of the items commonly classified as tax expenditures, the term “spending through the tax code” is a reasonably descriptive shorthand. I would quibble a bit, saying that the better distinction to draw is between (a) allocative policy aimed at resource allocation and (b) distributional policy aimed who has what. But “spending” is an acceptable lay shorthand for “allocative policy” (such as favoring the coal industry) – although this helps make the point that it is silly to apply the “spending” label to, say, child tax credits, which is a distributional adjustment for family size or the earned income tax credit.

(3) The need to use the budgetary gains from reducing or eliminating bad tax expenditures in the process of addressing the long-term U.S. fiscal gap and avoiding a catastrophic actual or implicit default or collapse in U.S. credit-worthiness.

On (2), a paper-to-be (for now, just PPT slides) by Len Burman and Marvin Phaup proposed that, as a matter of budgetary accounting and for purposes of various budget rules of the present or future, tax expenditures be treated as a tax payment plus an outlay.

To illustrate, consider what has long been my favorite canonical example, David Bradford’s fictional “weapons supplier tax credit” or WSTC. In my version of the example, the government decides to reduce both taxes and spending by $10 billion by (a) eliminating a $50 billion weapons program even though we still want the weapons, and (b) offering the supplier, Acme Industries, $10 billion in WSTCs if it continues furnishing the weapons. Acme uses these to eliminate the taxes it would otherwise have owed. The WSTC is tradable, so it can sell any that it exceed its tax liability to someone else who gets to claim them instead. But for simplicity, let’s assume that Acme would otherwise have owed $11 billion, so it can use the WSTCs itself and lower its federal income tax liability to $1 billion.

When the dust has settled, absolutely nothing has changed. Everyone has the same amount of money as before, and the government has the same weapons as before. But official accounting measures treat both spending and taxes as $10 billion lower than previously.

Under the Burman-Phaup proposal, accounting wouldn’t change either. Acme is treated as having paid $11 billion of tax and received a $10 billion payment from the federal government, even though the WSTC permitted it to net these two transactions without any need for the matching tax flows.

Same proposed treatment for, say, the income exclusion for employer-provided health insurance. Suppose my marginal rate is 35 percent, and that I get to exclude a $10,000 employer-paid health insurance premium from my taxable income. When the federal government computes for official accounting purposes (and any applicable budget rules in the Congress) the amount of overall taxes and spending, my little transaction figures in as a $3,500 tax collection plus a $3,500 federal outlay (i.e., a payment to me offsetting the tax I notionally paid).

Same accounting treatment, presumably, for my excluding the value of the premium for payroll tax purposes, though for the Social Security tax the exclusion won’t affect my liability if I am over the annual ceiling.

The idea is to prevent the use of tax expenditures instead of direct spending from causing the budget picture (or rules) to apply differently when they only differ in form (if the frame viewing this as disguised spending fully applies).

I think this is likely to improve both the informational content of budgetary reporting and the capacity of budget rules to constrain legislation as intended. An obvious application is to the House of Representatives’ new “CUTGO” rule, under which the differential treatment of “tax cuts” on the one hand and “spending” on the other invites legislators simply to repackage the types of “spending” proposals that the Republicans supposedly want to discourage, without changing their substance whatsoever, and thereby evade the rule. Indeed, to make the point clear, the hated “earmarks” that we often hear so much about should presumably get the same opprobrium even if they are cleverly restructured to operate via the tax system with no change in substance.

But now let’s think a bit more broadly about how the proposal invites us to restructure our thinking about tax reform proposals. Suppose we think of the 1986 Act as a budget-neutral, “revenue-neutral,” and distribution-neutral trade of tax expenditure repeal for lower rates. Using the improved frame has no effect on whether the 1986 Act was budget-neutral or distribution-neutral. But it tells us that viewing it as “revenue-neutral” was an illusion based on mistaking form for substance with respect to the repealed tax expenditures that one has agreed were “disguised spending through the tax code” (as the Bowles-Simpson Fiscal Commission Report tells us). Rather, the 1986 Act was a budget-neutral and distribution-neutral cut in both taxes and spending, as reformulated to take into account the point that tax expenditure analysis makes.

Same point for the Fiscal Commission’s proposal to repeal a lot of tax expenditures, but to give back almost all of the budgetary improvement achieved thereby by cutting tax rates (with the top individual rate perhaps declining to as little as 23%). In the cause of narrowing the fiscal gap, and in terms of their own description of tax expenditures, they are greatly reducing taxes – even if not “tax revenues” – with no budgetary motivation whatsoever. The only rationale for cutting the tax rates in this context is that (a) they would otherwise have overshot the target of fiscal balance, and (b) as a policy matter, they happen to like using some of the budgetary surplus thereby created in this particular way. No word on why this is better than some other way of using this surplus, other than that they happen to like lower taxes (a course that has clear efficiency advantages, but is merely one of many possible policy choices).

Based on this line of reasoning, the slides for my commentary at the conference included the following relatively strongly-worded statement:

“For Bowles-Simpson proponents to ponder budget-neutral tax reform is merely insane; for them to ponder revenue-neutral repeal of TEs is also incoherent.”

Insane because one should deploy the budgetary gain from tax expenditure repeal to reduce the fiscal gap, not to fund rate cuts in the face of a huge gap. Incoherent because they are forgetting their own point about tax expenditures if they focus on revenue-neutrality as formally defined.

And I argued that the same point applies to their idea of capping “revenue” as a percentage (such as 21%) of GDP. That feature of their plan also attracted well-deserved flak from other speakers at the conference, who viewed it as overly constraining when we need to be able to respond flexibly to our growing budgetary problems and the general political difficulties in addressing them.

During the question period after my panel, an offsite on-line participant submitted a question to me, based on noting that the above-quoted statement about Bowles-Simpson sounded a bit harsh. Was I saying that adopting it would be worse than where we stand otherwise?

In answering, I felt the questioner had a point, so far as the tone of the bullet point was concerned, because I don’t think Bowles-Simpson as a whole was as bad as I feel they were at that dimension. Yes, for all its limitations (such as underspecified ways of cutting spending growth) it is among the approaches one could reasonably consider deploying in any attempted march away from the cliff of catastrophic fiscal collapse. As it happens, among such plans it’s surprisingly to the right of center for a plan that comes out of a Democratic Administration. If the most liberal plan that could be within plausible responses stands all the way to the left at point 0, and the most conservative such plan stands all the way to the right at point 10, I would say Bowles-Simpson is at about point 8.

Fine, I’d prefer a lower number myself, but having plans at different points along the spectrum is generally constructive, albeit that one might have expected a Democratic Administration (especially one routinely attacked as left-wing) to generate something no higher than point 5. But that doesn’t make the Bowles-Simpson plan insane or incoherent, so from that perspective my comment (if generalized to an assessment of the plan as a whole) was too harsh.

But the error in the Commission’s thinking about and presentation of tax policy issues needs to be made forcefully and clearly if it is to be noticed – and all the more so because this wasn’t just the Commission’s distinctive error, which will cease to matter as such if (as seems likely) its report is swiftly forgotten. The idea that tax reform means a 1986-style trade of base-broadening for lower rates, even in the face of an approaching fiscal crisis that makes budget neutrality entirely the wrong approach, needs to be discredited before the next budget commission (and there will no doubt be many) gets to work.

13 comments:

Marc said...

Dr Shaviro,

Are you suggesting Bowles-Simpson is budget-neutral tax reform? It is not, not even close. In 2020, the base broadening is meant to raise $180 billion more than the rate-cutting loses. And that $180 billion is money on top of revenue from letting the upper-income tax cuts expire. Moreover, it's on top of $40 billion in estimated revenue from moving to chained-CPI and raising the gas tax, and another $30 billion in Social Security revenue. No one I know is trying to describe Bowles-Simpson as either budget neutral OR revenue neutral. That's why the House Republicans voted against it!

Marc said...
This comment has been removed by the author.
Marc said...
This comment has been removed by the author.
Marc said...

I also have to take issue with you rating Bowles-Simpon as an 8 on the liberal-conservative plan. I assume you would put Paul Ryan's roadmap as a 10 (though I might argue it is a 9.5 -- since I think there are a few areas where one could get more conservative). Compared to that #10, Bowles-Simpson has:

1) SUBSTANTIALLY more revenue. As I just showed, Bowles-Simpson asks for over $250 billion in 2020 BEYOND what one would need if they let the top income tax cuts (above $250k) expire. Revenue in 2020 under Bowles-Simpson is 20.6% of GDP, compared to an 18.6% of GDP target under Ryan's Roadmap (and some outside estimates that his plan would raise substantially less than that).

Marc said...

2) A much more progressive tax code than what we currently have, particularly due to the taxation of capital gains as ordinary income. Given your preference for consumption taxes, the higher capital gains and dividends rates (likely 28%) may not appeal to you -- but they certainly appeal to the "left". And check out the distributional effect of the whole revenue plan (with the illustrative tax reform): http://www.taxpolicycenter.org/numbers/displayatab.cfm?DocID=2857&topic2ID=40&topic3ID=&DocTypeID=2. The bottom quintile gets a tiny tax cuts, the middle three quntiles see tax increases between 1% and 1.5%. And the top quintile faces a 4% tax increase, driven by an 8% increase for the top one percent and a 12% increase for the top 0.1%.

Marc said...

3) A very (distributionally) progressive Social Security plan. Yes, the plan is more benefit cuts than revenue increases, but actually just barely. On a present value basis, 46% of the gap is closed by new revenue. Other measures have that number somewhat lower, but still a pretty damn good ratio for the Democrats considering that the Republican starting point is to do it all on the benefits side and the Democratic starting point is maybe 60/40 (revenue/benefits). Also, check out the distributional tables: http://www.ssa.gov/oact/solvency/FiscalCommission_20101201.pdf. Those in the two lowest income categories actually see their benefits INCREASE. Whereas a scaled maximum earner receives only two thirds of scheduled benefits if he/she retires in 2050 and 60% in 2080.

Marc said...

4) Defense cuts are pretty damn deep. People have really overlooked this, but defense spending is hit just as hard as non-defense. And there are "firewalls" to ensure this occurs. Considering that many of these non-defense cuts are going to happen anyway (the House Republicans want even deeper than Bowles-Simpson, though they probably won't get all they want), the Democrats on the commission got these defense cuts at a pretty low cost. But remember, Ryan's roadmap doesn't really touch defense at all. Nor does the Republican pledge.

Marc said...

So yes, it's fair to criticize the revenue cap at 21% of GDP. But I'm not sure why you are suggesting that this is revenue-neutral 86-style reform. And considering the AMOUNT of revenue, the DISTRIBUTIONAL EFFECT of the tax plan, the tax/spending mix and distributional effect of the Social Security plan, and the size of the defense cuts, I don't know how you describe this as an "8" on the liberal-conservative scale. Dick Durbin would have never signed up for an 8, and I doubt Kent Conrad would have either.

Daniel Shaviro said...

Marc, you make some good points. I was going on the 21% revenue cap and possible 23% top rate in my left-to-right ranking, but you make good arguments for moving it back towards the center.

Agreed about the $180B, I am more careful in making this point clearly in the article draft I'm working on. I think of Bowles-Simpson as having a revenue-neutral frame and then saying OK, we'll actually dedicate $180B in 2020 to deficit reduction despite this. (Repealing all the TEs raises way more than this, after all, even if not really a trillion.)

On the Social Security aspect, there's a weird dynamic that makes it hard to score in left-right terms. Their making Social Security more progressive is ironically something the right likes more than the left (a consistent paradox in entitlements for at least 40+ years). I actually like it and regard it as in principle on the more left side, but in fact most of the self-identified left hates it (because they think it will cause the programs to become less popular) while the right tends to like it, on the same view (or less cynically because they see these programs as rightly limited to serving special distributional needs).

Marc said...

Dr. Shaviro,

Thank you for your much-appreciated response. But I want to push back a bit more on all three points.

1) Yes, the fact the plan has a revenue cap is a point for the conservatives. But we should at least remember that 21% of GDP is higher than revenue has EVER been in the history of the country. And while in theory, the plan's instructions to Finance and Ways & Means allows for a top rate as low as 23%, we are more likely talking about 28% or so -- the top rate in the illustrative plan.

Marc said...

2) I disagree with you that something can be described as revenue neutral or budget neutral (and certainly disagree with you that the commission describes it that way) when it raises $180 billion. And again, remember this is on top of the $150 billion or so from letting the upper-income tax cuts expire and the $70 billion or so from gas tax, chained CPI, and Social Security revenue.

Yes, the plan uses much of the money from base broadening to lower the rates (though it also uses some of it to get rid of the AMT, PEP, and Pease -- which are in some ways tax expenditure claw-back policies). But a substantial amount of revenue -- not just an incidental amount -- goes to deficit reduction.

You could argue, I suppose, that there are two things going on (though I would never describe it this way). First, the commission came to an agreement about how much revenue they wanted and then broadened the base sufficiently. Then, the commission decided the rest of the tax code would do more for equity and efficiency (and growth) by having lower rates and a broad base than by having higher rates and a narrower base. Once the amount of revenue was decided (an amount equal to about 2% of GDP relative to current policy), would you disagree with this conclusion?

Marc said...

3) Yes, you are right that the left does get nervous about making Social Security more progressive -- both because of the belief that "programs for the poor make poor programs" and because the 'true believers' of social insurance are mostly on the left.

But I'd assert to you that the social insurance folks play a much weaker role on the left than they did ever a few years ago, and that the commission solves for the "programs for the poor" problem.

In terms of the social insurance folks, if you look back to the 90s many of them supported raising the payroll tax RATE, and doing across-the-board benefit cuts (some of the more conservative ones liked retirement age, even Ball supported chained CPI). Now, the left's preferred position is to raise the payroll tax cap significantly -- maybe even with a "donut hole", and often combined with a reduction in the top bendpoint or the addition of a new bendpoint above the current cap which is significantly lower than 15%.

In other words, both liberals and conservatives want to improve the system's finances by making the program more progressive in a way that reduces "equity" in order to preserve adequacy. It's just that Republicans prefer to do it on the benefit side and Democrats on the tax side. The Bowles-Simpson plan does both.

As for the "programs for the poor" argument, the Bowles-Simpson solves for that a bit by setting fixed PIA factors of 90-30-10-5 rather than doing "progressive price indexing". Progressive indexing eventually pushes the top bendpoint down to zero, which was one of the major concerns from the left. Bowles-Simpson avoids that.

(And as an aside, liberals are now very much supportive of a minimum benefit; and the Bowles-Simpson one is quite generous since it is wage-indexed).

Daniel Shaviro said...

Marc, thanks for the follow-up. Agreed, I've been unfair to the Bowles-Simpson plan. I meant to take aim at the assumption I viewed as strongly influencing it, that tax reform should be 86-style and revenue neutral (albeit that the plan went $180B net positive in 2020), because I view that as an ongoing meme that will influence future proposals and needs to be addressed. But while I want to make the broader point, I should have been more careful re. what I regarded as merely an illustration of the broader point (but which is more at the center of current debate, as Bowles-Simpson is still current news).