Tax policy colloquium, week 9: Alan Viard's "Progressive Consumption Taxation: The Choice of Tax Design"

Yesterday at the colloquium, we discussed Alan Viard's above-titled draft paper, which is a follow-up to his excellent book (co-authored with Robert Carroll), Progressive Consumption Taxation: The X-Tax Revisited.  But whereas the book focused in depth on how one should design an X-tax, if it were used to supplant the existintg income tax, the article is a comparative assessment of the X-tax versus the other main instrument for progressive consumption taxation, a personal expenditure tax or PET.

The X-tax would replace the individual and corporate income taxes (and potentially other taxes as well) with what is basically a VAT collected from all businesses, modified to include firm-level wage deductions and matching individual tax returns on which the same wages were included.  The idea is to add progressivity, which is missing from a VAT, by applying progressive rates (rising in stages from zero to the full business rate) via the wages.  So it's equivalent to a VAT plus a net wage subsidy that, in effect, partly and progressively rebates the VAT for workers, to the extent that they have income in lower brackets.

The PET would replace the same set of taxes with a levy on individuals that would look a bit like a conventional income tax, only with unlimited IRA-type deductions.  More specifically, all saving and investment would be deductible (and expensed) - most likely, as an administrative matter, through the use of "qualified accounts" that would help keep track of the dollars involved.  All withdrawals from the accounts would be added to income, as would loan proceeds (though these would be re-deducted if used to fund saving and investment).  To some eyes, taxing loan proceeds may look a bit odd, but of course a retail sales tax, just like the PET, reaches loan proceeds that one spent on consumption rather than adding to one's gross savings.

This is a topic that I know fairly well, and it has a pleasant personal association for me (wholly without regard to any policy merits or demerits) given the association of both with the great David Bradford, who was my colloquium co-convener for almost 10 years.  David (with the U.S. Treasury Tax Policy straff) spearheaded a classic 1980s study of a PET prototype called the Blueprints cash-flow tax, and he also developed the X-tax (which was based on the Hall-Rabushka flat tax but added greater progressivity).

I believe that Bradford's main rationale for shifting from the PET to the X-tax was optical: he felt that having part of the tax collected at the business level was politically necessary.  But there may also be an offsetting optical advantage to the PET, which is that the likes of Mitt Romney and Warren Buffett are visibly taxpayers under it even if they don't have current year wages.  (This is just an optical concern if one believes they are bearing, as consumers, the business-level tax on their purchases at the maximum rate, and that those paying the individual-level wage tax are actually getting a net subsidy on their wages.)  More on the optical issues at the end of this post.

I've never come to closure on which I would prefer as between the PET and the X-tax.  Making this less urgent is the fact that I don't expect either to happen.  Viard's paper at the colloquium yesterday usefully illuminated the tradeoffs that one should think about.  Pertinent aspects include the following:

--Both systems, like the current income tax, have an "averaging" problem.  That is, under consumption tax (lifetime-based) reasoning in particular, it may be undesirable to have the graduated marginal rates, applied on an annual basis, impose higher taxes over time on people who have up-and-down rather than smooth annual taxable amounts.  It's possible that the problem under the X-tax is worse in the absence of an averaging mechanism (if wages vary more on an annual basis than does consumption), yet it also might in practice be easier to solve.  David Bradford's Blueprints cash flow tax had a very clever mechanism to let people self-average under the PET, but it may have required greater awareness of the needed manipulations than one could realistically expect from most people with busy lives.  But I wonder if computers could solve this at some point (in effect, if everyone used a Turbo Tax-type program with full but secure on-line access to one's own prior years' return info).

--I would argue that the PET is more compatible with very steep high-end tax rates, if one favors them to address rising wealth inequality at the top.  Under the PET, you limit those rates to the very high end.  Under the X-tax, you more or less have to make the business tax rate equal the top individual rate, so businesses below the plutocratic level (on an individual owner basis) would need to go through the two-step of nominally paying out all the net cash flow as salary in order to get it into the lower rate brackets.  This might be a nuisance and be under-utilized by people who didn't understand how it worked, even though it is just a paper transaction (you can "pay" yourself a high salary for tax purposes and then reinvest the funds tax-free).

--Income and/or assets would still be used, under either proposal, to determine what needs-based benefits people got (TANF, Food Stamps, Medicaid, etc.).  This is interestingly in possible tension with the usual consumption tax approach that current assets, if not spent on consumption are irrelevant (e.g., because we will tax their use in consumption in future years).  One could view this reliance on assets or income either as contradicting the basic consumption tax philosophy or merely as responding to the adjustability of current year consumption or wages if, in effect, low wage/consumption levels would otherwise have a high implicit tax rate via the phaseout of transfers.

--Otherwise, the main virtues and vices of the PET, relative to the X-tax, come from its making greater use of the individual-level return.  Whether for good or ill, tax expenditures may be easier to provide (including to non-wage-earning rich people such as Romney and Buffett) in the PET than in the X-tax, at least if refundable credits are politically costly since they look more like "spending."  This point might alter the current partisan politics of refundable credits.  Household-based adjustments also are easier to make universally via the PET, if costly to do otherwise.  (By this I refer to the treatment of marital or couple status and of children.)

--The main virtues and vices of the X-tax, relative to the PET, come from its using the business level.  Optics aside, this is often a convenient place to collect tax revenues.  The Viard-Carroll book suggests that one defect is the need to police over-payment of wages to owner-employees or others.  E.g., a partnership pays its partners more than the value of their services, so as to max out on the lower rate brackets.  But I believe this probably isn't a problem outside of the household setting (e.g., I have my firm pay my spouse and kids big salaries, even though they are not doing significant work so we can multiply our use of the lower brackets).  Why not give true owners the full benefit of the lower rates, and presumably they won't want to overpay true third parties (leaving aside the personal vs. business problem, e.g., I pay for a meal and pretend it's to an employee for business services, that we already face under the existing income tax).

--A key dilemma under the X-tax is whether the business-level tax should be origin-basis or destination-basis.  All existing income taxes are origin-basis, which means that you include amounts received from foreigners for exports, and can deduct or capitalize amounts paid to foreigners for imports.  All existing VATs are destination-basis, aka border-adjusted, which means you wholly ignore cash flows to or from foreigners.  The horrible thing about origin basis is that it requires transfer pricing for the operations of multinational firms.  Hence, most experts prefer destination basis for the X-tax.  However, Carroll and Viard (along with David Bradford) prefer the origin basis, despite the transfer pricing problem, because of the transition wealth effect on Americans vs. foreigners of switching to a destination basis tax.  In their book, for reasons that are logically compelling but too complicated to explain here, they estimate that straight-out replacement of the existing income tax with an X-tax that used the destination basis might transfer almost $9 trillion from Americans to foreigners in the transition, although one could try to mitigate this by applying transition rules.  However one comes out, all can agree that it is going to be suboptimal either way.

The PET would also be effectively destination-basis, as it taxes national consumption rather than production, even though it doesn't have special rules for cash transactions with foreigners.  So that is a bad thing if you don't like the one-time wealth transfer, but at least it is a good thing if you hate transfer pricing.

--A final issue raised by Viard's article is: What about a hybrid?  That is, instead of having just an X-tax or PET, why not have both - say, at half the tax rates for each that you would have chosen in the standalone setting.  This may initially seem to be Herman Cain-style silliness, a la thinking that 9-9-9 is better than 27 if the taxes are actually in some general sense all the same.  But insofar as implementation problems with the X-tax and the PET cause them to create distinctive distortions (or fraud / tax planning / under-collection risks), using some of each not only diversifies the collection risks that the federal government faces, but might also reduce those distortions (which would be expected to rise faster than ratably with the tax rate).  Plus, you get to address both optical problems - there's a business level tax, and Romney / Buffett are taxpayers.