How To Tax Wealth Constitutionally
An annual wealth tax quite likely would have to be apportioned among the states — at least in part. But that’s not necessarily a fatal flaw. And anyway, we don’t need an annual wealth tax in order to tax wealth
Elizabeth Warren, the senior Democratic senator from Massachusetts and now a candidate for her party’s 2020 presidential nomination, has proposed an annual wealth tax of 2% on households with a net worth of at least $50 million, rising to 3% for households with a net worth of at least $1 billion. Quite a few commentators have questioned whether Warren’s proposed tax would run afoul of the constitutional requirement that any “direct tax” be apportioned among the states on the basis of population. Several leading scholars of constitutional law have signed letters saying that Warren’s wealth tax would not be subject to the apportionment requirement.
I have enormous respect for the scholars who have endorsed Warren’s plan — and for Warren herself, who has helped to spark a serious national conversation about wealth inequality and what to do about it. I also share Warren’s normative prior that the federal government should do much more to ensure a more equitable distribution of wealth. I am less convinced that an annual wealth tax is the way to do it. After a number of exchanges (on and off Twitter) over the last few days, I’ve come to a few (tentative) conclusions that seem worth writing up:
— Warren’s wealth tax, at least insofar as it applies to real property wealth (land and structures), probably would have to be apportioned among the states;
— Contrary to conventional wisdom, apportionment of the real property share of a wealth tax is probably not a deal-breaker;
— As economists have known for some time, we can (nearly) recreate the economic effects of a wealth tax through taxes on asset sales and estates, which would not have to be apportioned among the states;
— One caveat to the previous conclusion is that a wealth tax and a tax on asset sales and estates cease to be near-equivalents if the latter tax is temporary. Unless there is a durable progressive majority in Congress, a tax on asset sales and estates would quite likely be temporary; and
— The potential instability of a tax on asset sales and estates gives rise to a relatively weak argument for an annual wealth tax and a much stronger argument for a one-time wealth tax similar to that proposed by Donald Trump in 1999.
I should make clear that I am not calling for a one-time wealth tax along the lines of Trump’s 1999 proposal. One-time taxes have problems of their own, which I’ll briefly discuss at the end of this post. But a one-time wealth tax a la Trump circa 1999 has considerable advantages over Warren’s plan.
The Constitutional Objection to Warren’s Wealth Tax
Two clauses in the Constitution — Article I, section 2, clause 3, and Article I, section 9, clause 4 — require that “direct” taxes be apportioned among the states according to population. That means that if California amounts to 12% of the U.S. population, Californians must pay 12% of any direct tax — no more, no less. If the Warren wealth tax is a direct tax subject to the apportionment requirement, then it flunks that requirement because it would raise more revenue from wealthier states than poorer states.
The Constitution does not explain what a “direct” tax is, but it’s long been assumed that a tax on the value of land is a direct tax. This is the view adopted by Alexander Hamilton in Federalist №36, where he writes — with respect to taxes on “real property” or “houses and land” — that “the proportion of these taxes is not to be left to the discretion of the national legislature, but is to be determined by the numbers of each State.” The advantage of apportioning land taxes, according to Hamilton, is that assessing the value of real property “requires the knowledge of local details” and thus “must be devolved” to local elected or appointed officials. Without apportionment of a national land tax, local officials might have an incentive to undervalue land in their jurisdictions so as to reduce their residents’ tax liability. With apportionment, by contrast, residents of a particular state will pay the same amount in toto regardless of their individual assessments, and officials of that state will no longer have a perverse incentive to systematically undervalue their residents’ real property.
In the 1796 case Hylton v. United States, the Supreme Court ruled that a federal tax on passenger carriages was not a direct tax subject to apportionment. Along the way, Justices Samuel Chase, William Paterson, and James Iredell all indicated that they understood land taxes to be direct taxes. In the 1869 case Veazie Bank v. Fenno, Chief Justice Salmon Chase — writing for the majority — said that “taxes on land and appurtenances” are direct taxes. A unanimous Court said the same thing five years later in Scholey v. Rew: “Taxes on lands, houses, and other permanent real estate have always been deemed to be direct taxes.” Again in 1881, a unanimous Court said that “taxes on real estate” are “direct taxes.” And at least a half-dozen times in the first century of the Republic, Congress imposed taxes on land and required them to be apportioned among the states on the basis of population.
The Court adopted a much broader definition of direct taxes in the 1895 case Pollock v. Farmers’ Loan & Trust Co., when it said that taxes not only on land but also on personal property (including securities) are direct taxes subject to the apportionment requirement. The Pollock Court went even further when it held that a tax on income or rents from real or personal property is a tax on the property itself and so subject to apportionment. The Sixteenth Amendment, ratified in 1913, established that Congress has the power to tax “incomes, from whatever source derived,” without apportionment — effectively abrogating the portion of Pollock that held that taxes on income from real and personal property must be apportioned. The Sixteenth Amendment does not, however, directly address the holding in Pollock that taxes on the value of personal property are subject to apportionment, nor does it address the pre-Pollock consensus that land taxes are subject to apportionment.
It’s possible — though far from certain — that the Supreme Court would say today that Pollock was wrongly decided insofar as it held that taxes on personal property are direct taxes, thus reverting to the pre-Pollock view. It’s less likely that the current Court would allow an unapportioned tax on the value of land to stand. Pollock was an anomaly, but the century of pre-Pollock precedent and practice treating land taxes as direct taxes is much harder to explain away. Perhaps a highly motivated Court could do it, but it seems doubtful that the current conservative Court will go out of its way and dismantle centuries-old precedents in order to facilitate the enactment of a very progressive agenda.
Dawn Johnsen and Walter Dellinger, in a thought-provoking 2018 article, argue that a wealth tax that includes real property in the base does not implicate “the specific rationale that lay behind including real property in ‘direct’ taxes that must be apportioned: to protect the South from ‘slave taxes’ and per acre land taxes.” But that was not the rationale recited by Hamilton in Federalist No. 36 (which, like the other Federalist Papers, was addressed “[t]o the People of the State of New York,” though some of the Federalist Papers were read in southern states). Meanwhile, an esteemed group of constitutional law professors concludes in a letter to Senator Warren that a wealth tax is “plainly constitutional” under the 1900 precedent of Knowlton v. Moore. But Knowlton, which upheld the 1898 federal estate tax, also acknowledged that taxes on property are direct taxes subject to apportionment. The key intellectual move by the Knowlton Court was to say that the estate tax (or “death duty”) is “levied on the transmission or receipt of property” while a direct tax is “imposed upon property solely by reason of its ownership.” It is hard to see how Knowlton — which went to great lengths to distinguish death duties from taxes on the value of property — serves to validate an unapportioned tax on the value of property.
Apportionment of a Real Property Tax Is Not Necessarily a Deal-Breaker
Assuming that the Court reverts to the pre-Pollock understanding that taxes on real property (but not on other forms of wealth) are direct taxes, then apportionment among the states actually could be accomplished in a somewhat sensible manner. The portion of wealth tax revenue raised from real property holdings in each state could be rebated to the state government as a block grant, while the non-real property portion would be deposited in the U.S. Treasury. Net tax revenue from each state would thereby be precisely proportionate to population (it would be equal to population times zero). For households with a net worth of $50 million or more, personal residences and other real estate constitute only about 7% of assets, according to IRS data, so the share of all wealth tax revenues that would be subject to apportionment would be small. Households with a net worth of $50 million or more would not have a strong incentive to shift wealth to real property because they would pay the same 2% tax either way. Apportionment of the real property share would only affect where the revenues go. There might be a weak incentive to shift wealth to real property for high-net-worth households with lots of state pride (“Don’t Mess with Texas,” etc.) or who anticipate that some of their state’s block grant revenue would flow back to them. But this seems like a much smaller problem than exempting real property from a wealth tax entirely.
To get a quick sense of how significant the state-to-state variation in block-grant size might be, I quickly modeled a 2% wealth tax on households with a net worth of $5 million or more and imagined that revenues from the real property portion would be rebated to the government of each taxpayer’s state of residence (excluding the District of Columbia). I used this tax rather than the Warren wealth tax thresholds to match IRS wealth statistics by state of residence. Based on 2013 data, the largest per-capita rebate would be to Connecticut ($274 per person); the smallest would be to West Virginia ($2 per person). These state-to-state disparities could be offset, at least in part, by adjustments to other block grant formulas.
Four further notes:
— (1) The estimates in the previous paragraph give a sense of the order of magnitude of potential state-to-state disparities, but note the considerable differences between the tax I modeled and the Warren proposal. Hers sets in at a higher threshold, and the first $50 million would be exempt from taxation. All in all, these differences suggest that state-to-state variation would be considerably less if we applied apportionment to Warren’s proposal.
— (2) If apportionment applies to all wealth tax revenues and not just revenues from real property wealth, then the state-to-state disparities become less tolerable. Again imagining a 2% wealth tax on households with a net worth of $5 million or more, the difference in block grants would range from $71.80 per person for West Virginia to $1,151.29 per person for Connecticut.
— (3) The District of Columbia is potentially an outlier, as the percentage of high-net-worth households there is much larger than in any state. But since the District of Columbia is not a state, direct tax revenues from D.C. residents probably are not subject to the constitutional apportionment requirement.
— (4) For a different (and interesting!) apportionment proposal, see this 2014 article by John Plecnik.
We don’t need an annual wealth tax in order to tax wealth. Perhaps most straightforwardly, a consumption tax — like the value added tax (VAT) that more than 100 other countries already employ — imposes a tax on all existing wealth. A VAT is a tax on all future consumption of goods and services. Future consumption must be funded either out of future income or current wealth. By imposing a VAT while also cutting the tax on future income, we effectively tax only current wealth.
Let’s say I have $100 today and that Congress enacts a VAT tomorrow at a 20% (tax-inclusive) rate. Suddenly, my $100 today will finance only $80 of future consumption. Put differently: I can afford the same amount of future consumption as if Congress had taken 20% of my wealth today. Note that for constitutional purposes, a VAT would almost certainly be considered a “duty” or “excise” that is exempt from the apportionment requirement.
If Congress wants to tax only existing wealth and not future income, it could impose a 20% VAT in tandem with a 25% wage subsidy. If I earn $100 in wages next year, the federal government would kick in $25 more; and my $125 would finance $100 of future consumption (after accounting for the VAT of 20% x $125 = $25). The net burden of the VAT-plus-wage-subsidy would not fall on future wage-earners and would instead fall purely on existing wealth.
Another way to potentially achieve the economic effect of an annual wealth tax (and not just a tax on existing wealth) is to impose a retrospective wealth tax. The idea is as follows: Let’s say that the risk-free rate of return on investment is 10%. An asset worth $100 this year will be worth $110 next year and $121 the year after. With a 1% annual wealth tax, the taxpayer will owe $1 in year 1, $1.10 in year 2, and $1.21 in year 3.
Instead of imposing the wealth tax each year though, we can wait until the taxpayer sells her asset or dies. Let’s say it’s worth $121 then, and the taxpayer has held the asset for three years. We would then assume that the asset had grown in value at the risk-free rate, that the annual wealth tax had been imposed, and that — instead of paying the tax — the taxpayer had borrowed the amount of the tax from the federal government at the risk-free rate. We would require the taxpayer — at the time of sale — to pay back the $1 from year 1, the $1.10 from year 2, and now the $1.21 from year 3, plus interest on the amounts from previous years. The taxpayer ends up in the same economic position that she would be in if she had been paying the annual wealth tax all along.
Unlike an annual wealth tax, a retrospective wealth tax would almost certainly not need to be apportioned. The reason is that the retrospective wealth tax would be imposed only at the time of sale or death. Taxes on the transmission of property and death duties lie outside the definition of direct taxes in Knowlton. (For more on how a retrospective wealth tax might work, and why assuming growth at the risk-free rate is a sensible design decision, see this paper by Alan Auerbach and the late David Bradford or, for an explanation aimed at non-economists, this paper by James Kwak.)
A consumption tax or a retrospective wealth tax (or a combination of the two) could be made highly progressive by rebating payments to households below a certain income threshold. Unlike Warren’s wealth tax, though, a consumption tax or a retrospective wealth tax would not require tax authorities to assess the value of illiquid assets like artwork or real estate each year. In most cases, valuation would be relatively straightforward because there would be an arm’s-length sale to look to. If the retrospective wealth tax treats death as a realization event, then there still would be the need to assess the value of illiquid assets in a taxpayer’s estate, but doing that once is much easier than doing it year after year. (Stepped-up basis would be eliminated.)
The Problem of Political Instability
If Congress can accomplish the economic equivalent of a tax on existing wealth by imposing a consumption tax, and if Congress can accomplish the economic equivalent of a tax on all existing and future wealth by imposing a retrospective wealth tax, and if both of these methods are clearly constitutional and also easier to administer than an annual wealth tax, then why on earth would we want an annual wealth tax?
One answer is that a consumption tax only functions as a tax on all existing wealth — and a retrospective wealth tax only operates as the near-equivalent of an annual wealth tax — if these provisions remain in place. If a consumption tax such as a VAT is in place for five years and then repealed, holders of existing wealth can avoid the tax by waiting until year 6 to consume. Same for a retrospective wealth tax: a taxpayer can wait to sell her assets until the tax is repealed, provided that she does not die in the meantime.
This was essentially the problem that plagued the taxation of U.S.-based multinational corporations pre-2018. U.S.-based corporations had to pay corporate income tax at the standard 35% rate when they repatriated earnings from overseas. But they anticipated that the tax rate might drop in the future. As luck would have it, Congress enacted a repatriation holiday in 2004 that allowed U.S. multinationals to repatriate earnings from overseas and pay a tax of 5.25% rather than 35%. U.S.-based corporations continued to hold post-2004 earnings offshore in the hope that Congress would declare a repatriation holiday again. Likewise, a less progressive Congress might enact a “consumption holiday” or a “realization holiday” during which it would cut the statutory rate (or, better yet, repeal the consumption tax or the retrospective wealth tax entirely). In that case, a consumption tax would not in fact capture existing wealth, and a retrospective wealth tax would not operate as the near-equivalent of an annual wealth tax.
An advantage of a wealth tax over a consumption tax or retrospective wealth tax is that you can’t just “wait it out”: it applies year after year. If Warren wins the White House in 2020 and Democrats control both houses of Congress, they can enact an annual wealth tax and actually accomplish quite a bit of redistribution before the Democratic coalition gives way to divided government or Republican control. By contrast, a consumption tax might fail to tax all existing wealth — and a retrospective wealth tax might fail to replicate an annual wealth tax — if taxpayers correctly anticipate that these taxes ultimately will be repealed and if they time their consumption or their realization of assets strategically.
To be sure, Congress could — in theory — pass a constitutional amendment that enshrines a consumption tax or a retrospective wealth tax, and could perhaps win ratification from three quarters of the states (though this would require quite a dramatic change in the political climate in some places). But recall that one of the reasons to tax wealth through a consumption tax or a retrospective wealth tax rather than an annual wealth tax is that the former two options obviate the need for constitutional change. It turns out, though, that to truly approximate a tax on all existing or future wealth, a consumption tax or a retrospective wealth tax will require the support of a durable political coalition.
My own view is that the administrability advantages of a consumption tax and/or a retrospective wealth tax are sufficiently high that the best course for progressive politicians is to focus on building broad public support for the latter two policies — thereby convincing the wealthy and financially sophisticated that these policies aren’t going away. That is a lower threshold of support than constitutional ratification, but a much higher threshold than typical legislation. That said, one can make a reasonable argument for a wealth tax along the following lines: Apportionment of wealth tax revenues from real property is actually quite doable, so no constitutional amendment is required. A wealth tax with apportionment of revenues from real property could be passed with a narrow progressive majority (50 votes plus a vice presidential tiebreaker in the Senate if budget reconciliation is used to circumvent the filibuster, plus 218 votes in the House and the president’s signature). Administrability is a serious concern (especially with regard to valuation), but an imperfectly enforced wealth tax still might be preferable to no wealth tax at all.
The Trump Option
One approach that no presidential candidate is now proposing — but that might accomplish a lot more than Warren’s plan — comes straight out of Donald Trump’s playbook. In 1999, Trump — who was then considering a third-party presidential bid — floated the idea of a 14.25% one-time tax on households with a net worth of $10 million or more. Trump’s proposal actually has a number of advantages vis-à-vis Warren’s plan.
First, Trump’s proposal would accomplish a lot of redistribution fast — even if a different political coalition comes to power in a few years. Warren’s wealth tax would take longer to make a dent in wealth inequality.
Second, Trump’s proposal would involve one-time valuation, so the cost to the IRS of auditing high net worth taxpayers and reappraising their properties would be much lower than under Warren’s approach.
Third, many of the opportunities to game the Warren wealth tax take time to implement (e.g., shifting assets to family members without triggering gift tax). A Trump-like one-time tax could be assessed based on a household’s net worth as of January 27, 2019, cutting off these time-dependent workarounds. For example, if Warren’s wealth tax were enacted tomorrow, a couple in a community property state with $99 million of assets between them might choose to divorce, in which case they would each have $49.5 million of assets and would fall below Warren’s threshold. If Trump’s plan were enacted tomorrow but the tax were assessed as of yesterday, then any avoidance moves that individuals make in the future would not reduce their legal liability.
The Trump plan would run into the same constitutional objection as the Warren plan, and wealth tax revenues from real estate would likely have to be apportioned among the states. Those could then be block-granted, potentially helping some state governments (like Illinois!) emerge from fiscal crises. The primary problem — aside from selling this politically — is how to convince the country that we won’t do this again. A one-time wealth tax would not affect labor and savings incentives going forward, but the expectation of a 14.25% wealth tax applied year after year could have profoundly negative effects on growth. One option is to enact a one-time wealth tax in tandem with a constitutional amendment that prohibits the move from being repeated. It would be difficult — but not inconceivable — to cobble together a coalition of folks on the left who support the redistributive aspect of the one-time wealth tax and conservatives who want to make sure it never happens again.