The Yale Law Journal


The Taxing Power, The Affordable Care Act, and the Limits of Constitutional Compromise

05 Apr 2011

After a quiet century or so, the scope of Congress’s power “[t]o lay and collect taxes” is once again in the news. The taxing power was at issue when the Supreme Court issued a decision that President (and Chief Justice) Taft would later call the worst injury to the Court’s reputation ever, Pollock v. Farmers’ Loan & Trust, striking down the Income Tax Act of 1894. That decision was largely reversed by the 1913 enactment of the Sixteenth Amendment. Today the taxing power is one of three grounds on which the federal government defends the constitutionality of the Patient Protection and Affordable Care Act, particularly the individual responsibility requirement (IRR)—the portion of the Act requiring each individual to purchase insurance or pay a penalty tax.

As of this writing, two federal district courts have found that Congress lacks constitutional authority to enact the IRR, whether under the taxing power or otherwise. Recent commentators have also argued that the taxing power is inadequate to justify the IRR. While I have no doubt that the IRR is constitutional as an exercise of Congress’s power to regulate commerce, my comparative advantage, if any, lies in the tax field. I have covered this ground before, but I return to respond to the commentators who criticized my earlier effort and to the courts that have overlooked it.

In brief, my argument is that the IRR is unquestionably a tax because Congress repeatedly (if not universally) refers to it in that way and, in any event, that there is no requirement that Congress invoke the magic word “tax” to rely on its taxing authority. Those claims make up Part I of this Essay. Parts II and III take up the question of how one ought to interpret constitutional clauses, such as those that produced the existing limits on the taxing power, that are obviously the product of unprincipled compromise. Part II argues that the IRR is not a “direct” tax subject to a burdensome apportionment requirement, in part because that term should be read narrowly, as the Founders read it. And Part III argues that even if it is “direct,” the IRR is exempt from apportionment by the Sixteenth Amendment’s exception for “income” taxes because courts should not be in the business of writing their own tax code.

I. Is the IRR a “Tax”?

First, let me make clear that the IRR is an exercise of Congress’s power “[t]o lay and collect taxes.” The federal district courts in Virginia and Florida, as well as some commentators, argue that the IRR is not a tax at all, apparently because it is not clearly labeled as a “tax.” Ordinarily, Congress does not have to invoke specifically the source of authority for its enactments. But Randy Barnett and Erik Jensen both argue that courts give special deference to exercises of the taxing power and that the source of this deference is really just a refusal to look behind Congress’s choice of the “tax” label. So, on these accounts, if there is no label, there is no special deference.

It takes a particularly obstinate—even hostile—reading of the IRR provision to find that it is not labeled a “tax.” True, the result of a failure to obtain insurance is in some places called a “penalty.” But the letter t is followed by the letters a and x, in that order, forty-five times in the section of the Tax Code setting out the insurance requirement alone. Those who are subject to the requirement to provide insurance for themselves and their dependents are called “taxpayers.” The period for which they are required to carry insurance is called a “taxable year.” The amount payable for those who do not acquire qualifying insurance is in part determined according to a “percentage of . . . the taxpayer’s household income for the taxable year.” “Household income,” in turn, is defined as a slight modification of “adjusted gross income,” which is not defined in the statute, but which cross-references an important component of the federal income tax.

In any event, the claim that only a statute expressly labeled as a “tax” can be justified under the taxing power is false. In fact, since its earliest cases interpreting the taxing power, the Supreme Court has held that it is the effect of a statute as a tax, not its mere label, that controls. For example, confronted with the question of whether a federal requirement to obtain a license to engage in certain “immoral” activities was within Congress’s power, the Court in 1866 easily upheld the statute in question as an exercise of the taxing power. Although the statute made no mention of a “tax,” the Court held that “[t]he granting of a license must be regarded as nothing more than a mere form of imposing a tax.” Later, the Court would explain that the “scope and effect” of a statute determined whether it could be upheld as “within a granted power,” although in that particular case it noted the inquiry was unnecessary because the levy was called a tax on its face.

The Florida district court’s claim to the contrary is based on a logical fallacy. It states correctly that under Supreme Court precedent, if Congress uses the word “tax,” then the enactment is constitutionally a tax. To put that in logical terms, if A (“tax”), then B (tax). But the court then asserts that if Congress does not use the word “tax,” it follows that the enactment is not a tax: in logical terms, if not-A, then not-B. That is a formal logical fallacy, known as denying the antecedent.

The case law also deeply undercuts the suggestion by two courts that Congress’s decision to replace the word “tax” with “penalty” during the drafting of the statute demonstrates Congress’s “intent” to treat the IRR as something other than a tax for constitutional purposes. Since Congress does not have to use magic words to rely on its taxing power, the fact that it chose not to use those words sheds no light on its intent. Suppose, for example, that by default tenants in my state can obtain attorneys’ fees if they prevail in suits against their landlords. I have on my word processor a form lease that also provides for fees in that instance. I delete the fee-shifting clause, then print the lease and sign it. Have I waived my right to sue my landlord? If I know about the default rule, the answer is clearly not; all I have done is saved some printer toner by omitting a redundant clause. And the default rule here, since 1866, is that Congress does not need to use the word “tax” to rely on its constitutional power to tax. We have no reason to believe Congress was unaware of well-settled precedent. All of this presumes, as well, that Congress even has the power to prevent courts from sustaining a statute on some constitutional ground, a difficult proposition some of the district courts have simply assumed to be true with no explanation.

Nor is there any persuasive normative case for conditioning constitutionality on Congress’s definitively labeling an excise as a “tax.” Barnett and Jensen appear to suggest that the tax label will create some additional political constraint, perhaps on the theory that the label will increase the salience of the burden on the public. As I have argued, though, there is no evidence that decreasing the salience of a tax eases its passage. Public choice theory in fact implies the opposite. Few voters oppose obvious taxes because each free rides on the others. When taxes are partly hidden, though, those who are aware both of the tax and others’ ignorance of it increase their opposition because they know they cannot free ride.

Courts have sometimes also used “clear statement rules” of the sort Barnett and Jensen suggest to defend federalism values. The IRR, though, is an example of exactly the kind of legislation that, in a normatively sensible federalist structure, should be within federal authority. The structure of health care produces a race to the bottom that diminishes state autonomy. The fact that some states provide care for the uninsured creates a cross-border moral hazard, allowing neighboring states to offer fewer free services but permitting citizens of the low-service states to cross the border when they fall ill. Offering fewer free services means that paid services are cheaper or taxes are lower. So the pressure on each state is to free ride on the efforts of its neighbors; states that offer better services attract migrants that drive up prices, taxes, or both. As a result, states cannot set the policy that their citizens might prefer. Whether this collective action problem is solved through direct federal legislation, conditional spending, or, as here, conditional taxation, the answer should be the same: federal action enhances state autonomy and so should face few judicial barriers.

In addition to his clear-statement argument, Barnett also appears to argue that the IRR is not a tax because the “mandate cannot have been imposed to raise revenue.” His point seems to be that it is only the penalty for failing to follow the IRR that raises money and so the IRR itself cannot be justified under the taxing power. But the IRR, like any tax on a particular transaction, simply defines the transaction that is subject to taxation. If Barnett’s view were correct, then nearly all of the taxing power cases decided by the Supreme Court have been wrongly decided. When Congress properly imposed a tax on margarine colored to look like butter, the coloring of butter itself brought in no funds for the Treasury. Transferring marijuana to someone who has not obtained a license raises no money for the Treasury, but taxes on such transfers are part of the taxing power.

Finally, two federal district courts have mistakenly concluded that the IRR could be recharacterized as imposing a “penalty” and therefore falls outside of Congress’s power to impose a tax. The courts rely for this proposition on cases interpreting the rights of individuals subject to government punishment. It is true that Congress cannot escape the heightened due process standards to which criminal defendants are entitled, such as protection against double jeopardy, by the expedient of attempting to label a punishment as something else. But that fact does not in any way undermine other clear holdings of the Supreme Court that any tax that raises revenue, no matter how little, is within the grant of authority contained in Article I, Section 8.

II. Is the IRR a “Direct” Tax?

So if the IRR and its accompanying penalty provisions are a tax scheme, are they a proper exercise of Congress’s taxing power? Barnett and the Virginia court continue to claim that they are not because otherwise there would be no limit on Congress’s power to regulate the economy. As I have explained before, those arguments are wrong both normatively and as a matter of law. The Supreme Court has already rejected that identical argument in an almost identical context: Congress’s use of its Article I, Section 8, Clause 1 power to collect taxes and spend them for the general welfare. Further, the taxing and spending powers are not unlimited but rather face unique structural and judge-made limits that at least check, if not absolutely prohibit, congressional efforts to use the purse as a source of endless authority.

Let me focus here, then, on the seemingly more technical question, raised by Jensen and others, of whether the IRR meets the constitutional requirement that “direct” taxes be “apportioned.” Again, though, large questions of federalism arguably lurk in the background. “Direct” taxes have to be apportioned across states by population (including, infamously, three-fifths of a state’s slave population), which is to say that if a state has ten percent of the U.S. population, it would have to pay ten percent of the total federal revenues for that tax.

The apportionment requirement derives from negotiations at the Constitutional Convention over slavery. Southern states wanted slaves to count as persons for purposes of allocating representation in Congress; delegates from Northern states ultimately agreed, but only after the addition of compromise language providing that slaves would also increase the Southern burden for certain “direct” taxes by counting slaves as three-fifths of a person for apportionment purposes. Famously, no one knew what “direct” taxes were supposed to include, and historians generally think the term was just a way of papering over the controversy with language that had no real importance. Early authorities treat it as covering taxes on persons as such, and perhaps on the mere ownership of property, but no more.

Jensen argues for a more expansive view of “direct,” one that might include many forms of income tax and perhaps the IRR, by asserting that the Founders would never have intended an unlimited taxing power. He brushes aside a contrary 1796 Supreme Court decision as having been authored by “Federalists,” and, more recently, he claims that the Founders could never have imagined an income tax and so their failure to mention income taxes as direct can have no interpretive significance. Great Britain, however, instituted an income tax in 1799, and that tax had antecedents going back to 1758. And, as I have just argued, in fact the taxing power is subject to a series of limits, if not the ones preferred by Jensen. So Jensen’s syllogism—that “direct” taxes must be read broadly enough to cover some income taxes because otherwise the Taxing Clause would be unlimited—fails in almost every respect.

Nor is it the case that every constitutional clause must be read to embody some broader purpose. As John Manning argues, we should acknowledge that some provisions are simply the result of political horse trading and take them at their own terms. Manning’s controversial example is the Eleventh Amendment, but consider the more prosaic Article II, Section 1 requirement that only “a natural born citizen” can serve as President. Few would argue that John McCain, born in the Panama Canal Zone, was ineligible for the presidency; we simply read the provision as a somewhat inexplicable, thin, and formal requirement, easily satisfied.

Even on Jensen’s own terms, the IRR is not a direct tax. In his earlier work, Jensen claimed that a “direct” tax was one that could not be “shifted” by avoiding the untaxed activity. Picking up on some language in the Federalist papers authored by Alexander Hamilton, Jensen claimed that such shifting was an inherent limit on the power of government to tax because, if taxes were too high, they would distort behavior so much that they would collect no revenues. It was this limit, Jensen suggested, that made indirect taxes consistent with his view that the Founders would not have countenanced an unlimited tax.

By these standards, the IRR is obviously indirect. Anyone can avoid it by purchasing insurance. In his recent work on the IRR, though, Jensen claims that it is not indirect because the individual will have to pay either the government or the insurance company. But this flatly misreads his own argument. The shifting point, once more, is that consumer behavior will constrain congressional overreaching by limiting government revenues; it has nothing to do with how much individuals pay each other. In fact, in the shifting described by Hamilton, consumers often shift from a taxed product, such as tea, to an untaxed one, such as coffee. Just as with the IRR, consumers may still be paying for something.

Perhaps Jensen would say that his new version of the shifting analysis prevents government not just from overtaxing but also from overregulating. His revised stance on direct taxes would tend to constrain Congress from driving people toward a particular economic outcome. If so, then he too is arguing that all of the taxing power cases on the books are wrong. Under this version of his argument, taxes on colored margarine, gambling, guns, liquor, and marijuana, all of which have been upheld by the Court, would be “direct” because shifting cannot prevent the government from achieving its policy goal of deterring the activity.

III. Is the IRR an Income Tax?

Even if the IRR were a “direct” tax, it is still exempt from apportionment because it is an income tax, and the Sixteenth Amendment states that income taxes need not be apportioned. My view, as I have said elsewhere, is that the IRR is an income tax because those subject to the tax pay a percentage of their income, with a floor of $695 and a cap determined by the cost of obtaining qualifying insurance coverage. The floor has an additional exception that is itself based on income.

Here, again, the question is how broadly to read a mysterious political compromise enshrined as constitutional text. Jensen and others assert that, when Congress proposed a constitutional amendment exempting “income” but not other direct taxes from apportionment, they must have intended to preserve some aspect of the apportionment requirement. Thus, Jensen interprets “income” to exclude provisions that would resemble capitation taxes (taxes on persons as persons), apparently on the ground that to do otherwise would be to permit end-runs around the dregs of apportionment. Therefore, he suggests that a tax is not an “income” tax if “it may be measured by income for some persons, but will not be for many others.” According to Jensen, since the IRR imposes an equal tax on all those whose taxable amount exceeds the cap, their tax is not “measured by income” and hence the IRR is not an income tax.

Jensen admits that by his standard the Social Security tax is constitutionally questionable, but he fails to realize that under his test nearly every aspect of the income tax, from its inception to today, would fall as well. The income tax increases liability for many reasons other than income alone. It imposes higher taxes on those who live in the United States for the majority of each year; on those who pay no college tuition; on those who fail to save for retirement; on those who forgo compensation in the form of insurance premiums; and on those who finance their homes with a loan of more than $1 million. In all cases, two taxpayers with identical incomes pay different tax. Are these deviations from the constitutional definition of “income,” thereby requiring apportionment, or are they instead merely a way of measuring income, properly conceived? Jensen’s test would require that courts make that decision for nearly every provision of the Tax Code.

Consider, too, the rate structure of the tax system. In Jensen’s view, if too many people with differing incomes pay the same amount, a levy is no longer an income tax. Apparently, then, Jensen’s Constitution dictates the rate structure that Congress must choose in enacting an income tax. If it is too flat, resulting in too little difference between what those of different incomes pay, then it must be apportioned, which is to say that it cannot practically be implemented. Whatever the Sixteenth Amendment was supposed to accomplish, the context of its enactment as a response to a prior judicial invalidation must at a minimum be understood to foreclose endless judicial nitpicking over the form of the income tax—but that is precisely what Jensen’s view demands.

To be sure, the language of the Sixteenth Amendment should be read to have some content. Perhaps Congress was unwilling to expend the effort to win passage of an amendment that simply deleted the apportionment requirement, so we should honor that choice. But there are many ways to give content to “income” without installing the Supreme Court as a tax super-legislature. For example, the Supreme Court has distinguished income taxes from wealth taxes. An income tax measures accessions to, or changes in, net wealth, and those new accessions are taxed only once by an income tax system. A wealth tax taxes the whole value of the taxpayer’s property, and it can do so repeatedly. Under this view, so long as Congress does not attempt to use its income tax power as a wealth tax, any income tax would be presumptively valid. This rule has the strong appeal that it would avoid the need for the Supreme Court to create its own idealized notion of incomes and rate structures against which any congressional efforts would have to be judged.

In contrast, the normative appeal of Jensen’s approach is at best circular: it rests on his insistence that the taxing power must have some limits. If this represents an argument distinct from Barnett’s lament about the size of federal government, it is not clear how. And, as with Barnett, the response is that there are limits in place. The question for Jensen is: why are those limits inadequate? For instance, is there some reason why the limits must be judicially enforceable, rather than structural?

That Jensen and his cohorts have offered no answers thus far is not to say that no answers could be found. But before we unleash the Court to stride among the tax laws and smite them down where the Court will, it would be nice to know what principles should guide the smiting hand. For now, the Direct Tax Clause, and the income tax exception from it, remain constitutional text in search of any obvious underlying rationale.


Once again, the IRR is clearly constitutional under governing Supreme Court precedent, and normatively it should not be otherwise. The IRR solves major social problems, while the constitutional clauses that putatively restrain it serve no purpose at all except resolving long-dead political stalemates. Perhaps those compromises deserve to be honored, if for no other reason than to facilitate future compromise. But honor is fully paid by narrowly respecting the literal terms of the deals; critics who rely on those provisions propose instead some unspecified set of quasi-libertarian norms whose outcome would rest solely in the hands of judges.

Brian Galle is Assistant Professor, Boston College Law School. The author is grateful for helpful exchanges with Randy Barnett, Erik Jensen, and Ilya Somin, as well as comments and suggestions from Gillian Metzger and Trevor Morrison.

Preferred citation: Brian Galle, The Taxing Power, the Affordable Care Act, and the Limits of Constitutional Compromise, 120 Yale L.J. Online 407 (2011),