The only part of the Patient Protection and Affordable Care Act (hereafter, ‘the ACA’) struck down was a provision expanding Medicaid. We will argue that this was a mistake; the provision should not have been struck down. We’ll do this by identifying a test that C.J. Roberts used to justify his view that this provision was unconstitutional. We’ll defend that test against some objections raised by J. Ginsburg. We’ll then go on to argue that, properly applied, that test establishes the constitutionality of the Medicaid provision.
The only part of the Patient Protection and Affordable Care Act (hereafter, ‘the ACA’) struck down in National Federation of Independent Business (NFIB) et al. v. Sebelius, Secretary of Health and Human Services, et al. was a provision expanding Medicaid.1 We will argue that this was a mistake; the provision should not have been struck down. We’ll do this by identifying a test that C.J. Roberts used to justify his view that this provision was unconstitutional. We’ll defend that test against some objections raised by J. Ginsburg. We’ll then go on to argue that, properly applied, that test establishes the constitutionality of the Medicaid provision.
To say just what the provision in question is, it will help to have before us the distinctive structure of Medicaid. Each state runs its own Medicaid program, with substantial financial support from the federal government. There are several conditions that a state Medicaid program must satisfy in order to qualify for this federal support, and which all fifty states do currently satisfy. In particular, there have always been minimum coverage requirements.2 Before the ACA, these minimum coverage requirements were a bit of a hodgepodge. As J. Ginsburg notes,
To receive federal Medicaid funds, States must provide health benefits to specified categories of needy persons, including pregnant women, children, parents, and adults with disabilities. Guaranteed eligibility varies by category: for some it is tied to the federal poverty level (incomes up to 100% or 133%); for others it depends on criteria such as eligibility for designated state or federal assistance programs.3
The ACA introduced a broad new category of Medicaid eligibility. It said that the states must extend Medicaid eligibility to pretty much anyone whose income is below 133% of the federal poverty level.4 In addition, it provided quite generous federal support for these newly eligible claimants: over 90% of the costs of covering these individuals would be reimbursed to the states by the federal government.5 (For other categories of claimants, the reimbursement rate is considerably lower.6) But importantly to NFIB v. Sebelius, the ACA made provision of Medicaid services to these newly eligible individuals a condition of continuing federal support. That is, if a state did not expand its Medicaid program to accommodate these newly eligible individuals, the ACA gave the Secretary of Health and Human Services the authority to withhold all of the Medicaid funds the state would otherwise be entitled to.7
It is this last provision that the Supreme Court found to be unconstitutional in the current case. By a margin of 7–2, the Court ruled that it was unconstitutional to make federal support for continuation of the old Medicaid program conditional on states’ participation in this expansion. We’re going to argue that that was a mistake. We’ll mostly focus on the opinion written by C.J. Roberts (and joined, in this respect, by J. Kagan and J. Breyer), and the dissent written by J. Ginsburg (and joined by J. Sotomayor).
Our discussion in this paper takes place in a relatively constrained ideological space. Our aim is to argue against one part of the Court’s decision in this case, but to keep things manageable, we’ll do this while ignoring some very interesting philosophical and policy questions in the vicinity. Thus, for instance, we won’t concern ourselves with arguments against the constitutionality of the ACA’s Medicaid provision that, if successful, would also judge the old Medicaid program to be unconstitutional. And we’ll assume that there are constitutional constraints on what the federal government can do via its spending power, that go beyond what is explicitly stated in the U.S. Constitution. While we think these issues are very much worth pursuing, we’ll leave them aside as much as possible in this paper.
In what follows, it will be helpful to have some terminology for the various Medicaid requirements. We will use the following:
‘Old Medicaid’ refers to the set of requirements and funding levels that existed prior to the passage of the ACA.
‘Expanded Medicaid’ refers to the requirement that those not covered by Old Medicaid, but whose earnings fall below 133% of the federal poverty line, now be covered, plus the set of requirements and funding levels for these newly eligible individuals stipulated by the ACA.
‘New Medicaid’ refers to the conjunction of Old Medicaid and Expanded Medicaid, i.e., Medicaid as it was envisioned to work after the passage of the ACA.
‘The ACA’s Offer’ refers to the ACA’s offering the states the option of participating in (all of) New Medicaid, or not participating at all, but not the option of participating in just Old Medicaid.
One question that became surprisingly central to the ruling in NFIB v. Sebelius was the ontological relationship between Old Medicaid and New Medicaid. As we’ll see, J. Ginsburg held that these were the same program; the addition of Expanded Medicaid was just one of the many modifications that have been made to Old Medicaid over the years, without destroying its identity.8 C.J. Roberts disagreed, arguing that the Medicaid expansion provision “accomplishes a shift in kind, not merely degree.”9
We’re not going to take a strong stand on this ontological question. That’s partly because we think that questions like this are the wrong kinds of questions to be asking here. We don’t really have views on the criteria of identity through time for federal-state cooperative programs. And we’re not convinced that these questions have determinate answers. But even if they did, we still wouldn’t think that these answers were relevant to the constitutionality of proposed changes/supplements to those programs. Rather, on our view, what matters is the functional relationship between the old and new programs.
We’ll have more on this presently. But first, we need to look at why the Court thought the relevant sections of the ACA should be struck down.
The Spending Clause of the U.S. Constitution gives Congress the power “to pay the Debts and provide for the … general Welfare of the United States.”10 The justices in NFIB v. Sebelius agreed that the Spending Clause gives Congress a “broad authority” to interpret what that “general Welfare” consists in, and to apportion funds accordingly.11 This includes the power to offer the states funds as an inducement to take certain actions - such as establishing and operating certain programs - that accord with Congress’ understanding of the “general Welfare.” Old Medicaid is just such a federal-state cooperative program, established by Spending Clause legislation.
What the Spending Clause does not allow is for the federal government to require the states to implement a particular federal-state cooperative program, or to accept the associated funding package. That includes either directly ordering the states to do these things, or ‘indirectly’ coercing the states into doing them. As we’ll see in the next section, one of the main issues in NFIB v. Sebelius is whether the ACA’s Offer constitutes an attempt to unconstitutionally coerce the states into realizing a federal spending objective.12
The justices in NFIB v. Sebelius emphasized that Spending Clause legislation has the nature of a contract.13 The federal government offers the states funds conditional on their satisfying certain conditions. That is, it offers them money in exchange for doing something. That looks like a contract. And, arguably, some contracts are coercive.
When we talk about contracts being coercive, we don’t mean that there is coercion involved in getting one of the parties to accept the offer. Rather, we mean that there is something about the offer itself that makes it coercive. Call coercion of the first kind ‘coercion alongside the contract.’ In simple cases, where the parties have roughly equal power, there can be coercion of this sort - i.e., coercion alongside the contract - but it’s hard to see how the offer itself can be coercive. And in such cases, as long as there is no coercion alongside the contract, the party to whom the offer is made can simply refuse it. By contrast, when the parties are unequal in power, the mere making of the offer can sometimes amount to an abuse of the extra power. We’ll illustrate this point with some examples in section 5. For now we want to note two further points about the notion of coercion at issue here.
First, the question of whether a particular piece of Spending Clause legislation is unconstitutionally coercive should be distinguished from the question of whether that legislation runs contrary to the U.S. system of federalism. If the federal government offered each state a million dollars to introduce a filibuster rule into their state Senate procedures (perhaps because the U.S. Senate was embarrassed to be so idiosyncratic), that would arguably be unconstitutional. But that’s not because the offer would be unconstitutionally coercive. Rather, it would be because the Spending Clause isn’t an open invitation to let the federal government interfere with every power a state has, including powers over their own legislative procedures. To put it another way, the Spending Clause doesn’t allow the federal government to do an end run around the system of federalism.
Second, even if an offer is judged to be unconstitutionally coercive, it doesn’t follow that it will be voided. In NFIB v. Sebelius, the Supreme Court did not strike down the federal government’s attempt to expand Medicaid. Rather, it amended the terms of the offer so they were no longer unconstitutionally coercive. Rather than states having a choice between New Medicaid and nothing, the Court held that they would have a choice between Expanded Medicaid, Old Medicaid, and nothing. This isn’t the usual remedy when a contract is found to be coercive; usually, it’s simply voided.14
As we discussed in the previous section, the Spending Clause permits the federal government to offer the states financial inducement to implement particular programs, as long as the offer itself isn’t coercive. Typically, this just means that the states must have the option not to participate. As C.J. Roberts put the point,
In the typical case we look to the States to defend their prerogatives by adopting “the simple expedient of not yielding” to federal blandishments when they do not want to embrace the federal policies as their own. ... The States are separate and independent sovereigns. Sometimes they have to act like it.15
But sometimes, merely having the option not to participate is not enough. That’s the case if, for example, there’s something about the nature of the federal government’s offer that means that the states don’t have a genuine choice about participating. According to C.J. Roberts, there were three aspects of the ACA’s Offer that, together, made it the case that the states didn’t have a genuine choice about accepting. So, the offer was unconstitutionally coercive.
First, the ACA’s Offer conditioned the granting of funds for an already existing program (Old Medicaid) on states’ participation in another program (Expanded Medicaid). Call this phenomenon - i.e., conditioning the funding for an existing program on states’ participation in another program - ‘bundling.’ Bundling is importantly different from conditioning the funding for a program on states’ willingness to operate that very program in some particular way. The latter, according to C.J. Roberts, is just the federal government providing for the “general Welfare” as permitted by the Spending Clause.16
But second, even bundling may be permitted when it serves some legitimate purpose. The bundling involved in the ACA’s Offer, however, “serves no purpose other than to force unwilling States to sign up for the dramatic expansion in health care coverage effected by the Act.”17 To put the point another way, the only purpose of the bundling proposed by the ACA was to force the states to participate in Expanded Medicaid.18
Finally, the states’ financial stake in participating (or not) in the bundled programs was enormous. Federal support for Old Medicaid makes up more than 10% of the typical state’s budget.19 Threatening the states with losses of that magnitude was, according to C.J. Roberts, a kind of “economic dragooning.”20 When the financial stake at issue is so large, the states have no genuine choice about participating.
Why think that these features of the ACA’s Offer are sufficient to constitute it as unconstitutionally coercive? C.J. Roberts doesn’t spell out his reasoning here, but perhaps the thought is this. Bundling of federal-state cooperative programs can serve various purposes. For example, bundling can encourage the states to participate in a pair of programs, where the existence of each program helps the other one operate more efficiently. Or else, bundling can encourage the states to implement a new program that helps the existing program achieve its aims better. When bundling serves these (and other) legitimate public policy purposes, it’s plausible that the federal government is attempting to provide for the general Welfare in accordance with the Spending Clause.
If, on the other hand, the only purpose served by the bundling is to get the states to participate in one of the bundled programs, and further, the financial inducement offered is so large that it effectively serves as a “gun to the head” of the states, then that just amounts to the federal government requiring the states to participate in a particular program.21 And as we observed in the previous section, that’s not permitted under the Spending Clause. (We’ll have much more to say about this sort of reasoning in later sections of this paper.)
In sum, then, C.J. Roberts argued that the ACA’s Offer was unconstitutionally coercive because it satisfied the following three conditions:
The offer bundles two independent programs (Old Medicaid and Expanded Medicaid);
That bundling serves no other purpose than to force the states to participate in one of the bundled programs (Expanded Medicaid);
Failure to participate in the bundled programs exposes the states to enormous financial loss, and so, constitutes a kind of “economic dragooning.”
As we understand the Roberts Rule, conditions (1)-(3) are not only jointly sufficient for unconstitutional coerciveness, but each of them is individually necessary as well. To see why, we can consider the conditions in turn.
Without (1), the states could potentially opt out of any modification to an existing federal-state cooperative program. Imagine that the federal government established new standards for effectiveness of cancer treatments, and decided that Medicaid would henceforth only fund programs that complied with those new standards. The new parts of this Medicaid package are bundled together with the old parts to force the states to comply, and failure to participate in the bundled programs exposes the states to enormous loss. But this isn’t unconstitutionally coercive; the imagined policy could just be a good instance of quality control on government spending.22
Without (2), federal government actions that by hypothesis have a legitimate purpose could be ruled out. That would be a bad result. Here’s one sort of case that illustrates this point. Imagine that the federal government in its wisdom subsidizes widget production through conditional grants to the states. The government then discovers that widget production has serious environmental consequences. So it decides to fund cleanup operations, again through conditional grants to the states. It seems reasonable to bundle these two grants together, since the federal government has a legitimate interest in not subsidizing a certain industry without also subsidizing the elimination of its external costs. But that could be true even if widget production and the cleanup operations look like independent programs, and each is a signifiant component of state budgets.
Finally, (3) is needed to ensure that South Dakota v. Dole is not overturned.23 That case concerned the introduction of a new condition on federal highway funding, namely, that the states enforce a drinking age of 21. States that failed to comply with this condition stood to lose up to 5% of the highway funds they would otherwise be entitled to. Several states, including South Dakota, argued that this was unconstitutionally coercive. And they had a point; setting the drinking age and repairing highways look like quite independent programs. Further, it was clear that the point of the bundling was simply to get the states to comply with the federal government objective of raising drinking ages to 21. But C.J. Roberts held that this didn’t matter, because the threatened financial loss for noncompliance with the federal objective was small enough so as not to amount to unconstitutional coercion.24
Our reading of C.J. Roberts’ argument - and in particular, of his test for unconstitutional coerciveness - is quite similar to J. Ginsburg’s.25 But there’s one point of difference. While J. Ginsburg includes conditions very much like our (1)-(3) in her reading of the test, she adds a further condition as well.
We agree with J. Ginsburg that C.J. Roberts commits himself to the truth of (4).27 But we don’t think that (4) is relevant to the issue of whether the ACA’s Offer is unconstitutionally coercive (and it’s not clear to us that C.J. Roberts does either).28
As we see it, (4) speaks to a different question, namely, whether Medicaid program as envisioned by the ACA (New Medicaid) is the same program as the already existing one (Old Medicaid). And as we’ll argue next, that ontological question is irrelevant to whether the ACA’s Offer is unconstitutionally coercive.
Recall that the Roberts Rule says that an offer by the federal government to help establish a federal-state cooperative program is unconstitutionally coercive if the following conditions are satisfied:
The offer bundles two independent programs;
That bundling serves no other purpose than to force the states to participate in one of the bundled programs;
Failure to participate in the bundled programs exposes the states to enormous financial loss, and so, constitutes a kind of “economic dragooning.”
Condition (1) requires that that the programs be independent, not that their conjunction (in this case, New Medicaid) be distinct from the previously existing program (Old Medicaid). Thus, the Roberts Rule, as we understand it, places no emphasis on whether the federal government’s offer creates a new program, or merely modifies an already existing program. We’ll have a lot more to say about independence later (in section 6), but it should be clear that whether two programs are independent, and whether conjoining them creates a new program, are different questions. Independence has to do with how the programs function, what they do, not with their ontological status, what they are. So, even if New Medicaid turned out to be the same program as Old Medicaid, the two major parts of New Medicaid - Old Medicaid and Expanded Medicaid - could be quite independent of each other.
At first reading, it seems that both C.J. Roberts and J. Ginsburg take the ontological question (about whether New Medicaid is the same program as Old Medicaid) seriously. For instance, C.J. Roberts notes, in what seems like a positive way, the states’ claim that “the expansion is in reality a new program and that Congress is forcing them to accept it by threatening the funds for the existing Medicaid program.”29 And he says that Ginsburg’s reply, which assumes New Medicaid and Old Medicaid are the same program, “begs the question” against the states.30 But we think that while J. Ginsburg does take a stance on the ontological question, C.J. Roberts in the end does not. And on this point, we side with the latter, at least to the extent that we think that the ontological question is irrelevant to the issue of unconstitutional coerciveness.
To see why, consider a rival test to the Roberts Rule which says that what matters for unconstitutional coerciveness is that the programs in question (say, A and B) be numerically distinct. Would we get a rule that is better than the Roberts Rule? Actually, that breaks down into two questions. First, would the revised rule make for better law? And second, would the revised rule be a better interpretation of C.J. Roberts? We answer both questions negatively, with the first negative answer being some part of our reason for the second negative answer.
Questions about individuation criteria, and criteria of identity over time, for governmental programs are rather hard. We might think we could make progress by looking at the area where philosophers have made the most thorough investigation of identity criteria - namely, personal identity - and carrying the lessons from there over to debates about identity of governmental programs. But this would be useless twice over. For one thing, the debates about personal identity are so far from being settled that we have little to go on. For another, different views are going to be plausible in the two cases. A thoroughgoing conventionalism about personal identity is a rather unpopular view (though it is ably defended by Caroline West).31 But conventionalism about identity criteria for conventionally established programs, like Medicaid, seems much more plausible.
So in general we start knowing very little about identity criteria for governmental programs. But what we do know should give us pause before putting identity criteria into a substantial legal rule. It will often be indeterminate whether A and B are the same program, or different ones. (Governmental programs provide as clear an example of indeterminate identity as anything in Terrence Parsons’s study of indeterminate identity.)32 If we made the identity, or otherwise, of A and B relevant to whether a particular law was unconstitutional, we would risk concluding that it is indeterminate whether that law is unconstitutional. That doesn’t feel like an acceptable outcome.
In the next section, we’ll consider a thought experiment suggested by J. Ginsburg about what would have happened if Congress had repealed Old Medicaid and then enacted New Medicaid as a replacement.33 We’ll argue that the thought experiment isn’t particularly revealing, because it differs from what actually happened in a striking way. But perhaps we should question this assumption. Why should we say that the ACA merely enacted Expanded Medicaid, rather than saying that it actually repealed Old Medicaid, and enacted New Medicaid in its place? Indeed, if such a reading would make the ACA constitutional (as J. Ginsburg suggests), wasn’t the Court obliged to read the Act that way?34
The relevant reason is presumably that neither the ACA, nor the debate around it, reads like it was repealing and replacing Old Medicaid. That is, the rhetoric around the ACA reads like it was an expansion of Old Medicaid, not a replacement of it. And that’s enough to make it be the case that the ACA merely enacted an expansion of Old Medicaid, not a replacement of it. This talk about how the rhetoric matters to the ontological description of what the ACA did is part of what we meant above by saying that a conventionalist theory of identity for governmental programs is plausible. We all treated the ACA as expanding, not replacing, Old Medicaid, and hence it really was an expansion, not a replacement, of Old Medicaid. That’s so even though a functionally equivalent Act could have replaced Old Medicaid.
But while this kind of rhetoric can matter for ontological questions, it can hardly matter for the constitutional legitimacy of the ACA. C.J. Roberts held that one central part of the ACA, namely, the individual mandate, was a valid exercise of the taxing power, even though it was never marketed as such during Congressional debates.35 And that seems right to us. What matters for constitutionality is whether Congress has a power, not how they talk about their powers.36 But to make the ontological relationships between Old Medicaid, Expanded Medicaid, and New Medicaid relevant to the constitutionality of the latter would be to grant this talk, these “recitals of the powers,” too much significance. It’s hard to see that that is right, or that C.J. Roberts, in the very opinion where he upheld the individual mandate as an exercise of the taxing power, would do that.
So we conclude that the right way to read C.J. Roberts’ test is in terms of the relationship between what A and B do, not what A and B are. That’s how we take his comments that what matters is that the Medicaid provision in the ACA brought about a “shift in kind, not merely degree,” which would “transform it [i.e., Medicaid] so dramatically.”37 Looking at what Congress is trying to do is a much better guide to the constitutionality of its actions than looking at its talk.
Having said all that, we suspect that if one did take ontological distinctness to be important in testing for constitutionality, then the Medicaid provision of the ACA would be constitutional. It’s actually rather tricky to motivate a position on the ontology of Medicaid that makes that provision problematic. To see this, note that there are three ontological options here. (At least, there are three determinate options; there are also views on which the truth is indeterminate between these.)
First, it might be that New Medicaid and Old Medicaid are the same program, and adding Expanded Medicaid is just a familiar way for Medicaid to grow. That’s roughly J. Ginsburg’s position.38 Second, it might be that New Medicaid and Old Medicaid are distinct programs, with New Medicaid having two components - Old Medicaid and Expanded Medicaid - that run in parallel alongside each other. On this option, New Medicaid and Old Medicaid are no more one program than Social Security plus the Defense Department are one program. This feels like the option most in tune with the Court’s ruling. Third, it might be that Old Medicaid was repealed by the ACA, and New Medicaid is a new program put in its place. We’ve discussed this possibility a couple of times already, and will have more to say about it in the next section.
Now to get an ontological view that supports C.J. Roberts’s ruling (on the assumption that ontology is constitutionally relevant), we’d have to say that New Medicaid is sufficiently different from Old Medicaid that the first option is ruled out.39 But we couldn’t say that they are so different that the third option becomes the only plausible one.40 A middle ground has to be found, and that ground doesn’t look stable to us. Note in particular that Expanded Medicaid is a program whose eligibility is defined largely in terms of who is not eligible for Old Medicaid. It isn’t that Expanded Medicaid is for everyone earning less than 133% of the poverty line. Rather, it’s for everyone earning less than 133% of the poverty line, who wasn’t already covered by Old Medicaid. That’s a little odd. It’s especially odd because many Old Medicaid supported state programs were already more generous than the federal minimums, and so already covered many of the people who fall under Expanded Medicaid.41
Now ultimately none of these ontological speculations matter. But perhaps the underlying considerations do matter a little. The oddness of thinking of New Medicaid as a distinct program from Old Medicaid will be reflected in the oddness of operating it as one. And as we’ll consider in sections 6–7, that latter oddness is relevant to the test C.J. Roberts set out.
Before turning to applications of the Roberts Rule, however, we’ll consider an important objection to the Rule itself. We’ll argue that the objection is unsuccessful, but that it highlights some important aspects of the Rule, including the vitality of its third condition.
In her dissent, J. Ginsburg offered the following thought experiment.
Consider also that Congress could have repealed Medicaid ... Thereafter, Congress could have enacted Medicaid II, a new program combining the pre-2010 coverage with the expanded coverage required by the ACA. By what right does a court stop Congress from building up without first tearing down?42
We take the point of this thought experiment to be something like this. There would have been nothing unconstitutionally coercive about Congress repealing (Old) Medicaid, and enacting Medicaid II. But the end product of that repeal-and-replace effort (Medicaid II) would have been functionally equivalent to New Medicaid, though arrived at in a different manner. So, if enacting Medicaid II by repealing-and-replacing would have been constitutional, enacting New Medicaid by just expanding Old Medicaid (and so, altering an already existing agreement) must be constitutional as well. Therefore, any test that holds the ACA’s Offer to be unconstitutionally coercive - such as the Roberts Rule - should be rejected.
We think this is a bad way to assess whether a proposed use of Congress’ spending power is unconstitutionally coercive. To see why, it will help to consider two examples of agreements. The first involves contracts, as C.J. Roberts suggests we should understand agreements between the federal government and the states. The second is a more informal agreement. What the two agreements have in common is that they’re both continuing agreements. As we’ll argue below, that’s significant for assessing their coerciveness.
Supplier makes widgets, and their largest customer by far is MegaCorp. For many years, MegaCorp has had an annual order for one million widgets. The price that MegaCorp pays has basically tracked inflation since the deal was first established, and they now pay $10 per widget. This is a decent deal for Supplier, since it costs them $8 to make each widget. Although the supply contracts are explicitly only for a year at a time, it is generally understood that the contracts will be renewed, and the norm in the widget industry is that these contracts are renewed.
One year, MegaCorp says it is only interested in continuing the deal if Supplier also sells it a million gimcracks for $10 per gimcrack. This isn’t a great deal for Supplier, since it costs $11 to make each gimcrack. But Supplier will likely go out of business without the deal to sell widgets to MegaCorp.
We think MegaCorp’s proposal is coercive. Supplier has no real choice but to take on an extra supply contract that does not even cover their costs. And we think that the offer is coercive even though the combined offer MegaCorp makes, namely, $20 million for a million widgets and a million gimcracks, is not a bad deal for Supplier. Indeed, Supplier stands to profit on the combined deal. Nevertheless, the newly added part of the deal is basically a gift to MegaCorp, and MegaCorp is using their monopsony power to extract that gift. That makes the deal coercive.
The inequality in power between MegaCorp and Supplier matters here. Had Supplier been flooded with potential buyers for their widgets, MegaCorp could still make an offer of “Widgets and gimcracks, or nothing,” but they wouldn’t be in a position to make that offer credible. That’s because it isn’t credible, in the envisaged circumstances, that if Supplier had countered with an offer of “Widgets at the old price, and nothing more,” MegaCorp would have stuck to their guns and refused the mutually beneficial deal.43 Under those circumstances, the offer might not have been coercive. But those are not the circumstances in our example.
Many of the same points apply to our second example of a continuing agreement.
The Conditional Philanthropist
Child has recently graduated college, and has her first job. As with many first jobs, the pay is not fantastic. But it’s enough to afford a (barely) tolerable apartment in a safe enough neighborhood. Her Parent offers to pay the difference in rent that would allow her to live in a nicer apartment in a safer neighborhood, and Child takes up this offer. We assume that Parent is not under any obligation to do this; Child’s living situation without parental support is sub-optimal, but acceptable. It’s just a nice gift from Parent.
Some years later, after Child has established roots in the neighborhood that she can live in thanks to Parent’s gift, Parent informs her that he won’t keep providing financial support unless she agrees to assist with one of Parent’s political causes. As it happens, it is a cause that Child does not agree with.
We think this offer is also coercive. It would have been acceptable for Parent to simply never provide support for Child’s rental expenses. It would also have been acceptable for Parent to make clear from the start that the offer of financial support was conditional on reciprocal political support. Making an offer like that would be distasteful, and frankly strikes us as an appalling way to relate to one’s own child. But if Child could have an acceptable standard of living without this extra money, it isn’t coercive to offer her a little more money in exchange for political support. Once the arrangement has commenced, though, and Child has structured her life around it, threatening to take it away unless Child supports a political cause does seem coercive.
And this is why we think J. Ginsburg’s thought experiment fails. It’s true that it would be constitutional for Congress to repeal Old Medicaid. It’s also true that had Old Medicaid never existed, and Congress had enacted New Medicaid all at once in the ACA, there would be no constitutional question here. J. Ginsburg suggests that this is enough to show that the expansion is constitutional.
But the facts in J. Ginsburg’s thought experiment aren’t the facts at hand. The existing federal support for Old Medicaid creates an important kind of relationship between the states and the federal government. The states have structured a significant part of their operations around the (quite reasonable) assumption that this relationship would persist. Requiring the states to do something new in exchange for the preservation of that relationship is potentially coercive for just the same reasons that Parent and MegaCorp’s offers are coercive.
Put another way, in the context of a continuing relationship, particularly one in which there is an imbalance in power, we have to look at how the relationship is changing, and not just at the end result, to see whether we have a case of coercion. Even if the federal government’s offer of New Medicaid, appearing as a deux ex machina, would have been constitutionally acceptable, it doesn’t follow that adding Expanded Medicaid to Old Medicaid (in the way the ACA does) is acceptable.
Our argument in this section highlights the importance of condition (3) of the Roberts Rule. That condition focuses on the possibility of “enormous financial loss.” If Old Medicaid and Expanded Medicaid were both first enacted as part of the ACA, and they were bundled in the sense that a state was not free to participate in one but not the other, then this bundling would not expose the states to any losses. Rather, the bundling would just make it a little harder for the states to receive funds offered by the federal government. But the fact that Old Medicaid existed, and had been incorporated into the states’ financial planning, means that a threat to any state’s continued participation is a threat of loss to that state, as required under condition (3) of the Roberts Rule. And as we’ve been arguing, this is signifiant for assessing coerciveness.
When we said that MegaCorp’s and Parent’s offers were coercive, we were not offering an opinion about whether there are, or should be, legal remedies available to Supplier or Child. It could well be argued that the potential costs of involving the courts in relationships like these outweigh the costs of allowing some coercive offers to be made.
But when we look at legislation that alters the relationship between the states and the federal government, it is more plausible that there is a role for the courts in preventing coercion. If one level of government is coercing others, that is not something that should be allowed to stand. In fact, allowing it to stand seems incompatible with the U.S. system of federalism.
So we reject this argument of J. Ginsburg’s. As we read her, this is the only objection she makes to the test C.J. Roberts proposes. As we’ll see in the next section, she makes several further points that can be used as objections to his application of the test. We will endorse some of those objections. But she doesn’t appear to offer other objections to the test itself. And we too will assume, from here on, that the Roberts Rule is a good test for striking down proposed uses of Congress’ spending power on grounds of unconstitutional coerciveness.
This is primarily a paper on constitutional questions, so we’ll keep this digression brief. But we do want to note that our disagreement with J. Ginsburg here has wider ramifications. It is common in several walks of life to have agreements between two parties that are year-to-year on paper, but are expected by both parties to continue somewhat indefinitely. Many employment arrangements are like that. And, although less common in the United States, arrangements to rent housing in many parts of the world are also like that. In those cases, when considering whether proposed changes to the relationship by the more powerful party (usually the employer or the landlord) are coercive, we think it’s important to look at the changes themselves, and not just to whether the new agreement would be acceptable taken on its own. The same kind of reasoning should apply to continuing agreements between the states and the federal government.
While we agree that an expansion of Medicaid could be unconstitutionally coercive, we don’t think the expansion envisioned in the ACA actually is. Further, we think that the Roberts Rule, properly applied, gets this result. In this section, we’ll argue that Old Medicaid and Expanded Medicaid are sufficiently closely related that the first condition of the Roberts Rule is not satisfied. In the next section, we’ll argue that there is legitimate reason to bundle the two programs together, so the second condition is also not satisfied. Thus, even accepting the Roberts Rule as a good test for unconstitutional coerciveness, the ACA’s Offer turns out to not be unconstitutionally coercive.
Here’s the Roberts Rule once more. It says that an offer by the federal government to help establish a federal-state cooperative program is unconstitutionally coercive if the following conditions are satisfied:
The offer bundles two independent programs;
That bundling serves no other purpose than to force the states to participate in one of the bundled programs;
Failure to participate in the bundled programs exposes the states to enormous financial loss, and so, constitutes a kind of “economic dragooning.”
A pair of federal-state cooperative programs may be independent in at least two different senses: first, if the programs have substantially different purposes; and second, (even) if they have the same (or closely related) purposes, but attempt to achieve those purposes in substantially different ways.
When it comes to Old Medicaid and Expanded Medicaid, that the programs are not independent in either of the senses just outlined seems obvious on its face. After all, the two programs share an overall purpose; both have the aim of improving access to health services for the neediest Americans. In fact, as we’ve already mentioned (in section 4), eligibility criteria for one of the programs (Expanded Medicaid) is defined partly in terms of ineligibility for the other one. That suggests that the programs are designed to work together to achieve their overall purpose. Moreover, they try to achieve this purpose in the same way, via the same circuitous means; both feature the federal government encouraging the states to provide health care to their poorest residents by paying a (large) percentage of the costs, conditional on the states meeting certain conditions for minimum care. That looks like enough to make it the case that the programs are closely related, contra condition (1) of the Roberts Rule.
Obviously, C.J. Roberts disagreed with this assessment. We find in his opinion four considerations that might be used to argue that the programs are not suitably related.44 The first two purport to identify significant differences in purpose between Old Medicaid and Expanded Medicaid, while the third and fourth point to differences in how they’re intended to operate. None of these strikes us as persuasive. (Three of these considerations were also discussed by J. Ginsburg in her dissent, and as will be clear below, we largely sympathize with her responses.45)
The first consideration is that Old Medicaid covered discrete “categories of the needy: the disabled, the blind, the elderly, and needy families with dependent children.”46 But Expanded Medicaid had a somewhat blunter condition of eligibility, namely, that the claimants be earning less than 133% of the federal poverty level (and not already be covered by Old Medicaid). It’s hard to see how this amounts to a source of unrelatedness. It’s as if C.J. Roberts thought of Old Medicaid as not merely having a disjunctive essence, but as being essentially disjunctive. He seems to be suggesting that any program that didn’t have a long list of eligibility criteria could not be closely related to Old Medicaid. This strikes us as absurd, especially since the list of eligibility criteria for Old Medicaid were hardly arbitrary or ad hoc. It was meant to reflect, as C.J. Roberts recognized, ways of being especially needy. A simpler means of determining need would accomplish the same thing.
This brings us to the second consideration, that by covering up to 133% of the poverty level (rather than merely covering the discrete categories mentioned in the previous paragraph), Expanded Medicaid was no longer reserved for the neediest among us. But a family of four has to earn less than $31,322 to qualify under this condition.47 To exclude those earning so little from among the neediest seems bizarre.48
Of course, it would be possible to keep expanding Medicaid, in something akin to the manner envisioned by the ACA, until it no longer covers just the neediest, but more closely resembles a program of universal health care. And it may even be indeterminate where the line between these lies. But C.J. Roberts offers no reason to think that Expanded Medicaid crosses this line. Moreover, in light of well-known criticisms of the federal poverty measure - criticisms which charge that the measure, developed in the 1960s, is now outdated and significantly undercounts poverty in the U.S. - there’s good reason to think that 133% of the federal poverty level is not an especially generous threshold.49
The third consideration is that the federal subsidies for Expanded Medicaid are more generous than those for Old Medicaid.50 But it’s hard to see how the generosity, or lack thereof, of the federal government speaks to the relatedness of the two programs. Would more stinginess on the part of the federal government have made the two programs more closely related?
And the final consideration C.J. Roberts offers is that the conditions imposed on the states by Old Medicaid and Expanded Medicaid are different. But as J. Ginsburg notes, things aren’t quite so simple. While it’s true that the conditions imposed under Expanded Medicaid are different from those traditionally required under Old Medicaid, they aren’t any different from what has been required under Old Medicaid since 2006. So Expanded Medicaid isn’t different, in this sense, from Medicaid as it was at the time ACA was passed.51
None of this is meant to deny that there are differences between Old Medicaid and Expanded Medicaid. But by itself, that doesn’t tell us much about whether the ACA’s Offer was unconstitutionally coercive. Condition (1) of the Roberts Rule requires not merely that programs in question be different in some way(s), but that they be independent. We’ve been arguing that the differences cited by C.J. Roberts don’t speak to independence of the two programs in any relevant sense.
In this section, we describe a further way in which the ACA’s Offer fails to be unconstitutionally coercive under the Roberts Rule. We argue that the federal government has at least three legitimate reasons for bundling the programs together. Thus, it’s not true that the only reason for the bundling is to force the states to participate in Expanded Medicaid, contra condition (2) of the Roberts Rule.
The first (and more minor) reason concerns the complexity that will arise when future modifications are made to Medicaid. As we’ve already had reason to note, Medicaid requirements are not set in stone.52 They change, often substantially, from year to year. But the Court’s decision in NFIB v. Sebelius has just made it considerably more complex to make any such changes. As things stand, if Congress wants to enact changes to both Old Medicaid and Expanded Medicaid, it will have to apply the change to each program separately.
Of course, this point is related to our discussion in the previous section; if the two programs were genuinely unrelated, we wouldn’t expect many changes that would apply to both programs. But given that the two programs work in substantially similar ways, at least some such changes - maybe even many - will be forthcoming.
The second reason concerns potential savings for the states. Though both C.J. Roberts53 and the authors of the Joint Dissent54 emphasized the extra financial burden Expanded Medicaid would impose on the states, there’s ample research to suggest that that burden is in fact quite minor.55 For example, the Center on Budget and Policy Priorities estimated that the expansion would cost the states just 2.8% more than they would otherwise spend on Medicaid between 2014 and 2022.56 Even more strikingly, that figure overstates the increase in state spending once we add in the savings to the states from no longer having to provide uncompensated care to those currently uninsured (or underinsured). The Urban Institute estimated that the states would end up saving money - anywhere from $92 to $129 billion between 2014 and 2019 - by taking up the ACA’s Offer.57 Those are savings that can be used by the states to bolster their other health care programs, including Old Medicaid. So, implementing Expanded Medicaid might in fact put the states in a position to run Old Medicaid better.
The third reason has to do with improving health outcomes generally. There’s research suggesting that high rates of uninsurance in a community adversely affect health outcomes for everyone there, including those with insurance.58 The difficulty and expense involved in treating the uninsured when they are finally driven to seek health care tends to reduce the quality of care that might otherwise be available for the insured. So, by reducing the ranks of the uninsured, Expanded Medicaid can help states achieve better health outcomes via their other health care programs, including Old Medicaid.
In assessing the costs and benefits of programs like Medicaid, it’s important to remember the ways in which basic health care is unlike many other goods. In particular, even if a state chooses not to participate in something like Expanded Medicaid, it still must share the financial burden of providing health care to the population that would have been covered by such a program. States still have to provide emergency rooms, and in practice, are not compensated for a significant portion of the care these provide to the uninsured. If some of the people who move onto Expanded Medicaid were previously uninsured, a state could see an overall reduction in its health care spending. Some of these savings would come from moving state costs onto the federal component of Medicaid. But some would come from the fact that patients would be moving from a form of health care that’s very expensive to provide, i.e., emergency room care, to the more efficient forms that are mostly available to the insured.59 These savings could strengthen Old Medicaid, and the health system as a whole. So expanding Medicaid could make Old Medicaid more financially self-sufficient.
Again, the fact that the programs are not unrelated is relevant. It’s because these are both health care programs - and moreover, health care programs for those who might otherwise be uninsured - that it’s plausible to think that savings incurred in one program strengthen the other program, rather than just strengthen the federal government’s balance sheet.
So the expansion of Medicaid fails two of Roberts’ three criteria for being unconstitutionally coercive. That’s why we can agree with his test, but disagree with the conclusion that the expansion was unconstitutionally coercive.
We’ve argued in this paper for several conclusions:
The ontological relationship between Old Medicaid and New Medicaid is irrelevant to the constitutionality of the ACA.
The facts that Old Medicaid was well-established prior to the passage of the ACA, and that the states relied on its continuation, are relevant to the constitutionality of the ACA, and, in fact, open up the possibility that the ACA’s Offer is unconstitutionally coercive.
But that Offer is not, after all, unconstitutionally coercive, for at least two reasons: first, because it bundles together two programs (Old Medicaid and Expanded Medicaid) that are closely related, and second, because there are legitimate reasons for that bundling.
Our discussion has left open several interesting questions. That’s partly for space reasons, and partly because we’re not sure what the right answers are. We’ll end our discussion by listing three of those questions.
First, are offers by the federal government to help establish federal-state cooperative programs unconstitutionally coercive only when the states are exposed to significant losses? We are inclined to think that this is not the case, and that this poses a problem for C.J. Roberts’ attempt to distinguish the current case from South Dakota v. Dole via condition (3) of the Roberts Rule.
Second, is the Roberts Rule a good test for unconstitutional coerciveness of offers by the federal government in other federal jurisdictions? For example, does it throw light on whether the Australian High Court ruled correctly in the Uniform Tax Cases?60
Third, is the Roberts Rule a good test for coerciveness of proposed changes to continuing relationships in business, or in residential tenancy? If so, this case could have implications for areas far removed from constitutional law. Although the previous two sections have provided a number of reasons to doubt that the ACA’s Medicaid provision is unconstitutionally coercive, the issues raised here are relevant to the broader question of when the more powerful party in a continuing relationship can force changes to that relationship.61
567 U.S. ___ (2012), available at http://www.supremecourt.gov/opinions/11pdf/11--393c3a2.pdf.↩︎
Jonathan Engel, Poor People’s Medicine: Medicaid and American Charity Care since 1965 (Durham, NC: Duke University Press, 2006), 48–51.↩︎
NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 38 (2012).↩︎
42 U.S.C. 1396a(a)(10)(A)(i)(VIII).↩︎
42 U.S.C. 1396d(y)(1). More specifically, the ACA required the federal government to bear 100% of the costs of covering these newly eligible claimants through 2016. The level of federal support was then allowed to gradually decline, but to no lower than 90% of these costs.↩︎
In fiscal year 2012, federal funds offset between 50 and approximately 74% of states’ costs of covering claimants eligible for Medicaid prior to the passage of the ACA. Kaiser Commission on Medicaid and the Uninsured, “An Overview of Changes in the Federal Medical Assistance Percentages (FMAPs) for Medicaid,” July 2011, available at http://www.kff.org/medicaid/upload/8210.pdf, 2.↩︎
42 U.S.C. 1396c.↩︎
NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 41–44 (2012).↩︎
Id., Roberts, C.J., slip opin. at 53.↩︎
U.S. Constitution, Article I, Section 8, Clause 1.↩︎
NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 50 (2012). See also Roberts, C.J., slip opin. at 45–46, and Joint Dissent, slip opin. at 29–32.↩︎
In our discussion, we’ll leave open the question of whether an offer by the federal government can be coercive without being unconstitutionally coercive.↩︎
Id., Roberts, C.J., slip opin. at 46, and Joint Dissent, slip opin. at 33.↩︎
This is actually a surprising fact about the Court’s decision. Some prominent court watchers seemed to assume, before the decision, that if the court found the expansion to be constitutionally problematic, it would simply be voided. See, for instance, the pre-decision discussions in Kaiser Family Foundation, “The Health Reform Law’s Medicaid Expansion: A Guide to the Supreme Court Arguments,” March 2012, available at http://www.kff.org/healthreform/upload/8288.pdf, and Lyle Denniston, “Argument Preview: Health Care, Part IV - The Medicaid Expansion,” 23 March 2012, available at http://www.scotusblog.com/2012/03/argument-preview-health-care-part-iv-the-medicaid-expansion/. Both of these discuss the possibility the expansion will be struck down, but neither seems to mention the possibility that the expansion will be made voluntary. We suspect they were assuming the court would either find the expansion non-coercive, and hence let it stand, or coercive, and hold that coercive offers are void.↩︎
Id., Roberts, C.J., slip opin. at 49, citation omitted.↩︎
Id. at 50.↩︎
We’ll talk throughout this paper about purposes served by bundling, rather than the purposes Congress had in mind by bundling. This is in keeping with both C.J. Roberts’ and J. Ginsburg’s discussions.↩︎
In fiscal year 2012, spending on Old Medicaid comprised nearly 24% of total spending by the states. The National Association of State Budget Officers, “State Expenditure Report: Examining Fiscal 2010–2012 State Spending,” available at http://www.nasbo.org/sites/default/files/State%20Expenditure%20Report_1.pdf, 44. Federal support offset at least 50% of that state spending, and for some states, quite a lot more. See note 6.↩︎
NFIB v. Sebelius, 567 U.S. ___, Roberts, C.J., slip opin. at 52 (2012).↩︎
Id. at 51.↩︎
Note that Old Medicaid has evolved in just this way, with modifications to the program being bundled together with unchanged parts. For example, the Balanced Budget Act of 1997 (P.L. 105–33) required the states to expand their coverage of home health visits required by Medicare beneficiaries. The federal government met all of the costs of this expansion, but did not make it voluntary. For more on this, see Melvina Ford, Richard Price, and Jennifer Neisner, “Medicaid: 105th Congress,” Congressional Research Service, February 4 1998, available at http://www.policyarchive.org/handle/10207/508, 12–13.↩︎
483 U.S. 203 (1987).↩︎
NFIB v. Sebelius, 567 U.S. ___, Roberts, C.J., slip opin. at 50–51 (2012). But in a footnote, C.J. Roberts seems to argue that any amount of threatened financial loss makes the federal government’s offer coercive. Id. at 52n2. Perhaps he intended to draw a distinction between an offer being coercive, and it being unconstitutionally coercive. But that distinction doesn’t figure anywhere else in his opinion. So, we’re left at a loss about how to square this footnote with the rest of C.J. Roberts’ argument.↩︎
Id., Ginsburg, J., slip opin. at 39.↩︎
Id., Roberts, C.J., slip opin. at 54. “A State could hardly anticipate that Congress’ reservation of the right to”alter" or “amend” the Medicaid program included the power to transform it so dramatically."↩︎
The result of adding (4) to the Roberts Rule is implausible as a test for unconstitutional coerciveness. Imagine that the federal government conditions federal highway spending on states enforcing a drinking age of 21 (as in South Dakota v. Dole), but this time, threatens to withhold 100% of highway funds from any states that fail to comply. That offer seems unconstitutionally coercive. And this verdict isn’t changed if the federal government threatens to add such a condition to federal highway spending for many years before actually doing so. So the foreseeability (or lack thereof) of the condition doesn’t seem to affect whether the offer is unconstitutionally coercive.↩︎
Id. at 52.↩︎
Caroline West, “Personal Identity: Practical or Metaphysical?” in Kim Atkins and Catriona Mackenzie (eds.), Practical Identity and Narrative Agency (New York: Routledge, 2008), 56–77.↩︎
Terrence Parsons, Indeterminate Identity: Metaphysics and Semantics (Oxford: Oxford University Press, 2000).↩︎
NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 51 (2012).↩︎
Id., Roberts, C.J., slip opin. at 33–40.↩︎
“[T]he constitutionality of an action taken by Congress does not depend on recitals of the powers which it undertakes to exercise.” Woods v. Cloyd W. Miller Co., 333 U.S. 138, 144 (1948), emphasis added. Quoted in NFIB v. Sebelius, 567 U.S. ___, Roberts, C.J., slip opin. at 39 (2012).↩︎
Id. at 53, 54.↩︎
Id., Ginsburg, J., slip opin. at 41–44.↩︎
For if ontology were constitutionally relevant, and New Medicaid and Old Medicaid were substantially similar, then it’s hard to see why the former would be unconstitutionally coercive but the latter not.↩︎
As J. Ginsburg suggests, there’s little doubt about the constitutionality of the third option. Id. at 51.↩︎
Kaiser Commission on Medicaid and the Uninsured, “Federal Core Requirements and State Options in Medicaid: Current Policies and Key Issues,” April 2011, available at http://www.kff.org/medicaid/upload/8174.pdf.↩︎
NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 51 (2012), citation omitted.↩︎
We are drawing here on the literature in game theory on credible threats. For more detailed discussion see, for example, Avinash Dixit and Susan Skeath, Games of Strategy, second edition (New York: Norton, 2004), Chapter 10.↩︎
Id., Roberts, C.J., slip opin. at 53–54.↩︎
Id., Ginsburg, J., slip opin. at 50–51. J. Ginsburg treats these considerations as comprising C.J. Roberts’ argument for the view that Old Medicaid and New Medicaid are distinct programs. We’ve explained (in section 4) why we don’t take C.J. Roberts to subscribe to this view.↩︎
Id., Roberts, C.J., slip opin. at 53.↩︎
J. Ginsburg makes a similar point. NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 50 (2012).↩︎
Jared Bernstein, “More Poverty than Meets the Eye,” Economic Policy Institute, April 11 2007, available at http://www.epi.org/publication/webfeatures_snapshots_20070411/. Bernstein writes, “When it comes to poverty in America, almost every analyst agrees that the official measure is terribly out-of-date and no longer provides a valid indication of economic deprivation.” See also David M. Betson, Constance F. Citro, and Robert T. Michael, “Recent Developments for Poverty Measurement in U.S. Official Statistics,” Journal of Official Statistics, vol. 16, no. 2, 2000, 87–111.↩︎
See notes 5 and 6.↩︎
NFIB v. Sebelius, 567 U.S. ___, Ginsburg, J., slip opin. at 50–51 (2012).↩︎
See note 22.↩︎
Id., Roberts, C.J., slip opin. at 52n12.↩︎
Id., Joint Dissent, slip opin. at 45–46.↩︎
See, for example, January Angeles, “How Health Reform’s Medicaid Expansion Will Impact State Budgets: Federal Government Will Pick Up Nearly All Costs, Even as Expansion Provides Coverage to Millions of Low-Income Uninsured Americans,” Center on Budget and Policy Priorities, July 25, 2012, available at http://www.cbpp.org/files/7--12--12health.pdf, and John Holahan and Irene Headen, “Medicaid Coverage and Spending in Health Reform: National and State-by-State Results for Adults at or Below 133% FPL,” Kaiser Commission on Medicaid and the Uninsured, May 2012, available at http://www.kff.org/healthreform/upload/medicaid-coverage-and-spending-in-health-reform-national-and-state-by-state-results-for-adults-at-or-below--133-fpl.pdf.↩︎
Angeles, op. cit., 1.↩︎
Matthew Buttguens, Stan Dorn, and Caitlyn Carroll, “Consider Savings as well as Costs: State Governments Would Spend at Least $90 Billion Less With the ACA than Without It from 2014 to 2019,” The Urban Institute, July 2011, available at http://www.urban.org/UploadedPDF/412361-consider-savings.pdf, 1. This projection includes savings from moving some adults currently covered by various state Medicaid programs onto federal subsidies via the health care exchanges to be established under the ACA.↩︎
For an useful overview of some research on this point, see the Institute of Medicine, America’s Uninsured Crisis: Consequences for Health and Health Care (Washington, D.C.: National Academies Press), especially Chapter 4.↩︎
In its first full year of implementing health reform substantially similar to the ACA, Massachusetts saw an astounding 38% drop in its spending on uncompensated care. Angeles, op. cit., 5.↩︎
South Australia v Commonwealth 65 CLR 373 (1942) and Victoria v Commonwealth 99 CLR 575 (1957).↩︎
Thanks to the editors and referees of this journal for many helpful comments.