Part I of this series explained that “[e]ven though the [Affordable Care Act’s] penalty is [now] set to $0, the individual mandate still plays an important social function to reduce adverse selection, and plays a role in the operation of employer-based coverage.” It still has a legal effect. Part II “focus[ed] on how the recent tax reform legislation affects Chief Justice Roberts’s saving construction in NFIB v. Sebelius.” Specifically, because the mandate can no longer fall within the NFIB saving construction—because the penalty is $0—it is no longer constitutional. This third installment will address the issue of severability: if the mandate is now unconstitutional, should other portions of the law also be set aside.

A “Convergent Constitutional Violation” in Frost

Prior to December 22, 2017, Section 5000A(a) of the Affordable Care Act—that is, the individual mandate—was constitutional by virtue of Chief Justice Roberts’s saving construction in NFIB v. Sebelius. However, when President Trump signed the Tax Cuts and Jobs Act of 2017 (TCJA) into law, thereby reducing the penalty to $0, a conflict arose. Neither the mandate, nor the tax cut, by themselves, pose any problems. But cannot both exist in harmony. A new article from two recent Yale Law School graduates describes this conflict as a “convergent constitutional violation.” James Durling and E. Garrett West contend that if the TCJA of 2017 renders the individual mandate of 2010 unconstitutional, then the “court should strike the repeal of the tax penalty in the Tax Cuts Act.” When I first read their article, I quickly assumed that this argument was insane. (Perhaps you had the same reaction). But, after letting the concept marinate for a few weeks, I began to see its insight. I encourage you to do the same, and read the eighteen-page essay in its entirety. Even if you ultimately disagree with the authors, you will see this case in a new light.

The starting point for their severability analysis is Frost v. Corporation Commission of Oklahoma. 278 U.S. 515 (1929). W.A. Frost was allowed operated a cotton gin, pursuant to a permit granted by the Oklahoma Corporation Commission, because he made a “satisfactory showing of public necessity.” 278 U.S. at 517. Subsequent to the issuance of Frost’s permit, Oklahoma modified the pre-existing statutory regime with a new “proviso” that required the Commission to issue a license for a “gin to be run cooperatively” if a petition was “signed by one hundred (100) citizens and tax payers of the community where the gin is to be located.” Id. Critically, the “proviso made it mandatory to grant the permit applied for without regard to [public] necessity.” Id. at 518 (emphasis added). The Supreme Court, in an opinion by Justice Sutherland, found the proviso violated the Fourteenth Amendment, because “its effect is to relieve all corporations organized under the [pre-existing legislation] from an onerous restriction upon the right to engage in a public business which is imposed by the statute.” Id. at 521-22. In other words, the original statutory regime was sound; the proviso, however, rendered that regime unconstitutional. (The similarities to the facts in Texas v. U.S. quickly become apparent).

Next, Justice Sutherland turned to the question of severability by considering two different scenarios. First, what would happen if the same legislature had enacted both the unconstitutional proviso and the otherwise-constitutional substantive provision? That is,

the original statutory scheme excluded cooperative gins from having to demonstrate a public necessity. In such a case, he wrote, “the effect would be to render the entire section invalid,” for it would not be sufficient to simply strike out the proviso. Id. at 525. The Court would not “extend the scope of the law in that regard so as to embrace corporations which the Legislature passing the statute had, by its very terms, expressly excluded.” Id. (Justice Scalia’s dissenting opinion in Legal Services Corp. v. Velazquez, which I will discuss infra, relies on this first scenario.)

In the second scenario, which described the facts at issue in Frost, “the proviso here in question was not in the original section.” Id. at 526. Rather, “[i]t was added by way of amendment many years after the original section was enacted.” Id. In other words, the proviso “repeal[ed] by implication the requirement of the existing statute in respect of public necessity,” rendering that old statute unconstitutional. The Court’s conclusion is worth quoting in full:

“Here it is conceded that the statute, before the amendment, was entirely valid. When passed, it expressed the will of the Legislature which enacted it. Without an express repeal, a different Legislature undertook to create an exception, but, since that body sought to express its will by an amendment which, being unconstitutional, is a nullity and, therefore, powerless to work any change in the existing statute, that statute must stand as the only valid expression of the legislative intent.” Id. at 526-27.

To clarify, because the later-in-time proviso rendered the original statute unconstitutional, it is a “nullity” and cannot “work any change in the existing statute.” What is the “only valid expression of the legislative intent”? The original statute, enacted by the earlier-in-time legislature.

My initial reaction was to question whether Frost was still good law. As a threshold matter, this was not a new rule. Justice Sutherland’s majority opinion favorably cited an 1866 opinion from the Michigan Supreme Court by Chief Justice Thomas M. Cooley involving a “similar situation.” Id. at 527. In that case, the renowned jurist wrote, “nothing can come in conflict with a nullity, and nothing is therefore repealed by this act on the ground solely of its being inconsistent with a section of this law which is entirely unconstitutional and void.” Id. The Supreme Court expressed a similar sentiment in United States v. Jackson: a recently-added “clause authorizing capital punishment [was] severable from the remainder of the kidnaping statute and that the unconstitutionality of that clause does not require the defeat of the law as a whole.” 390 U.S. 570, 586 (1968).

I could not find any case that cast Frost in doubt. Several federal and state court decisions favorably cited and applied it. See e.g., White Motor Corp. v. Citibank, N.A., 704 F.2d 254, 261 (6th Cir. 1983) (“It has long been held that a statute which is unconstitutional does not repeal a prior statute on the subject when a contrary construction would create a void in the law which the legislative body did not intend. The prior statute is ‘revived’ to avoid a chaotic hiatus in the law.”); In re Benny, 812 F.2d 1133, 1148 (9th Cir. 1987) (Norris, J., concurring) (“My second reason for disagreeing with the Department’s position that Congress in 1984 impliedly or explicitly repealed the prior holdover provision is that any attempt by Congress to terminate the holdover tenure of incumbent judges would itself have constituted an independent constitutional violation.”); Matter of Certification of Questions of Law from U.S. Court of Appeals for Eighth Circuit, Pursuant to Provisions of SDCL 15-24A-1, 1996 S.D. 10, ¶ 88, 544 N.W.2d 183, 204 (“An act or amendment to an act which violates the Constitution has no power, and can neither build up nor tear down. It can neither create new rights nor destroy existing rights.”).

The most detailed discussion of Frost by the Supreme Court appeared in Justice Scalia’s dissent in Legal Services Corp. v. Velazquez.531 U.S. 533, 538 (2001). This 2001 opinion warrants a careful study. Through § 504(a)(16) of the Omnibus Consolidated Rescissions and Appropriations Act of 1996, Congress prohibited the Legal Services Corporation from funding any organization

“that initiates legal representation or participates in any other way, in litigation, lobbying, or rulemaking, involving an effort to reform a Federal or State welfare system, except that this paragraph shall not be construed to preclude a recipient from representing an individual eligible client who is seeking specific relief from a welfare agency if such relief does not involve an effort to amend or otherwise challenge existing law in effect on the date of the initiation of the representation.”

This provision was challenged in court. On appeal, the Second Circuit declared unconstitutional the “the qualification that representation could ‘not involve an effort to amend or otherwise challenge existing law,’ because it ‘clearly seeks to discourage challenges to the status quo.’” Id. at 539. This, the court of appeals found, constituted an “impermissible viewpoint discrimination.” Id. Because “congressional intent regarding severability was unclear,” the Second Circuit “decided to ‘invalidate the smallest possible portion of the statute, excising only the viewpoint-based proviso rather than the entire exception of which it is a part.’”

On the merits, the Supreme Court agreed, and found that the “funding condition” in § 504(a)(16) violates the First Amendment, and is “invalid.” With respect to severability, the Court noted that the Second Circuit “reached the reasoned conclusion to invalidate the fragment of § 504(a)(16) found contrary to the First Amendment, leaving the balance of the statute operative and in place.” Id. at 549. However, Justice Kennedy’s majority opinion “decline[d] to address” anew the question of severability because “[t]hat determination was not discussed in the briefs of either party or otherwise contested here.” Id. Instead, the majority simply affirmed the judgment of the Second Circuit, and left the remainder of § 504(a)(16) in place. In other words, the Legal Services Corporation could now fund organizations that represent an individual eligible client who is seeking specific relief from a welfare agency even if such relief involves an effort to “amend or otherwise challenge existing law in effect on the date of the initiation of the representation.”

Justice Scalia, joined by the Chief Justice, Justice O’Connor, and Justice Thomas, dissented from the majority’s severability analysis. He posed the relevant question succinctly: “whether, without the restriction that the Court today invalidates, [would] the permission for conducting welfare litigation . . . have been accorded.” Id. at 1059. That is, would Congress have allowed any funding for individualized welfare benefits litigation unless the organizations were also prohibited from challenging existing welfare laws? “One determines what Congress would have done,” Justice Scalia explained, “by examining what it did.”

And what did Congress do in 1996? The dissent found that the amendment linked together the “funding of welfare benefits suits and its prohibition on suits challenging or defending the validity of existing law.” These provisions, Justice Scalia concluded, are “conditions, considerations [and] compensations for each other” that cannot be severed.” Id. at 558-59 (quoting Warren v. Mayor and Aldermen of Charlestown, 68 Mass. 84, 99 (1854) (Shaw, C.J.)). Stated differently, the dissent imagines a legislative bargain: Congress would fund individuals who seek to challenge the application of welfare laws, so long as welfare laws themselves were not challenged. “The statute concocted by the Court of Appeals,” he wrote, “bears little resemblance to what Congress enacted, funding without restriction welfare-benefits litigation that Congress funded only under the limitations of § 504(a)(16).” Id. at 559. And so, the majority opinion errs. Under Justice Scalia’s analysis, “[t]o strike the restriction on welfare benefits suits is to void § 504(a)(16) altogether.” Id. at 561. Why? “To remove that limit [on litigation] is to repeal subsection (a)(16) altogether, and thus to eliminate a significant quid pro quo of the legislative compromise.”

Justice Scalia cited Frost v. Corporation Commission of Oklahoma. He noted that the Court has “in some cases stated that when an ‘excepting proviso is found unconstitutional the substantive provisions which it qualifies cannot stand,’ for ‘to hold otherwise would be to extend the scope of the law . . . so as to embrace [situations] which the legislature passing the statute had, by its very terms, expressly excluded.’” Id. at 560 (quoting 278 U.S. at 525). Again, this discussion from Frost considered the first scenario where the same legislature enacted both the “excepting proviso” and the “substantive provision which [the proviso] qualifies.” These were the facts in Velazquez, because the same Congress in 1996 enacted the objectionable, and unobjectionable portions of the statute. Such were not the facts of Frost itself, where legislatures at different points in time enacted the “excepting proviso” and the “substantive provision which [the proviso] qualifies.” For the latter scenario, the Frost Court set aside the proviso itself, and not the original substantive provision.

Under Justice Scalia’s analysis, if a single Congress enacted a statute—part of which is unconstitutional and part of which is constitutional—than that entire statute should be set aside if Congress would not have intended the constitutional part to operate independently. Justice Scalia’s analysis does not speak to the second Frost scenario where a later-in-time Congress took an action that rendered an earlier-in-time statute unconstitutional. That is, the 1996 amendments did not render the original 1974 Legal Services Corporation Act unconstitutional

A “Convergent Constitutional Violation” in Texas

Frost illustrates why Durling and West are, at least in the abstract, correct. The 2010 individual mandate remained constitutional until the 2017 TCJA eliminated the revenue-generating tax that saved its constitutionality. The two provisions cannot coexist. Durling and West write “[w]hen Congress enacts an amendment that renders a broader statutory scheme unconstitutional, the default rule should be to strike down the amendment and restore the law to the pre-amendment status quo.” The key word here is “render.” Neither the mandate nor the tax cut are independently unconstitutional, but the latter renders the former unconstitutional.

Durling and West explain that Congress could have resolved any doubts by drafting the 2017 TCJA differently:

For example, a legislature could write that, if the amendment is unconstitutional, then the rest of the statutory scheme should also be held unconstitutional—what is often called an “inseverability clause.” After all, if Congress includes that language in the statute, then it’s just as much a part of the legislative bargain as the first-order statutory language. And even if the initial amendment is an unconstitutional “nullity,” the fallback provision can itself be a constitutional revision to the law. In these two ways, legislatively adopted fallback rules do not undermine the separation-of-powers and judicial- restraint values that underlie severability doctrine. (p. 11).

When zeroing out the penalty in 2017, Congress did not include an “inseverablity” clause. Indeed, it expressed no concern, whatsoever, that it was about to render another statute unconstitutional, even though “Congress is presumed to act with full awareness of existing judicial interpretations.” United States v. Fausto, 484 U.S. 439, 460 n.6 (1987) (citing Rodriguez v. United States, 480 U.S. 522, 525 (1987) (per curiam)). Yet, Congress passed a statute that flouted the terms of NFIB’s saving construction. (Many members were quite vocal about their intent to repeal the individual mandate, even though they could not do so under the reconciliation process they adopted.)

The New Jersey Attorney General—perhaps after reading Durling and West’s essay—suggested this unconventional remedy to any convergent constitutional violations: “The proper remedy is to strike the amendment that reduced the tax liability to $0 and revert back to the prior tax penalty found constitutional in NFIB.” Brief at 2.

While I agree in the abstract with Durling and West’s proposed default rule, I cannot agree with New Jersey’s argument in this case for a simple reason: Frost considered a challenge to the unconstitutional proviso, while Texas does not. Because of how Texas structured its challenge, the district court is presented with a narrower menu of options with respect to severability. Federal courts do not have a roving license to flip through the U.S. Code with a red pencil to strike out one statute to save another. Invalidating the 2017 tax cut is simply not an option in the Texas litigation, because it has not been challenged.

To understand this dynamic, it is important to carefully parse the relevant statutory framework, and what Texas has, and has not challenged. Section 11081 of the TCJA, titled “elimination of shared responsibility payment for individuals failing to maintain minimum essential coverage,” provides in full:

(a) In General.—Section 5000A(c) is amended—

(1) in paragraph (2)(B)(iii), by striking “2.5 percent” and inserting “Zero percent”, and

(2) in paragraph (3)—

(A) by striking “$695” in subparagraph (A) and inserting “$0”, and

(B) by striking subparagraph (D).

(b) Effective Date.—The amendments made by this section shall apply to months beginning after December 31, 2018.

The following table illustrates how the relevant provisions of Section 5000A(c) of the ACA appeared before the TCJA, and how those provisions were amended by Section 11081 of the TCJA. (I’ve highlighted the changed portions in red).

Section 5000A(c) before TCJA	Section 5000A(c) as Amended by TCJA
(2)(B) Percentage of income An amount equal to the following percentage of the excess of the taxpayer’s household income for the taxable year over the amount of gross income specified in section 6012(a)(1) with respect to the taxpayer for the taxable year: (i) 1.0 percent for taxable years beginning in 2014. (ii) 2.0 percent for taxable years beginning in 2015. (iii) 2.5 percent for taxable years beginning after 2015.	(2)(B) Percentage of income An amount equal to the following percentage of the excess of the taxpayer’s household income for the taxable year over the amount of gross income specified in section 6012(a)(1) with respect to the taxpayer for the taxable year: (i) 1.0 percent for taxable years beginning in 2014. (ii) 2.0 percent for taxable years beginning in 2015. (iii) 2.5 Zero percent for taxable years beginning after 2015.
(3) Applicable dollar amount For purposes of paragraph (1)- (A) In general Except as provided in subparagraphs (B) and (C), the applicable dollar amount is $695.	(3) Applicable dollar amount For purposes of paragraph (1)- (A) In general Except as provided in subparagraphs (B) and (C), the applicable dollar amount is $695 $0.
(D) Indexing of amount In the case of any calendar year beginning after 2016, the applicable dollar amount shall be equal to $695, increased by an amount equal to- (i) $695, multiplied by (ii) the cost-of-living adjustment determined under section 1(f)(3) for the calendar year, determined by substituting “calendar year 2015” for “calendar year 2016” in subparagraph (A)(ii) thereof. If the amount of any increase under clause (i) is not a multiple of $50, such increase shall be rounded to the next lowest multiple of $50.	(D) Indexing of amount In the case of any calendar year beginning after 2016, the applicable dollar amount shall be equal to $695, increased by an amount equal to- (i) $695, multiplied by (ii) the cost-of-living adjustment determined under section 1(f)(3) for the calendar year, determined by substituting “calendar year 2015” for “calendar year 2016” in subparagraph (A)(ii) thereof. If the amount of any increase under clause (i) is not a multiple of $50, such increase shall be rounded to the next lowest multiple of $50.

In the Amended Complaint, Texas prayed for the court to “Declare the ACA, as amended by the Tax Cuts and Jobs Act of 2017, to be unconstitutional either in part or in whole.” Texas challenged the right-hand column of the above table. It did not in any way challenge Section 11081 of the TCJA.

The notation in the U.S. Code illustrates this point:

Pub. L. 115–97, title I, §11081, Dec. 22, 2017, 131 Stat. 2092 , amended subsection (c) of this section, applicable to months beginning after Dec. 31, 2018. After amendment, subsection (c) reads as follows:

After December 31, 2018, Section 5000A(c), standing by itself, will now always calculate a penalty of $0. Section 11081 of the TCJA is not part of the ACA, and is not subject to a lawsuit here.

There is, however, another option that could give rise to the remedy the intervenors seek: if New Jersey (or another similarly situated party) challenged the constitutionality of the tax cut as rendering the individual mandate unconstitutional. In that collateral attack—assuming there is standing to challenge the elimination of a tax burden—a federal court could in fact set aside the TCJA, leaving the mandate in place. After giving this issue some thought, I was frankly surprised that none of the usual suspects have filed suit in a favorable forum, arguing that the tax cut was unconstitutional. Nothing came of threats that blue states would challenge the modification to deductions for state and local taxes (SALT). But a challenge to the elimination of the penalty enforcing the individual mandate could be viable. Such a suit, however, would pose a distinct risk. Conflicting rulings may set up a circuit split, presenting the Supreme Court with a tough choice: increase the taxes for millions of Americans, or invalidate the individual mandate—and potentially other ACA provisions. This risk may explain why a progressive group or a Democratic Attorney General has not challenged the constitutionality of the tax cut.

Durling and West offer a more generalized response to the argument that the Court is unable to set aside Section 11801 of the TCJA. They note “the Court does not confine itself to the plaintiff’s requested relief.” For example, in Session v. Morales-Santana, the Court considered a provision of the Immigration and Nationality Act that imposed a gender-based classification. 137 S.Ct. 1678 (2017). The respondent in this case could only derive citizenship if his unwed father had been physically present in the United States for five years.  However, citizenship could be derived from an unwed mother who was “continuously present in the United States for one year at any point in her life prior to the child’s birth.” Id. at 1698. This “gender-based differential,” Justice Ginsburg found for the Court, violated the “equal protection principle implicit in the Fifth Amendment.” Id. at 1686. The Respondent urged the Justices to level-down, and apply the one-year limit to unwed fathers. While “the preferred rule in the typical case is to extend favorable treatment,” Justice Ginsburg observed, “this is hardly the typical case.” Id. at 1701. As a result, the Court left in place the five-year requirement for unwed fathers, and increased the requirement for unwed mothers from one year to five years. This level-up remedy was proposed by the Solicitor General: “The proper way to cure any equal protection violation would be to apply, on a prospective basis, the longer physical-presence requirements in Section 1401 to children born out of wedlock to U.S.-citizen mothers.” Petitioner Brief at 12. Congress was left with one option: “settle on a uniform prescription that neither favors nor disadvantages any person on the basis of gender.” Id. The Respondent’s victory was pyrrhic.

Without question Durling and West are correct that in cases involving severability, “the Court does not confine itself to the plaintiff’s requested relief.” Indeed, Morales-Santana’s victory was pyrrhic: he could not benefit from the one-year requirement, and, prospectively, neither would other children of unwed mothers. Morales-Santana, however, does not stretch far enough to resolve the question presented in Texas v. United States: if an earlier-in-time statute is rendered unconstitutional by a later-in-time amendment, can the court invalidate the latter, even if the latter was never challenged?

The individual mandate and the tax cut can exist independently, but cannot exist simultaneously. If the tax cut is invalidated, the individual mandate remains constitutional. That is, the right-hand column is eliminated, and we revert back to the left-hand column. Conversely, if the individual mandate is invalidated, the tax cut remains constitutional. Call it Schrödinger’s Mandate. Morales-Santana presented a very different scenario. The one-year requirement for unwed mothers and the five-year requirement for unwed fathers cannot exist independently. The presence of one provision, by itself, amounts to an unconstitutional gender based classification. Justice Ginsburg’s majority opinion determined that the only viable remedy was to put everyone on the same page. Morales-Santana does not address the question presented in Texas.

Congress’s Intent in 2017

Part II of New Jersey’s brief contends that if the court finds that the mandate is no longer constitutional, the correct remedy is to reinstate the prior tax amount. I view this position as its much stronger argument—it is no surprise that the intervenors led off with it. Part III offers an argument in the alternative: even if the individual mandate is unconstitutional, the remainder of the ACA should be severable. The severability analysis, as always, turns on Congress’s intent. But which Congress: the Congress that enacted the ACA in 2010 or the Congress that enacted the TCJA in 2017?

New Jersey argues that what matters is the “intent of the Congress that passed that amendment” in 2017. In a footnote, the brief adds:

Plaintiffs focus exclusively on the intent of the Congress that passed the ACA. But that is the wrong focal point. None of Plaintiffs’ cases involved a statutory provision amended by a subsequent Congress in a manner that purportedly makes the amended provision unconstitutional. Under these circumstances, the intent of the Congress that amended the provision should govern.

The American Medical Association, represented by former Solicitor General Don Verrilli, agreed with New Jersey that the only the intent of Congress in 2017 matters; not the intent of Congress in 2010:

Both the Plaintiffs and the Federal Defendants nonetheless argue that two ACA provisions—the guaranteed-issue and community rating requirements—are inseverable from § 5000A. But their entire analysis turns on a congressional finding from 2010—not 2017. See 42 U.S.C. § 18091(2)(I); Patient Protection and Affordable Care Act, Pub. L. No. 111-148 § 1501 (2010). While it is true that one Congress found that the “guaranteed issue and community rating requirements would not work without” the minimum essential coverage provision, King v. Burwell, 135 S. Ct. 2480, 2487 (2015), a later Congress manifestly disagreed. And that Congress’ intent is the only one that matters. Brief at p. 21 (emphasis in original).

Likewise, five law professors—Jonathan Adler, Nicholas Bagley, Abbe Gluck, Ilya Somin, and Kevin Walsh—argued in an amicus brief that the intent of Congress in 2010 no longer matters:

This time-shifting of congressional intent misapplies severability doctrine. By expressly amending the statute in 2017 and setting the penalty at zero while not making other changes, Congress eliminated any need to examine earlier legislative findings or to theorize about what Congress would have wanted. Congress told us what it wanted through its 2017 legislative actions—“One determines what Congress would have done by examining what it did.” Legal Servs. Corp. v. Velazquez, 531 U.S. 533, 560 (2001) (Scalia, J., dissenting). Whatever Congress may have believed about the connection among these provisions in 2010, the relevant question now is what Congress intended in 2017 when it took the action that provides the basis for plaintiffs’ challenge, i.e., when it reduced the mandate’s penalty to zero. And Congress demonstrated that intent not through mere findings but through amendments to the operative provisions of the ACA. It repealed the penalty while leaving the insurance reforms in place. Cf. Association of Am. Railroads v. Costle, 562 F.2d 1310, 1316 (D.C. Cir. 1977) (“A preamble no doubt contributes to a general understanding of a statute, but it is not an operative part of the statute . . . . The operative provisions of statutes are those which prescribe rights and duties and otherwise declare the legislative will.”).

As an intuitive matter, this argument makes sense, but I have been unable to find any authority to support the proposition that the earlier findings of Congress in 2010 are irrelevant. Texas challenged the ACA as amended (with the penalty dropped to $0). It did not challenge the TCJA. Indeed, the closest case I can find is Frost, which involved–to quote New Jersey’s brief–“a statutory provision amended by a subsequent Congress in a manner that purportedly makes the amended provision unconstitutional.” I welcome submissions for authority that rejects the “time-shifting of congressional intent” approach to severability. There may be something, but I haven’t found it yet.

The law professors’ citation to Justice Scalia’s dissent in Velazquez is not on point. In that case, he was considering the first scenario from Frost where the same legislature enacted the unconstitutional proviso and the otherwise constitutional provision. No “time-shifting” was involved . When Justice Scalia referred to “Congress” in Velazquez, he was referring to a single statute enacted by the same legislators. He did not consider the second scenario in Frost, which mirrors the facts of Texas, in which the unconstitutional proviso was enacted later-in-time. In any event, Justice Scalia’s statute delenda est approach stands for a much broader conception of severability: he would throw out the baby, the bathwater, the bathtub, the bathroom, and even the kitchen sink. (And in hindsight, we know he was willing to jettison the entire ACA in NFIB v. Sebelius.) To the contrary, Frost supports the “time-shifting of congressional intent.”

Congress’s Intent in 2010

If the Supreme Court follows the reasoning in Frost, then the proper remedy is to invalidate the tax cut, and restore Section 5000A(c) to its original form. From a pragmatic perspective, the Supreme Court may not be keen on reimposing a tax on millions of Americans–especially if this case takes some time to trickle upstairs. Confronted with the choice of invalidating the mandate–which has minimal legal effect–and raising taxes, the Court could choose the former. (Of course, the Court could, as it often does, punt the issue by finding there is no standing.)

If the Court does choose to leave the tax cut in place, Frost still informs which intent to follow: the Congress in 2010 or the Congress in 2017. Frost speaks to this specific issue of which legislature’s intent matters.

“Here it is conceded that the statute, before the amendment, was entirely valid. When passed, it expressed the will of the Legislature which enacted it. Without an express repeal, a different Legislature undertook to create an exception, but, since that body sought to express its will by an amendment which, being unconstitutional, is a nullity and, therefore, powerless to work any change in the existing statute, that statute must stand as the only valid expression of the legislative intent.” 278 U.S. at 526-27 (emphasis added).

The “only valid expression of the legislative intent” is the “will of the Legislature” which enacted “the statute, before the amendment.” That is, the Congress that enacted the ACA in 2010. What about the “will” of the “different Legislature” that enacted the amendment in 2017? That “will” was “a nullity and, [is] therefore, powerless to work any change in the existing statute.” The ACA reverts back from the right-hand column to the left-hand column Because of Frost’s bite, the legislative intent of the Congress that enacted the TCJA is not legally irrelevant for purposes of severability.

New Jersey wisely argues in the alternative that “even if it were proper to consider the legislative intent of the 2010 Congress that passed the minimum coverage provision in its original (and fully constitutional) form—and to graft that intent onto a statutory amendment passed by a different Congress—that would still be of no assistance to Plaintiffs.” If Frost is the appropriate rule, then the “will of the Legislature” in 2017 with respect to the zeroing out the penalty is truly a “nullity.” In short, the “expression of legislative intent” in 2017 was “powerless to work any change in the existing statute” from 2010. And that includes Congress’s findings in Section 18091 of the Affordable Care Act.

As I discuss in my first book, Unprecedented, this section was added in order to bolster the government’s defense of the individual mandate against any potential constitutional challenge:

Ultimately, the 2,700-page Affordable Care Act contained three pages of constitutional findings to show that the “requirement to maintain minimum essential coverage” was constitutional. First, “the individual responsibility requirement provided for in this section . . . is commercial and economic in nature, and substantially affects interstate commerce.” Second, the findings listed a number of “effects on the national and interstate commerce” that resulted from uninsured people shifting costs. Third, the “findings” stated that “the requirement regulates activity that is commercial and economic in nature: economic and financial decisions about how and when health care is paid for, and when health insurance is purchased [emphasis added].” The word “activity” would prove decisive. Fourth, “in United States v. South-Eastern Underwriters Association, the Supreme Court of the United States ruled that insurance is interstate commerce subject to Federal regulation.”

(Ironically enough, the taxing power was not mentioned anywhere in the findings, which were put together by the Constitutional Accountability Center.) There was a fifth finding which I did not mention in Unprecedented that has now become relevant once again. Section 18091(2)(I) provides in part:

if there were no [individual responsibility] requirement, many individuals would wait to purchase health insurance until they needed care. By significantly increasing health insurance coverage, the requirement, together with the other provisions of this Act, will minimize this adverse selection and broaden the health insurance risk pool to include healthy individuals, which will lower health insurance premiums. The requirement is essential to creating effective health insurance markets in which improved health insurance products that are guaranteed issue and do not exclude coverage of pre-existing conditions can be sold. (emphasis added).

The emphasized portion refers to the ACA’s guaranteed issue and community rating provisions which require (to grossly oversimplify) that insurers do not charge more for, or deny coverage to, customers with pre-existing conditions. This finding played an important role in the Obamacare litigation. For example, Chief Justice Roberts cited Section 18091(2)(I) in NFIB v. Sebelius:

The individual mandate’s regulation of the uninsured as a class is, in fact, particularly divorced from any link to existing commercial activity. The mandate primarily affects healthy, often young adults who are less likely to need significant health care and have other priorities for spending their money. It is precisely because these individuals, as an actuarial class, incur relatively low health care costs that the mandate helps counter the effect of forcing insurance companies to cover others who impose greater costs than their premiums are allowed to reflect. See 42 U.S.C. § 18091(2)(I) (recognizing that the mandate would “broaden the health insurance risk pool to include healthy individuals, which will lower health insurance premiums”). If the individual mandate is targeted at a class, it is a class whose commercial inactivity rather than activity is its defining feature. 132 S.Ct. at 2590.

As I noted in Part I of this series, the TCJA did not repeal, expressly or impliedly, this statutory finding. DOJ made a similar argument in its brief: “Those findings cannot be deemed to have been impliedly repealed by Congress’s mere elimination of the financial penalty.” Brief at 15-16. Indeed, because Congress enacted the tax cut through the budget reconciliation process, under its “rules of proceeding”—to which courts ought to defer—the other provisions of the ACA, including the findings, could not be amended. The question, then, is what is the relevance of these findings?

New Jersey counters that these findings only concern Congress’s constitutional authority to enact the mandate, and do not implicate severability:

For starters, these congressional findings were designed to show that the requirement to maintain minimum essential coverage “is commercial and economic in nature, and substantially affects interstate commerce . . .” 42 U.S.C. § 18091(1) (emphasis added). In other words, these findings were drafted to demonstrate that Congress had constitutional authority under the Commerce Clause to require that most Americans purchase health insurance. Id. They do not reflect Congress’s judgment as to whether the community-rating and guaranteed-issue provisions should cease to exist if the minimum coverage requirement were invalidated. And in light of NFIB—which held that Congress lacked authority under the Commerce Clause to require individuals to purchase insurance—these congressional findings are no longer relevant to the constitutional analysis for which they were crafted. Brief at 37.

New Jersey is correct that Congress did not include these findings in a section about severability. Rather, they were expressly added to bolster the constitutional defense of the mandate. This fact did not trouble then-Solicitor General Verrilli, who relied on the findings Section 18091 in his severability brief:

The question of severability is one of congressional intent, and Congress expressly found that the minimum coverage provision is “essential” to the guaranteed-issue reforms. 42 U.S.C.A. 18091(a)(2)(I). Brief at p. 52.

Verrilli made the point more directly in his reply brief:

In this case, however, it is evident that Congress’s finding on the “essential” connection between the provisions, 42 U.S.C.A. 18091(a)(2)(I), “though directed at the antecedent constitutional question, can also be read to answer the severability question.” Court-Appointed Amicus Br. 33. This is so because the finding rested on evidence showing that, unless paired with a minimum coverage provision, the guaranteed-issue and community-rating provisions would actually undercut Congress’s goals because they would cause premiums to rise and coverage to decline. As both a logical and practical matter, therefore, Congress’s finding on the “essential” role of the minimum coverage provision in effectuating the guaranteed-issue and community-rating provisions effectively serves as an inseverability clause—albeit one limited to only those two provisions, given that Congress did not find the minimum coverage provision to be “essential” to any other part of the Act. Brief at p. 10.

Verrilli had the better argument in his NFIB brief. The findings in Section 18091, though directed at the constitutional question, also serve as reliable indicia of Congress’s intent in 2010 concerning the relationship between the individual mandate, and the guaranteed issue and community rating provisions.

Finally, New Jersey argues that Congress’s concerns in Section 18091(a)(2)(I) should be discounted because they have not come to fruition:

Any concern about adverse selection is not well founded in 2018. First, as Congress stated at the time, the three-prong approach that it adopted was intended to assist in “creating effective health insurance markets. . .” 42 U.S.C. § 18091(2)(I) (emphasis added). Congress was attempting to create brand new insurance markets from scratch, a major undertaking that involved tremendous uncertainty. But those markets were successfully created years ago, and even Plaintiffs do not assert that the minimum coverage provision is essential to maintaining those already-created health insurance markets. In fact, Plaintiffs themselves acknowledge that the “death spiral” scenario is far-fetched when they cite a 2017 CBO report about the effect of eliminating the shared responsibility payment. CBO found that repealing the minimum coverage requirement would cause average premiums in the nongroup market to rise by about 10%, but that “nongroup insurance markets would continue to be stable in almost all areas of the country throughout the coming decade.” Brief at 40.

To borrow a phrase from the law professors’ brief, “[t]his time-shifting of congressional intent misapplies severability doctrine.” Hindsight is always 20/20. Of course, if Congress knew in 2010 what it knows today about health care markets, it would have acted in a very different fashion. Congress may not even have bothered to enact the individual mandate, which proved to not be nearly as essential as was thought. But it did not know today what it knows now. The relevant question is, what would Congress in 2010 have intended to happen if the individual mandate was invalidated. The findings in Section 18091(a)(2)(I) provide a strong indication of that intent. Indeed, these findings may be the best evidence we have of that intent, as they are in the actual text of the statute, and not in the legislative history.

The Attorney General’s Severability Decision

In Part II of this series, I discussed the Attorney General’s decision not to defend the constitutionality of the individual mandate. As a matter of first principles, I generally agree with Sai Prakash and Neal Devins that the President, through his Attorney General, has no duty to defend laws he has determined are unconstitutional—even in the absence of a judicial declaration. (Michael McConnell takes the opposite position on this question.) With respect to the individual mandate, where the Attorney General has determined that the saving construction no longer applies, and the mandate is now unconstitutional, his decision is defensible.

A different analysis applies to his opinion concerning severability, which the Attorney General discussed in the following paragraph:

In their lawsuit the plaintiffs further argue that Section 5000A(a) [the individual mandate] is also inseverable from the rest of the ACA, and therefore that the statute and all of its implementing regulations should be invalidated. In NFIB, the Department previously argued that if Section 5000A(a) is unconstitutional, it is severable from the ACA’s other provisions, except those “guarantee[ing] issuance of coverage in the individual and group market” (“guaranteed issue”), 42 U.S.C. 300gg-1, 300gg-3, 300gg-4(a), and “prohibiting discriminatory premium rates” (“community rating”), id. 300gg(a)(I), 300gg-4(b). I concur in the Department’s prior determination. Post-Jobs Act, Congress’s express findings in the ACA continue to describe Section 5000A(a) as “essential” to the operation of the guaranteed-issue and community-rating provisions. because otherwise individuals could wait until they become sick to purchase insurance, thus driving up premiums for everyone else. See 42 U.S.C. 18091(2)(1). This question of statutory interpretation does not involve the ACA’s constitutionality and therefore does not implicate the Department’s general practice of defending the constitutionality of federal law. Outside of these two provisions of the ACA, the Department will continue to argue that Section S000A(a) is severable from the remaining provisions of the ACA. (emphasis added).

As the emphasized portion explains, his decision concerning severability does not implicate the Department’s obligation to defend the constitutionality of laws. (This point has been lost in most commentary about the Attorney General’s decision.) The standards articulated in 28 U.S.C. § 530D(a)(B)(ii) are simply irrelevant with respect to decisions concerning severability. For example, in 2011, the Holder Justice Department argued that if the individual mandate was invalidated, the Court should also set aside the guaranteed issue and community rating provisions. See Fla. ex rel. Bondi v. U.S. Dep’t of Health & Human Servs., 780 F. Supp. 2d 1256, 1299 (N.D. Fla. 2011) (“I note that the defendants have acknowledged that the individual mandate and the Act’s health insurance reforms, including the guaranteed issue and community rating, will rise or fall together as these reforms “cannot be severed from the [individual mandate].”). There was no need for DOJ to notify Congress about this decision, even though it is tantamount to a decision not to defend the constitutionality of other provisions. Accordingly, Bartow Farr-the Court appointed amicus—explained that “Petitioners and the United States are asking this Court to invalidate perfectly lawful provisions of a federal statute.” That is, the guaranteed issue and community rating provisions. Brief at 2.

Precisely because this decision does not implicate the Attorney General’s independent constitutional judgment, he has a much, much stronger obligation to defend Congress’s handiwork. As my analysis in this post illustrates, the severability doctrine is extremely murky. It is not well-settled, and much of the resolution may turn on how the Court interprets a 90-year old decision by Justice Sutherland. There is more than enough ground for the Attorney General to make arguments along the lines that New Jersey advanced. With respect to severability, Sessions’s decision is not defensible.

I pause to note briefly one related issue that results from the Attorney General’s decision. The District Court found that New Jersey satisfied the requirements of Rule 24(b), and could intervene. The order, however, did not address whether New Jersey has Article III standing to intervene in light of Windsor. Though the Attorney General submitted a 530D letter, there has been no indication that Congress intends to authorize counsel to defend the mandate, such as through the Bipartisan Legal Advisory Group (BLAG). Windsor found that the BLAG had standing to defend DOMA, but there is no indication that a state attorney general has such authority. This issue has not been raised in the District Court, as far as I can tell—Windsor is not cited in New Jersey’s motion to intervene. Because jurisdiction can be raised at any time, it should be addressed in future pleadings.

United States v. Ardoin

I will close this already-too-long post with a brief rejoinder to an argument raised by the AMA brief: “Plaintiffs ignore binding Fifth Circuit precedent that expressly rejected Plaintiffs’ contention that a law cannot be upheld as an exercise of Congress’s tax power unless it actually raises revenue. United States v. Ardoin, 19 F.3d 177 (5th Cir. 1994).” The Supreme Court already held that Congress did not exercise its taxing power in enacting Section 5000A. The Chief Justice only found that the exaction could be construed as a tax for purposes of the saving construction. If the saving construction no longer applies, then Section 5000A cannot be supported by Congress’s taxing powers. Rather, Congress can only fall back on its Commerce and Necessary and Proper Clause powers to justify the mandate. Five Justices already held that Congress lacks the power to impose the mandate through these provisions. The argument based on Ardoin is foreclosed by NFIB.

Conclusion

In this third installment, I raised many questions about the state of severability doctrine, and only answered some of them. Why? Because I haven’t made up my mind yet. This case presents a lot of really, really difficult issues that have been dismissed far too easily. My goal here is not to persuade anyone how the case should turn out. I’m still thinking it through. Rather, my purpose has been to raise issues that have not yet been addressed in the briefing, and hopefully will be addressed in the future.