Article

Beyond Severability

I.     Introduction

In one of the most consequential cases heard by the Roberts Court—National Federation of Independent Business v. Sebelius1—four justices in dissent attacked not only the specific outcome reached by the majority, but the very approach those justices took to their judicial role. Accusing the majority of “vast judicial overreaching,”2 the four argued that the Court had more than simply erred; it had engaged in “judicial usurpation.”3 What, according to the dissent, was the nature of this judicial crime? It was construing the Affordable Care Act in a manner that permitted most of the statute to continue in operation. More specifically, it was the majority’s application of the so-called “severability” doctrine—a framework for analysis requiring a court to disregard an unconstitutional statute in whole or in part—that triggered the impassioned attack. What, then, was the dissent’s preferred, more judicially restrained alternative? It was to strike down the entire Act. According to the dissent, complete invalidation of the Affordable Care Act was the only response consistent with the values that should guide a court engaging with a congressionally enacted statute: “caution,” “minimalism,” and “judicial modesty.”4

The tension is startling. On the one hand are calls for judicial restraint and modesty; on the other, a willingness to reach conclusions about statutes that destroy their operation. Despite the inherent friction in the dissent’s position, attempts by the majority in National Federation to defend its opinion against these attacks fell flat—or, at least, they lacked the rhetorical power churning through the dissent. This was not due to oversight or neglect by the five in the majority. It turns out it is surprisingly difficult to explain why taking a more flexible approach to a constitutionally defective statute might not be, in the words of the dissent, “a more extreme exercise of the judicial power than striking the whole statute.”5 It is similarly difficult to justify why a court, when engaging in a more accommodating form of severability analysis, is not impermissibly “impos[ing] on the Nation, by the Court’s decree, its own new statutory regime, consisting of policies, risks, and duties that Congress did not enact.”6 To the contrary, as soon as one accepts that the “severability” framework controls, it proves extraordinarily difficult to resist accusations like those lodged by the National Federation dissent. Understanding the fundamental error of the criticism—and of the National Federation dissent—therefore requires taking a step back: asking whether “severability” should apply at all.

This Article embraces this overlooked question. And it reaches what may seem like a radical conclusion. Notwithstanding the ranks of courts and scholars who reflexively accept severability as a framework for construing constitutionally flawed statutes,7 this Article concludes, quite to the contrary, that severability should lose its prominent place in the doctrine. It is a deeply flawed framework for analysis that, in the last three years alone, has imperiled the Voting Rights Act,8 the Bankruptcy Code,9 and the Affordable Care Act,10 to name but a few examples. Even when taken on its own terms, the severability framework cannot justify what it does to statutes.

This Article is the first to reach this result. While elaborate and impassioned scholarly debates continue to unfold over severability’s specifics—how exactly the doctrine should be articulated and applied11—near silence reigns with respect to whether severability’s fundamental framework is appropriate for the problems that it purports to resolve. Instead, courts and scholars take most of severability for granted. They accept, nearly without fail, that courts should turn to “severability” when grappling with constitutionally defective statutes,12 that severability’s analytical framework requires courts to disregard statutes either in whole or in part,13 and that legislative intent must play a central role in the courts’ conclusions.14 Courts and scholars also tend to agree on what motivates severability’s framework: the often stated but rarely defined principle of judicial restraint.15

These points of consensus over severability’s general contours mask its deep and consequential flaws. Seemingly innocuous in its basic operation, severability in actuality limits courts to an artificial and highly restrictive inquiry: whether a court should disregard an unconstitutional statute in whole or in part. Once triggered, it does not permit any other judicial response to what this Article terms “constitutionally disrupted statutes”: statutes that, due to some conflict with the Constitution, cannot operate as enacted. More specifically, severability forbids courts from employing a range of responses that are permissible in other legal contexts. These include construing a statute in a way that expands the reach of its applications,16 disregarding, or otherwise altering, a portion of the statute that is not considered to be the most immediate source of the unconstitutionality,17 and construing the statute in a nonstandard manner.18

Emerging trends in severability seek to narrow these options even further. These developments, which signal a newer and more severe form of severability, prohibit a court from construing a constitutionally disrupted statute in any manner that affords the court too much discretion.19 They similarly resist the use of so-called “application severability,” a well-established and previously uncontroversial approach to statutory construction that works by invalidating a statute’s unconstitutional applications rather than portions of its text.20 Severability has long accommodated these two methods. Yet objections to both appear to be growing increasingly prominent, and this newer form of severability threatens massive interference with the work of the legislature. Such interference will only compound severability’s already destructive effects.21

This Article identifies and describes these limitations in an effort to demonstrate how severability drastically restricts what a court can do in its efforts to “save” constitutionally disrupted legislation. It reveals, moreover, that these restrictions have their effects at the expense of the democratic will. Despite the near consensus over the central role that legislative intent plays in severability analysis, legislative intent does not overcome severability’s restrictions. The restrictions exist even if Congress would have preferred otherwise.22 What this means is that severability, which claims to effectuate legislative intent, actually contravenes it. Its framework limits the analytical approaches a court can take, and it does so irrespective of legislative intent.

Severability’s aggressively restrictive framework also creates unsettling discrepancies in the case law. Though courts routinely employ certain analytical approaches in other legal contexts,23 these same analytical approaches are deemed, in the context of severability, to be more than merely disfavored. They are regarded as beyond the judicial function and an encroachment on the legislative sphere.24

Despite severability’s many quandaries, reflexive acceptance remains the norm. Questioning severability’s basic framework begins to expose these restrictions and the contradictions they represent. It confirms, moreover, that the stakes are high. Severability’s restrictive framework potentially affects any exercise of judicial review,25 and among its effects are doctrinal confusion, a refusal to effectuate legislative intent, and the invalidation of wide swaths of Congress’s work.26 These costs are both abstract and practical: without severability’s restrictions, the statutory landscape across a range of subjects would look very different. In light of these effects, this Article demands that severability’s destructive effects be justified—and it concludes that they cannot be.27

All this analysis leads to the same, inexorable result, which is the need for an improved regime. Severability, at its core, does not “limit the solution to the problem,”28 as it promises to do; it simply limits the solutions. An improved regime would return to one of severability’s animating principles: that courts confronting a constitutional flaw in a statute should respond by minimizing the disruption to that statute. This Article argues that legislative intent is what properly drives such analysis. By complying with unencumbered legislative intent, rather than the restrictive framework of severability, courts would interfere less drastically with the statutory code and, as such, more successfully respect the legislative will. The broader approach might, for example, have saved much of the Voting Rights Act in Shelby County,29 it might have produced a less disruptive result in National Federation (while protecting the Affordable Care Act against the dissent’s insistence that the statute be even more drastically dismantled),30 and it almost certainly would have obviated the attacks being lodged at the bankruptcy code in the follow-up cases to Stern v. Marshall.31 These three examples are quite recent and high-profile, but they are not anomalous. To the contrary, the adjustment potentially would affect any statute subject to constitutional challenge.

In making the case that the severability framework should be retired and replaced, this Article proceeds in three parts. Part I situates the severability framework in contemporary case law and the scholarly literature. In so doing, it reveals that this framework tends to enjoy the unquestioning acceptance of both courts and scholars, who, at best, cite vague and undertheorized principles of judicial restraint in support. Part II explores the constraining effect that severability has on the courts and the statutes they are charged with construing. On the judicial branch, severability imposes a restrictive structure that prohibits courts from relying on a host of otherwise permissible approaches to statutory construction. On the legislature’s work, it imposes a disruptive regime that generates doctrinal confusion and stifles legislative intent. Part III insists that this aggressive regime justify its existence and effects. Concluding that it cannot, this Article closes with a proposal for reform. The proposed regime would ensure greater fidelity to legislative intent by removing severability’s artificial constraints and, in so doing, avoid its destructive effects.

II.     Severability as a Blithely Accepted Tool

Severability is controversial on the margins. Debates rage over how it should be articulated and applied.32 Yet as an overarching framework for analysis, severability enjoys a rare and enviable position: courts and scholars tend to accept it reflexively.33 They claim to do so to fulfill a commitment to judicial restraint—a commitment that, in this context, is as vaguely defined as it is frequently invoked.34

This Part identifies and describes these related phenomena. It begins with a brief description of the doctrine itself. At its core, severability is a response to a difficult problem that arises when a statute is deemed to be, in some respect, unconstitutional. The question is how to apply that statute going forward. Though “hardly a model of clarity,”35 current doctrine nevertheless can be coaxed into providing a working definition: severability is a measure of whether, in response to a statute’s constitutional defect, the legislature would have intended for the court to disregard only the statute’s unconstitutional portions (in which case, the statute is “severable”), or for the court to disregard the entire statute (in which case, the statute is “inseverable”).36 Stated at this level of generality, this definition is consistent with virtually every leading case on severability.37

It is important to recognize the centrality of legislative intent to this well-established definition.38 Although the test for severability is a moving target, with each of the cases on severability tending to “formulate the test a little bit differently,”39 legislative intent remains a constant; it is the “touchstone.”40 To illustrate just how deeply engrained this principle is, all of the members of the current Supreme Court—as well as the recently deceased Justice Scalia, who was among the most insistent on a textualist approach to statutory interpretation and the most resistant to broad inquiries into legislative intent—repeatedly have acknowledged the central role that legislative intent plays in severability analysis.41

Outside of these general principles, the standards for severability fuel a heated debate.42 Still, these general principles—relating to severability’s basic definition and its central reliance on legislative intent—provide a framework for severability that runs consistently throughout both the case law and the literature.43 Despite such ubiquity, few challenge the privileged status that this framework enjoys. Few question whether it is correct.

A.     The Unquestioning Acceptance of the Severability Framework by Courts and Scholars

The severability framework, as defined above, enjoys widespread acceptance. Courts tend to turn to it without hesitation or deliberation when deciding how a statute should apply after a ruling of unconstitutionality.44 In the landmark 2012 health care decision, for example, the Supreme Court granted certiorari specifically to address “the issue of severability” in light of the constitutional challenges to the ACA, and it scheduled a separate hour of oral argument to address the matter.45 In United States v. Booker,46 the Court granted certiorari to hear two questions: first, whether the sentencing regime was unconstitutional and, second, “[i]f the answer to the first question is ‘yes,’” how the Sentencing Guidelines should apply “as a matter of severability analysis.”47 Despite the asserted centrality of severability to both cases, in neither case did the Court order briefing on the predicate question of whether the severability framework should apply in the first place.

Severability also has dominated court discussions even when not raised or briefed by the parties. Recently, when the Court heard oral arguments in an important follow-up case to Stern v. Marshall,48 Justice Kagan interrupted a discussion of the proper interpretation of the Bankruptcy Code to ask if what was facing the Court was “really a problem of severability.”49 Justice Kagan’s question quickly became a “focus” of the argument.50 In Shelby County v. Holder,51 the case declaring portions of the Voting Rights Act of 1965 to be unconstitutional, Justice Ginsburg dedicated a critical portion of her dissent to the argument that a different outcome had been required pursuant to principles of severability—even though neither the majority nor the parties appear to have raised the issue.52 Several Supreme Court justices have gone so far as to attribute “the absence of explicit severability reasoning” in prior cases not to reliance on some alternate framework, but rather to “implicit severability analysis.”53

What courts have not done, however, is provide an explanation for why they so routinely resort to severability. Instead, when faced with a constitutionally disrupted statute, courts tend to apply severability without discussion, much less consideration, of other options.54

Scholars have followed suit. Characterizing the issue as “the problem of what to do with partially unconstitutional laws,” one commentator has acknowledged that, according to both “[c]urrent law and scholarship . . . severability doctrine is the exclusive way to deal with partial unconstitutionality.”55 Similarly, in the course of arguing that “no workable system of judicial review could function without a large role for severability,” Michael Dorf has referred to “the role that severability plays in the background of every successful constitutional challenge.”56

This is not to say that scholars have responded uncritically to each of the innumerable articulations and applications of the severability doctrine. To the contrary, “established doctrine on the severability of unconstitutional statutory provisions has drawn criticism on almost every conceivable basis.”57 A sophisticated collection of works also explores related themes: the ways in which severability interacts with other doctrines.58 What none of these works does, however, is question the underlying assumption that the severability framework should be what guides a court’s analysis when it is engaging with a constitutionally disrupted statute.

Still, on rare occasion, a scholar will push back on the dominant view and question whether the severability framework should play such a prominent role.59 Kevin Walsh, for example, has advocated for the return of what he considers to be “the original approach to partial unconstitutionality,” which he identifies as “displacement without . . . fallback law.”60 While Professor Walsh’s proposal represents a significant conceptual shift in how courts might interact with constitutionally disrupted statutes, as a practical matter it tends to accept the doctrinal status quo and therefore fails to address the question of whether a severability-type framework should be a default response to constitutionally disrupted statutes.61 Eric Fish, by contrast, has resisted the notion—so prominent in the doctrine and the literature—that severability should play such a dominant role. Instead, Fish treats severability analysis as something triggered in quite limited circumstances: namely, once a court already has concluded it should disregard a statute in part and is simply trying to determine what else, if anything, to disregard.62 As a normative matter, Fish’s characterization makes good sense.63 Yet courts often engage in severability analysis without first determining whether its rigid framework of invalidation is the appropriate treatment of the statutes before them.64 As a result, Fish’s works explore a different question than does this Article, which asks whether the severability doctrine can justify the predicate conclusion implicitly reached by any court engaging in severability analysis. More specifically, this Article asks whether it is proper for such a court to have concluded that the offending statute should indeed be invalidated in some respect—rather than treated in some other manner.

Throughout these discussions, a difficult question arises: is severability best understood as a species of statutory construction, or as a remedy?65 This debate, while important, does not need resolution in this Article. Whether operating as construction or remedy, severability is plagued by the same fundamental flaws. To the extent this Article assumes that severability is a matter of statutory construction, this is merely for purposes of explication; the same conclusions adhere if severability is considered to be a remedy.66 However characterized, what severability does is impose a restrictive framework on courts that cannot be justified—and therefore should be replaced.

B.     The Appeal to Judicial Restraint that Courts and Scholars Offer in Support

The dominance of the severability framework in both the case law and the literature might lead one to assume a sophisticated set of theories or justifications supporting its use. If anything, however, the opposite is true.67 It is nevertheless possible to identify a dominant theme purporting to justify severability’s operation, and that is the theme of “judicial restraint.”68

This theme encompasses two more specific expressions in the context of severability. The first is that severability minimizes the statutory disruption caused by a court’s constitutional ruling. The Supreme Court has referred to this principle as helping the court “not to nullify more of a legislature’s work than is necessary.”69 Justice Stevens has echoed the sentiment in describing severability as “limit[ing] judicial power by minimizing the damage done to the statute by judicial fiat.”70

An example of this expression of judicial restraint emerged in Ayotte, a case on Supreme Court review after the Court of Appeals had concluded that an abortion regulation lacked a constitutionally required health exception.71 According to the Supreme Court, the Court of Appeals had erred by invalidating the state regulation “in its entirety” without first analyzing legislative intent and declaring the statute to be either severable or inseverable.72 The Court, in other words, insisted that the Court of Appeals employ the dominant severability framework. In justifying this insistence, the Court provided what soon would become an oft-cited articulation of why such an approach was necessary: it helped “to limit the solution to the problem.”73

The second theory underlying the judicial-restraint justification is that the severability framework, properly understood, helps a court to avoid “judicial legislation.”74 Implicit in this theory is the idea that there is a line separating certain judicial responses to a statute (“judicial legislation”) from other forms (for example, “judicial interpretation” or “judicially fashioned remed[ies]”75), and that only the latter set of responses is permissible.

An example of what nearly every court and scholar likely would deem “judicial legislation” might be if the Supreme Court, in response to constitutional challenges to the Affordable Care Act, completely rewrote the statute—which is based integrally on the participation of private insurance companies—to transform it into a single-payer scheme run entirely by the federal government. This would require a drastic and exhaustive overhaul of a complex and interrelated statute that runs hundreds of thousands of words long,76 and which, to this author’s knowledge, nowhere contains an indication that Congress would have preferred a court-created single-payer system over the invalidation of the ACA as written.77 An example, by contrast, of what nearly every court and scholar likely would deem “judicial interpretation” (or “construction”78) as opposed to judicial legislation would be a straightforward, uncontroversial conclusion regarding how one of the ACA’s many provisions applies to the particular facts of a routine dispute among parties.

Despite the near ubiquity of this theory of judicial restraint—that is, the theory that severability helps to separate “judicial legislation” from the routine interpretative work of a court—it suffers from a profound degree of imprecision. In large part this is because it is not clear what, exactly, distinguishes “judicial interpretation” (or construction) from “judicial legislation.”79 Stated otherwise, how does one draw the line between the single-payer example cited above, on the one hand, and the routine work of a court, on the other? The lack of precision may relate to the level of generality adopted in the discussions. While each invocation of this theory of judicial restraint seems to provide a slightly different characterization of the relevant line, nearly all are stated in very general terms. Usually, the rhetoric involves references to text, as courts are accused in a conclusory way of “tamper[ing] with the text,”80 “supplement[ing] the text,”81 “drafting legislation,”82 or “rewriting” statutes.”83 These are, of course, all metaphors: in the American system of judicial review, the courts have no power to change statutory text. In response to any judicial ruling, the text remains in the Code, exactly the same as before.

Despite the vagueness characterizing these discussions, courts and scholars have offered at least a few attempts, in the context of severability, at more precisely drawing the line between impermissible and permissible judicial conduct. These attempts tend to adopt one of following three substantive approaches, which this Article terms the “Blue Pencil Test,” the “Complexity Test,” and the “Primary Drafter Test.” Given the importance of these tests in both justifying and defining the severability framework,84 each is described below.

The Blue Pencil Test. The first of these tests—that is, of those purporting to provide a guide for determining which of a court’s severability-related actions are “legislative” versus “judicial” in nature—will be referred to in this Article as the “Blue Pencil Test.” Though an implicit version of this test appears occasionally in discussions of severability, it originally derives from contract law. In the latter context, it is a judicial standard that permits offending words in a contract to be “invalidated” if, and only if, “it would be possible to delete [those words] simply by running a blue pencil through them.”85 This standard deems any other approach to the contractual language—including “changing, adding, or rearranging words”—to be impermissible.86 Applied in the context of a constitutionally disrupted statute, this test has intuitive appeal. Even the expression “severability” (derived from “to sever”) is consistent with this idea, as the verb tends to invokes images of merely excising parts of a thing, rather than otherwise altering that thing in any way. Under the Blue Pencil Test, if the court were to do anything other than disregard certain words from a statute, then it would not be engaging in severability. In the words of the joint dissent in National Federation, it instead would be construing the statute “to say what it does not say,”87 and “[s]uch a judicial power would not be called the doctrine of severability but perhaps the doctrine of amendatory invalidation.”88 This is one of many examples of the Blue Pencil Test appearing in substance, though not necessarily by name, in the context of constitutionally disrupted statutes.

The Complexity Test. A second attempt at drawing this line—between judicial responses to constitutionally defective statutes that are either legislative or judicial in nature—may be referred to as the “Complexity Test.” This is because it suggests that the line in question tracks the complexity or creativity of the relevant analysis. To this end, a court will be accused of impermissibly acting as a legislature when it responds to a ruling of unconstitutionality through “line-drawing [that] is inherently complex,”89 or when it construes a statute in a manner that is perceived as “creative,”90 “imaginative,” 91 or “novel.”92 By contrast, a court tends to avoid such criticism when its severability analysis can be considered a “‘relatively simple matter’”93: when the court, for example, is merely “invalidat[ing] an application of a statute”94 rather than engaging in an “atextual” application.95 It is an understanding of the judicial function that resists complexity, creativity, or nuance.96

This approach appears, in one form or another, with some frequency. It is, for example, consistent with the rhetoric the Chief Justice employed in Free Enterprise Fund v. Public Company Oversight Board when he objected to what he perceived as a court’s “editorial freedom.”97 Justice Scalia also appeared to champion this idea at the National Federation oral argument when he suggested the Court should adopt the approach because severability “reduces our options the most.”98 In Ayotte, a unanimous court seemed to endorse this idea when it asserted that the Court’s “ability to devise a judicial remedy that does not entail quintessentially legislative work often depends on how clearly we have already articulated the background constitutional rules at issue and how easily we can articulate the remedy.”99

It is true that this test suffers from a lack of precision and, more fundamentally, a lack of theorization.100 It is also possible that this inquiry into complexity means to be a proxy for another concern,101 though its role as a proxy does not appear to have been expressly articulated. It nevertheless plays a recurring role in discussions of severability.

The Primary Drafter Test. A third attempt at drawing the line between the judicial and the legislative spheres looks to the degree to which a court, rather than Congress, is determining a constitutionally disrupted statute’s meaning or scope. Professor Fallon has offered one of the most lucid articulations of this idea. As he has explained, under this theory, “a federal court may sever a statute, or impose a limiting construction on one,” only “if the particular subrules that a court would need to specify to ‘save’ part of a statute would . . . sufficiently reflect the structure and history of the statute to be attributed to Congress, rather than the court.”102

This Article has termed this test the “Primary Drafter Test” because it seeks to differentiate between, on the one hand, statutes that “the legislature actually adopted,”103 even if they reflect some relatively minor judicial modification, and on the other, statutes so severely affected by judicially modification that, in effect, “Congress did not write” them.104

Although the Primary Drafter Test remains “vague,” as Professor Fallon has acknowledged,105 it is nevertheless more analytically sophisticated than the other two tests. Indeed, it may even help to explain the other two tests. In other words, it is possible that either or both of the Blue Pencil Test and the Complexity Test are meant to serve as proxies for the Primary Drafter Test—that is, as proxies for ensuring that Congress remains the primary drafter of a given statute.

As the latter observation indicates, the three identified tests are not exclusive, but rather may overlap or be considered together. One might conclude, for example, that a court engaging in severability avoids judicial legislation only if it satisfies both the Complexity Test and the Primary Drafter Test. Moreover, these three tests are not the only possible tests.106 They nevertheless do appear to be the most prevalent.

Whatever the particular taxonomy, articulation of these tests helps to explain why severability has garnered such widespread acceptance and use. It at least purports to ensure that the courts limit their activity to judicial work, thereby avoiding encroachment on “the legislative domain.”107 Yet, as this Article will explain, severability’s promise of judicial restraint cannot withstand closer analysis. The doctrine not only fails to accomplish its own intended purposes; it undermines them.108

 III.     Severability as a Destructive Force

Severability purports to protect statutes. Yet as it is employed it undermines them. As this Part explains, the severability framework prohibits courts from employing a range of otherwise permissible analytical approaches when they are responding to constitutionally disrupted statutes. These restrictions directly interfere with the effectuation of legislative intent. Despite the widespread acceptance of the severability framework, neither courts nor scholars have acknowledged this destructive dynamic, much less explained how it can be justified. Nor have either courts or scholars addressed a related analytical concern: why the analytical approaches in question are impermissible in the context of severability but permissible in other areas of the law.

The troubling effects that emerge from this regime include doctrinal confusion, disregard of legislative intent, and a sweeping invalidation of swaths of the U.S. Code. Moreover, severability may be headed down an even more destructive path: toward a stricter form of severability, with effects that are even more disruptive of Congress’s work.

A.     Legislative Intent: An Unacknowledged Casualty of Severability’s Restrictive Framework

Neither courts nor scholars have acknowledged severability’s subtly destructive effect on legislation. More specifically, the severability framework responds to the problem of constitutionally disrupted statutes by limiting courts’ options, which in turn undermines their ability to follow legislative intent.

These limitations arise because “severability,” as generally understood, sets forth a restrictive framework that provides a court with only two options: invalidate a constitutionally disrupted statute’s unconstitutional portions, or invalidate that statute in full. So defined, this framework does not permit a court to employ such alternative responses as: (1) construing a constitutionally disrupted statute in a way that expands its reach;
(2) disregarding (or otherwise altering) a portion of the statute that is not considered to be the most immediate source of the unconstitutionality; or
(3) treating some portion of the statute as having a nonstandard meaning.

And this is not all. Recently, jurists and commentators have set forth an even more severe form of severability, one that imposes even more drastic limitations on courts. These limitations forbid courts from: (4) construing or applying a statute in a way that gives the court too much “editorial freedom”; or (5) engaging in so-called “application severability,” a well-established form of severability that invalidates certain statutory applications in a manner that does not track the text.109 Each of these five approaches is prohibited (or, at least, threatened) once a court decides to proceed through the framework of severability. To illustrate how this approach restricts courts’ ability to effectuate legislative intent—that is, to illustrate how severability narrows the courts’ options, even when legislative intent directs otherwise—this Part explores each of the five imperiled approaches in turn.

1.     Construing a statute in a way that expands its reach

Severability, as commonly understood, does not permit a court to treat a constitutionally disrupted statute as having an expanded reach, even if that approach would cure the unconstitutionality. Imagine, for example, a statute that is unconstitutionally underinclusive. Perhaps a statutory scheme provides benefits to families in response to the unemployment of the father but denies those same benefits in the case of unemployment of the mother.110 The court may conclude that this discriminatory treatment of parents runs afoul of the Equal Protection Clause.111 If the court then responds to the constitutional disruption through the framework of “severability,” it has two only options: it can disregard the entire statute, or it can disregard its unconstitutional portion. The severability framework does not permit the court to take a third path—for example, to construe the employment-related statute as granting benefits to families with an unemployed father or mother.

This is because “severability,” as generally understood, “may proceed ‘by striking out or disregarding words [or applications] that are in the [challenged] section,’” but “it may not proceed ‘by inserting [applications] that are not now there.’”112

To the extent courts or commentators even have attempted to justify this limitation on the courts, they have done so through conclusory appeals to judicial restraint—by asserting, for example, that a contrary approach “would constitute legislation beyond [the Court’s] judicial power.”113 In so doing, however, they have offered no explanation for why this same judicial response may be permissible in other legal contexts, such as in the context of “remedy.”114 A classic case illustrating the latter is Califano v. Westcott, which provides the inspiration for the hypothetical discussed above.115 In Califano, the Court declared unconstitutional a provision of the Social Security Act that provided aid to families with an unemployed father, but not, all else being equal, to those with an unemployed mother.116

In discussing the appropriate remedy, the Court explained that two alternatives exist for underinclusive statutes: “a court may either declare [the statute] a nullity and order that its benefits not extend to the class that the legislature intended to benefit, or it may extend the coverage of the statute to include those who are aggrieved by the exclusion.”117 Indeed, in such a case, “extension, rather than nullification, is the proper course.”118

Yet such an approach, which is “proper” as a remedy,119 is prohibited in the context of severability. Neither the case law nor the commentary has attempted to explain this discrepancy. Nor has either provided an answer to an even more fundamental question: why, in the context of severability, is it permissible for a court to reject such a response if that is what the legislature would have intended? The received view of severability provides no explanation.

2.     Disregarding (or otherwise altering) a portion of the statute that is not considered to be the most immediate source of the unconstitutionality

A second approach prohibited in the context of severability is the courts’ acceptance of what are, in a sense, “substitute” statutory provisions.120 One manifestation of this approach, if it were permitted, would allow a court to go beyond what it considers to be the most immediate source of a statute’s unconstitutionality and, in so doing, “invalidate any part or parts of a statute (and add others)” to cure the constitutional defect.121 This approach theoretically is implicated every time a statute is unconstitutional with respect to certain statutory portions, but would remain within constitutional bounds if a court were to disregard or otherwise alter some separate set of statutory portions.

It is true that this distinction (between the most immediate source of a statute’s unconstitutionality and some other implicated portion) is vague and problematic. This Article does not defend the distinction as conceptually sound. It nevertheless exists in the doctrine. To illustrate the phenomenon, the prime example (or, perhaps more precisely, the exception that proves the rule) is the Sentencing Reform Act challenged in United States v. Booker.122 In conjunction with the promulgated guidelines, this Act required courts to sentence defendants based on facts found not by a jury. As applied to a subset of defendants—that is, those receiving sentences higher than they otherwise could have received—this regime violated the Sixth Amendment. As a result, the Court (as is so often the case) turned to severability.

Under the prevailing severability framework, the Court was permitted two options: it could disregard this particular subset of statutorily applications (that is, it could vacate the sentences of those who had been subject to the higher sentences, and not allow any such sentencing going forward), or it could disregard all statutory applications (that is, it could vacate all the sentences calculated under the Act and not allow the Act to dictate any sentences going forward). What was not available to the Court, under the prevailing conception of severability, was a third approach: that is, choosing to disregard a separate provision in the Act, even though it did not consider that separate provision to be the immediate source of the unconstitutionality.123

Ironically, a fractured majority in Booker actually did take this third approach. Over the vigorous dissent of four Justices, the Court decided it would disregard the separate provision of the Act: the provision making the Guidelines mandatory rather than advisory.124 This cured the statute’s unconstitutionality by ensuring that all necessary facts would be found by the jury. In adopting this approach, however, the Court accepted a framework for severability that is inconsistent with virtually every articulation that the Court has set forth before or since.

Tellingly, Justice Stevens, who wrote the majority opinion on the substantive constitutional claim but dissented with respect to severability, refused to accept that the Court had engaged in “severability analysis” at all.125 Rather, he accused the majority of adopting “entirely new law”126 through “a novel and questionable method of invalidating statutory provisions,” which in his mind amounted to “a wholesale rewriting” of the statute.127 Justice Thomas, in a separate dissent, “agree[d] with [Justice Stevens] that [the majority opinion] grossly distorts severability analysis by using severability principles to determine which provisions the Court should strike as unconstitutional.”128

In at least one respect, the dissents won this argument: notwithstanding the majority’s attempt to summon supporting precedent,129Booker’s articulation of the severability framework—which, as a preview, this Article ultimately will defend130—does appear to constitute a stark anomaly in the case law.131 It is in this sense that Booker constitutes the exception that proves the “rule”132: that is, that the severability framework only permits one of two narrow responses. As such, it does not allow a court to invalidate just any portion of a statute in a more creative effort to cure constitutional defects.

Yet outside the context of severability, courts feel empowered to respond to a statute’s constitutional disruption by adopting substitute provisions. Often couched as a “remedy,”133 such a result can adhere even when it requires altering statutory portions other than those considered to be the most immediate cause of the statute’s unconstitutionality. In the context of redistricting, for example, courts regularly reject district lines drawn by statute, in light of some constitutional flaw, before redrawing those lines themselves.134 And courts openly acknowledge their acceptance of “substitute” provisions so long as those provisions are contained explicitly in a statute.135

The Court in Bowsher v. Synar, for example, declared unconstitutional the Balanced Budget and Emergency Deficit Control Act before ordering, as a “remedy,” that substitute provisions be enforced.136 Referring to these substitute provisions—which were express provisions enacted by Congress—as “fallback provisions,”137 the Court took pains to clarify that, by effectuating these provisions, it was not engaging in severability analysis. Characterizing itself as refusing to “perform the type of creative and imaginative statutory surgery urged by appellants” (who had been urging the Court to “sever” only a small portion of the statute), the Court explained that severability was a “thicket we need not enter,” given the existence of the so-called fallback provisions.138 Freed of the limiting effects of the severability framework, in other words, the Court was able to “simply permit[] the fallback provisions to come into play.”139 At no point did the Court explain why, once in the “thicket” of severability, it would have been impermissible to turn to substitute provisions.

Nor, of course, did the Court address the same, fundamental question identified above: why impose such limitations if the legislature would have preferred otherwise? If the answer boils down to whether the substitute provisions are explicit or not (a response that nevertheless fails to address the redistricting example), that merely modifies the question: why must a court disregard legislative intent whenever the legislature has failed to make its substitute provisions explicit? Neither case law nor the commentary has attempted to provide an explanation.

3.     Treating some portion of the statute as having a nonstandard meaning

Another approach a court might take, in response to a ruling of unconstitutionality, is to treat some offending portion of a statute as having a nonstandard meaning—that is, a meaning that the court would not have found in the absence of the constitutional disruption. In other words, a court might construe a statute to have a meaning curative of the unconstitutionality, even if it would not have adopted such a meaning without the constitutional disruption.

This is, essentially, what Chief Justice Roberts did in the National Federation case with respect to the litigants’ broad (and ultimately unsuccessful) challenge to the ACA’s individual mandate. The Chief began by concluding that “[t]he most straightforward reading of the [ACA] mandate is that it commands individuals to purchase insurance” but that such a reading would produce an unconstitutional result.140 Had the Chief then engaged in traditional severability analysis, he would have been required to disregard the ACA either in whole or in part. Instead, however, the Chief asked a different question: “whether the Government’s alternative reading of the statute—that it only imposes a tax on those without insurance—is a reasonable one.”141 He concluded this alternative reading was, indeed, reasonable and, citing doctrines of constitutional avoidance, adopted that alternative reading. In so doing, the Chief strongly implied that he would not have adopted this reading had he not first reached a ruling of unconstitutionality.142

The dissent, disagreeing strongly with the Chief’s alternate reading, argued that construing the mandate as a tax did “violence to the fair meaning of the words used.”143 Yet had the Chief, having concluded that the Act was unconstitutional under its most natural reading, then resorted to “severability” analysis (rather than familiar canons of constitutional avoidance), the dissent would have had a much more forceful objection. For under no commonly held understanding of severability is a court permitted to adopt a nonstandard construction of a statute it has deemed to be constitutionally disrupted.

Outside the framework of severability, however, courts routinely assign statutes nonstandard meanings in response to constitutional concerns. This approach is so common, in fact, that it has risen to the level of a well-established canon of statutory interpretation—perhaps the “preeminent canon.”144 This is the canon of constitutional avoidance.145 Despite this pedigree, the analytical approach that underlies the doctrine of constitutional avoidance—that is, its reliance on nonstandard meanings—is prohibited once a court concludes that “severability” applies. Again, the courts police this prohibition without explanation either for the discrepancy or for the interference with legislative intent.

4.     Construing or applying the statute in a way that gives the court too much “editorial freedom”

Each of the first three limitations—relating to the expansion of applications, the altering of statutory portions that are not considered to be the most immediate source of the unconstitutionality, and the reliance on nonstandard constructions—are well established in the context of severability. Yet there appears to be a movement afoot; the list appears to be expanding. Recently, courts and commentators have advanced an even more restrictive form of severability—one that imposes additional limitations on the courts by relying on similarly vague concerns over judicial restraint. One of these newer restrictions seeks to limit how much discretion a court can exercise when determining which portions of a given statute should be disregarded. Hostility with a court’s discretion in this context overlaps quite a bit with the “Complexity Test” described above, as both track an insistence that a court’s severability determinations somehow remain uncomplicated or even mechanical in nature.

The Court recently illustrated its resistance to discretion and complexity in the previously discussed Free Enterprise Fund.146 In this case, the Supreme Court concluded that a combination of statutory provisions violated the Constitution’s separation-of-powers principles. The Court then considered how the statute should apply going forward. Concluding that certain provisions (those associated with “good-cause removal”) were “severable from the remainder of the statute,” the Court held that those provisions would no longer be given effect.147 It then discussed the possibility of the statute applying in some other way. It explained:

It is true that the language providing for good-cause removal is only one of a number of statutory provisions that, working together, produce a constitutional violation. In theory, perhaps, the Court might blue-pencil a sufficient number of the Board’s responsibilities so that [the constitutional problem would be resolved]. Or we could restrict the Board’s enforcement powers, so that [the constitutional problem would be resolved]. Or the Board members could in future be made removable by the President, for good cause or at will[, thereby satisfying the relevant constitutional requirements]. But such editorial freedom—far more extensive than our holding today—belongs to the Legislature, not the Judiciary.148

As this passage makes clear, the Court is rejecting the latter options not based on any conclusion about legislative intent, but rather because of its conclusion that their adoption would have given the Court too much freedom in deciding how the statute should operate going forward. (Presumably this would include freedom to decide, for example, which of the Board’s responsibilities or enforcement powers should be “blue-pencil[ed].”) Importantly, the Court did not reject these options because it had concluded that their adoption would have been different in nature from adopting the option it did take. Stated otherwise, the Court did not reject these options because it considered them different in nature from the two approaches normally permitted in the context of severability (disregarding the statute in part, or disregarding it in whole).149 Instead, the problem came down to an excess of discretion.150

Taken at face value, this limitation—a rejection of any severability analysis that accords the court too much “freedom” or too many “options”151—is troubling. While some jurists and scholars have expressed skepticism toward courts engaging in any sort of complicated legal analysis (and this skepticism tends to be part of an overarching judicial philosophy that seeks more generally to resist creativity in judicial analysis),152 even the most aggressive of skeptics does not go so far as to suggest that analytical complexity, on its own, constitutes judicial error. Justice Thomas, for example, acknowledges that intricate statutes often will require a court to engage in complicated analysis and that it is proper to do so.153 And even Justice Scalia had agreed that one of the most creative (and, as such, controversial) forms of judicial analysis—that is, federal common law making—is appropriate in certain circumstances, including when a statute directs a court to engage in it.154 So even for those who are quick to question any perceived creativity or complexity in judicial analysis, it is far from self-evident why creativity or complexity, without more, should be prohibited—whether or not in the context of severability. Nor is any answer evident in response to the more fundamental question again posed by this limitation: with respect to a statute reflecting contrary legislative intent—that is, intent that a court respond to a statute’s constitutional disruption with complicated, innovative, or creative forms of analysis—why is it permissible for the court to disregard the will of Congress? No answer is evident. Yet this limitation continues to gain traction in the doctrine.

5.     Engaging in “application severability”

This Article has saved what very well may be the most drastic and disruptive limitation for last. Recently, jurists and commentators have expressed anxiety over the use of what, for decades, has been relied on largely without controversy: so-called “application severability.”

For background, traditional severability analysis can be broken down into two subsets: “text severability” and “application severability.”155 “Text severability” occurs when courts “sever [a statute’s] problematic portions while leaving the remainder intact.”156 “Application severability,” by contrast, occurs when courts “enjoin only the unconstitutional applications of a statute while leaving other applications in force.”157

A hypothetical example helps to illustrate. A newly enacted statute provides that “any person convicted of murder shall be incarcerated for thirty years.”158 The question then becomes: does this statute apply to those who already have been convicted and already have served a sentence of less than thirty years?159 “If so, a serious question of constitutionality arises under the Ex Post Facto Clauses,” and the court will “rule[] the statute unconstitutional as applied to those already convicted but constitutional as applied to future convictions.”160 In other words, the court will sever the unconstitutional applications from the constitutional ones, and refuse to enforce the former.

So described, application severability may seem unremarkable. “It is axiomatic that a ‘statute may be invalid as applied to one state of facts and yet valid as applied to another,’”161 and courts engage in this sort of line-drawing all the time.162 And, indeed, there is substantial doctrinal and academic authority in support of the propriety of application severability. The Supreme Court repeatedly “has acknowledged the severability of applications in striking down some applications of a statute while leaving others standing.”163 Or, as Professor Dorf has explained, “the formal tests [for severability] treat applications as no less severable than provisions.”164

Yet, as Professor Dorf himself acknowledged in 2007, this particular form of severability nevertheless leads to significant judicial anxiety. “[J]udges and Justices find application severability more troubling in practice [than text severability],” he explains, “because it can require them to craft substantive provisions where the legislature has crafted none.”165

Since the publication of Professor Dorf’s article, both the rhetoric and the doctrine appear to have become even more hostile to application severability. In National Federation, for example, the dissent aggressively attacked the legitimacy of the majority’s decision to apply a straightforward form of application severability.166 The dissent’s criticism was so forceful—and, given the generally well-established use of application severability, therefore so startling—that it is worth providing a brief summary of the constitutional defect. In this case, a majority of Justices had found a violation of the Spending Clause arising out of the ACA’s Medicaid expansion, which grants federal funding for States that provide certain types of health care. The constitutional violation arose because the Medicaid expansion had implications for an already existing statutory provision, 42 U.S.C. § 1396c. The latter provision granted the federal government authority to withhold Medicaid payments from a state determined to be out of compliance with any Medicaid requirement. As a result of the interaction between the ACA and
§ 1396c, the federal government became empowered to withhold all Medicaid funding from any state refusing to comply with the obligations of the new Medicaid expansion. Concluding that this newly created financial threat amounted to coercion, the Court concluded it was unconstitutional.167

This constitutional flaw presented the Court with a difficult problem: how should the ACA apply going forward? It is worth noting that the Court could not easily engage in “text severability.” While a simple refusal to follow § 1396c indeed would have solved the constitutional problem, that approach would significantly disrupt the entire Medicaid program, including any number of provisions that (like § 1396c itself) had been enacted well prior to the ACA. The majority therefore did not engage in text severability. Instead, the majority disregarded a particular subset of the ACA’s applications—i.e., it engaged in application severability. As the Chief Justice explained, “[i]n light of the Court’s holding, the Secretary cannot apply § 1396c to withdraw existing Medicaid funds for failure to comply with the requirements set out in the expansion.”168

This response to a constitutionally disrupted statute was unremarkable—except for the response it provoked. The dissent excoriated the majority for this decision. In so doing, it made clear that, with respect to the Medicaid issue, it was not the Court’s substantive holding, but rather its reading of the constitutionally disrupted ACA, that it found objectionable. The dissent asserted that:

The Court today decides to save a statute Congress did not write. . . . [I]t changes the intentionally coercive sanction of a total cut-off of Medicaid funds to a supposedly noncoercive cut-off of only the incremental funds that the Act makes available.

The Court regards its strained statutory interpretation as judicial modesty. It is not. It amounts instead to a vast judicial overreaching.169

The dissent made clear, moreover, that it considered the majority’s treatment of the statute to lie outside what “severability” will allow. It presented this argument as follows:

The Government cites a severability clause codified with Medicaid . . . . But that clause tells us only that other provisions in Chapter 7 should not be invalidated if § 1396c, the authorization for the cut-off of all Medicaid funds, is unconstitutional. It does not tell us that § 1396c can be judicially revised, to say what it does not say. Such a judicial power would not be called the doctrine of severability but perhaps the doctrine of amendatory invalidation—similar to the amendatory veto that permits the Governors of some States to reduce the amounts appropriated in legislation. The proof that such a power does not exist is the fact that it would not preserve other congressional dispositions, but would leave it up to the Court what the “validated” legislation will contain. . . . The Court severs nothing, but simply revises § 1396c to read as the Court would desire.170

In short, use of application severability did not shield the majority from criticism. To the contrary, the dissent lambasted the majority for engaging in precisely what “severability” purports to prevent: judicial legislation.

National Federation’s criticism of application severability emerged from a joint dissent authored by four Justices, including Justice Thomas, who had previously authored a spirited defense of application severability.171 In another recent, high-profile case, the criticism came from the majority. In the previously discussed Free Enterprise Fund, the Supreme Court rejected the use of application severability in response to the constitutionally defective Sarbanes–Oxley Act.172 In explanation for this rejection, it explained (similar to the National Federation dissent) that such an approach “belongs to the Legislature, not the Judiciary.”173

The implications of this seeming nascent trend in the doctrine—that is, the trend away from application severability on the grounds that it constitutes impermissible judicial legislation—are difficult to overstate. Courts routinely enforce only a subset of a statute’s applications while refusing to enforce the others. A court does so, for example, whenever it concludes that, on account of the Supreme Court’s constitutional ruling in Roper v. Simmons,174 a criminal provision may be applied to those who committed the requisite crime when eighteen years or older, but not to those who committed the crime when under eighteen years old.175

More broadly, a court demonstrates its willingness to engage in this practice—to enforce only a subset of a statute’s applications—whenever it concludes that a litigant has failed to establish a successful facial challenge to some portion of a statute but nevertheless may be able to prevail on an as-applied challenge.176 This occurred, for example, in Crawford v. Marion County,177 where the Court rejected a constitutional challenge to a statute requiring voters to show photo identification prior to voting.178 Though the Court rejected the facial challenge (which would have prohibited the statute from taking effect in every possible application), it did leave open for future litigation the possibility of as-applied challenges (which, if successful, would prohibit the statute from taking effect in certain applications).179

Far from an outlier approach, as-applied challenges are very well established, with ample support suggesting that they are actually preferred to facial challenges.180 Ironically, this stated preference for as-applied challenges is intended to advance the very same thing that severability is intended to promote: “the fundamental principle of judicial restraint.”181

Yet even this is not all. A willingness to engage in application severability can, in a sense, be inferred whenever a court rejects on the merits any constitutional challenge brought by a plaintiff.182 This is because the court in such a circumstance has not concluded that the statute is constitutional in every conceivable application. Rather, it merely has concluded that the relevant statute is constitutional as applied to the litigant before the court. It is leaving to another day (and another lawsuit) the possibility that it is unconstitutional as applied to someone else—and, as such, is implicitly acknowledging at least the possibility that the statute constitutionally could be enforced in some applications but not others.

Stated succinctly, a refusal to permit application severability—and to do so on the ground that so deviating from the text would constitute an impermissible “rewriting [of] the statute”183—threatens not only well-established and fundamental doctrines in the context of severability, but across any number of implicated doctrines.184 Yet the emerging trends in severability seem to be headed in precisely this direction—again without acknowledging (much less explaining or affirmatively justifying) how this can be reconciled with doctrines outside the context of severability, or with the possibility of contrary legislative intent.

*         *         *

This Article has identified five approaches to constitutionally disrupted statutes that, once subject to the severability framework, become threatened. Stated otherwise, as soon as a court accepts that “severability” applies, that court risks reversal (or, at a minimum, impassioned criticism) if it employs any of the following five approaches: (1) construing a statute in a way that expands its reach; (2) disregarding or otherwise altering a portion of the statute that is not considered to be the most immediate source of the unconstitutionality; (3) treating some portion of the statute as having a nonstandard meaning; (4) construing or applying the statute in a way that gives the court too much “editorial freedom”; or (5) merely engaging in application severability.

Critically, resistance to these approaches exists even if effectuation of legislative intent would require their use. In other words, it does not matter if an inquiry into legislative intent reveals that, had Congress known about a statute’s constitutional defect, it would have wanted a court to expand the reach of that statute, to disregard some separate portion, or otherwise to veer outside the severability norm. Severability simply does not permit such an outcome.

It is in this way that the severability framework drastically limits a court’s ability to effectuate legislative intent. The prevailing framework for severability does not account for the possibility that, had the legislature known of a statute’s constitutional flaw, it would not have intended for the court to disregard the statute in whole or in part, but instead would have intended for the court to take some third path. This analytical conundrum exists in the background every time the court engages in severability. This is no surface-level problem. It instead results predictably from the combination of severability’s two defining features: first, its requirement that the court follow legislative intent, and, second, its requirement that, regardless of legislative intent, the court may only take only one of two narrow approaches.185

As indicated above, there is virtually no acknowledgment, much less analysis or critical scrutiny, of this central tension embedded in the severability framework.186 The unanimous Ayotte provides a vivid illustration of how this serious tension is simply glossed over:

After finding an application or portion of a statute unconstitutional, we must next ask: Would the legislature have preferred what is left of its statute to no statute at all? . . . All the while, we are wary of legislatures who would rely on our intervention, for “[i]t would certainly be dangerous if the legislature could set a net large enough to catch all possible offenders, and leave it to the courts to step inside” to announce to whom the statute may be applied. “This would, to some extent, substitute the judicial for the legislative department of the government.”187

This startling juxtaposition—the Court’s endorsement of legislative intent followed immediately by an indication that a court should at times refuse to follow legislative intent—is presented without acknowledgement or explanation.

Once severability’s problems are revealed, the tensions grow even more strained. There is, for one, no answer to the question of what a court should do if it determines that the legislature’s preference reflects an impermissible “rel[iance] on [the Court’s] intervention.”188 Presumably the court should, contrary to legislative intent, insist on invalidating the entire statute. Yet one of severability’s central purposes is “not to nullify more of a legislature’s work than is necessary.”189

Nor is there any answer to the question of why a court may, on its own accord, reject legislative intent once it determines that the legislature has somehow “rel[ied] on [its] intervention.”190 Such a scenario is particularly startlingly not only in light of severability’s purported interests in “limiting judicial power by minimizing the damage done to the statute by judicial fiat,”191 but also in light of its insistence that legislation be dictated by Congress, not by the courts. A court is, in a sense, accomplishing precisely the opposite (i.e., it is allowing legislation to be dictated by the courts, not Congress) when it rejects legislative intent in order to engage in severability analysis.192

These contradictions (at best, tensions) exist without acknowledgment or explanation. Yet they do not exist without consequence. Yet as Part III.B reveals, these subtle limitations have profound effects.

B.     Statutory Damage: How Severability’s Disregard of Legislative Intent Wreaks Havoc on the Law

The limitations imposed by the severability framework have drastic effects on constitutionally disrupted statutes. At the outset, they result in deep doctrinal confusion, as illustrated by the previously discussed passage in Ayotte.193 By simultaneously endorsing and resisting a court’s inquiry into legislative intent, the Court in Ayotte tacitly acknowledged the tension inherent in that combination—but, as usual, it failed to provide any means by which to resolve it. Such undertheorized complexity contributes to the muddled and constantly shifting rules governing severability analysis.194 It also provides a troubling illustration of what Eric Berger has termed “stealth constitutional decision making,”195 a phenomenon characterized by “unpredictability, imprecision, and opacity” with corrosive effects on the case law.196

Yet the effects run still deeper. By preventing courts from effectuating legislative intent, these limitations result in greater statutory invalidation than Congress would have preferred. Three doctrinally prominent examples—involving the Voting Rights Act, the legislative veto, and the Sentencing Reform Act—help to illustrate this phenomenon, though this list is far from exhaustive. This dismantling of Congress’s work becomes even more startling, moreover, when one considers where the doctrine may be headed. Emerging trends in severability threaten to demolish wide swaths of the U.S. Code.

The recent dismantling of the Voting Rights Act provides one illustration of severability’s effects. A brief description of this blockbuster case—and the landmark piece of legislation it shattered—provides the relevant context. In Shelby County v. Holder, the majority invalidated section 4(b) of the Voting Rights Act.197 Yet section 4(b), as relevant, did not itself impose any substantive mandate. Rather, it provided the “coverage” formula for section 5, which in turn required covered jurisdictions to receive permission from the federal government prior to enforcing any voting-related changes. The combination of section 4(b) and section 5, as a result, subjected certain state and local jurisdictions to federal oversight of voting practices. So designed, the effects of this statute were extraordinary: the combination of sections 4(b) and 5 was a central reason why many consider the Voting Rights Act to be the “most successful piece of civil rights legislation ever enacted.”198

The Shelby County majority concluded that section 4(b) was unconstitutional because its coverage formula failed to take “current conditions” into adequate account.199 Responding to the majority’s conclusion that section 4(b) must fall in its entirety, the dissent wondered why principles of severability would not allow the majority to keep some section 4(b) jurisdictions covered—and therefore keep section 5 at least partially in force. In the words of Justice Ginsburg, “even if the VRA could not constitutionally be applied to certain States—e.g., Arizona and Alaska—[its severability provision] calls for those unconstitutional applications to be severed, leaving the Act in place for jurisdictions as to which its application does not transgress constitutional limits.”200 The majority’s response to this severability-based argument was curt. Essentially, it stated that it would be inappropriate for the Court to respond in such a manner to a constitutionally disrupted statute.201

The majority’s response was consistent with the doctrinally dominant view of severability—and certainly with the more restrictive view of severability that has been emerging in the doctrine.202 What the majority’s response appears not to have been consistent with, however, is legislative intent.203 Tellingly, the majority did not argue otherwise. Were severability employed in a manner more consistent with legislative intent (and therefore less consistent with severability’s restrictive framework), the Court could have responded to the constitutional disruption through a more tailored reading of the Voting Rights Act—and, as such, allowed some of the applications of sections 4(b) and 5 to remain in effect over offending jurisdictions.204

A second illustration of severability’s effects comes in response to the so-called legislative veto, which appears in hundreds of federal statutes and which the Supreme Court declared to be unconstitutional in Immigration & Naturalization Service v. Chadha.205 This substantive constitutional ruling has required the courts to wade through numerous statutes to determine whether the legislative veto in each is severable (and therefore, under the dominant view of severability, invalid on its own) or inseverable (and therefore invalid along with the rest of the statute).206 What the dominant framework does not allow is some third option. To take but one example, Congress might have intended for the legislative veto to be treated as severable with respect to certain grants of authority to the executive branch but inseverable with respect to other such grants—even where the statute’s text would not allow for such a result through straightforward excision.207 As Glenn Smith has argued, this third approach may most closely track congressional intent.208 Yet due to its tension with severability’s restrictions, the courts do not even consider it.209

A third illustration of the effects of severability’s narrowing framework already has appeared in this Article. In Booker, the Court concluded that the Sentencing Reform Act violated the Constitution.210 The Court then concluded, in response to this constitutional disruption, that Congress would have intended the following judicial responses, in descending order of preference:

First, for the Court to “make the Guidelines system advisory” through the invalidation of two provisions that, on their own, did not impose any unconstitutionality.211

Second, for the Court to invalidate the entire Sentencing Reform Act.212

Third, for the Court, while continuing to enforce much of the Sentencing Reform Act, to invalidate a subset of its applications (that is, those imposing sentences that had been increased based on facts not found by a jury).213

As discussed above, any of these approaches would have cured the constitutional defect—but only the second and the third were permissible under the dominant severability framework.214 In perhaps the Supreme Court’s only overt rejection of the dominant severability framework, the majority adopted the first approach. In so doing, it avoided what severability otherwise would have required: total invalidation of the Act.215

These three examples help to illustrate the profound effects of severability’s restrictive framework. Without it, one of the most important civil rights provisions might still be in partial effect; hundreds of statutes might still include a more tailored grant of authority to the executive; and Booker’s “audacious” approach to constitutionally disrupted statutes might be the standard rather than the exception.216 And, of course, this list is far from exhaustive. An abundance of examples help to illustrate how the statutory landscape might change in the absence of severability’s restrictive effects. The National Federation majority, for example, might have further minimized the disruption caused to the ACA by the flaw in its Medicaid expansion.217 Courts’ attempts to construe the Bankruptcy Code in light of the Court’s opinion in Stern v. Marshall might more naturally track legislative intent—rather than threaten the practical collapse of the entire code based on an “isolated” constitutional disruption that, according to the Stern majority, did not “change all that much.”218 And, of course, innumerable lower-court opinions might have come out differently.219 Indeed, elimination of severability’s restrictive effects raises the possibility that any given successful constitutional challenge to a statute might have resulted in a different treatment of that same statute.220

Yet even this summary understates the potential effects of severability’s narrowing framework, for it has not yet acknowledged where the doctrine appears to be headed. As dissents like those in National Federation have made clear, an even more restrictive form of severability—with its resistance to editorial freedom and its rejection of application severability—would result in the full invalidation of statutes with far more frequency.221 At least one scholar has taken this approach to its logical conclusion, arguing that any constitutional flaw (regardless how small, and regardless of legislative intent) requires the complete invalidation of the implicated statute.222 Tellingly, this scholar’s argument has been cited multiple times in briefs filed with the United States Supreme Court.223

In short, while the narrowing framework of severability may suffer from a lack of theorization (and acknowledgement), it does not suffer from a lack of practical consequence. To the contrary, these limitations pack a powerful punch.

IV.     Severability as an Unjustified Doctrine

The dominant framework for severability restricts the courts’ options at the expense of legislative intent. This dynamic has serious practical consequences on the continued operation of constitutionally disrupted statutes. As this Part reveals, these effects need justification to retain legitimacy. Yet no such justification exists.

This Part begins by explaining that while any interference with legislative intent is troubling, the disruptive effects of severability are particularly problematic in light of the ways in which they are inconsistent with fundamental principles purporting to underlie the severability doctrine. This Part then attempts to identify such a justification—but it cannot. Even the most plausible justifications for severability’s constraints fail on their own terms.

In the absence of justification, the severability framework should be replaced. This Part therefore concludes with an outline for an improved regime. This new regime allows courts to effectuate legislative intent without grappling with severability’s unjustified restrictions.

A.     The Lack of Justification for Severability’s Destructive Effects

Severability’s restrictive framework requires justification for two overarching reasons: first, because its effects are inconsistent with fundamental principles purporting to support the severability doctrine; and, second, because courts routinely engage in those same practices, largely without controversy, in other legal contexts. Taken either independently or together, these two concerns should be considered to shift the burden. Either some justification affirmatively supports these discrepancies, or one should seek to eliminate their cause.

Severability’s limiting effects are inconsistent with at least three fundamental principles driving the severability doctrine. At the outset, severability doctrine purports, at its core, to effectuate legislative intent.224 As a result, there is a deep and troubling conflict that arises when the severability framework actually prevents a court from following legislative intent.

A second frequently articulated goal of severability is to treat constitutionally disrupted statutes in a way that “limit[s] the solution to the problem.”225 Stated otherwise, severability is intended to help “minimiz[e] the damage done to the statute by judicial fiat.”226 Yet when the severability framework interferes with a court’s efforts to effectuate legislative intent, it is not “minimizing” the damage done to the statute. Quite to the contrary, it is limiting the solutions potentially available to the court in its efforts to achieve this end. As a result, it is actually increasing the damage done to the statute. Such an approach not only nullifies the legislature’s work as it relates to the constitutional defect; it also nullifies the legislature’s work as it relates to how the legislature intended for a court to respond to the constitutional disruption.227

Finally, severability is supposed to promote judicial restraint by preventing a court from engaging in so-called judicial legislation. Yet at best, this goal is in serious tension with severability’s restrictive framework. This is because adherence to legislative intent—which, to be sure, is defined differently depending which jurist or commentator one asks228—normally is what ensures a court is effectuating Congress’s statutes rather than creating its own law.229 By rejecting legislative intent in favor of its own conception of how best to construe and apply a constitutionally disrupted statute, a court opens itself to criticism that it is doing precisely what it purporting to avoid: that is, substituting its own judgment for that of the legislature.

In addition to requiring justification due to inconsistencies with respect to its own fundamental principles, severability requires justification because it relies on an incongruous characterization of which practices lie within the judicial sphere. As explained above, severability limits courts’ options by discouraging or prohibiting certain practices on the grounds that they do not lie within the judicial function.230 Yet courts routinely engage in those same practices, typically without controversy, in other legal contexts. Specific illustrations of this phenomenon are illustrated in the discussion above.231 It is true that some of these approaches may inspire criticism from certain jurists and commentators even when employed outside the context of severability. For example, certain justices are more likely than others to be willing to expand a statute’s applications when fashioning a “remedy.”232 The underlying discrepancy nevertheless remains. At a minimum, this incongruity undermines the idea that the practices in question obviously, inherently, or necessarily constitute “judicial legislation” or otherwise lie outside of what a court is permitted to do. Yet that is precisely the explanation provided for why, in the context of severability, these same practices are considered off-limits.

Combined, these two overarching observations—relating to inconsistencies with severability’s fundamental principles and with other areas of the law—force a defender of the severability framework to find some offsetting justification in support of its limiting effects. Otherwise, the framework should be replaced. As the remainder of this Part explains, no such justification appears to exist.

Certainly, no justification appears in the case law itself. Perhaps this is because no reasoned justification ever motivated this aspect of the doctrine. Stated otherwise, it may be that the limiting effects of severability emerged, in effect, accidentally, as courts and commentators have either overlooked how the doctrine narrows court’s options or passively accepted the correctness of the underlying assumption (that is, that Congress would have wanted a court to disregard the statute in whole or in part).

Even in the absence of expressly stated justifications, it is appropriate to try to identify justifications that could support severability’s limiting effects. For in the absence of critical analysis there are, for many, deeply held intuitions about severability: that it simply makes sense for a court to try to limit the solution to the problem in response to a constitutionally disrupted statute; that it makes sense for a court to refrain from rewriting a statute or otherwise from adopting the legislative role; in short, that it makes sense for a court to apply the severability framework in a form similar to how it now stands. Despite the appeal of these intuitions, however, normative support for severability’s limiting effects begins to collapse once the possible justifications are subject to critical analysis.

In an effort to illustrate the deficiencies of possible justifications, this Article attempts to identify and analyze the most persuasive. All five of these possible justifications rely on theories of judicial restraint. More specifically, all rely on a conception of judicial restraint that seems most reasonably to justify limiting a court’s options in the face of contrary legislative intent: namely, that courts should not engage in legislative work.

The first three of these possible justifications already have appeared in this Article, as they are captured by the three possible tests for drawing the line, in the context of severability, between the legislative and the judicial spheres. This Article has termed these tests the Blue Pencil Test,233 the Complexity Test,234 and the Primary Drafter Test.235 Underlying each of these tests is the same general principle: a court must stay in its proper lane and not veer into legislative territory.236 It is this principle of judicial restraint that purports to provide a justification for the three tests’ restrictive effects. As such, the strength of each test as a possible justification for severability depends, at a minimum, on how convincingly each polices the line between the judicial and the legislative spheres.

The fourth possible justification goes to a more general concern: institutional competence. More specifically, this fourth possible justification derives from the idea that courts lack the ability to engage with constitutionally disrupted statutes without the restrictions imposed by the severability framework. The fifth and final possible justification is even more general. Essentially, it proposes that, regardless of the specific test used, a court simply cannot engage in judicial rewriting of a constitutionally disrupted statute. As such, the court’s options must be so limited. As this Part of this Article demonstrates, none of these five possible justifications adequately supports the ways severability limits the courts’ options. Each possible justification is addressed in turn.

The Blue Pencil Justification. The first justification is reflected in the Blue Pencil Test. This test requires that a court respond to a constitutionally disrupted statute by (metaphorically) striking out statutory provisions or other bits of statutory language. Any other response is impermissible. The line drawn by the Blue Pencil Test, as such, deems “text severability” permissible but refuses to accept the validity of “application severability.”237

Despite the intuitive appeal of this test, it suffers from a host of deficiencies in purporting to police the judicial/‌legislative line. At the outset, it quickly leads to what many would consider arbitrary results. To illustrate, it would treat the following two statutes differently:

Statute 1: Any person convicted of murder committed at any time shall be incarcerated for thirty years.238

Statute 2: Any person convicted of murder committed before, on, or after the date of this statute’s enactment shall be incarcerated for thirty years.239

Assume, as suggested above, that both statutes violate the Ex Post Facto Clause as applied to someone who already has been convicted.240 Assume further that legislative intent indicates that Congress would have preferred, in the case of a constitutional defect, for as many applications as possible to remain in force.241 If the Blue Pencil Test properly policed the line between permissible and impermissible judicial responses, then the court would be forced to invalidate the entirety of Statute 1, for there is no other way to “blue pencil” the statute to remove the offending applications. By contrast, the court would be able to “blue pencil” Statute 2 in a way that permitted a significant fraction of its applications (i.e., all those relating to murders occurring on or after the date of the statute’s enactment) to continue in force. Despite the fact that the two statutes, in substance, are identical, the Blue Pencil Test requires radically different treatment in response to the constitutional disruption.

A related problem with the Blue Pencil Test is that it fails to account for application severability. As discussed above, application severability is a historically well-established use of the severability doctrines.242 However, it is one that cannot be accomplished merely by running a blue pencil through the relevant statutory language. The recent hostility to application severability should not be mistaken for a reflection of the doctrine as it has been long understood. To the contrary, any refusal to employ application severability marks a quite drastic turn in the case law.243

And the fundamental problems with the Blue Pencil Test run still deeper. This is because, as discussed above, a strict adherence to this position—that is, a refusal to accept the possibility of application severability, which is the essence of the Blue Pencil Test—would have deeply disruptive ripple effects across all statutes subject to constitutional challenge.244 It is not clear that such a system would even be workable.245 At best, the system would be internally inconsistent, as an insistence on judicial restraint purports to justify both
(1) the Blue Pencil Test’s refusal to allow application severability and (2) the courts’ stated preference for as-applied challenges over facial challenges. Application severability and as-applied challenges are, in many cases, two sides of the very same coin.246

In short, the conception of the judicial function characterized by the Blue Pencil Test fails to provide adequate support for severability’s limiting functions. It leads to inconsistent, arbitrary results; it fails to track the longstanding case law; and its adoption would require severe and unjustified changes to the way courts respond to any statute potentially subject to a constitutional challenge. It is, of course, possible that the Blue Pencil Test reflects a deeper concern: that courts simply should not be rewriting statutes, no matter the cost. Yet that concern, which is discussed in detail below,247 also is not able to justify severability’s restrictive framework.

The Complexity Justification. The second possible justification is captured by the Complexity Test. The Complexity Test purports to draw a line between, on the one hand, judicial treatment of statutes that is relatively straightforward (and therefore, under this approach, permissible) and, on the other hand, judicial treatment of statutes that is too complicated or creative to constitute “judicial work” and is therefore properly understood to constitute work that is legislative in nature. References to this idea appear with great frequency in the doctrine.248 Yet this test also fails on closer analysis.

At the outset, this approach does not—and, importantly, cannot—track existing doctrine. It is inherently incapable of doing so. This is because the standard it purports to advance is so vague and subjective that it provides essentially no guidance to courts or observers.

The Supreme Court’s decision in United States v. National Treasury Employees Union249 provides one illustration of how the Complexity Test fails to provide such guidance. In National Treasury Employees Union, the plaintiffs had challenged certain provisions of the Ethics in Government Act, which prohibited federal employees from accepting honoraria for engaging in specified types of expressive activity. The Court found the Act, as written, violated the First Amendment. This led to the question: how should the state be applied going forward?250

The Court made clear that, as a matter of substantive constitutional law, at least four readings would have responded adequately to the constitutional defect it had identified.251 While these four readings all would have cured the constitutional defect, each would have done so in a different way. The first was (arguably) the least complicated, and the fourth the most.252 If severability doctrine meaningfully tracked the complexity of the relevant analysis, the question concerning the court’s options in National Treasury Employees Union would be straightforward: at what point does this proposed treatment of the statute become too complicated? Yet severability does not (and could not) reliably track this line—a fact confirmed by the judicial chaos that ensued in response to this statute.

The Court of Appeals, planting the first stake, rejected the fourth reading as a “purely legislative act” before concluding that principles of severability required it to adopt the second reading.253 Judge Sentelle disagreed. In dissent, the judge insisted that the first reading was proper and that any other reading was “‘nothing less than judicial legislation.’”254 After the case went to the Supreme Court, the majority insisted on a still different reading—the third reading—citing “our obligation to avoid judicial legislation” as its reason for rejecting the fourth reading and overruling the Court of Appeals as to the second.255 Meanwhile, Justice O’Connor drafted a separate opinion accusing the majority of misunderstanding the severability doctrines and arguing in favor of the fourth reading, which at least eight other jurists already had rejected as invading the legislative sphere.256 Chief Justice Rehnquist, for his part, similarly endorsed the fourth reading, but refused to join Justice O’Connor’s opinion.257 Instead, he wrote his own, in which he accused the majority of (what else?) “rewrit[ing] the statute.”258

The divided and unconvincing manner in which National Treasury Employees Union came to a conclusion—one dominated by contradictory accusations of legislating from the bench—helps to demonstrate how severability analysis is fundamentally incommensurate with a distinction based on complexity of analysis. The majority’s analysis was nothing if not complicated, just like that of all the other myriad, conflicting opinions issued in this case. This failure to implement the “complexity” line is a result, at least in part, of the complexity line being too subjective and vague to be meaningful. As the finger-pointing in National Treasury Employees Union confirmed, one cannot draw a clear and consistent line between readings of a statute that are “imaginative”259 or “creative”260 or “novel”261 rather than “easily . . . articulat[ed]”262 or “relatively simple,”263 particularly when each case and challenged statute has its own details and complexities. (And, as Professor Dorf has explained, even seemingly simple statutes have their own inherent complexities.264)

Even if the Complexity Test did somehow provide meaningful guidance, its enforcement would lead to radical effects in the ways courts construe and apply statutes. As discussed above,265 a court’s refusal to engage in complicated analysis would implicate not only severability analysis, but also its ability to construe or apply any statute that happened to present complications. Yet under no conception of the judicial role are the courts limited to the application and enforcement of simple statutes.

A blanket resistance to complexity in statutory analysis also would call into question the ability of courts to engage in federal common law making. Under even the most restrictive understandings of the federal courts’ lawmaking powers, however, courts are able to develop rules of decision under certain conditions.266 In Justice Scalia’s words, courts may engage in “federal common law (in the sense of judicially pronounced law)” in areas where “‘Congress has given the courts the power to develop substantive law’”267 and also, at times, when an area of the law involves “uniquely federal interests.”268 While scholars and jurists may disagree with respect to what is necessary to trigger such conditions, the fundamental point remains. Unless one is willing to challenge settled law on the propriety of “judicially pronounced law,” one cannot justify severability’s limiting effects as a means to avoid creativity or complexity in statutory construction (both hallmarks of federal common law making). Something more is needed.

To this end, it is possible that the Complexity Test means to accomplish something slightly different. Perhaps it is meant to prevent courts from engaging in creative, complicated, or otherwise freewheeling forms of analyses due to concerns that such analysis, by definition, cannot conform with congressional intent. While that concern may be motivating the Complexity Test, it is more precisely captured in the Primary Drafter Test.

The Primary Drafter Justification. The Primary Drafter Test, which reflects the third possible justification for severability’s restrictive framework, permits a court to sever a statute, or “impose a limiting construction” on one, only if the result would “sufficiently reflect the structure and history of the statute to be attributed to Congress, rather than the court.”269 The problem with the test is not in its articulation. The Primary Draft Test, as articulated, is entirely consistent with legislative intent, and, as such, it does not pose many of the problems identified above. Rather, the problem with the Primary Draft Test is that it fails to explain why a court would ever deviate from legislative intent when construing or applying a constitutionally disrupted statute.

Stated otherwise, it is not clear how this justification supports the restrictions imposed by severability’s framework. If a court construing a statute merely is following Congress’s intent, then it is difficult to understand how the result could do anything other than “reflect the structure and history of the statute to be attributed to Congress.” If this is correct—in other words, if it is correct to conclude that Congress is the primary drafter of the statute so long as the court is effectuating legislative intent—then the Primary Drafter Test collapses into a legislative-intent test, and the former test has no independent meaning. Courts grappling with a constitutionally disrupted statute simply must effectuate legislative intent.

It nevertheless is possible that, as with the other tests, the Primary Draft test means to respond to a deeper concern: here, a concern over the propriety of a court engaging in a truly untethered, open-ended sort of decision-making. The concern tracks anxieties about overly aggressive or otherwise illegitimate forms of federal common-law making,270 and it arises in particular when a court is formulating rules of decision that lack a clear source of legislative authority.271 Under this theory, a court formulating these sorts of rules has crossed the “vague line that divides judicial interpretation from judicial legislation.”272

Although this concern over freewheeling decision-making does appear to reflect an anxiety motivating the Primary Drafter Test, it again fails to explain how a court’s treatment of a constitutionally disrupted statute should ever deviate from the effectuation of legislative intent. As discussed above, few would argue that federal courts lack lawmaking powers when Congress has so empowered them,273 which is precisely the scenario envisioned in this context.

The collapse of this third test into a legislative-intent test precludes it from justifying the ways that severability precludes courts from following legislative intent. It nevertheless helps to bring to the fore an additional concern, which relates to institutional competence.

The Institutional Competence Justification. Accommodation of institutional competence motivates the fourth possible justification. Pursuant to this theory, courts lack the competence to follow legislative intent with respect to certain constitutionally disrupted statutes. As a result, their options must be narrowed. To illustrate, the Court in Ayotte appears to be alluding to this concern when noting, in response to a constitutionally disrupted statute, that courts should be “mindful that [their] constitutional mandate and institutional competence are limited.”274

This concern is certainly a fair one. There is little question that discerning legislative intent—always a tricky endeavor—is particularly difficult in the context of a constitutionally disrupted statute.275 Yet this justification also fails to carry the weight it must. In part, this is for the same reasons that the prior justifications involving the Complexity and the Primary Drafter tests fall short: namely, legislative intent often is considered to be what defines a court’s competence. And a court normally will not refuse even to try to effectuate such intent on the grounds that the task is difficult. This has not been the courts’ standard response, for example, when faced with statutes requiring the courts to delve deeply into difficult questions of a legislature’s policies, purposes, or priorities.276

In addition, if the concern is that the courts will get the answer wrong, that concern exists even if the courts’ options are narrowed. (Indeed, if anything, it applies with greater force if court’s options are narrowed, given the possibility that Congress would have intended one of the prohibited responses.) Some jurists and commentators nevertheless treat less controversial forms of severability as not posing these same concerns. The dissent did so in National Federation, for example, when it criticized the majority’s “revis[ion]” of the statute as one that

creates a debilitated, inoperable version of health-care regulation that Congress did not enact and the public does not expect. It makes enactment of sensible health-care regulation more difficult, since Congress cannot start afresh but must take as its point of departure a jumble of now senseless provisions, provisions that certain interests favored under the Court’s new design will struggle to retain. And it leaves the public and the States to expend vast sums of money on requirements that may or may not survive the necessary congressional revision.277

This all may be true. Yet the dissent’s proposed solution—declaring the ACA inseverable and invaliding it in full278—also would have “created a debilitated, inoperable version of health-care regulation that Congress did not enact and the public does not expect.”279 It also would have replaced Congress’s enacted statutory regime with a different statutory regime (i.e., the comprehensive legal regime against which the ACA was enacted), one that “certain interests favored under the Court’s new design [would] struggle to retain” and with respect to which others would “expend vast sums of money” to achieve compliance, notwithstanding the possibility of congressional revision.280 And despite the dissent’s attempt at differentiating between the situations facing Congress in the case of partial invalidation versus total invalidation,281 the tool Congress has at its disposal to remedy judicial errors in statutory construction remains the same: it can enact statutes that fix the court’s errors.

As a final point, it may be that, for those concerned about institutional competence, it is simply inconceivable to accept the propriety of a court engaging in truly freewheeling, untethered reconceptualizing of a constitutionally defective statute. Imagine, for example, that a majority of Supreme Court justices had agreed with the dissent that the Constitution could not support the ACA’s individual mandate. One response to cure that constitutional defect would be, as discussed above,282 to completely rewrite the statute (which is founded on the participation of private insurance companies) to transform it into a single-payer scheme run entirely by the federal government. Those concerned about institutional competence might insist that it can never be correct to expect a court to engage in such a complicated reconceptualization of an already extraordinarily intricate statute.

That response is almost certainly right—but not because the court necessarily lacks competence to concoct such a regime if Congress really had intended it. Rather, it is because it is not plausible to think that the ACA could be read to reflect such legislative intent. Stated otherwise, it is not plausible to think that this statute—the ACA—can support the conclusion that, in response to a constitutional disruption, Congress would have wanted the courts to engage in a wholesale reconceptualization of the entire health-care regime. Such an extreme example, then, is less likely to support severability’s restrictive framework than it is to illustrate that, with respect to any given statute, Congress would not have desired certain responses by the courts.

The Intuitive Justification. In any event, these four identified justifications—each attempting to draw the line between what a court should and should not be doing—all fail to provide adequate support for severability’s restrictive framework. This leaves at least one final possible justification, one largely based on intuition. The idea is simple: courts cannot judicially rewrite a statute. As a result, their options when grappling with a constitutionally disrupted statute must be limited to invalidating that statute in whole or in part.

Despite the near ubiquity of this idea in the rhetoric of both courts and scholars,283 it is based on a false assumption: that limiting the courts’ options in the face of a constitutionally disrupted statute somehow affects whether they can be said to be “rewriting” that statute. Depending on how one defines this idea of “rewriting” a statute, it is either impossible for courts to engage in such behavior or impossible for them to avoid it.

On the one hand, courts have no power actually to change statutory text. No matter what the court does, the text remains in the Code, exactly the same as before. In this sense, it is impossible for a court to rewrite a statute.

On the other hand, the idea of “rewriting” a statute might refer to a court’s decision to construe or apply a statute in a nonstandard way—that is, in a way that deviates from the way it normally construes or applies a statute.284 For a textualist, for example, this might be engaging in “atextual” analysis.285 Pursuant to this understanding, a court is not literally enacting a revised statute when it engages in atextual analysis. But it is figuratively doing so. It is this understanding that frequently underlies the anxiety over “judicial legislation.”

The problem with this understanding is that there is no way for a court, faced with a constitutionally disrupted statute, to avoid nonstandard statutory analysis. Even the most serious constitutional flaw does not cause a statute simply to disappear. To the contrary, all statutes—even constitutionally disrupted ones—require construction and application. By definition, of course, constitutionally disrupted statutes cannot be construed and applied in the way they normally would be; such analysis would lead to an unconstitutional result. So the court must engage in some other form of construction and application.

Say, for example, a court engages in a very simple form of severability: it deems a provision of statute to be unconstitutional and then decides to disregard that provision when applying the statute in the future. This is a nonstandard (and atextual) treatment of the statute. Although the disregarded part of the statute remains right there in the Code, the court is treating it as though it does not exist. Alternatively, say the court decides to respond to the constitutional flaw by disregarding the statute in its entirety. The same result holds; this too is atextual. Even these two simple examples—involving the most straightforward and least controversial applications of the severability framework—therefore involve a so-called “rewriting” of the relevant statute.

Stated succinctly, if “rewriting” a statute is understood, as it commonly is, to refer to atextual or otherwise nonstandard statutory analysis, then there is no way for a court to avoid “rewriting” a constitutionally disrupted statute. As such, a desire to avoid judicial rewriting of statutes, without more, cannot support severability’s restrictive framework.

None of the identified justifications, then, provides adequate cover for severability’s restrictive framework. And neither the case law nor the academic literature appears to have provided any additional justification. Given that severability’s framework nevertheless undermines legislative intent and the statutes that are implicated, this unjustified doctrine should be abolished.286

B.     A Proposal for a New Framework

In severability’s place, courts should turn to the principle purporting to animate this same doctrine: legislative intent. But instead of accepting severability’s artificial limitations on how to effectuate this intent, a court should conduct a broader inquiry. The court should determine, first, the various constructions of a constitutionally defective statute that would diffuse its constitutional defect and, second, which among these options the legislature would prefer. The court then should implement the legislatively preferred option.

Adoption of this regime would produce results more closely aligned with those described above in Part II.B. As such, it would resist pernicious trends in the case law toward greater and more frequent invalidation of statutes. The Court’s controversial treatment of the Sentencing Reform Act in Booker would become the standard,287 rather than the anomaly, and blockbuster opinions such as Shelby County, Chadha, and Stern would be less likely to wreak havoc on the statutes they are scrutinizing.288 If this approach had been applied to preexisting cases, Part 4 of the Voting Rights Act very well might still be in partial force; executive power might be balanced in a manner that more closely tracks the intentions of the legislature that originally granted that power; and innumerable individuals attempting to rely on the national system of bankruptcy courts might with more confidence trust the Stern Court’s promise that its constitutional ruling was far from earthshattering in its effects.289 The list goes on.290

With respect to form, one might see more rulings akin to that reached in a case like United States v. Thirty-Seven Photographs.291 In this case, the Court recognized the constitutional need for an obscenity statute to include certain time limits that were lacking in the statute’s plain text. Construing this problem as an issue of constitutional avoidance—perhaps to avoid the criticism that would arise if it had invoked severability—the Court refused to declare the entire statute invalid. (Tellingly, it cited the statute’s severability clause in support.) It instead concluded that it would develop and impose the missing time limits itself. Explaining that it saw “no reason for declining to specify the time limits which must be incorporated into [the statute]—a specification that is fully consistent with congressional purpose and that will obviate the constitutional objections raised by claimant,”292 the Court set forth a detailed regime for the statutorily unspecified time limits,293 and began applying the statute, so construed, where it could be applied.294

The approach of Thirty-Seven Photographs is characterized by judicial creativity and a willingness to expand on the statutory text. As such, it is reminiscent of the “judicial legislation” that so many find offensive and makes this 1971 case feel like an anachronism. Yet in a world where everyone on the Supreme Court agrees that “legislative intent” must control severability analysis,295 the approach of Thirty-Seven Photographs is normatively defensible where the current regime is not.

Under the improved regime, of course, many rulings would remain the same. In response to a given constitutional disruption, Congress very well may prefer a response authorized by the traditional severability framework: that is, invalidation of the challenged statute either in whole or in part. And on the flip side, rulings under this improved analytical regime would rarely, if ever, lead to results so radical as to become unrecognizable under the current regime. To return to the ACA-based hypothetical discussed above,296 this improved analytical regime would not justify a court’s development of single-payer health-care regime to cure the ACA’s constitutional defects. The explanation for why it would not, however, is not based on arbitrary restrictions. As discussed above, it boils down to legislative intent. Under any reasonable reading of the ACA, it seems clear that the legislative preference would be for the implementation of other options (up to and including complete invalidation of the ACA) over the requirement that the courts develop, from scratch and in unprecedented fashion, an extraordinarily complicated single-payer regime.297

Discussion of this ACA hypothetical nevertheless helps to provide greater nuance to the improved analytical regime. Courts employing this new regime reasonably could exercise a presumption, for example, that Congress does not intend wholesale revisions of a statute in response to that statute’s constitutional disruption. Congress would need to overcome this presumption before a court would wade into such rocky and difficult waters. A so-called severability clause,298 by contrast, might provide at least some evidence of legislative intent to preserve as many applications of a statute as possible, even when that preservation would require construing the statute in a more creative (or less text-bound) manner. These sorts of presumptions and understandings would help courts work through the inevitably difficult task of discerning legislative intent in the context of constitutionally disrupted statutes.

There nevertheless remain at least three concerns raised by the alternative regime.299 The first relates to whether it would give Congress an undesirable incentive to enact unconstitutional statutes. Rather than force Congress to draft its statutes within constitutional bounds or suffer the consequence of either partial or total invalidation, the improved regime would allow Congress to draft flawed statutes and then rely on the courts to implement its preferred alternative.300 To be clear, the concern is not that the new regime would require courts to enforce a statute’s unconstitutional applications. Rather, it is that the new regime would not adequately, in a sense, punish Congress—through greater damage to statutes than otherwise would be necessary—for failing to draft legislation that is in accord with court-developed constitutional law. Under this theory, a more accommodating regime might lead to poorly drafted statutes and confusion with respect to lines of accountability.

At the outset, this first concern assumes it is the job of the courts to discipline Congress with respect to its efforts to comply with the Constitution. Accepting this assumption for the sake of argument,301 it quickly becomes clear that severability is an exceedingly clumsy tool for such a task. It is overinclusive, as it threatens to dismantle even those statutes that Congress clearly has enacted in an effort to comply with the Constitution.302 It also ignores the reality that constitutional law doctrines, which are constantly changing, occasionally develop in ways that Congress may be incapable of predicting.303 Given that it is unrealistic to expect Congress accurately to predict every future constitutional ruling, punishing Congress for failing at this task becomes both chilling and arbitrary.

Using severability as a tool to discipline Congress also is underinclusive. As revealed above by the hypothetical statute violating the Ex Post Facto Clause,304 a legislature hoping to punt a constitutional determination to the court often can do so, consistent with severability’s restrictive framework, simply by drafting the statute in a manner that allows for partial invalidation. And if the lines of accountability are blurred under the new regime, then they are blurred under severability as well—under either regime, the court is forced to articulate the relevant line.305

While it is beyond the scope of this Article to develop or defend a better way of policing Congress (assuming, again, this is even properly within the courts’ role), a more sensible way of accomplishing this end might be through substantive constitutional doctrines that reject overly broad or vague statutes when the imprecision itself is what threatens to cause some particular harm. The courts have developed such a doctrine in the context of the First Amendment, for example, which helps to explain why so many successful First Amendment challenges result in the total invalidation of the implicated statute rather than its partial invalidation306—and why the new regime likely would have little effect on statutes challenged on First Amendment grounds.307 In short, the first concern not only depends on a contested assumption about the role of the courts vis-à-vis Congress; it also relies on a model for policing Congress’s work that is both damaging and ineffectual.

A second concern raised by this Article’s alternative regime relates to abuse not by Congress, but by the courts. Under this theory, severability’s restrictions make it more difficult for judges to implement their personal policy preferences under the guise of effectuating legislative intent.

Like the first concern, this second concern rests on a disputed assumption about courts and the doctrines they employ. Here, the assumption is that one role doctrine must play is to prevent courts from acting in bad faith. Assuming (again for the sake of argument) that this assumption is appropriate, severability again proves itself inadequate. Severability does not even attempt to discriminate between, on the one hand, judges who are actually trying to effectuate legislative intent and, on the other, those who would use that same project as mere cover. Severability instead ties the hands of all jurists, including those who try their hardest to remain faithful to the will of the legislature.308

Severability also fails to rein in disingenuous conduct. Judges willing to apply doctrines strategically to achieve certain statutory results have many tools at their disposal: they can insist on labelling their analysis not severability but something else (such as “remedy”) that allows them more freedom; they can develop substantive constitutional law doctrines that, in effect, compel the results they desire; and they, of course, can use severability itself as a way of achieving the destructive ends criticized in this Article. For courts committed not to following the law but rather to reaching certain outcomes, severability’s restrictions are both easy to circumvent and subject to manipulation—yet, with respect to this second concern, this is precisely the sort of conduct that severability purports to restrain.

The third concern over the new regime relates to the difficulty of the task facing the court. Attempting to discern legislative intent with respect to a constitutionally disrupted statute is never easy. Expanding the range of possible outcomes (a step that is integral to the improved regime) may help to ensure that legislative intent actually can be effectuated, at least in theory. But on a practical level, it makes the judicial task more difficult.

Under the new regime, courts would be able to use the analytical tools that always have been at their disposal in their attempts to discern legislative intent. A shift in the doctrine, moreover, would help courts develop new tools for effective statutory construction, as it would enlist litigants, judges, scholars, and others in the joint project of figuring out how best to answer these difficult questions. Despite all this, one must admit: the exercise very well may be irreducibly challenging.

While such challenges are important to recognize, they are no defense of the current regime. Difficulty of analysis is hardly grounds for refusing to respond properly to a constitutionally disrupted statute. As Judge Katzmann recently has explained in a related context, “to jettison the inquiry altogether, because of the difficulty in particular cases, means that judges will interpret statutes unmoored from the reality of the legislative process and what the legislators were seeking to do.”309 In the context of severability, such unmooring comes at the direct cost of the democratic will.

Still, one might remain worried about the possibility of error. Particularly when faced with such a difficult task, there is always the possibility that the court will fail to reach the correct conclusion. In this context, error results in the implementation of a statute in a manner that Congress never intended.

Again, this is a legitimate concern. Yet here the shift away from severability actually betters the situation, as the improved regime opens up the possibility that the courts will answer the question of legislative intent correctly whenever Congress prefers an approach that the severability framework refuses even to consider. Moreover, the improved regime allows Congress the same tool it always has in response to statutory errors by the courts: the “congressional ‘override.’”310 While practical realities may at times thwart Congress’s employment of this tool,311 the question then becomes one of preferable defaults. Is it better to accept, as a default, a court’s attempt at discerning legislative intent through an arbitrarily restrictive lens, such as that imposed by severability? Or is it better to demand that the court attempt to discern legislative intent without those arbitrary constraints? If the goal is to minimize judicial error, the answer is clear. The proposed regime—one dedicated to the effectuation of legislative intent without the arbitrarily restrictive framework of severability—is an improved regime.

IV.     Conclusion

Severability elicits widespread criticism. Yet rarely do courts or scholars question whether its framework should apply in the first place. In tackling this underexamined question, this Article has attempted to show why, contrary to near universal consensus, the framework of severability should be retired from the doctrine and replaced by framework a more accommodating of legislative intent.

It is true that courts routinely employ severability analysis, without objection or even deliberation, after they conclude that a statute suffers from some constitutional flaw. Many scholars have joined the courts in reflexive acceptance. Once examined, however, the severability framework reveals its deeply troubling nature. It is an aggressively disruptive approach to statutory construction that encourages courts to dismantle legislation without adequate justification, all under the guise of judicial restraint. The source of the problem ultimately lies in the insistence—implicit but nevertheless fundamental to the severability framework—that a properly restrained court may respond to a constitutionally flawed statute in only one of two ways. Either it must disregard the statute either in whole, or in part. This artificial dichotomy limits the broad array of analytical tools that otherwise would be available to a court seeking to effectuate legislative intent.

In light of its unjustifiably disruptive effects, severability should be abandoned. Courts instead should turn to a central goal purporting to animate the severability doctrine: limiting the “solution to the problem” through resort to legislative intent. A fundamentally broader inquiry into legislative intent—one that forgoes severability’s artificial restrictions—would require that a court identify, first, the constructions of a constitutionally defective statute that would diffuse its constitutional defects, and second, which among these options the legislature would prefer. The court should then implement the legislatively preferred option.

Adoption of this regime would effectuate legislative intent, rather than reject it in favor of severability’s unjustified prohibitions on what the court may do. As such, it would resist pernicious practices in the case law, including those worsened by emerging trends toward greater and more frequent invalidation of statutes. It would, in short, better fulfill severability’s broken promise: to minimize the damage done to our statutes by judicial fiat.

  1. [1]. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012).

  2. [2]. Id. at 2676 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).

  3. [3]. Id. at 2668.

  4. [4]. Id. at 2676.

  5. [5]. Id. at 2668.

  6. [6]. Id.

  7. [7]. See, e.g., Kevin C. Walsh, Partial Unconstitutionality, 85 N.Y.U. L. Rev. 738, 740 (2010) (“Current law and scholarship [maintain] that severability doctrine is the exclusive way to deal with partial unconstitutionality.”); see generally infra Part II.A (exploring the reflexive acceptance that severability enjoys from both scholars and the courts).

  8. [8]. See Shelby Cty. v. Holder, 133 S. Ct. 2612 (2013); infra Part II.B.

  9. [9]. See Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165 (2014); infra Part II.B.

  10. [10]. See Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2566; infra Part II.B.

  11. [11]. See, e.g., Mark L. Movsesian, Severability in Statutes and Contracts, 30 Ga. L. Rev. 41,
    41–42 (1995) (compiling list of scholarly works that, collectively, have criticized severability “on almost every conceivable basis”); see also Tom Campbell, Severability of Statutes, 62 Hastings L.J. 1495, 1497 n.3 (2011) (same).

  12. [12]. See infra notes 44–59 and accompanying text.

  13. [13]. See infra notes 35–43 and accompanying text.

  14. [14]. This third area of consensus may strike some as surprising, given that it addresses the role that legislative intent (an oft-maligned concept in the world statutory interpretation and construction) plays in the analysis. See, e.g., Conroy v. Aniskoff, 507 U.S. 511, 528 (1993) (Scalia, J., concurring) (“[The Court] should not pretend to care about legislative intent (as opposed to the meaning of the law)....”); see also id. at 519 (“We are governed by laws, not by the intentions of legislators.”). Yet legislative intent plays a central role in every articulation of the severability test, and the eight members of the current Supreme Court (as well as the most recently departed member of the Court, Justice Antonin Scalia) agree that severability requires a court to discern and effectuate legislative intent. See infra note 41 and accompanying text.

  15. [15]. See, e.g., Michael C. Dorf, Facial Challenges to State and Federal Statutes, 46 Stan. L. Rev. 235, 292 (1994) (discussing principles of “legislative supremacy” and “judicial restraint” and their interaction with the “function-oriented severability test”); see generally infra Part II.B (describing how courts and scholars invoke judicial restraint in the context of severability).

  16. [16]. See infra Part III.A.1.

  17. [17]. See infra Part III.A.2.

  18. [18]. See infra Part III.A.3.

  19. [19]. See infra Part III.A.4.

  20. [20]. See infra Part III.A.5.

  21. [21]. See infra Part III.B.

  22. [22]. See infra notes 146–50 and accompanying text (discussing Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477 (2010)).

  23. [23]. See infra Part II.A (identifying these analytical approaches and explaining how each is used in other legal contexts).

  24. [24]. See, e.g., Richard H. Fallon, Jr., As-Applied and Facial Challenges and Third-Party Standing, 113 Harv. L. Rev. 1321, 1333 (2000) (referring, in the context of severability, to the “well-known principle [that] establishes that any separation of a statute through its specification into subrules must not cross the vague line that divides judicial interpretation from judicial legislation”); David H. Gans, Severability as Judicial Lawmaking, 76 Geo. Wash. L. Rev. 639, 641 (2008) (“Although [many cases] and much of the scholarly writing make the point that courts should not save statutes by rewriting them, the decisions in the more recent cases seem to do just that.” (footnote omitted)); see also infra Part II.A (identifying analytical approaches and explaining how each, in the context of severability, is treated as outside the judicial function).

  25. [25]. See, e.g., Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165 (2014); Shelby Cty. v. Holder, 133 S. Ct. 2612 (2013); Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012); see also infra Part II.B.

  26. [26]. See infra Part III.B.

  27. [27]. See infra Part III.A.

  28. [28]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 328 (2006).

  29. [29]. See Shelby Cty., 133 S. Ct. at 2612; see also infra Part III.B.

  30. [30]. Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2566; see also infra Part III.B.

  31. [31]. Stern v. Marshall, 564 U.S. 462 (2011); see also Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165 (2014); infra Part III.B.

  32. [32]. See infra note 57 and accompanying text.

  33. [33]. See infra notes 44–59 and accompanying text.

  34. [34]. See infra Part II.B.

  35. [35]. United States v. Nat’l Treasury Emps. Union, 513 U.S. 454, 487 (1995).

  36. [36]. See, e.g., Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 506 (1985) (inseverable); Immigration & Naturalization Serv. v. Chadha, 462 U.S. 919, 934 (1983) (severable). It is worth acknowledging that it is not obvious ahead of time which statutory unit is at issue—particularly when the statute in question is complex and has been subject to amendment—and the analysis can be conducted multiple times.

  37. [37]. See, e.g., Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012); Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477 (2010); Alaska Airlines, Inc. v. Brock, 480 U.S. 678 (1987); Brockett, 472 U.S. at 491; Regan v. Time, Inc., 468 U.S. 641 (1984); Chadha, 462 U.S. at 919. But cf. United States v. Booker, 543 U.S. 220 (2005). See infra notes 124–32 for a discussion of how Booker is an outlier in this area. As discussed below, Eric Fish has offered a more nuanced definition of severability that renders its use more limited. See infra note 62 and accompanying text.

  38. [38]. In this context, of course, legislative intent is most often understood in the conditional, as severability asks “what the legislature would have done, not what the legislature actually did [do],” in response to the statute’s constitutional defect. Walsh, supra note 7, at 740. One might refer to this subspecies of legislative intent as “contingent legislative intent,” a term this Article proposes and which in concept has been recognized by many. See, e.g., Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2607 (articulating the relevant legislative intent standard in the conditional); Walsh, supra note 7, at 740–41. However, because this Article does not wade into the distinction between contingent legislative intent and non-contingent legislative intent, it uses the generic term “legislative intent” to refer to the inquiry into intent that governs under the leading definitions of severability analysis.

  39. [39]. Transcript of Oral Argument at 9, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (No. 11-393).

  40. [40]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006); see also Transcript of Oral Argument at 9, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (No. 11-393) (“[E]very one of [the Supreme Court’s cases on severability] talks about congressional intent.”).

  41. [41]. See supra note 14 and accompanying text for a general discussion of objections to legislative intent. With respect to how, by contrast, legislative intent is accepted in the context of severability, National Federation of Indep. Bus. provides a recent illustration. In this case, every Justice signed onto an opinion confirming the central relevance of legislative intent. See Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2668–69 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting) (describing the “framework for severability analysis” as requiring the Court to “determine[] whether the now truncated statute will operate in the manner Congress intended” and to determine whether, “even if the remaining provisions can operate as Congress designed them to operate,... Congress would have enacted them standing alone and without the unconstitutional portion.”); id. at 2608 (plurality opinion) (basing its severability decision on its conclusion that “Congress would have wanted to preserve the rest of the Act”). It is interesting to note that Justice Scalia helped to author the joint dissent even after pushing back during oral argument on the reliance on legislative intent. See Transcript of Oral Argument at 9, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (No. 11-393) (counsel for challengers noting that “every one of [this Court’s cases on severability] talks about congressional intent”); see also id. at 10 (Justice Scalia expressing agreement with this characterization of the doctrine before asking “but is it right?”).

  42. [42]. See infra note 57 and accompanying text.

  43. [43]. See generally Emily Sherwin, Rules and Judicial Review, 6 Legal Theory 299, 299–308 (2000) (providing overview of doctrine).

  44. [44]. This may not always have been the case. According to Professor Walsh, for example, a regime of displacement preceded the “modern hypothetical-intent-based severability doctrine.” Walsh, supra note 7, at 742.

  45. [45]. See Florida v. Dep’t of Health & Human Servs., 132 S. Ct. 604 (Mem.) (2011).

  46. [46]. United States v. Booker, 543 U.S. 220 (2005).

  47. [47]. Id. at 229 n.1.

  48. [48]. Stern v. Marshall, 564 U.S. 462 (2011). The follow-up case was Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165 (2014).

  49. [49]. Transcript of Oral Argument at 13, Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165 (2014) (No. 12-1200).

  50. [50]. Id.

  51. [51]. Shelby Cty. v. Holder, 133 S. Ct. 2612 (2013).

  52. [52]. Id. at 2648 (Ginsburg, J., dissenting).

  53. [53]. Walsh, supra note 7, at 745; see also id. at n.27 (describing several cases that fall into this category).

  54. [54]. See, e.g., Immigration & Naturalization Serv. v. Chadha, 462 U.S. 919, 931 (1983) (accepting, without explanation, that severability analysis applies when determining the effect that a ruling of unconstitutionality has on a statute); Buckley v. Valeo, 424 U.S. 1, 108 (1976) (same); cf. Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 508 (2010) (providing, as explanation for its turn to severability, that “[g]enerally speaking, when confronting a constitutional flaw in a statute, we try to limit the solution to the problem,” and that “[t]he unconstitutionality of a part of an Act does not necessarily defeat or affect the validity of its remaining provisions”).

  55. [55]. See Walsh, supra note 7, at 739–40 (emphasis added). As discussed below, Professor Walsh is one of the very few who has actually questioned severability’s propriety as a framework. See infra notes 60–61 and accompanying text.

  56. [56]. Michael C. Dorf, Fallback Law, 107 Colum. L. Rev. 303, 370–71 (2007).

  57. [57]. Movsesian, supra note 11, at 41. See also id. at 41–42 (“Commentators have condemned severability doctrine as too malleable and as too rigid; as encouraging judicial overreaching and as encouraging judicial abdication. They have criticized the doctrine’s reliance on legislative intent and its disregard of legislative intent; its excessive attention to political concerns and its inattention to political concerns; its lack of any coherent explanation.” (footnotes omitted)).

  58. [58]. See, e.g., Dorf, supra note 15 (exploring interactions among severability and related doctrines); Dorf, supra note 56 (exploring “fallback law,” which includes, in its most common manifestation, severability clauses); Fallon, supra note 24 (providing insight into the ways that principles of severability interact with principles of substantive constitutional law to affect the courts’ treatment of so-called “facial challenges”); Richard H. Fallon, Jr., Fact and Fiction About Facial Challenges, 99 Calif. L. Rev. 915, 955 (2011) (same); Scott A. Keller & Misha Tseytlin, Applying Constitutional Decision Rules Versus Invalidating Statutes in Toto, 98 Va. L. Rev. 301, 347 (2012) (resisting efforts to conflate questions of severability analysis with questions of as-applied, facial, and overbreadth challenges); Gillian E. Metzger, Facial Challenges and Federalism, 105 Colum. L. Rev. 873, 894–913 (2005) (exploring how severability informs courts’ treatment of facial challenges in the context of challenges to legislation as exceeding Congress’s powers); Sherwin, supra note 43, at 299 (exploring the extent to which competing principles of judicial review, including principles of severability, are able to fulfill their promise of “a modest conception of the role of courts in government”); Adrian Vermeule, Saving Constructions, 85 Geo. L.J. 1945, 1945–78 (1997) (connecting severability to the canon of constitutional avoidance).

  59. [59]. On similarly rare occasions, a scholar will reach a conclusion largely consistent with this Article’s, but without questioning severability’s prominent role. Professor Glenn Smith, for example, has proposed that courts might better effectuate legislative intent—particularly in the context of statutes containing a legislative veto—by engaging in “alternative approaches” to severability, including more precise forms of text severability and more flexible forms of application severability, rather than insist on the most rigid forms of the severability framework. See Glenn Chatmas Smith, From Unnecessary Surgery to Plastic Surgery: A New Approach to the Legislative Veto Severability Cases, 24 Harv. J. on Legis. 397, 400, 476 (1987).

  60. [60]. Walsh, supra note 7, at 755, 778. More specifically, Professor Walsh proposes an abandonment of legislative-intent-based severability analysis in favor of a regime whereby, regardless of legislative intent, the Constitution simply displaces as any statutory application that is in conflict. Id. at 778.

  61. [61]. Id. at 790 (“[T]he impact on case outcomes would be modest....”). A second scholar who, in a sense, has questioned the propriety of the severability framework is Tom Campbell. Professor Campbell takes an extreme position of arguing that separation-of-powers principles require that courts treat all statutes as inseverable, under the theory that anything else constitutes an encroachment on the legislative sphere. Campbell, supra note 11, at 1496–97.

  62. [62]. Eric S. Fish, Choosing Constitutional Remedies, 63 UCLA L. Rev. 322, 351 (2016).

  63. [63]. Stated otherwise, to the extent that the severability framework applies only after the court already has decided that it should disregard some or all portions of a statute—and, as such, only after the court already has concluded that the proper response to a constitutional disruption is not to take one of the approaches prohibited by the severability framework—that characterization avoids the fundamental flaws plaguing current doctrine and is largely consistent with this Article’s normative conclusions. See generally supra Part III.

  64. [64]. See supra notes 44–56 and accompanying text.

  65. [65]. Compare Gans, supra note 24, at 688 (arguing that severability should be considered a question of remedy), with John Harrison, Severability, Remedies, and Constitutional Adjudication, 83 Geo. Wash. L. Rev. 56 (2014) (rejecting characterization of severability as a question of remedy and concluding that “[s]everability analysis is statutory construction in light of a conclusion of unconstitutionality”). See generally, Eric S. Fish, Severability as Conditionality, 64 Emory L.J. 1293 (2015) (discussing three theories of severability and proposing a fourth).

  66. [66]. The arguments advanced by this Article are, in a sense, actually easier to make if severability is considered to be an analysis regarding remedy. This is because the severability framework purports to impose restrictions on courts that normally do not apply in the context of “remedy.” For illustrations of this phenomenon, see infra notes 114–19 and accompanying text.

  67. [67]. See generally supra Part II.A (describing the reflexive acceptance severability enjoys). Perhaps it is the seeming dearth of other options that explains the lack of analysis addressing whether severability is the appropriate framework for analysis. See supra notes 55–56 and accompanying text.

  68. [68]. As a court addressing the constitutional challenge to the Affordable Care Act put it, “[s]everability is a doctrine of judicial restraint.” Florida ex rel. Bondi v. U.S. Dep’t of Health & Human Servs., 780 F. Supp. 2d 1256, 1307 (N.D. Fla. 2011), aff’d in part, rev’d in part sub nom. Florida ex rel. Attorney Gen. v. U.S. Dep’t of Health & Human Servs., 648 F.3d 1235 (11th Cir. 2011), aff’d in part, rev’d in part sub nom. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012). The judge in this case offered this observation in the course of striking the ACA down in its entirety. See also United States v. Booker, 543 U.S. 220, 303 (2005) (Stevens, J., dissenting in part) (discussing, in the context of severability, “the tradition of judicial restraint that has heretofore limited our power to overturn validly enacted statutes”); Dorf, supra note 15, at 292.

  69. [69]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 329 (2006).

  70. [70]. Booker, 543 U.S. at 282 (Stevens, J., dissenting in part). See also id. at 284 (arguing that the unusual approach taken by the majority “expands, rather than limits, judicial power”).

  71. [71]. Ayotte, 546 U.S. at 330–31.

  72. [72]. Id. If the Court of Appeals had concluded it was the former, according to the Supreme Court, it should have issued “an injunction prohibiting the statute’s unconstitutional application.” See id. at 331–32. If the latter, it should have expressly held “that consistency with legislative intent requires invalidating the statute in toto.” See id. at 332.

  73. [73]. Id. at 328; see also Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 508 (2010) (quoting Ayotte, 546 U.S. at 328–29); Booker, 543 U.S. at 302 (Stevens, J., dissenting in part) (“No judicial remedy is proper if it is ‘not commensurate with the constitutional violation to be repaired.’” (quoting Hills v. Gautreaux, 425 U.S. 284, 294 (1976))).

  74. [74]. See, e.g., United States v. Nat’l Treasury Emps. Union, 513 U.S. 454, 479 (1995); Vivid Entm’t, LLC v. Fielding, 774 F.3d 566, 574 (9th Cir. 2014) (“Because a court may not use severability as a fig leaf for judicial legislation, courts have fashioned limits on when a statute may be severed.”).

  75. [75]. See, e.g., Merrill Lynch, Pierce, Fenner & Smith Inc. v. Dabit, 547 U.S. 71, 85 (2006) (referring favorably to “judicial interpretations” (internal quotation marks omitted)); Seminole Tribe of Fla. v. Florida, 517 U.S. 44, 87 (1996) (Stevens, J., dissenting) (referring to a Bivens claim as a “judicially fashioned remedy”); United States v. Allen, 625 F.3d 830, 836 (5th Cir. 2010) (referring to the exclusionary rule as a “judicially fashioned remedy”).

  76. [76]. Read the Affordable Care Act, HealthCare.gov, https://www.healthcare.gov/where-can-i-read-the-affordable-care-act (last visited May 1, 2016) (containing the full text of the Affordable Care Act).

  77. [77]. For further discussion of this hypothetical, see infra note 270 and accompanying text; see also infra note 297.

  78. [78]. For a helpful discussion of the distinctions between “statutory interpretation” and “statutory construction,” see generally Lawrence B. Solum, The Interpretation-Construction Distinction, 27 Const. Comment. 95 (2010).

  79. [79]. See supra note 78 (addressing distinction between statutory construction and statutory interpretation). Another layer of confusion exists because it is not clear if this theory is meant merely to justify severability analysis, or if it is meant to impose an independent bar on what a court can do under the guise of severability analysis. In other words, it is not clear if adherents to this theory believe that the basic test for severability ensures that a court is not engaged in legislative activity—or if they instead believe that while the severability test helps to police this line, it occasionally does not, in which case the court still must avoid engaging in so-called judicial legislation. The phrasing of certain judicial and academic discussions suggests that the latter understanding may be more prevalent. See, e.g., Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006) (suggesting that severability analysis cannot unfold without an independent inquiry into whether a legislature is impermissibly “rely[ing] on [the court’s] intervention” to sever the relevant statute). Yet this only leads to more fundamental confusion: if a straightforward application of the test for severability occasionally leads to impermissible judicial legislation, what is a court supposed to do in that circumstance? No authority appears to have provided a clear answer.

  80. [80]. United States v. Nat’l Treasury Emps. Union, 513 U.S. 454, 478 (1995).

  81. [81]. United States v. Booker, 543 U.S. 220, 310 (2005) (Scalia, J., dissenting in part).

  82. [82]. Id. at 279 (Stevens, J., dissenting in part).

  83. [83]. Id. at 284.

  84. [84]. See infra Part III.A.

  85. [85]. Blue-pencil test, Black’s Law Dictionary (10th ed. 2014); cf. Dada v. Mukasey, 554 U.S. 1, 27 (2008) (joint opinion of Scalia, Roberts, and Thomas, JJ., dissenting) (deeming it “not unusual for the Court to blue-pencil a statute [by] directing that one of its provisions, severable from the rest, be disregarded”).

  86. [86]. Blue-pencil test, Black’s Law Dictionary (10th ed. 2014).

  87. [87]. Nat’l Fed’n of Indep. Bus. v. Sebelius,132 S. Ct. 2566, 2667 (2012) (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).

  88. [88]. Id.

  89. [89]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006).

  90. [90]. Bowsher v. Synar, 478 U.S. 714, 736 (1986).

  91. [91]. Id.

  92. [92]. United States v. Booker, 543 U.S. 220, 284 (2005) (Stevens, J., dissenting in part).

  93. [93]. Ayotte, 546 U.S. at 329–30 (quoting United States v. Nat’l Treasury Emps. Union, 513 U.S. 454, 479 n.26 (1995)).

  94. [94]. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2607 (2012) (plurality opinion).

  95. [95]. Id. at 2677 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).

  96. [96]. Professor Fallon has described this phenomenon. See Fallon, supra note 58, at 955 (“[W]hen substantial severing would be necessary to save a statute from invalidity, the Court will [deem the statute severable] if it can identify a relatively surgically precise way of curing the defect that an applicable test has identified... but not otherwise.”); id. at 956 (“Crucially, however—and contrary to the usual understanding of the presumption of severability—the Court ordinarily will not [deem a statute severable] when the statute would require relatively substantial severing in order to survive unless it can foresee reasonably precise lines along which severance could occur.”). Eric Fish provides an insightful description of the intuitions that underlie this approach in his analysis of what he refers to as “constitutional remedies.” See Eric S. Fish, Choosing Constitutional Remedies, 63 UCLA L. Rev. 322, 339 (2016).

  97. [97]. Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 510 (2010).

  98. [98]. Transcript of Oral Argument at 74, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (No. 11-393).

  99. [99]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 329 (2006).

  100. [100]. See infra notes 251–69 and accompanying text.

  101. [101]. See infra notes 270–74 and accompanying text (discussing the Primary Drafter Test, for which the Complexity Test may be intended to be a proxy).

  102. [102]. Fallon, supra note 24, at 1333–34. If this criterion is met, then the statute can be considered “readily susceptible to such a construction” and therefore on the proper side of “the vague line that divides judicial interpretation from judicial legislation.” Id. at 1333.

  103. [103]. United States v. Booker, 543 U.S. 220, 292 (2005) (Stevens, J., dissenting in part) (quoting Sloan v. Lemon, 413 U.S. 825, 834 (1973)).

  104. [104]. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2676 (2012) (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).

  105. [105]. Fallon, supra note 24, at 1333–34.

  106. [106]. See, e.g., infra notes 274–82 and accompanying text (discussing a line possibly based on the institutional competence of the court); cf. Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2668 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting) (suggesting that a court engages in judicial legislation if its severability analysis is “automatic or too cursory”).

  107. [107]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006) (quoting United States v. Nat’l Treasury Emps. Union, 513 U.S. 454, 479 n.26 (1995)).

  108. [108]. See infra Part III.A.

  109. [109]. See infra notes 157–62 and accompanying text for a more thorough definition of application severability.

  110. [110]. A timely variation on this example might involve a law granting benefits to parents of the opposite sex, but not to parents of the same sex. Cf. Obergefell v. Hodges, 135 S. Ct. 2584 (2015) (holding that same-sex couples cannot be deprived of fundamental right to marry).

  111. [111]. This hypothetical is taken from Califano v. Westcott, 443 U.S. 76 (1979), which is discussed in more detail below.

  112. [112]. United States v. Booker, 543 U.S. 220, 325 (2005) (Thomas, J., dissenting in part) (emphasis added) (quoting United States v. Reese, 92 U.S. 214, 221 (1875)).

  113. [113]. Id.

  114. [114]. For a discussion regarding whether severability analysis itself should be considered a question of remedy, see supra notes 65–66 and accompanying text.

  115. [115]. Califano, 443 U.S. at 89–91; see also supra note 111 and accompanying text.

  116. [116]. Id. at 89.

  117. [117]. Id. (emphasis added) (quoting Welsh v. United States, 398 U.S. 333, 361 (1970)).

  118. [118]. Id. It is true that remedies of this sort may have been more common prior to the Roberts Court. Yet precedents such as Califano remain good law, with Justices continuing to rely on the underlying remedial principles. See, e.g., Levin v. Commerce Energy, Inc., 560 U.S. 413, 427 (2010). In any event, a shift in frequency does not explain the continued discrepancy in the doctrine: namely, that what is permissible in the “remedy” framework is impermissible in the “severability” framework.

  119. [119]. See supra notes 65–66 and accompanying text for a discussion regarding whether legal scholars should consider the severability analysis as a question of remedy.

  120. [120]. This language is inspired by Professor Dorf’s use of the term in the context of what he refers to as fallback law. See Dorf, supra note 56, at 305.

  121. [121]. See United States v. Booker, 543 U.S. 220, 282 (2005) (Stevens, J., dissenting in part) (emphasis added); see also id. at 322 n.10 (Thomas, J., dissenting in part).

  122. [122]. Id.

  123. [123]. Id. at 246 (plurality opinion) (describing the third approach).

  124. [124]. Id. at 246–49 (summarizing conclusions).

  125. [125]. Id. at 281–83 (Stevens, J., dissenting in part). Scholars have echoed this understanding of what occurred in Booker. See, e.g., Dorf, supra note 56, at 306 (“Although nominally using the language of severability, the Court in effect substituted advisory for mandatory Guidelines.”).

  126. [126]. Booker, 543 U.S. at 282 (Stevens, J., dissenting in part).

  127. [127]. Id. at 284.

  128. [128]. Id. at 322 n.10 (Thomas, J., dissenting in part).

  129. [129]. Id. at 247.

  130. [130]. Id. at 322 n.10 (Thomas, J., dissenting in part); see also infra Part III.B for a defense of this approach.

  131. [131]. Justice Stevens appeared accurately to describe the precedents, in other words, when he asserted that:

    There is no case of which I am aware . . . in which this Court has used ‘severability’ analysis to do what the majority does today: determine that some unconstitutional applications of a statute, when viewed in light of the Court’s    reading of ‘likely’ legislative intent, justifies the invalidation of certain statutory sections in their entirety, their constitutionality notwithstanding, in order to save the parts of the statute the Court deemed most important.

    Booker, 543 U.S. at 282 (Stevens, J., dissenting in part).

  132. [132]. Id. at 280–81 (“Our normal rule, however, is that the ‘unconstitutionality of a part of an Act does not necessarily defeat or affect the validity of its remaining provisions. Unless it is evident that the legislature would not have enacted those provisions which are within its power, independently of that which is not, the invalid part may be dropped if what is left is fully operative.’” (first emphasis added)).

  133. [133]. See supra notes 65–66 and accompanying text.

  134. [134]. See Lisa Marshall Manheim, Redistricting Litigation and the Delegation of Democratic Design, 93 B.U. L. Rev. 563, 566 (2013). In so doing, redistricting courts very well may alter district lines beyond those that may be considered to be the most immediate cause of the unconstitutionality.

  135. [135]. For the relationship between courts and explicit substitute provisions, see generally Dorf, supra note 56.

  136. [136]. Bowsher v. Synar, 478 U.S. 714, 734–36 (1986).

  137. [137]. Id. at 735–36.

  138. [138]. Id.

  139. [139]. Id. at 736.

  140. [140]. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2593 (2012) (plurality opinion).

  141. [141]. Id. at 2593.

  142. [142]. See id. at 2593–94.

  143. [143]. Id. at 2651 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting) (quoting Grenada Cty. Supervisors v. Brogden, 112 U.S. 261, 269 (1884)).

  144. [144]. Vermeule, supra note 58, at 1948.

  145. [145]. In its modern form, the canon of constitutional avoidance states that “where an otherwise acceptable construction of a statute would raise serious constitutional problems, the Court will construe the statute to avoid such problems unless such construction is plainly contrary to the intent of Congress.” Id. at 1949 (quoting Edward J. DeBartolo Corp. v. Florida Gulf Coast Bldg. & Constr. Trades Council, 485 U.S. 568, 575 (1988)). It is, of course, true that this canon requires a court to consider and, as appropriate, to accept alternate constructions of a statute prior to any formal finding of unconstitutionality. Id. at 1959. Severability analysis, by contrast, normally occurs after the finding of constitutionality. As a result, the court’s refusal to engage in this sort of analysis in the context of severability is unlikely to make a difference in any given case, given that any constitutionally curative nonstandard construction presumably already has been considered and rejected. It nevertheless provides even further support for the more general point: in the context of severability, courts and commentators dismiss otherwise-accepted analytical approaches as existing outside the judicial sphere.

  146. [146]. Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 510–13 (2010).

  147. [147]. Id. at 508–09.

  148. [148]. Id. at 509–10 (emphasis added).

  149. [149]. To the contrary, by referring to one of the approaches as “blue-pencil[ing,]” the Court appears to be acknowledging that such a step would constitute nothing more than disregarding a statute in part, which was the same approach it did take with respect to the language providing for good-cause removal. Id. at 509.

  150. [150]. This same sentiment has been echoed recently in other contexts, as when Justice Scalia, during oral argument for National Federation, suggested the Court should adopt whatever approach to severability “reduces our options the most.” Transcript of Oral Argument at 74, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (No. 11-393). See also, e.g., Ayotte v. Planned Parenthood N. New Eng., 546 U.S. 320, 329–30 (2006).

  151. [151]. Transcript of Oral Argument at 74, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (No. 11-393).

  152. [152]. See, e.g., Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts xxii (2012) (“[T]he more the interpretive process strays outside a law’s text, the greater the interpreter’s discretion.”); A Conversation on the Constitution: Perspectives from Active Liberty and a Matter of Interpretation, Justices Breyer and Scalia Converse on the Constitution, Am. Const. Soc’y (Dec. 5, 2006), http://www.acslaw.org/news/video/justices-breyer-and-scalia-converse-on-the-constitution; Dahlia Lithwick, Justice Grover Versus Justice Oscar, Slate (Dec. 6, 2006 4:31 PM), http://www.slate.com/articles/news_and_politics/jurisprudence/2006/12/
    justice_grover_versus_justice_oscar.html (summarizing debate over judicial methods with the following description: “[f]or each time Breyer says his own constitutional approach is ‘complicated’ or ‘hard,’ Scalia retorts that his is ‘easy as pie’ and a ‘piece of cake’”).

  153. [153]. See, e.g., Pharm. Research & Mfrs. of Am. v. Walsh, 538 U.S. 644, 676 (2003) (Thomas, J., concurring in the judgment) (criticizing the majority’s analysis of the Medicaid Act on the ground that the analysis “ignore[d] [the] complexity” of the complicated statute).

  154. [154]. See, e.g., Caleb Nelson, The Legitimacy of (Some) Federal Common Law, 101 Va. L. Rev. 1, 20–21 (2015) (“In Justice Scalia’s words, a court that articulated this sort of federal common law would be exercising ‘substantive lawmaking power,’ and federal courts enjoy such power only to the extent that something in written federal law delegates it to them.” (quoting Sosa v. Alvarez-Machain, 542 U.S. 692, 742 (2004) (Scalia, J., concurring in part and concurring in the judgment))).

  155. [155]. See Metzger, supra note 58, at 886; see also United States v. Booker, 543 U.S. 220, 320 (2005) (Thomas, J., dissenting in part) (“The severability issue may arise when a court strikes either a provision of a statute or an application of a provision. These forms of less-than-total invalidation are typically (if awkwardly) referred to as ‘application severability’ and ‘provision severability’ respectively.”).

  156. [156]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 329 (2006) (citation omitted).

  157. [157]. Id. (citation omitted).

  158. [158]. This hypothetical is adopted from Professor Vermeule’s use of the same. Vermeule, supra note 58, at 1955.

  159. [159]. Id.

  160. [160]. Id.

  161. [161]. Ayotte, 546 U.S. at 329 (quoting Dahnke-Walker Milling Co. v. Bondurant, 257 U.S. 282, 289 (1921)).

  162. [162]. See, e.g., United States v. Booker, 543 U.S. 220, 321–22 (2005) (Thomas, J., dissenting in part) (discussing the use of application severability, both explicit and implicit, in Supreme Court case law).

  163. [163]. Id. at 321.

  164. [164]. Dorf, supra note 56, at 313 n.32.

  165. [165]. Id. In support of this observation, Professor Dorf cites Ayotte v. Planned Parenthood of Northern New England, a case in which the Court “warned... in the context of application severability” that a concern over “limited ‘institutional competence’ leads the Justices to ‘restrain’ themselves ‘from [rewriting] state law to conform it to constitutional requirements[] even as [they] strive to salvage it.’” See id. at 311–12 (quoting Ayotte, 546 U.S. at 329).

  166. [166]. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2667–71 (2012) (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).

  167. [167]. Id. at 2604–07 (plurality opinion).

  168. [168]. Id. at 2607. In other words, the Court required that, going forward, § 1396c be read not to authorize certain applications (i.e., any application by the Secretary that would deprive a state of existing funds because it had failed to comply with a certain requirements set by the ACA) but nevertheless be read to authorize other applications.

  169. [169]. Id. at 2676 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting).

  170. [170]. Id. at 2667 (emphasis added).

  171. [171]. See supra notes 162–63 and accompanying text.

  172. [172]. Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 509–10 (2010).

  173. [173]. Id. at 510. More specifically, the Court rejected the idea that the Court could restrict the Board’s enforcement powers (which could be done through application severability) to the extent necessary to render it in compliance with the Constitution. Importantly, the Court rejects this possibility not because Congress would have preferred the approach the Court did take, but rather because, in the Court’s view, application severability of this sort would invade the legislative sphere.

  174. [174]. Roper v. Simmons, 543 U.S. 551 (2005).

  175. [175]. See generally id.

  176. [176]. See, e.g., United States v. Salerno, 481 U.S. 739, 745 (1987) (discussing facial challenges); see generally Fallon, supra note 24 (discussing distinctions between facial and as-applied challenges).

  177. [177]. Crawford v. Marion Cty. Election Bd., 553 U.S. 181 (2008).

  178. [178]. See generally id.

  179. [179]. Id. at 188–89, 200 (limiting the holding to petitioners’ facial challenge).

  180. [180]. See, e.g., Wash. State Grange v. Wash. State Republican Party, 552 U.S. 442, 450 (2008) (“Facial challenges are disfavored for several reasons.”). But cf. Fallon, Fact and Fiction About Facial Challenges, supra note 58 (questioning whether the Supreme Court’s purported resistance to facial challenges accurately describes how the Court actually decides its cases).

  181. [181]. Wash. State Grange, 552 U.S. at 450.

  182. [182]. Professor Dorf, among others, has recognized this connection. See Dorf, supra note 15, at 249.

  183. [183]. Id. In this passage, Professor Dorf does not necessarily criticize this practice on this basis; he simply characterizes it as such.

  184. [184]. As Professor Fallon has recognized, “[i]n the absence of severability, all challenges to statutes would necessarily be facial, for a nonseverable statute that was invalid in some cases would necessarily be invalid as to all.” Fallon, Fact and Fiction About Facial Challenges, supra note 58, at 953. Although Professor Fallon was referring to all forms of severability, the point also holds with respect to application severability on its own. In the absence of application severability, all challenges to statutes would necessarily be facial—except with respect to those statutes that happened to have been drafted in a manner that allows, in response to the particular constitutional challenge, for text severability.

  185. [185]. See supra notes 35–37 and accompanying text.

  186. [186]. See supra Part III.A.

  187. [187]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006) (emphasis added) (citations omitted) (quoting United States v. Reese, 92 U.S. 214, 221 (1876)).

  188. [188]. Id. at 330.

  189. [189]. Id. at 329.

  190. [190]. Id. at 330.

  191. [191]. United States v. Booker, 543 U.S. 220, 282 (2005) (Stevens, J., dissenting in part) (emphasis omitted).

  192. [192]. See infra Part III.A (addressing these tensions in greater detail).

  193. [193]. See supra notes 187–92 and accompanying text.

  194. [194]. See supra note 39 and accompanying text; see also supra note 57.

  195. [195]. Eric Berger, Deference Determinations and Stealth Constitutional Decision Making, 98 Iowa L. Rev. 465, 471 (2013).

  196. [196]. Id. at 532.

  197. [197]. See generally Shelby Cty. v. Holder, 133 S. Ct. 2612 (2013).

  198. [198]. Extension of the Voting Rights Act of 1965: Hearing Before the S. Subcomm. on Constitutional Rights of the Comm. on the Judiciary, 94th Cong. 119–21 (1975) (statement of Nicholas Katzenbach, Exec. Comm. for Civil Rights Under Law).

  199. [199]. Shelby Cty., 133 S. Ct. at 2629.

  200. [200]. Id. at 2648 (Ginsburg, J., dissenting). It is true that, had the Court taken Justice Ginsburg’s approach, the timing of the severability analysis would have depended on whether the statute’s constitutional applications included its application to Shelby County. If it did not, then the Court would have severed that application while indicating that some separate set of applications could remain in force. If it did constitutionally apply to Shelby County, however, then Shelby County would have simply lost its case—but some future litigant (perhaps Arizona or Alaska, per Justice Ginsburg’s suggestion) might have been able to bring a successful challenge, thereby requiring severability to occur at that point in time.

  201. [201]. Id. at 2629 (plurality opinion) (“We cannot pretend that we are reviewing an updated statute, or try our hand at updating the statute ourselves, based on the new record compiled by Congress.”).

  202. [202]. Implementing Justice Ginsburg’s proposed approach would require, at a minimum, a very complicated form of application severability. In addition, if one were to take Justice Ginsburg’s proposal a step further and wade into the reformulation of section 4 (rather than simply its partial invalidation), that more radical approach likely would require some combination of expanding applications (to include offending jurisdictions not covered by section 4 prior to its invalidation), employing nonstandard meanings of the statute (at least, insofar as the constitutional disruption of section 4 might be treated as an invitation to create a new formula in its absence), or relying on implicit substitute provisions. See supra Part III.A (describing resistance to all these forms of analysis in the context of severability).

  203. [203]. The 2006 reauthorization of sections 4 and 5 included, among other things, the severability clause discussed by Justice Ginsburg and 15,000 pages of legislative record setting forth “countless ‘examples of flagrant racial discrimination’ since the last reauthorization” of the Voting Rights Act as well as “systematic evidence that ‘intentional racial discrimination in voting remains so serious and widespread in covered jurisdictions that section 5 preclearance is still needed.’” Shelby Cty., 133 S. Ct. at 2636 (Ginsburg, J., dissenting) (quoting Shelby Cty. v. Holder, 679 F.3d 848, 866 (D.C. Cir. 2012)). Both these inclusions are evidence of legislative intent to retain some portion of section 4(b) in response to the constitutional disruption identified by the Shelby County majority.

  204. [204]. It is worth noting that the Shelby County majority might have denied the propriety of such an outcome not only as a matter of severability, but also as a matter of substantive constitutional law. Cf. id. at 2629–30 (implying, cryptically, that the Constitution could allow no application of section 4(b) to remain in effect because the way section 4(b) selected jurisdictions—through reliance on a previously constitutional formula that itself had been based on now decades-old data—was as incurably unconstitutional as imposing invidious restrictions on a suspect class). If that is the correct reading of Shelby County’s holding, then the majority should have been pressed to articulate this novel theory of constitutional law. As it stands, severability’s restrictive framework allows the entire debate to be, in effect, swept under the rug.

  205. [205]. Immigration & Naturalization Serv. v. Chadha, 462 U.S. 919, 954 (1983); see also id. at 959–60 (Powell, J., concurring) (“Congress has included the veto in literally hundreds of statutes, dating back to the 1930s.”).

  206. [206]. See, e.g., Smith, supra note 59, at 398–99; see also id. at 398 n.3 and accompanying text (cataloging some of these cases).

  207. [207]. Id. at 465–66.

  208. [208]. Id. at 464–66 (discussing Alaska Airlines, Inc. v. Brock, 480 U.S. 678 (1987)).

  209. [209]. Id. at 465 (“The courts deciding the Alaska Airlines severability dispute conceived their task as deciding whether the legislative veto provision Congress had included was severable from all of the employee protection rulemaking grant, or none of it.”).

  210. [210]. United States v. Booker, 543 U.S. 220, 226–27 (2005).

  211. [211]. Id. at 246.

  212. [212]. Id. at 249.

  213. [213]. Id. at 246. See generally supra notes 122–32 and accompanying text (discussing Booker).

  214. [214]. See supra notes 35–37 and accompanying text.

  215. [215]. This conclusion assumes that the majority’s conclusions about legislative intent were correct. Compare Booker, 543 U.S. at 249 (plurality opinion) (summarizing majority’s conclusions about legislative intent), with Booker, 543 U.S. at 295–96 (Stevens, J., dissenting in part) (explaining the contrary conclusion that “given the choice between the statute created by the Court today or a clean slate on which to write a wholly different law, Congress undoubtedly would have selected the latter”). This Article does not take a position on which of these competing conclusions about legislative intent was correct. Instead, it explores the effects of failing to follow legislative intent in fidelity to the severability doctrine.

  216. [216]. Walsh, supra note 7, at 750 (referring to “the Court’s audacious use of [severability] in United States v. Booker”). Among the many others who have criticized Booker’s approach to severability is David Gans. See Gans, supra note 24, at 666 (arguing that “Booker illustrates the dangers that occur when a court uses severability doctrine to make fundamental changes to a legislature’s work and the need for better doctrinal tools in this area”). But see Gillian E. Metzger, Facial Challenges and Federalism, 105 Colum. L. Rev. 873, 893 (2005) (“[A]ccepting arguendo the majority’s reading of the Sentencing Act, its remedial approach of facial invalidation was perfectly legitimate.”).

  217. [217]. As the dissent observed, “[t]he Court today opts for permitting the cut-off of only incremental Medicaid funding, but it might just as well have permitted, say, the cut-off of funds that represent no more than x percent of the State’s budget.” Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2667 (2012) (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting). Absent the restrictive framework of severability, the dissent’s observation is correct: the Court very well might have considered the “x percent” approach, which it could have adopted if it more closely aligned with legislative intent.

  218. [218]. Stern v. Marshall, 564 U.S. 462, 502 (2011). The practical collapse of the Bankruptcy Code was threatened, for example, in Exec. Benefits Ins. Agency v. Arkison, 134 S. Ct. 2165 (2014). Arkison was a follow up to Stern. In Stern, the Court had concluded, first, that the Code granted bankruptcy courts the authority to issue final judgments with respect to certain matters and, second, that this same grant of authority was unconstitutional. The question in Arkison therefore became what authority, if any, the bankruptcy courts retained over those same matters. Though congressional intent seemed clearly to point in one direction (that is, to allow bankruptcy courts to resolve these matters by issuing proposed findings of fact and conclusions of law), several members of the Court refused to accept that severability would permit such a result. See Transcript of Oral Argument at 48, Arkison, 134 S. Ct. 2165 (No. 12-1200) (“CHIEF JUSTICE ROBERTS: That’s what I thought severability was. If you carve—if you find part of it unconstitutional, you ask whether what is left can stand. You don’t say that we’re going to rewrite what is left.”); id. at 15 (“JUSTICE SCALIA: Congress might have—might have provided that if it had known about [Stern], right?... But do we sit here to write the statutes that Congress would have written... if they knew about some future events? I don’t think so.”). The Court ended up resolving the issue by concluding that the text of the Code allowed the matters in question to be classified differently and, as a result, avoided any need to “rewrite what is left.” Arkison, 134 S. Ct. at 2173 (noting that “[t]he conclusion that the remainder of the statute may continue to apply to Stern claims accords with our general approach to severability”). It is not clear what the Court would have done if the text of the statute did not happen to permit such a conclusion—nor what it will do in the future when post-Stern problems cannot be so easily avoided.

  219. [219]. See, e.g., Act Now to Stop War & End Racism Coal. v. District of Columbia, 905 F. Supp. 2d 317, 350 (D.D.C. 2012) (rejecting alternative construction of the constitutionally disrupted statute because “[i]n doing so, the Court would go beyond merely construing the statute, and would engage in lawmaking”).

  220. [220]. The uncertainty is there because courts rarely even consider whether Congress might have preferred some option other than the two traditionally permitted under the severability framework. With the elimination of severability’s limiting effects, the analysis with respect to any given statute might reveal a strong legislative preference for some third option.

  221. [221]. See, e.g., Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2668 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting) (advocating for the full invalidation of statutes rather than what it referred to as the “more extreme exercise of the judicial power” that can be implied by partial invalidation).

  222. [222]. Campbell, supra note 11, at 1496–98.

  223. [223]. See, e.g., Brief of the Family Research Council and 27 Members of the U.S. House of Representatives as Amici Curiae in Support of Petitioners and Reversal on the Issue of Severability at 26–27, Nat’l Fed’n of Indep. Bus. 132 S. Ct. 2566 (Nos. 11-393 & 11-400).

  224. [224]. See, e.g., Transcript of Oral Argument at 9, Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566 (2012) (Nos. 11-393 & 11-400) (“[Though each of the Supreme Court’s cases on severability] formulate[s] the test a little bit differently,... every one of them talks about congressional intent.”); see also supra notes 35–37 and accompanying text.

  225. [225]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 328 (2006); see also supra note 73 and accompanying text.

  226. [226]. United States v. Booker, 543 U.S. 220, 282 (2005). (Stevens, J., dissenting in part) (emphasis omitted).

  227. [227]. The phrase “legislature’s work” is taken from Ayotte, 546 U.S. at 329 (stating that severability analysis helps “not to nullify more of a legislature’s work than is necessary”).

  228. [228]. Compare Scalia & Garner, supra note 152, at 30 (“Traditional authorities on interpretation, while repeating the mantra that the objective of interpretation is to discern the lawgiver’s or the private drafter’s intent, would add that this intent is to be derived solely from the words of the text. We would have no substantive quarrel with the search for ‘intent’ if that were all that was meant.” (footnote omitted)), with Stephen Breyer, On the Uses of Legislative History in Interpreting Statutes, 65 S. Cal. L. Rev. 845, 864 (1992) (“Conceptually,... one can ascribe an ‘intent’ to Congress in enacting the words of a statute if one means ‘intent’ in its, here relevant, sense of ‘purpose,’ rather than its sense of ‘motive.’”); see also Lawrence M. Solan, Private Language, Public Laws: The Central Role of Legislative Intent in Statutory Interpretation, 93 Geo. L.J. 427, 458 (2005) (arguing that “[t]ext-based canons of construction” are not “an alternative to considering legislative intent” but rather “default rules for assessing the likely intent”).

  229. [229]. This is particularly true in the context of severability, where even those wary of calls to legislative intent (such as the late Justice Scalia) agree that legislative intent must be what controls the analysis. See, e.g., Nat’l Fed’n of Indep. Bus., 132 S. Ct. at 2668 (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting). But cf. Gans, supra note 24, at 688 (arguing that severability should be considered a question of remedy). See also supra notes 65–66 and accompanying text.

  230. [230]. See supra Part III.A.

  231. [231]. Id.

  232. [232]. See supra note 120 and accompanying text.

  233. [233]. See supra notes 85–88 and accompanying text.

  234. [234]. See supra notes 89–101 and accompanying text.

  235. [235]. See supra notes 102–08 and accompanying text.

  236. [236]. See generally supra Part II.B.

  237. [237]. See supra notes 155–59 and accompanying text for definitions of text severability and application severability.

  238. [238]. As noted above, this example originally was adopted from Adrian Vermeule, Saving Constructions, supra note 58, at 1955.

  239. [239]. Id.

  240. [240]. See id. (explaining challenge).

  241. [241]. Evidence of such intent might, for example, include a severability clause. See Gans, supra note 24, at 646 n.28 (“Though not dispositive, courts regularly accord severability clauses significant weight.”).

  242. [242]. See supra notes 164–77 and accompanying text.

  243. [243]. See supra notes 160–63 and accompanying text.

  244. [244]. See supra notes 174–82 and accompanying text.

  245. [245]. See supra notes 180–82 and accompanying text.

  246. [246]. See supra notes 174–79 and accompanying text.

  247. [247]. See infra notes 272–74 and accompanying text.

  248. [248]. See supra notes 89–99 and accompanying text.

  249. [249]. United States v. Nat’l Treasury Emps. Union, 513 U.S. 454 (1995).

  250. [250]. See id. at 477.

  251. [251]. The four constructions were as follows. First, the Court could have disregarded all the relevant provisions of the Act in all their applications. Second, the Court could have disregarded the relevant provisions of the Act insofar as they applied to employees in the Executive Branch. Third, the Court could have disregarded the relevant provisions of the Act insofar as they applied to employees in the Executive Branch who were of a sufficiently low rank. Fourth, the Court could have disregarded the relevant provisions of the Act insofar as they applied to employees in the Executive Branch who were of a sufficiently low rank and whose expressive conduct lacked a sufficient “nexus” to the relevant employment. See id. at 477–80.

  252. [252]. See the preceding footnote for the four constructions listed in order of apparent complexity. Of course, one could argue for a different complexity hierarchy if one phrased the options in a different way—a point that itself helps to confirm the fundamental failings of the Complexity Test.

  253. [253]. Nat’l Treasury Emps. Union v. United States, 990 F.2d 1271, 1277 (D.C. Cir. 1993).

  254. [254]. Id. at 1296 (Sentelle, J., dissenting) (quoting Ballard v. Miss. Cotton Oil Co., 34 So. 533, 554 (Miss. 1903)).

  255. [255]. United States v. Nat’l Treasury Emps. Union, 513 U.S. 454, 479 (1995).

  256. [256]. Id. at 480 (O’Connor, J., concurring in part and dissenting in part).

  257. [257]. Id. at 489 (Rehnquist, Scalia, and Thomas, JJ., dissenting).

  258. [258]. Id. at 502 n.8.

  259. [259]. Bowsher v. Synar, 478 U.S. 714, 736 (1986).

  260. [260]. Id.

  261. [261]. United States v. Booker, 543 U.S. 220, 284 (2005) (Stevens, J., dissenting in part).

  262. [262]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 329 (2006).

  263. [263]. Id.

  264. [264]. See Dorf, supra note 56, at 370 (“[L]et us ask what it means to say, in some particular case, that a plaintiff challenges the application to him of some ‘law.’ Consider federal statutes. Article I, Section 7 of the Constitution uses the word ‘Law’ to refer to bills that have been successfully enacted, a category that, in modern times, includes omnibus legislation concerning disparate subjects codified in different sections of the U.S. Code. In addition, courts frequently and sensibly use the term ‘law’ to refer to an uninterrupted string of text appearing somewhere in the U.S. Code, even if that text emerged from successive enactments of Congress, each one amending what came before.” (footnote omitted)).

  265. [265]. See supra notes 151–53.

  266. [266]. See Nelson, supra note 154, at 8 (acknowledging that even “skeptics of federal common law” carve out exceptions when Congress has conferred on the courts “a special delegation of lawmaking authority”).

  267. [267]. Sosa v. Alvarez-Machain, 542 U.S. 692, 741 (2004) (Scalia, J., concurring in part and concurring in the judgment) (quoting Tex. Indus., Inc. v. Radcliff Materials, Inc., 451 U.S. 630, 640 (1981)); see also supra note 154.

  268. [268]. Boyle v. United Techs. Corp., 487 U.S. 500, 504 (1988) (quoting Tex. Indus., Inc. 451 U.S. at 640).

  269. [269]. Fallon, supra note 24, at 1333–45 (quoting Reno v. ACLU, 521 U.S. 844, 844 (1997)).

  270. [270]. See, e.g., Nelson, supra note 154.

  271. [271]. Id.

  272. [272]. Fallon, supra note 24, at 1333.

  273. [273]. See supra notes 266–68 and accompanying text.

  274. [274]. Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 329 (2006).

  275. [275]. See Walsh, supra note 7, at 752–53 (“Severability doctrine requires focus on hypothetical intent because there often is no actual, expressed legislative intent to be found. The hypothetical legislative intent test gets around this absence of any actual legislative intent to discern but does so by posing a question whose answer often calls for rank speculation. The inquiry ‘can perhaps be given some content by reference to legislative purpose, or to the ‘enterprise’ to which the statute belongs. But it can easily deteriorate into a question of what ought to happen, in which case ‘legislative intent’ adds nothing.’” (footnotes omitted) (quoting Sherwin, supra note 43, at 304–05)).

  276. [276]. See, e.g., Leegin Creative Leather Prods., Inc. v. PSKS, Inc., 551 U.S. 877, 885, 899 (2007) (discussing Section 1 of the Sherman Act and noting that “the Court has never ‘taken a literal approach to [its] language,’” but rather “has treated the Sherman Act as a common-law statute” that “evolve[s] to meet the dynamics of present economic conditions” (first alteration in original) (quoting Texaco Inc. v. Dagher, 547 U.S. 1, 5 (2006))); Thornburg v. Gingles, 478 U.S. 30, 60 (1986) (endorsing the district court’s intricate analytical framework in response to general language set forth in section 2 of the Voting Rights Act, as amended); Abbe R. Gluck, Imperfect Statutes, Imperfect Courts: Understanding Congress’s Plan in the Era of Unorthodox Lawmaking, 129 Harv. L. Rev. 62 (2015) (exploring the role of the courts vis-à-vis complicated modern statutes); see also Gluck, supra at 64 (describing King v. Burwell, 135 S. Ct. 2480 (2015), as the Supreme Court’s “most explicit recognition ever of modern statutory complexity”).

  277. [277]. Nat’l Fed’n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2676 (2012) (joint opinion of Scalia, Kennedy, Thomas, and Alito, JJ., dissenting) (emphasis added).

  278. [278]. It is worth noting that the dissent’s proposed solution (full invalidation) was associated with its conclusion not only that the Medicaid expansion was unconstitutional, but also that the individual mandate was unconstitutional. Were the dissent to agree with the plurality that only the first was unconstitutional, it is not clear what its proposed solution would be, though it seems likely it would be to disregard all portions of the ACA that addressed the Medicaid Expansion. Id. at 2667. (“The most natural remedy would be to invalidate the Medicaid Expansion.”). If true, this only makes the underlying point stronger: this Medicaid-free revision of the ACA would suffer from all of the same “version” flaws flagged by the dissent.

  279. [279]. Id. at 2676.

  280. [280]. Id.

  281. [281]. The above-quoted passage by the dissent suggests that Congress can “start afresh” only through total invalidation of the statute. Among other things, this ignores the comprehensive legal regime predating the enactment of the ACA.

  282. [282]. See supra notes 76–77 and accompanying text.

  283. [283]. See, e.g., Campbell, supra note 11, at 1496; Gans, supra note 24, at 645; see also supra notes 249–58 and accompanying text (using United States v. Nat’l Treasury Emps. Union, 513 U.S. 454 (1995), to illustrate, among other things, the extent to which opposing jurists accuse one another of impermissible judicial legislation).

  284. [284]. A more precise way of thinking about this distinction is to assume that there is some standard way of analyzing statutes. (Of course, jurists and scholars will disagree—vigorously—on whether such standards exist and, if so, how they should be understood.) The court might be said to be “rewriting” a statute, rather than “interpreting” or “construing” a statute, if it interprets or construes a statute in a manner that deviates from standard practice—that is, in a nonstandard way. For example, if a jurist believes that a statute should be interpreted based strictly on its text (standard analysis), then a court engages in “judicial rewriting” of that statute if it interprets it in a manner that does not derive strictly from its text (nonstandard analysis).

  285. [285]. John F. Manning, What Divides Textualists from Purposivists?, 106 Colum. L. Rev. 70, 110 (2006).

  286. [286]. Counterarguments, including those related to the incentives of Congress, abuse by the courts, difficulty of analysis, and stare decisis, are discussed below. See infra notes 301–13 and accompanying text.

  287. [287]. See supra notes 212–17 and accompanying text.

  288. [288]. See supra notes 199–211, 220 and accompanying text.

  289. [289]. See supra note 220 and accompanying text.

  290. [290]. See supra notes 219, 221–22 and accompanying text.

  291. [291]. United States v. Thirty-Seven Photographs, 402 U.S. 363 (1971).

  292. [292]. Id. at 372–73.

  293. [293]. See id. at 373–74 (“Accordingly, we construe § 1305 (a) to require intervals of no more than 14 days from seizure of the goods to the institution of judicial proceedings for their forfeiture and no longer than 60 days from the filing of the action to final decision in the district court. No seizure or forfeiture will be invalidated for delay, however, where the claimant is responsible for extending either administrative action or judicial determination beyond the allowable time limits or where administrative or judicial proceedings are postponed pending the consideration of constitutional issues appropriate only for a three-judge court.”).

  294. [294]. Id. at 374 (“So construed, § 1305(a) may constitutionally be applied to the case before us.”).

  295. [295]. See supra note 41 and accompanying text.

  296. [296]. See supra notes 76–77 and accompanying text.

  297. [297]. In a hypothetical world where Congress had expressed the converse set of preferences, thereby charging the court with the development of such a complicated regime, the court would have an exceptionally difficult task on its hands—but under well-accepted principles of federal common law making, it is not clear how it could justify a refusal even to try. See supra notes
    268–69 and accompanying text.

  298. [298]. See, e.g., 19 U.S.C. § 1652 (2012) (“If any provision of this chapter, or the application thereof to any person or circumstances, is held invalid, the remainder of the chapter, and the application of such provision to other persons or circumstances, shall not be affected thereby.”).

  299. [299]. In addition to these three concerns, one might voice an objection on the grounds of stare decisis. This concern is valid. Yet the principles of stare decisis—to the extent they can be meaningfully understood, see Randy J. Kozel, Stare Decisis as Judicial Doctrine, 67 Wash. & Lee L. Rev. 411, 424 (2010)—allow reversal when the rule in question is, among other factors, unworkable as a practical matter. See Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 854 (1992). Severability’s unworkability is demonstrated by its arbitrarily destructive effects as well as the inconsistencies and confusion it introduces to related doctrines. Still, to the extent that one nevertheless wishes to follow stare decisis, a more moderate response to severability’s failings may be to turn to other legal contexts and labels (for example, the doctrines relating to “remedy”) rather than severability whenever case precedent allows for either option. See supra Part III.A (describing acceptability of analytical approaches in contexts outside of severability).

  300. [300]. Sources raising this concern include Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 330 (2006), and Gans, supra note 24, at 675–83.

  301. [301]. The contention underlying this assumption is the subject of rigorous scholarly debate. See, e.g., Matthew C. Stephenson, The Price of Public Action: Constitutional Doctrine and the Judicial Manipulation of Legislative Enactment Costs, 118 Yale L.J. 2, 62 (2008) (exploring the extent to which “reviewing courts may implement constitutional guarantees more effectively by crafting doctrines that raise the costs to government decisionmakers of enacting constitutionally problematic policies”); cf. Brianne J. Gorod, The Collateral Consequences of Ex Post Judicial Review, 88 Wash. L. Rev. 903, 959 (2013) (exploring ways in which the courts might assist Congress’s attempts at discerning constitutional law rather than punish its errors ex post).

  302. [302]. See, e.g., Cal. Pub. Emps. Ret. Sys. v. WorldCom, Inc., 368 F.3d 86, 97 (2d Cir. 2004) (acknowledging that Congress enacted the 1984 revisions to the Bankruptcy Code—later invalidated in Stern v. Marshall, 564 U.S. 462, 502–03 (2011)—in an effort to comply with recent developments in Supreme Court jurisprudence).

  303. [303]. See, e.g., John K. DiMugno, The Affordable Care Act After the Supreme Court’s Ruling: An Overview, 22 Experience, no. 3 at 10, 15 (2013) (“The Court’s willingness to review Congress’s power under the Spending Clause [with respect to the ACA’s Medicaid program] took many by surprise....”); cf. David A. Hyman, Why Did Law Professors Misunderestimate the Lawsuits Against PPACA?, 2014 U. Ill. L. Rev. 805 (2014) (discussing the incredulity with which many law professors initially had responded to the theory of the Commerce Clause eventually adopted by the Supreme Court in National Federation).

  304. [304]. See supra note 240 and accompanying text.

  305. [305]. Gans, supra note 24, at 644 (arguing that severability, as it currently stands, “breeds an unhealthy dependency on courts and results in a loss of accountability”).

  306. [306]. See, e.g., Citizens United v. Fed. Election Comm’n, 558 U.S. 310, 329 (2010) (“[T]he Court cannot resolve this case on a narrower ground without chilling political speech, speech that is central to the meaning and purpose of the First Amendment.”).

  307. [307]. This discussion touches on a related concern. To the extent that the improved regime gives courts greater flexibility in construing constitutionally disrupted statutes, one might worry about statutes becoming unconstitutionally vague. This concern derives from the doctrines governing criminal and First Amendment law—see, e.g., City of Akron v. Akron Ctr. for Reprod. Health, 462 U.S. 416 (1983); Coates v. City of Cincinnati, 402 U.S. 611 (1971)—and in the context of severability, these same doctrines provide the answer. This is because the vagueness-related doctrines impose substantive constitutional limits, and those limitations would apply in the same manner under the improved regime. By definition, a construction of a statute that produces an unconstitutionally vague result is not one that cures the statute’s constitutional defects—and therefore is not an construction of a statute adoptable by a court.

  308. [308]. It is worth mentioning, in this context, the irony of United States v. Booker, 543 U.S. 220 (2005). In this singular example of the Supreme Court employing a more flexible form of “severability” in response to a constitutionally disrupted statute, the so-called “remedial majority” was composed of four justices who had dissented with respect to the substantive constitutional claim. Stated otherwise, four out of the five justices who had determined the proper construction of the constitutionally disrupted Sentencing Reform Act had concluded that, in their opinion, it contained no constitutional flaw at all. In light of their support of the Act as it was enacted, it seems likely that those four were particularly committed to effectuating legislative intent to the full extent possible under the Constitution. And, tellingly, they did so not by reflexive resort to severability but by resisting severability’s restrictive framework.

  309. [309]. Robert Katzmann, Judging Statutes 35 (2014).

  310. [310]. See William N. Eskridge, Jr., Overriding Supreme Court Statutory Interpretation Decisions, 101 Yale L.J. 331, 332 n.1 (1991).

  311. [311]. See, e.g., Matthew R. Christiansen & William N. Eskridge, Jr., Congressional Overrides of Supreme Court Statutory Interpretation Decisions, 1967–2011, 92 Tex. L. Rev. 1317 (2014); Richard L. Hasen, End of the Dialogue? Political Polarization, the Supreme Court, and Congress, 86 S. Cal. L. Rev. 205, 209 (2013).

*

Assistant Professor, University of Washington School of Law. 



I am indebted to those who have offered insights in furtherance of this project, including Kate Andrias, Eric Berger, Ryan Calo, Eric Fish, Brianne Gorod, Sanne Knudsen, Anita Krug, Shannon Weeks McCormack, Elizabeth Porter, Zahr Said, Kathryn Watts, and David Ziff, as well as the participants at the Seattle Junior Faculty Forum. Grateful acknowledgement is due to Dane Westermeyer, Thomas Miller, and the research librarians at the Gallagher Law Library for outstanding research assistance.