Friday, October 04, 2019

Response to Comments on Rationing the Constitution

Andrew Coan

I am grateful to Jack Balkin for organizing this symposium and to the commenters on Rationing the Constitution for their close, careful, and generous engagement with my book. Their wide-ranging posts raise more valuable and interesting points than I can address here. I will therefore focus on four recurrent topics that strike me as primary: (1) extensions, applications, and friendly amendments to my judicial capacity model of Supreme Court decision-making; (2) the place of judicial capacity in the legal process tradition; (3) the internal logic of the judicial capacity model; and (4) causal inference and the problem of observational equivalence.

To start, it will be helpful to restate the core claims of the book. At least since Henry Hart’s famous 1958 Harvard Law Review Foreword, it has been well established that the Supreme Court can decide only so many cases per year—150-200 at most. It is also familiar wisdom that this limited capacity influences the Court’s decisions and development of constitutional doctrine, albeit in diffuse and indeterminate ways. Building on these foundations, Rationing the Constitution makes two novel claims: First, judicial capacity limits have especially significant bite—and real predictive power—in many of the most important constitutional domains, spanning federalism, separation of powers, and individual rights. Second, these limits are the product not just of limited time and resources but also, crucially, of widely shared judicial norms. Chief among these is the Court’s felt need to review virtually all lower-court decisions that invalidate federal statutes.

These core claims are the premises for my judicial capacity model, which I spend most of the book attempting to substantiate using a multiple case-study research design. In constitutional domains implicating a large volume of federal statutes, this model predicts that the Supreme Court will be strongly constrained to choose between some combination of strong deference to the political process and hard-edged categorical rules. This, in turn, limits the Court’s ability to challenge dominant political forces and undermines its ability to competently resolve constitutional controversies—categorical rules being generally too crude to sensibly define the contours of federalism, separation of powers, equal protection, etc.

Judicial capacity is obviously not the only factor that influences Supreme Court decision-making in these domains. Ideology, strategic calculations, and legal principle all play important roles, too. But the judicial capacity model helps to explain the pattern of post-New Deal Supreme Court decisions in ways that none of these other factors can. Or so Rationing the Constitution contends. On this contention, the book must stand or fall.  

One final terminological clarification: When I use the term judicial capacity, I mean the total volume of cases that the court system—and, in particular, the Supreme Court—is capable of handling. I do not mean the capacity of courts to produce reliably good decisions, which I shall call judicial competence. Nor do I mean the capacity—or inclination—of courts to produce social change against the tide of dominant political forces, which I shall call judicial independence.

Extensions, Applications, and Friendly Amendments

In illuminating and complementary posts, David Marcus and Aaron-Andrew Bruhl draw attention to judicial capacity issues in the lower federal courts that I do not discuss at any length in my book. The focus of Rationing the Constitution is on the relationship between judicial capacity and constitutional doctrine. And constitutional doctrine is made, for the most part, in the Supreme Court. Moreover, through the hierarchical system of precedent, the Court projects its capacity-management decisions through the federal judiciary as a whole. But Marcus and Bruhl helpfully demonstrate the payoffs of bringing lower courts into the story—payoffs that both enrich and reinforce the judicial capacity model. 

As Bruhl points out, there is a substantial literature demonstrating that lower courts respond to caseload pressures by limiting oral argument, delegating decision-making authority to law clerks, and reviewing trial court decisions more deferentially. Bruhl does not cite specific examples, but highlights of this impressive body of work include William Richman & William Reynolds, Injustice on Appeal (2012); Bert Huang, “Lightened Scrutiny,” 124 Harv. L. Rev. 1109 (2011); and Marin Levy, “Judicial Attention as a Scarce Resource,” 81 Geo. Wash. L. Rev. 401 (2013). To this list might be added Bruhl’s own “Deciding When to Decide,” 96 Cornell L. Rev. 203 (2011), an insightful case study of docket management in the U.S. Courts of Appeal after Blakely v. Washington (2004).

Together, these scholars have persuasively shown that lower federal courts are simultaneously under much more constant capacity pressure than the Supreme Court and much more limited in the tools available to them to manage this pressure. Unlike the Supreme Court, the lower courts cannot manage their dockets by turning away cases. They also have limited flexibility to make doctrine more deferential or rule-like. Laboring under such constraints, the lower courts have had no choice but to compromise the minimum professional standards—extensive oral argument; lengthy published opinions; little or no delegation of actual decision-making to law clerks and administrative staff—that Supreme Court justices have the luxury to treat as sacrosanct.

Indeed, in most legal domains, the Supreme Court’s discretionary jurisdiction enables it to maintain these minimum professional standards without much consideration of the volume of litigation its decisions will invite down the road. Bruhl and Marcus discuss two such domains, vividly demonstrating how differently the Supreme Court behaves outside of the capacity-constrained domains discussed in my book. Bruhl focuses on the void-for-vagueness doctrine and its application to federal sentencing laws in U.S. v. Johnson (2015), which led to thousands of new filings in the federal district courts and courts of appeals. It is impossible to imagine the Court issuing the same decision in a world where it was compelled to review even ten percent of these new cases itself. But because federal sentencing is a normal domain, one in which the Court is content to leave all but a handful of cases to the lower courts, judicial capacity did not substantially constrain its doctrinal options. The contrast with capacity-constrained domains, where the Court feels compelled to review virtually every lower court decision invalidating a federal law, is striking.

Marcus points to a different body of lower-court cases to make a different point: “Components of dozens of state prison systems remain under federal judicial supervision.  In recent decades, litigants have successfully sued dozens of state and city foster care agencies, winning broad injunctive remedies for tens of thousands of children.  Federal judges play key roles in the administration of disability policy in cities across the country.” As Marcus recognizes, “[t]his litigation does not fall within the domains that require intensive Court supervision. The Court can tolerate doctrinal mushiness and whatever volume of litigation it invites when a lawsuit successfully challenges the constitutionality of state law or local government policy administration.” Yet this pattern of social reform litigation underscores an important qualification to the judicial capacity model: The sharp limits that judicial capacity imposes on judicial power extend only to capacity-constrained domains.

Many of the most important constitutional domains fall into this category, but not all of them. And in domains where the Court is content to review only a small fraction of cases, the lower courts retain significant freedom to deviate from the deference and categorical rules predicted by the judicial capacity model in capacity-constrained domains. I believe this qualification is implicit in the book, but Marcus is right to flag the “sweeping intimations about the limits of federal judicial power that surface here and there in Rationing the Constitution.” These intimations should be read to apply solely to capacity-constrained domains.

Adrian Vermeule’s comments are in a similar vein. While pronouncing Rationing the Constitution “generally admirable, clear, and correct,” he proposes three characteristically helpful refinements in the form of “second-order complexities about the idea of rationing judicial capacities.” I am happy to consider all of these as friendly amendments, though I think Vermeule may overestimate their practical significance.

First, Vermeule reframes the volume of litigation invited by the Supreme Court’s doctrinal choices as an externality—a cost the Court’s discretionary jurisdiction permits it to partially shift onto lower courts and the bar. This reframing crisply captures the dynamic Bruhl noted in U.S. v. Johnson, where the Supreme Court’s void-for-vagueness ruling buried the lower courts under an avalanche of new sentencing challenges. Since this is a normal domain, the Court reviewed only a tiny handful of these cases, leaving lower courts to bear nearly all of the costs.

Even in capacity-constrained domains, where the Court feels compelled to review a large fraction of cases, the Court can externalize some of the costs of its doctrinal choices. The Court does not feel compelled to review every lower court decision in any of these domains. In particular, it does not feel compelled to review decisions invalidating state and local legislation or decisions upholding federal statutes against constitutional challenge. The costs of these cases are borne wholly by the lower courts. On the other hand, what distinguishes capacity-constrained domains from normal domains—their very sine qua non—is the Court’s inability to avoid reviewing an overwhelmingly large fraction of cases. (Usually, this inability is the product of the judicial norm requiring the Court to review virtually every lower-court invalidation of a federal statute.) In other words, these are domains in which the Court is forced to internalize much of the cost of its doctrinal choices. In this sense, Vermeule’s point about externalities is not so much a refinement of the judicial capacity model as an incisive relabeling of its central underlying mechanism.

Vermeule next points out that the Supreme Court may have the power to expand its judicial capacity budget in a variety of ways, including through its own constitutional decisions. If this is true, the Court’s “capacities may then themselves be partly endogenous to judicial choices over time.” Vermeule offers numerous potential examples. The most straightforward is “judicial decisions approving administrative tribunals or legislative courts for workers’ compensation cases and routine bankruptcy matters.” But the list could go on at great length. At a minimum, it would certainly include the Court’s restrictive modern standing doctrine, the heightened pleading standards established in Bell Atlantic v. Twombly and Ashcroft v. Iqbal and decisions like U.S. v. Morrison that invalidate statutory rights of action.

All of these decisions are significant, in their own right and for their impact on the judicial capacity budget. Yet none changes the fundamental reality on which the judicial capacity model is premised. In capacity-constrained domains, the Court’s felt compulsion to review nearly every lower-court invalidation of a federal statute sharply constrains the menu of doctrinal options. As Vermeule correctly observes, this constraint is not an inflexible law of nature. It cannot, however, be easily evaded without sacrificing deeply rooted judicial commitments to minimum professional standards and the uniformity of federal law.

Vermeule’s final suggested refinement is the most interesting. He calls it “the warning shot,” and I cannot improve on his elegant explanation: “The threat is stronger than the execution. While the Court cannot do very much, the standing threat to override political decisions anywhere potentially affects the anticipated reaction of institutions everywhere.” The upshot is that the Supreme Court’s capacity limits may be less significant than they seem. If the mere threat of judicial invalidation can deter the political branches from acting, it is a mistake to view the Court’s power solely in terms of the decisions it renders. One must also consider the threat of further action that each decision implies and the reaction of political actors to that threat.

I have no quarrel with Vermeule’s logic. The threat of judicial invalidation may well have a broader restraining effect than would the Court’s actual exercises of constitutional review. Like Marcus’s point about social reform litigation in the lower federal courts, this is a valuable caveat to my account of judicial capacity’s implications for judicial power. Still, I am skeptical that this effect is of great practical import in capacity-constrained domains, for two reasons. First, a threat that is never—or almost never—carried out ceases to be credible at some point. The Court’s stark record of deference in capacity-constrained domains, punctuated by a smattering of narrow categorical rules, seems unlikely to keep members of Congress up at night. Second, the history of federal legislation in these domains does not suggest to me a Congress that is cowed into submission by the remote threat of judicial review. Admittedly, this is an impressionistic judgment. Perhaps there is some muted effect not visible to the naked eye, but I do not believe any such effect is sufficient to undermine the book’s central conclusions.

To be clear, Vermeule does not actually argue the contrary. He is careful to acknowledge that the threat is not always stronger than the execution. His point is “merely that it can be, and the analyst always has to consider whether it is.” To this theoretical point, I am happy to accede.

Judicial Capacity and Legal Process

Maggie Blackhawk and Mariah Zeisberg both fault Rationing the Constitution for failing to engage adequately with important prior scholarship. Zeisberg is bothered by what she sees as my “false” and immodest claims to novelty. More substantively, she thinks I ought to have mined the existing literature more thoroughly for testable hypotheses that might compete with the judicial capacity model. She offers the work of Alexander Bickel, Lawrence Sager, and Gerald Rosenberg as illustrative examples. Blackhawk believes that a fuller engagement with the legal process tradition could and should have made my judicial capacity model more sensitive to the interplay between the Supreme Court and other institutional actors.

I share Zeisberg’s esteem for Bickel, Sager, and Rosenberg, as my citations to their work attest. It is also true that I am following in their footsteps in the broad sense that Adrian Vermeule identifies when he writes that Rationing the Constitution “develops seamlessly out of a venerable line of work that considers constitutional theory … in light of the capacities of judges and the resource constraints under which they labor, especially constraints of time, attention, information, and political capital.” On the other hand, Rationing the Constitution is a very different kind of book from The Least Dangerous Branch, Justice in Plainclothes, or The Hollow Hope. Bickel and Sager both offer normative—or, more precisely, interpretive—accounts of U.S. constitutional practice focused on judicial competence, democratic legitimacy, and, in Bickel’s case, the risk of public backlash to unpopular decisions. Neither devotes any substantial attention to judicial capacity, as I use that term. Rosenberg’s account is positive, rather than normative, like Rationing the Constitution, but his overwhelming focus is judicial independence, rather than judicial capacity.

This is not to suggest that my judicial capacity model lacks intellectual antecedents. Richard Posner’s Federal Courts: Crisis and Reform and Neil Komesar’s Imperfect Alternatives were the direct inspirations for my book and supply the foundation on which it rests. Both address precisely the concept of judicial capacity that is my focus in Rationing the Constitution and explain the crucial role of judicial hierarchy in limiting the capacity of the court system as a whole. But neither Posner nor Komesar is concerned principally with constitutional law. The same is true of Peter Strauss, whose article “One Hundred Fifty Cases a Year” was another important inspiration and influence. For all of these reasons, I stand by my claim that Rationing the Constitution “is the first book-length work on constitutional law to take judicial capacity seriously,” though it might have been better to say “the first systematic examination of the influence of judicial capacity on U.S. constitutional doctrine.”    

Zeisberg’s more substantive complaint is that Rationing the Constitution overlooks a goldmine of testable empirical hypotheses in the work of Bickel, Sager, Rosenberg, and others. I do not doubt that an imaginative reader could generate many testable hypotheses from this literature. The universe of such hypotheses, as Zeisberg surely understands, is effectively limitless. But she does not identify even one positive hypothesis that is (a) actually defended by a prominent legal process theorist and (b) could plausibly explain the pattern of Supreme Court decisions discussed in Rationing the Constitution better than the judicial capacity model. Nor am I aware of any hypothesis that satisfies both of these criteria.  

The two examples Zeisberg cites, drawn from Rosenberg and Bickel, are both normative rather than positive claims. The “hypothesis” she derives from Rosenberg is illustrative: “Courts concerned with capacity should time their major interventions in ways that are supported, rather than blocked, by the other branches” (my emphasis). Rosenberg makes no such normative claim, but he most certainly does not make the positive claim that courts actually do time their interventions in this way. Indeed, the major evidence for his thesis that courts are incapable of spearheading social change consists of failed judicial attempts to overcome political resistance.  

For these reasons, I chose to focus my analysis of the judicial capacity model on the competing models of judicial decision-making that are most widely and vigorously advocated in the existing literature. I do not claim to have shown that the judicial capacity model dominates all potential alternatives. I do not even claim that the judicial capacity model dominates the leading models I compare it to in the book as to all features of Supreme Court decision-making. I claim only that it is capable of explaining some of those features better than those alternative models. Given the wide influence of the legalist, attitudinalist, and strategic models, this is strong evidence in favor of the judicial capacity model, though Zeisberg is surely right that there is more work to be done.

Maggie Blackhawk’s critique is more limited, and her comments also have a constructive dimension. While recognizing the judicial capacity model as “valid and valuable,” she suggests that greater engagement with the legal process tradition would have made the model stronger. More specifically, she cites the work of Henry Hart & Albert Sacks, Phil Frickey, and William Eskridge for the proposition that the Supreme Court does not operate in an institutional vacuum. This is certainly correct and important, as is Blackhawk’s narrower point that the behavior of the political branches influences the prospective volume of litigation. As I explain in the book, “significant political opposition” to the Court’s constitutional decisions in capacity-constrained domains “tends to produce a high volume of litigation that taxes the limits of judicial capacity.” It is in this sense that judicial capacity helps to explain the Court’s limited ability to challenge dominant political forces.

Yet the opposite is also true, as Blackhawk points out: If dominant political forces welcome aggressive judicial review, as they sometimes do, judicial capacity might operate as less of a constraint. Instead of litigating vigorously up to the Supreme Court, the government might simply refuse to defend broad swaths of legislation against constitutional challenge. In theory, this could allow the Court to announce stringent new limitations on federal legislation without inviting much litigation that it would feel compelled to review. This is an astute and valuable point; Blackhawk is right that I could—and perhaps should—have done more to develop it.

On the other hand, I suspect its practical import is quite limited, for two reasons. First, in most capacity-constrained domains, it would be difficult for the Supreme Court to articulate broad constitutional limits on federal legislative authority without goring quite a few oxen on both sides of the political spectrum. Second, even when the government refuses to defend legislation challenged as unconstitutional, other interested parties will often have standing to step in and litigate in the government’s stead. Think of the challenge to the Affordable Care Act currently pending in Texas, which the Trump administration has refused to defend. At any rate, I am unaware of any capacity-constrained domain in which the political branches have engaged in sufficiently broad acquiescence to eliminate—or even substantially weaken—the constraints of judicial capacity.

One possible exception, suggested by Blackhawk, is the constitutional prohibition on sex discrimination, which the Solicitor General famously opposed, in part, on judicial capacity grounds. Hundreds of federal laws on the books at the time discriminated on the basis of sex. Yet no flood of cases materialized that the Supreme Court felt compelled to review. The judicial capacity model offers a partial explanation. The constitutional prohibition on sex and discrimination hardened into a virtual per se rule quite quickly—a mere five years elapsed between Reed v. Reed and Craig v. Boren—rendering the outcome of most cases a foregone conclusion. (More on this general dynamic below.) But Blackhawk is right to suggest that the women’s movement is a crucial part of the story, both in triggering doctrinal change and in shaping social attitudes to reduce the need for litigation. This point is fully compatible with the judicial capacity model and indeed strengthens it, as Blackhawk suggests. 

The Logic of Judicial Capacity

Aziz Huq and Mariah Zeisberg raise different but overlapping objections to the internal logic of my judicial capacity model. Huq questions my definition of capacity-constrained domains; my contention that judicial capacity tends systematically toward deference; and the relationship between categorical rules and the volume of litigation. Zeisberg makes a similar point about categorical rules and also questions whether my model generates falsifiable predictions. I believe all of these questions are answered in the book, though perhaps not as clearly as they should have been if they did not register with readers as perceptive as Huq and Zeisberg.

Capacity-constrained domains are those in which judicial capacity strongly constrains the Supreme Court to employ some combination of strong deference and categorical rules. Examples discussed in the book include the commerce power, the spending power, the nondelegation doctrine, presidential administration, the equal protection clause, and regulatory takings. Conceptually, I suggest that capacity-constrained domains come in two basic flavors—high-volume and high-stakes. In the former, the sheer volume of potential litigation is what constrains the Court’s doctrinal choices. In the latter, the principal constraint comes from the high fraction of prospective cases in which the Court would feel compelled to grant review. Some domains fall into both of these categories; I call this third set hybrid domains.

As I explain in the book, this three-part framework is meant as a heuristic: “What ultimately matters is the number of potential cases in which the Supreme Court would feel compelled to grant review. This number is inevitably and always a function of both stakes and volume.” For example, neither the Fourth Amendment nor Title VII qualifies as a capacity-constrained domain, even though both involve thousands of actual and potential cases per year. The reason is simple: the Court feels compelled to review only a tiny fraction of these cases; therefore, neither of these domains is likely to overwhelm the Court’s limited capacity.

A parallel point holds for something like the Emoluments Clause. The Court is likely to feel compelled to review every successful Emoluments Clause challenge—a very high fraction of cases—but no plausible interpretation of that clause is likely to invite a large enough volume of litigation to tax the Court’s capacity. Thus, the Emoluments Clause is not a capacity-constrained domain.

The Commerce Clause is different. The volume of potential litigation is probably smaller than the existing volume of Fourth Amendment and Title VII litigation. But the Supreme Court feels compelled to review virtually every lower court decision invalidating a federal statute. As such, even a hundred successful commerce-clause suits per year would severely tax the Court’s limited capacity. Other capacity-constrained domains—like Equal Protection and regulatory takings—implicate preponderantly state and local laws, which the Court does not feel compelled to review in large numbers. But these domains also implicate federal laws, whose invalidation the Court does feel compelled to review. This, together with the large potential volume of litigation, brings Equal Protection and takings within the ambit of the judicial capacity model.

I hope this clears up Huq’s mistaken impression that volume is doing all of the heavy lifting in the judicial capacity model. If anything, stakes—the fraction of cases the Court feels compelled to review—are more important. This explains why the book focuses so heavily on the invalidation of federal statutes, which is the next question Huq raises. In theory, a large enough volume of statutory cases or constitutional cases involving the actions of individual executive officials could pose a threat to the Supreme Court’s limited capacity. But the fraction of such cases that the Court feels compelled to review is so low that this is hard to imagine in practice. Huq points out that nothing in the Court’s formal rules gives precedence to cases involving federal statutes, but David Marcus gets this right: “no observer of the Court could plausibly doubt that [this] commitment exists.”[1] Contrary to Huq’s intimation, this is a purely positive observation about the Court’s certiorari practice, not a normative one about how that practice should operate.

As the book explains, this feature of the Court’s practice is also contingent, rather than necessary. Huq is correct that the justices might alter or abandon it, perhaps out of frustration with the limits of judicial capacity. In theory, the justices might also decide to delegate much greater decision-making authority to law clerks or to eschew written opinions or even to decide cases by coin flip. Any of these changes would enable the Supreme Court to decide far more cases and perhaps some of these changes will eventually come to pass. But the professional norms undergirding the judicial capacity model have functioned as fixed stars of Supreme Court administration from at least the New Deal era. No significant change seems likely in the near term.

Huq next points out that categorical rules can be intrusive as well as deferential: “No federal rule-making with the force of law,” for instance. This is correct and fully consistent with the judicial capacity model. In fact, Chapter 11 discusses Huq’s precise example as an illustration of judicial capacity’s impact on the constitutional choice set: “[S]weeping reforms” like Philip Hamburger’s proposal to abolish binding administrative rules “might be consistent with the limits of judicial capacity, due to their starkly categorical character, but nevertheless worse than the disease they were intended to cure.” Thus far, most of the justices have considered sweeping categorical limits too unpalatable to embrace in most capacity-constrained domains. The limits of judicial capacity have therefore generally compelled the Supreme Court to adopt categorically deferential rules like the rational basis test, supplemented by the occasional and narrow categorical limit like Clinton v. New York’s prohibition on line-item vetoes.

Of course, the preferences of the justices on this score could always change. If they do, judicial capacity could conceivably push the Court toward more sweeping interventions than the justices would adopt in a world without capacity constraints. I explicitly note this possibility in the book. Still, it strikes me as more theoretical than real. The types of sweeping rules Huq has in mind would be extraordinarily disruptive and would, in many cases, represent an incredibly crude mechanism for achieving their underlying purposes. Even with a significant ideological shift, the probability of the Court adopting sweeping constitutional limits of this character seems quite low. Nor would such limits offer ironclad security against a judicial capacity overload. Lochner era limits on federal legislative power were largely categorical in nature. Yet their sweeping scope led more than 100 federal district judges to issue more than 1,600 injunctions against New Deal legislation in the summer of 1935 alone. Had the Court not retreated from these limits in 1937, it is doubtful how much longer it could have borne the strain.

Huq’s final point, echoed by Zeisberg, is that categorical rules themselves may prove malleable and unpredictable and therefore fail to inoculate the Court against capacity overload. To illustrate this point, Huq and Zeisberg both point to the economic/noneconomic distinction of Lopez and Morrison and the activity/inactivity distinction of NFIB v. Sebelius. I hold no brief for either of these decisions or the wisdom of the rules they embrace. But the important point for the judicial capacity model is that both clearly insulated the vast majority of federal legislation against constitutional challenge. They did so not through the intrinsic clarity or precision of the categorical distinctions they employed, but by clearly signaling that those distinctions imposed extraordinarily narrow limits on federal power. This signaling was explicit in Chief Justice Roberts’s insistence in NFIB that the Affordable Care Act was the only federal law in history to regulate economic inactivity. It was implicit in Lopez’s reaffirmation of Wickard v. Filburn on the ground that consumption of home-grown wheat counts as economic activity.

The clarity of these signals did not deter all entrepreneurial litigators, but for almost 10 years after Morrison, the Courts of Appeals rejected every commerce-power challenge they heard, and the Supreme Court denied review. When the Ninth Circuit finally broke this pattern in Gonzales v. Raich, the Court reversed in a belt-and-suspenders opinion that defined economic activity with sweeping breadth. The Court also reinvigorated the rational basis test and the necessary and proper clause as alternative grounds for upholding federal legislative power. In dissent, Justice O’Connor accused the majority of using “a dictionary definition of economics to skirt the real problem of drawing a meaningful line between ‘what is national and what is local.’” She was exactly right, and this is just what the judicial capacity model would predict. It is too early to assess the legacy of NFIB v. Sebelius, but the preliminary evidence suggests a strikingly parallel story.

Finally, Zeisberg asks what it would take to falsify the judicial capacity model and suggests that the Court’s adoption of the conceptually malleable commerce power tests discussed above should count as falsification. I tried to be as explicit on this point as possible: In capacity-constrained domains, “the judicial capacity model predicts that the court will be strongly constrained to employ some combination of deference and categorical rules such that the government will almost always win, or the application of the Court’s test will almost always be readily predictable by judges or litigants, or both. This prediction is obviously not perfectly precise and its content will vary to some extent by context. Nevertheless, it is quite demanding and easily falsifiable.” For the reasons explained above, I believe the Court’s recent commerce-power decisions are fully consistent with the judicial capacity model. Indeed, they are some of its best illustrations.

Causal Inference and Observational Equivalence

Establishing the consistency of the judicial capacity model with the Supreme Court’s decisions is the easy part. The far more difficult question is how to disentangle the causal role of judicial capacity from other competing explanations for the pattern of Supreme Court decisions in capacity-constrained domains. If one or more of those explanations is equally consistent with the Court’s decisions, how is one to adjudicate between them? This is the problem of observational equivalence, and it is an extremely thorny one. Frederick Schauer, Victoria Nourse, and Mariah Zeisberg all raise probing questions on this front. In response, I am tempted to quote Philip Roth, quoting Joe Louis: “I did the best I could with what I had.” Causal inference is exceedingly tricky, no research design is perfect, and no single study can realistically aspire to be definitive. But I believe the methodological choices in Rationing the Constitution were well-justified and that the book supplies strong evidence to support the judicial capacity model.

To explain this belief, I must address two distinct questions raised by the commenters. The first, emphasized by Schauer and Nourse, is whether valid causal inference requires comparison of capacity-constrained domains with normal domains. The answer is no.

Any consideration of this question must start with the crucial premise that case studies employ a different kind of causal inference from that employed by statistical analysis, contra the work of King et al. cited by Schauer. Rather than patterns of co-variation across independent and dependent variables, causal inference in case research is based on “the match between what empirical evidence we would hypothesize that the [causal] mechanism should leave and what we actually find in the case.”[2] (Some older qualitative approaches—such as Mill’s method and its various offshoots—do attempt to draw causal inferences from variance across small numbers of cases, but these approaches are now widely viewed as suspect when used for this purpose.) This explains why some case-study researchers prefer the terms “causal condition” and “outcome” to “independent” and “dependent variables.” Within a given case, neither the value of the causal condition nor the value of the outcome actually varies. Instead, a causal condition, such as limited judicial capacity, either does or does not cause a particular outcome, such as deference or categorical rules.

From this account of “within-case” causal inference, it follows that the rationale for studying multiple cases is not to examine variation across cases—for instance, in the frequency of categorical rules or deferential decisions across capacity-constrained and normal domains. Instead, the reason for studying multiple cases is to identify the bounds of the population in which a given causal inference—made within individual cases, rather than across them—holds. For this purpose, the only relevant cases are those in which the causal condition (or the corresponding value of the independent variable) is present. Here that means the capacity-constrained domains in which the judicial capacity model predicts that the limits of judicial capacity will strongly constrain Supreme Court decision-making.

Within this category, the latest qualitative methods literature recommends studying as broad and diverse a range of cases as possible. The greater the number and diversity of cases in which a causal inference holds, “the better we are able to infer [that] what was found in the chosen cases should also be present in other typical cases throughout the population.”[3] As such, the case studies in Rationing the Constitution encompass a broad and diverse range of capacity-constrained domains, but no normal domains, since the causal condition on which the judicial capacity model is premised is not present in those domains. This method amounts to selecting on the independent variable (or causal condition), not the dependent variable, as Schauer suggests. It does not produce selection bias because its goal is not to generate a representative sample of the general population of cases, but rather to identify the subset of that population in which a causal inference drawn from individual cases holds.

This is not to suggest that normal domains are completely irrelevant to the judicial capacity model. Schauer is correct that a close examination of domains in which the Supreme Court is not constrained by judicial capacity could “tell us something about” the model’s generalizability. The symposium contributions of David Marcus and Aaron Bruhl are good examples, illustrating that the Supreme Court feels far freer to employ mushy, non-deferential standards in (at least some) normal domains than it does in capacity-constrained domains. Many other examples could be adduced to the same effect. A systematic examination of such domains would certainly be valuable and could help to more clearly mark the boundary between capacity-constrained domains, where the judicial capacity model applies, and normal domains, where it does not. But this was beyond what I could undertake in a single book. Its omission does not affect the strength of the causal inference in the domains I discuss, merely the extent to which that inference can be generalized beyond those domains.

Schauer is also correct that the judicial capacity model needs to account for “those instances in which the alleged causal agent was present but for which the hypothesized effect was not present.” Nourse puts the same point more concretely: Coan “must explain variance”—i.e., capacity-constrained domains “in which high capacity does not yield the bright line rule, or deference.” If there were capacity-constrained domains in which the model’s predictions did not hold, this would indeed be a problem requiring explanation. If no satisfactory explanation were available, the existence of such domains would falsify the model.

Fortunately for the model, there are no such domains, at least not that I am aware of. The possible examples suggested by Schauer—the dormant commerce clause, the establishment clause, the “time, place, and manner” branch of free speech doctrine, and Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993)—are all normal domains in which the Supreme Court feels perfectly comfortable denying review in virtually all cases. This is largely because the Court’s decisions in these areas threaten few—and, in the case of Daubert and dormant commerce, no—federal laws. In this sense, these domains are similar to the void-for-vagueness and social reform cases discussed by Bruhl and Marcus. Far from undercutting the judicial capacity model, they vividly illustrate how much freer the Supreme Court feels to employ vague, non-deferential standards outside of capacity-constrained domains.

This still leaves the question of observational equivalence. As Schauer and Zeisberg observe, judicial deference could be explained by a legalist commitment to judicial restraint, rather than judicial capacity. It is true that such a commitment—which might be grounded in democratic legitimacy or institutional competence—could explain any given deferential decision. It almost certainly does explain the deference embraced by at least some of the justices some of the time. But most of the legalist literature on judicial restraint is normative, rather than positive. I am aware of no significant positive theory of judicial behavior predicting that Supreme Court justices will systematically behave deferentially over long stretches of time out of a legalist commitment to judicial restraint. Even if there were such a theory, it could not easily explain the Court’s occasional willingness to invalidate federal legislative action in capacity-constrained domains using categorical rules.

Of course, the Court’s use of such rules might be explained by a complementary legalist commitment to rule-bound decision-making, grounded in administrative convenience or predictability. But like the literature on judicial restraint, the legalist literature touting the virtues of rules is primarily normative, rather than positive. A principled commitment to these virtues might explain why some justices employ categorical rules some of the time. But this literature supplies no convincing reason to expect that the Court would systematically employ categorical rules, rather than standards, when invalidating government action in capacity-constrained domains. If anything, as Adrian Vermeule points out, the Court in general has an incentive to over-produce vague standards, which “externalize the costs of uncertainty onto litigants and lower courts.” This incentive disappears—or is at least greatly reduced—in capacity-constrained domains where the Court feels compelled to review a large fraction of the litigation generated by vague standards. This is why the judicial capacity model predicts that the Court will seldom invalidate government action in these domains except in the form of categorical rules, which is exactly what the evidence shows.

To be sure, there are cross-cutting incentives for the justices to employ categorical rules, many of them relating to the effective control of lower courts. These incentives are the subject of a small but growing positive literature on rules and standards. But as I summarize in the book, “the only prediction that [these models] support is that the choice of doctrinal form will be unpredictable.” They cannot, therefore, explain why the Supreme Court would exhibit a stable and predictable preference for rules when invalidating government action in capacity-constrained domains.

To sum up, there is obviously some overlap between the predictions of the judicial capacity model and legalist commitments to judicial restraint and categorical rules. Legalist commitments do not predict the systematic deference or embrace of categorical rules that the judicial capacity model does, but they do predict some judges will defer to the democratic process or employ categorical rules some of the time. To the extent that this prediction overlaps with the predictions of judicial capacity model, it represents a case of observational equivalence. The evidence is consistent with both models and is therefore uniquely explained by neither.

The upshot is that some—though not all—of the Supreme Court’s deference and rule-bound decisions in capacity-constrained domains could be the product of legalist commitments. Alternatively, this pattern of decisions might be the product of judicial capacity limits, or it might be the joint product of both judicial capacity and legal principle. The available evidence is consistent with all of these possibilities. Yet the judicial capacity model still stands alone in its power to explain the Supreme Court’s systematic deference in capacity-constrained domains and the Court’s unwillingness to depart from such deference except in the form of hard-edged categorical rules. There is much more to be said on this subject, but I have already gone on far too long. I refer any interested readers to Chapter 12 of Rationing the Constitution, especially pages 182-86.

I will close where I began, by thanking all of the symposium participants for their thought-provoking contributions and Jack Balkin for organizing. This discussion has given me a great deal to think about and much grist for future work. I hope others will join me in carrying the project forward.


[1] See also Justice Elena Kagan on Supreme Court and Constitutional Law, C-SPAN (Aug. 31, 2016), (“[A]nytime another court invalidates a federal statute, we’ll always take that case . . . on the theory that Congress’s statutes shouldn’t be invalidated, except by us, that we’re the only court that gets to do that in the end.”); H.W. Perry, Jr., Deciding to Decide: Agenda Setting in the United States Supreme Court 2949-50 (ebook 1991).
[2] Derek Beach & Rasmus Brun Pedersen, Causal Case Study Methods: Foundations and Guidelines for Comparing, Matching, and Tracing 116 (2016); see also Gary Goertz, Multimethod Research, Causal Mechanisms, and Case Studies 89 (2017); Gary Goertz & James Mahoney, A Tale of Two Cultures: Qualitative and Quantitative Research in the Social Sciences (2012); Alexander L. George & Andrew Bennett, Case Studies and Theory Development in the Social Sciences (2005).
[3] Brun & Pedersen, supra, at 549.

Older Posts
Newer Posts