When Congress enacts command-and-control regulation, it chooses between implementation through litigation and courts, through bureaucracy, or through a hybrid regime. Since the late 1960s, the frequency with which Congress has relied on civil litigation for frontline enforcement of statutes grew dramatically, and with it grew rates of federal statutory litigation and the role of courts in federal regulatory policy. By the late 1970s, and with increasing intensity over the decades, a critique of these developments emerged that included two core themes. Relative to administrative implementation, direct enforcement through civil litigation (1) weakens democratic control over public policy because litigants and federal judges are harder for the elected branches to control than bureaucracy, and (2) degrades the quality of public policy because the judiciary is a less capable policy-making infrastructure than bureaucracy.
This Article argues that Congress’s reliance on frontline enforcement through civil litigation is associated with how specifically it articulates substantive policy in the statute, versus how much policy-making discretion it delegates to implementing agents. When legislative coalitions rely heavily on civil litigation for implementation, they have incentives to focus more attention and effort on developing and articulating policy substance in the statute, and to leverage more mandatory and specific administrative rulemaking power. The institutional attributes of litigation and courts that make them more challenging to supervise and influence during postenactment implementation, and that render them a less capable policy-making apparatus, create these incentives. This theoretical account contradicts existing arguments offered by the relatively few scholars to consider the relationship between the legislative choice of enforcement through civil litigation, and how much policy substance Congress lays down in the statute.
This Article deploys original data to investigate this theory and its rivals. The data contain granular information about the policy content of significant federal regulatory legislation passed between 1947 and 2008, and about the level of attention and effort legislators and witnesses in committee hearings focused on it. Empirical analysis demonstrates that Congress focused more than twice as much attention in legislative hearings on parts of regulatory statutes relying heavily on civil litigation for implementation, and elaborated policy in those parts of statutes in about twice as much detail. When relying substantially on civil actions, Congress was also much more likely to delegate administrative rulemaking authority, thereby leveraging more administrative expertise and enlarging congressional capacity to influence substantive elaboration of the statute via agency oversight powers.
Ultimately, this Article argues that meaningful assessment of the democratic and public policy consequences of legislative reliance on civil litigation for enforcement must reckon with the fact that—in addition to dislocating some power from bureaucracy to litigants and courts—reliance is associated with a materially enlarged policy-making role for Congress.
To read the full Article with all attached Appendices, please download as PDF here.
The United States has become a “republic of statutes” in an “age of statutes.” Statutory law is the dominant engine for the creation and evolution of rights, obligations, and norms that structure social and economic life in the country. We also live in an age in which “adversarial legalism,” “legalized accountability,” and the “litigation state” are central features of regulatory governance, with civil litigation serving as an important vehicle for enforcing statutes and courts playing a critical role in fleshing out statutes’ meaning in the course of implementation. The American “republic of statutes” is one in which civil litigants and courts are major players in carrying statutes into effect. In 2017, 67 percent of civil actions filed in US District Courts asserted statutory claims. Of those, over 95 percent were privately prosecuted.
The large role of civil litigation in frontline implementation of federal statutes is substantially a function of legislative design. Since the late 1960s, the frequency with which Congress has accorded a substantial role to civil actions in statutory implementation escalated sharply, with express private rights of action coupled with fee shifting or multiple damages provisions playing a notable role. Empirical scholarship has shown that the conjuncture of divided government and party polarization, which grew markedly after the late 1960s, was one important engine that drove this transformation. From Congress’s point of view—primarily Democratic Congresses facing Republican presidents—statutorily-provided opportunities and incentives for private enforcement, as an alternative or supplement to bureaucracy, offered valuable enforcement insurance when Congress distrusted presidential commitment to robust implementation of legislative mandates. In response, from the late 1960s to the mid-1990s, the population-adjusted rate of private statutory suits in federal court exploded by a factor of ten.
Not everyone has been happy about this course of events. By the late 1970s, and with increasing intensity over the ensuing decades, a critique emerged that included two now-familiar themes. Both arguments pivot on comparisons between courts and agencies. First, relative to bureaucracy, channeling more policy implementation into litigation and courts weakens democratic control over public policy because private litigants and life-tenured federal judges are far more difficult for the elected branches and the polity to control than bureaucrats. Second, relative to bureaucracy, implementation through piecemeal litigation before nonexpert judges in a decentralized judiciary produces lower quality public policy. For more than forty years, these democratic and public policy critiques have figured prominently in debates over the consequences of the growing role of civil litigation in public policy implementation.
This Article intervenes in this old debate from a fresh theoretical angle and with new evidence. It investigates the relationship between the legislative choice to rely on civil litigation for frontline implementation, and legislative choices about policy substance in statutory law. It offers a novel theoretical argument that when Congress relies substantially on litigation for implementation, it resolves more policy issues in the legislature, elaborating substantive statutory law in greater detail, and leverages more administrative rulemaking expertise. The theory advanced is that legislative coalitions do this because they are attentive to institutional properties of litigation and courts that render them more difficult for the elected branches to oversee and control in the future, and less capable at making public policy. That is, legislative coalitions anticipate and respond to the concerns identified by the democratic and public policy critiques in the way they fashion the policy substance of legislation.
Relatively few scholars have theorized about linkages between legislative reliance on civil litigation for implementation and legislative choices about policy substance in statutes. Those that have done so reach very different conclusions than the one advanced here. Further, existing theory generates hypotheses or makes assumptions pointing in every possible direction: legislative substance will be vaguer when Congress chooses civil litigation for implementation, it will be more specific, and it will be unaffected by the choice of civil litigation. None of this past work subjected the question to sustained theoretical examination, and none anticipated the argument made here. Moreover, no empirical scholarship, either qualitative or quantitative, has sought to gain traction on which among the competing and contradictory theoretical possibilities is consistent with actual legislative behavior.
This Article deploys original data to investigate its theory that Congress will resolve more policy issues in statutes, and leverage more administrative expertise, when relying significantly on civil litigation for enforcement. The new data are the product of a large-scale empirical study, and they contain granular information about the policy content of federal regulatory legislation passed between 1947 and 2008, the substance of committee hearings on that legislation, and how implementation powers were distributed between courts and agencies. Analysis of the data reveals that Congress focuses more attention in legislative hearings on the parts of regulatory statutes relying substantially on civil litigation for implementation, and elaborates policy in those parts of statutes in greater detail. It is also more likely to delegate mandatory and specific substantive rulemaking power, harnessing more administrative expertise and expanding congressional influence on substantive elaboration of the statute through agency oversight. Private rights of action are distinctively important in producing the effect.
In addition to supporting the theory advanced in this Article, the empirical results reject other existing theories in the literature. They also reject the democratic and public policy critiques’ premise that the policy-making role of Congress is unrelated to the design choice of civil actions. Going forward, the enlarged congressional role must be part of any evaluation of the consequences for democracy and public policy of reliance on civil litigation in statutes.
A. The Choice of Civil Litigation
It is important to be clear at the outset about the meaning of “legislative choice of civil litigation,” and how it fits with the existing literature. When Congress enacts command-and-control regulation, it faces a choice between implementation through bureaucratic machinery, implementation through lawsuits in court, or some combination of the two in a hybrid regime. In the National Labor Relations Act of 1935, for example, Congress did not permit aggrieved parties to file lawsuits to initiate enforcement. Rather, Congress lodged all frontline implementation power in an administrative body to make rules elaborating the meaning of the Act, hold administrative adjudications to evaluate claims, and issue cease and desist orders. In contrast, the job discrimination provisions in Title VII of the Civil Rights Act of 1964 denied bureaucrats the power to make substantive rules, hold adjudications, and issue cease and desist orders. Instead, implementation would occur through lawsuits in court, the only vehicle to secure an order granting relief. The Clean Air Amendments of 1970, in a sense, combined these approaches, providing a substantial role for lawsuits in court—both agency-prosecuted and “citizen suits”—and a variety of significant administrative powers, importantly including substantive rulemaking. In each of these three laws, Congress made a choice distributing power between courts and agencies.
As the Clean Air Act example makes clear, the notion of a choice “between” courts and agencies, adopted in much of the literature that this Article builds on, is highly stylized. It is not a dichotomous choice. Actual implementation designs in the US are very often hybrid, with courts and agencies both playing important roles. This is especially true if one considers that even where all implementation powers of first instance in a statute are administrative, in the federal system the prospect of appellate review by courts under both the Administrative Procedures Act and constitutional due process limitations always looms. In fact, in the National Labor Relations Act, the predominantly administrative design relied on federal courts for an appellate role; and in Title VII of the Civil Rights Act, the predominantly judicial framework also gave bureaucrats a prosecutorial role and some additional modest administrative powers.
Nevertheless, there is substantial statute-level variation in how implementation power is distributed between courts and agencies. One very important source of that variation is whether statutes provide rights of action for de novo civil litigation to enforce against alleged violators. When invoking the stylized formulation of a choice “between” courts and agencies, scholars have often focused on de novo civil litigation¾particularly whether private rights of action are provided¾in conceptualizing when statute-level design choices accord a central role to courts in policy implementation. This line of work frequently invokes the stylized choice “between” courts and agencies when examining the costs and benefits of reliance on civil litigation as compared to the alternative of bureaucracy. Building on this literature, I partly adopt this stylized formulation with the purpose of understanding legislative coalitions’ incentives when they rely on civil litigation. In Parts II and III, I discuss some important theoretical and empirical limits of this stylized formulation. This Article refers to substantial reliance on civil actions for enforcement and implementation as a “litigation regime.”
Part I first describes the concept of “statutory specificity,” which, under a variety of labels, has long figured in important debates about the delegation of legislative power. It then reviews existing theories, implicit and explicit, about linkages between the legislative choice of litigation regimes, and the degree of substantive specificity with which Congress legislates. It shows that the assumption that substantive specificity is unrelated, or exogenous, to the choice between courts and agencies is pivotal to two broad critiques of delegation to courts that have been repeated for more than forty years: that delegation to courts, as compared to bureaucracy, is (1) undemocratic, and (2) produces bad public policy. This Article further examines one set of scholars who have also theorized or assumed that statutory specificity is unrelated to the choice between courts and agencies. In their view, the extent that Congress resolves policy issues in a statute (rather than delegating the power to resolve them) is determined by a legislative cost-benefit calculus, and this calculus is not influenced by whether the delegation will be to a court or an agency. According to this “institutional neutrality” perspective, the institutional choice between courts and agencies is neutral with respect to legislative incentives for statutory specificity.
Another set of scholars examined in Part I argue or imply that, in fact, statutory specificity is related, or endogenous, to the choice between courts and agencies in the context of controversial issues, with the choice of courts associated with vaguer statutes. In the face of policy controversy and conflict, Congress seeks to avoid responsibility for specific policy decisions by writing vague statutes that punt politically risky issues. It also tends to delegate those statutes to courts because their weak electoral accountability, low public visibility, apparent objectivity, and seeming independence from the legislature will accomplish a greater responsibility shift away from the legislature for the inevitable, controversial decisions that will ensue. According to this “courts as dumping grounds” view, courts are magnets for vague statutes regulating politically conflictual issues.
Drawing on institutional theory, Part II advances a rival theory that generates a prediction that flatly contradicts existing scholarship. It argues that there are deep institutional differences between courts and agencies that will lead Congress to invest more in developing and specifying policy substance when deploying litigation regimes. Courts are far harder than agencies for legislators to control postenactment. Legislators have a weaker apparatus to monitor the policy-making activity of courts; lesser capacity to communicate their preferences to courts; virtually no opportunity for the types of informal legislative influence and intervention in courts that are common in bureaucracy; materially weaker capacity for the use of procedural devices to control substantive policy-making by courts; and a comparatively feeble arsenal of plausible sanctions to threaten courts into submission or punish them for transgressions. In addition, many have argued, courts have lesser policy-making capacity. Judges are rarely trained experts in any particular policy area, and bureaucrats are. Further, in a decentralized judiciary, piecemeal litigation of discrete issues before hundreds of judges, relying primarily on information presented by adversarial litigants, may inhibit the development of a unified, coherent, and effective regulatory scheme. Administrators working within hierarchical agencies with far deeper capacity to gather and assimilate relevant information are more capable policy-makers.
Many of these institutional differences between courts and agencies are well known. However, no theory has unified them into an explanation for how much substantive policy Congress decides and how much it delegates, conditional on how it distributes power between courts and agencies. Finally, Part II argues that existing empirical scholarship shows that Congress recognizes that courts will be harder to control than agencies postenactment, and that they possess less policy-making capacity; that Congress intends to delegate less discretion to courts than to agencies; and that legislative coalitions fashion the details of legislation with concern about controlling downstream implementation, and with an understanding of differences between courts and agencies. This view of Congress is in strong tension with the “institutional neutrality” and “courts as dumping groups” perspectives, which contemplate a Congress indifferent to statutory detail and downstream policy consequences. Part II concludes that, in light of the institutional argument laid out and the extensive body of empirical scholarship reviewed, it is difficult to fathom, on average, that legislative coalitions would not resolve more policy issues in statutes, writing more specific laws, when delegating to courts. Finally, Part II argues that the same institutional logic also increases legislative incentives to delegate administrative rulemaking powers when deploying litigation regimes, harnessing more administrative expertise and enlarging legislative influence over statutory elaboration via congressional oversight of agencies.
Part III turns to original evidence to evaluate the contending theoretical claims set out in Part II. If theoretical attention to the core question addressed by this Article has been scant, efforts to study its empirical reality have been entirely absent. The evidence shows that when Congress deploys litigation regimes, it focuses more than twice as much attention and effort on developing policy substance in committee hearings, as measured by a combination of testimony received, documents submitted, and legislator questioning. It also shows that a substantial role for civil litigation is associated with about twice the level of substantive policy specificity. The role of private rights of action is particularly important in driving the results. This association between litigation regimes and substantive specificity is distinctively large in the context of policy controversy. Finally, reinforcing these results, Part III shows that when Congress makes civil actions central to implementation, it is much more likely to delegate mandatory and specific administrative rulemaking authority, thereby leveraging more administrative expertise, more actively controlling and shaping administrative rules, and enlarging congressional capacity to influence substantive elaboration of the statute through agency oversight.
The Article concludes by arguing that its theory and evidence shed new light on old and enduring debates about the relationship between courts, public policy, and democracy. Those debates have failed to account for the possibility that delegation to courts is associated with a greater policy-making role for Congress. Finally, the conclusion links the Article’s results to work on divided government and party polarization. It highlights how the growth of these conditions has simulated greater congressional reliance on civil actions for statutory enforcement, and through that pathway has, ironically, led to greater congressional participation in regulatory policy-making.
I. Prevailing Views about Specificity and Delegation to Courts versus Agencies
A. What Statutory Specificity Is
Statutory specificity is not a technical concept. Its opposite is statutory vagueness or indeterminacy. At bottom, statutory specificity concerns how much policy-making power is exercised in the legislature and how much is delegated to other actors and institutions, such as courts or agencies. Statutory specificity has been an enduring subject of inquiry for students of law and politics for a half century. Understanding the dynamics that determine the degree of statutory specificity can help elucidate who is making the rules that govern the nation and how this varies under different political and institutional conditions, such as different configurations of judicial and administrative power. This raises fundamental questions of law and politics in a democratic society.
When courts or agencies are given power to implement statutes, they are often required to make policy judgments about the statute’s meaning and application, thereby exercising policy-making discretion and power. The level of power delegated to implementing agents is importantly a function of the level of specificity in the statute. Vaguer statutes, such as the Mann Act of 1910’s famous prohibition of “immoral practice,” delegate extremely broad interpretive power to implementing agents. Exactingly specific statutes, such as the tax code, delegate much less power. As Huber and Shipan explain: “Legislative statutes are blueprints for policy-making. In some cases, legislatures provide very detailed blueprints that allow little room for other actors . . . to create policy on their own. In other cases, legislatures take a different approach and write statutes that provide only the broad outlines of policy,” giving implementing agents “the opportunity to design and implement policy.” 
This Article’s focus is on substantive policy specificity, which concerns how much detail the legislature supplies regarding the resolution of the ultimate policy problems targeted by the statute. In regulatory laws such as environmental, civil rights, or antitrust laws, substantive specificity concerns the portions of statutes that actually identify what behaviors, under what conditions, and by whom, constitute illegal pollution, discrimination, or anticompetitive actions. For example, a substantively vague statute barring gender discrimination in employment may simply prohibit “discrimination” on the basis of gender in the terms and conditions of employment. A more substantively specific statute may further address whether independent contractors are covered, or only persons meeting the traditional definition of employee; whether religious institutions are exempted; whether gender discrimination must be intentional to violate the law; and whether affirmative action is permitted to remedy gender imbalances in a workforce disfavoring women. These are all questions of policy substance.
A statute that answers all of them, as compared to one that simply prohibits “discrimination” in employment based on gender, has a higher level of substantive specificity. In it, the legislature has made all of these critical policy decisions. In a statute that simply prohibits “discrimination,” the legislature has delegated these policy decisions. In the fullness of time, an implementing agent with interpretive authority—whether judicial or administrative or both—will be called on to resolve issues on which the statute is silent. The implementation details of the statute, such as whether it provides for civil actions or administrative rulemaking, will substantially determine which implementing agents will fill out statutory meaning.
For purposes of the analysis that follows, it is important to distinguish a statute’s substantive policy specificity from its procedural specificity. Congress sometimes relies on extensive procedural controls and requirements that implementers must satisfy, such as deadlines, necessity of public notice, evidentiary requirements, consultation with designated public officials, public hearings, and appeals procedures. Much of the literature just cited argues that such procedural requirements are used in legislation as a strategy to control the substance of policies that implementers will ultimately make. Legislative specification of actual policy substance in the statute, and legislative specification of detailed procedural requirements, are two different ways in which legislatures exercise control over policy. The use of procedural constraints as a strategy to control policy has received vastly more attention from scholars. This Article focuses on substantive policy specificity.
B. Why Statutory Specificity Matters: The Democratic and Public Policy Critiques, and the Assumption of Exogenous Substantive Law
The assumption that substantive specificity is unrelated to the legislative choice of civil litigation is central to two broad and well-known critiques of the role courts play in policy implementation. In discussing these critiques, my purpose is twofold: to demonstrate the widespread assumption that substantive specificity is unrelated to the choice between courts and agencies, and to show that the assumption is central to major and recurrent debates about the role of litigation and courts in American government.
1. Democratic Critique
The democratic critique of civil litigation in statutory implementation has two dimensions¾one focused on comparing agencies and courts as statutory interpreters, and the other focused on comparing private and public control of the prosecutorial function. Before discussing each, it bears emphasis that the scope of the democratic critique, as conceived in this Article, concerns the role of courts in statutory implementation, not judicial review of the constitutionality of governmental actions. If one accepts the legitimacy of judicial review in the American constitutional framework, it can be both appropriate and desirable for federal courts to act in contravention of the majoritarian will as expressed through Congress. Something like the democratic critique discussed in this Article is sometimes leveled against federal courts’ exercise of the judicial review function. However, in that context, the critique is typically couched as an argument that federal judges should be sparing and cautious in their exercise of judicial review powers, rather than as an argument against judicial review itself.
The democratic critique as applied to statutory implementation regimes, in contrast, operates on a very different normative terrain. It concerns the appropriate means to implement ordinary statutory policy where no questions of constitutionality are implicated. The American constitutional order—where the first clause of the first section of the first article of the Constitution provides that “All legislative Powers herein granted shall be vested in a Congress of the United States”—contemplates democratic control of legislative rules. The democratic critique discussed in this Article is focused on the implementation of legislation whose constitutionality is not challenged; in this sphere, the critique has its greatest normative force and widest support.
a. Democracy and Substantive Lawmaking
The democratic critique starts with the recognition that judicial and administrative interpretation of statutes entails policy-making. The critique hinges on the contention that, particularly in the federal system, judges are less democratically legitimate and accountable policy-makers than administrators. As applied to the domain of federal statutory law, the democratic critique complains that legislative delegations to courts have increasingly given them excessive power, fostering “judicial imperialism” by “activist” judges who interpret statutes in ways that would never succeed in a politically accountable institution. Appointed and life-tenured, federal judges are not electorally accountable, the elected branches cannot effectively supervise them, and they will stray from the democratic will precisely because they cannot be disciplined by it.
Proponents of this view argue that, by comparison, agencies are far more democratically legitimate, accountable, and responsive. At bottom, this argument is founded on the ability of presidents, Congress, and voters to hold administrators accountable in ways that life-tenured judges cannot be. Administrative bodies generally are led by people who are appointed by the President. They can be influenced by presidential oversight instruments and often can be removed by the President. Congress has available to it a variety of mechanisms to oversee administrators, including investigations, oversight hearings, and control over agency budgets. Ultimately, voters can replace presidential administrations in elections, bringing new presidents to power who will appoint new bureaucratic leaders with preferences closer to the polity. In sharp contrast with federal judges, both the elected branches and the electorate itself can oversee bureaucrats. Thus, delegation to federal courts versus agencies is associated with less democratic control over policy.
b. Democracy and Prosecutorial Power
While the democratic critique of federal judges as policy-makers applies to implementation through civil actions regardless of whether prosecutors are public or private, a second dimension of the democratic critique focuses specifically on private lawsuits. This line of criticism starts with the premise that the exercise of prosecutorial discretion is an important form of legal power. Government prosecutors in civil litigation are policy-makers who make bureaucratic judgments about whether it is appropriate to press a legal claim through litigation, and if so, how to frame it and carry out the prosecution. Delegation of this power to private attorneys and litigants is dangerous because they will exercise prosecutorial discretion in the service of private, particularistic, and often economic interests, which may conflict with the public interest. As Derthick puts it, “[p]olicymaking through litigation, engaging as contestants the parties principally at interest,” is “deeply problematic” for democracy compared to administrative power.
Like life-tenured federal judges, the elected branches cannot effectively supervise private litigants and lawyers if the private parties exercise the prosecutorial function in a manner injurious to the public interest. In contrast, public officials in the bureaucracy who are empowered to enforce regulatory policy through lawsuits are under a duty to exercise the prosecutorial function in the public interest; they are not guided by personal economic incentives; and, most important, the President, Congress, and the voting public can supervise them with the tools of bureaucratic oversight discussed above.
2. Public Policy Critique
Critics of delegation to courts also argue that it leads to substantively bad policy. Two dimensions of this critique are most relevant to the present discussion. The first focuses on judges’ lack of policy expertise, and the second focuses on the nature and structure of the judicial process. Regarding expertise, of course, a primary justification for delegation of policy implementation to bureaucracy is to leverage the expertise—informational resources, analytical competence, etc.—of policy-makers within an administrative body. The public policy critique emphasizes that federal judges are generalists by training, and in the course of judging they deal with a multitude of policy areas, one after another, developing a depth of knowledge in none. This makes federal judges, on balance, far less informed and expert than administrators as policy-makers.
Regarding structure and process, this body of work argues that courts are a structurally inferior policy-producing machinery, leading to policy cacophony. Litigants set courts’ agendas by selecting and framing issues while pursuing highly particularized interests. Because the court system is decentralized, non-expert judges make policy piecemeal, one case at a time, often without adequate consideration or understanding of the larger policy scheme. On an ideologically heterogeneous federal bench, the multitude of judges authoring policy often work at cross-purposes, seeking to advance conflicting and even contradictory agendas. As compared to a more centralized, unified, and integrated administrative scheme, orchestrated by an administrator at the top of a hierarchical agency with powers of national scope, the American judicial process and structure produces regulatory policy that is less consistent and coherent, and therefore of lower quality.
3. The Assumption of Exogenous Substantive Law
The democratic and public policy critiques assume that substantive statutory law is unrelated, or exogenous, to the delegate chosen to implement it. That is, the critiques suppose that when Congress delegates to courts rather than to administrators, some discrete block of policy-making power that might have been given to more politically accountable and expert administrators, operating within a more centralized policy machinery, is instead given to less politically accountable and expert judges, who operate within in a more decentralized policy machinery. This assumption is essential to the logic of both arguments, which pivot on institutional comparisons between courts and agencies. Of course, many regard Congress to be superior to agencies on the dimension of democratic legitimacy and accountability. Some also suggest that Congress possesses institutional capacity, through committee specialization, for expert policy-making comparable to agencies, even if it is costly to exercise. Conventional wisdom is that Congress is superior to federal courts on both dimensions. While these propositions are surely contestable, this much is clear. If Congress focuses more attention and effort on developing substantive policy when it relies on civil litigation, and specifies policy in greater detail, then assessing delegation to courts versus agencies against the metrics of democratic control and policy quality must take the enlarged role of Congress into account. If Congress more actively leverages and directs administrative rulemaking when deploying litigation regimes, this too must be factored into the analysis.
A relatively small number of scholars have taken up the theoretical possibility that substantive statutory law may be associated with the legislative choice of civil litigation. Of those who have, all that I am aware of reach conclusions that broadly support the assumptions or concerns of the democratic and public policy critiques. I turn to this work next.
C. The “Institutional Neutrality” of Claiming Credit, Avoiding Blame, and Minimizing Costs: The Argument for Exogenous Substantive Policy
A number of scholars—including law professors Matthew Stephenson, Margaret Lemos, Frank Cross, David Spence, and political scientist Thomas Burke—have theorized that, in fact, substantive policy specificity is unrelated to the choice between courts and agencies, or have explicitly embraced this assumption. I refer to this as the “institutional neutrality” view because it regards institutional differences between courts and agencies as neutral with respect to Congress’s incentives to create substantive policy specificity. Read together, this work identifies two key vagueness-inducing mechanisms and argues or explicitly assumes they will operate in the same way whether Congress is delegating to courts or agencies. The two mechanisms—political conflict and transaction costs—are among the dominant (though not the only) explanations given for statutory vagueness by legal scholars and political scientists.
First, scholars most frequently use political conflict over an issue to explain statutory vagueness. In the context of political divisiveness in the legislative process, vague and ambiguous laws allow legislators to respond to public demand for legislative action and claim credit for passing a law with a high-sounding title, while avoiding responsibility for difficult specific decisions, sloughing them off on implementing agents, whether they be courts or agencies. Moreover, quite aside from avoiding blame for specific decisions, vagueness may also allow passage of a law in the face of political conflict where a more specific law simply could not have commanded sufficient support to pass.
Second, statutory vagueness is a way to avoid, or delegate, the work and costs associated with writing specific policy. Epstein and O’Halloran characterize legislators’ choice of the degree of statutory specificity as analogous to a firm’s “make or buy” decision, and their account is as follows. Legislators, particularly through committee specialization, are in fact capable of “making” detailed legislation that addresses complex issues. However, to the extent that Congress elects to make policy itself, there will frequently be uncertainty about what policy should be made. In the face of this uncertainty, the information and expertise necessary to produce high quality policy will often be costly to obtain and assimilate, requiring extensive time and effort, such as through committee hearings. These transaction costs associated with making detailed policy, Epstein and O’Halloran argue, will be driven up further because the institutionally fragmented structure of Congress—particularly its many veto points and pervasive logrolling—makes lawmaking there cumbersome, difficult, and expensive. On the other hand, Congress can “buy” policy by delegating to bureaucracy, leveraging its expertise and institutional capacity, but only at the cost of losing a measure of control over policy since Congress will not be able to control bureaucrats perfectly.
The key move of scholars who treat substantive policy specificity as exogenous to the choice between courts and agencies is to explicitly assume, or to affirmatively argue, that these factors (and others) affecting specificity will operate in the same way whether legislators are delegating to courts or agencies. According to the “institutional neutrality” view, whether Congress delegates to courts or agencies, it will want to avoid the transaction and political costs of achieving specificity, and will want credit for vague and aspirational legislation. The degree to which this is true will not be affected by the choice of delegate.
D. “Courts as Dumping Grounds” for Vague Statutes: The Argument for Endogenous Substantive Law
Other important scholarship—including work by Mark Graber, Eli Salzberger, and Paul Frymer—suggests that substantive specificity will be internally related, or endogenous, to the choice between courts and agencies, with delegation to courts associated with vaguer and ambiguous statutes under conditions of political controversy. I refer to this as the “courts as dumping grounds” perspective. It focuses on institutional differences between courts and agencies, suggesting that courts are a more attractive delegate than agencies when Congress wishes to evade blame for politically conflictual policy decisions, and claim credit for responding to public demand for congressional action. As discussed in the last section, conditions of political conflict have been widely identified as producing statutory vagueness. The “courts as dumping grounds” perspective does not maintain that statutory drafters write vaguer laws because they recognize that a statute will be implemented in courts. Instead, the dynamic suggested is that public demand for a law, alongside political conflict over what its substance should be, simultaneously increase, from legislators’ point of view, the political utility of implementation in courts, and the political utility of substantive vagueness.
Read together, this perspective makes several institutional claims about courts. First, when controversial issues must be addressed, the judiciary tends to be a less visible and politically safer policy-making site, lowering the probability of public notice and electoral repercussions. Second, the public tends to perceive the judiciary, often wrapped in the cloak of legality, procedure, and technicality, as a more objective, neutral, and legitimate decision-maker than other governmental bodies. Third, the public perceives that the judiciary is, and should be, more independent of control by legislators than more politically accountable bureaucratic decision-makers. If judges make specific unpopular decisions, legislators can criticize the decisions as misguided, while deferring to them out of respect for the “rule of law” and “judicial independence.” Fourth, the judiciary’s greater independence from electoral politics and political control, and its apparent objectivity and neutrality, increase its capacity to achieve settlement of issues that could remain unsettled and politically disruptive to the ruling coalition if not absorbed by the judiciary.
On balance, therefore, legislators perceive that when legislating on an issue that is politically risky, delegating to the courts will yield a greater “responsibility shift,” as Salzberger puts it, than delegating to agencies. The “courts as dumping grounds” perspective suggests that in the context of politically controversial issues, legislative impulses toward credit calming and blame avoidance will simultaneously produce the twin effects of vague statutes and delegation to courts.
II. Legislative Incentives for Greater Substantive Policy Specificity When Delegating to Courts
This Part argues that, contrary to the forgoing views, institutional differences between federal courts and agencies provide strong incentives for Congress to resolve more issues of policy substance when it delegates to courts. While I am aware of no scholar making an argument to this effect, some scholarship assumes (without explanation) that it is true. The argument sees more substantive policy specificity as a legislative response to the courts’ greater independence from congressional control, and their weaker policy-making capacity. This Part argues that legislative coalitions, like academic proponents of the democratic and public policy critiques, recognize these problems, and mitigate them by resolving more issues in Congress, specifying statutory policy in greater detail, and leveraging more administrative expertise when delegating to courts.
Before turning to the details of the argument, two questions need to be addressed in advance. First, if courts are inferior policy-makers and harder to control, why would Congress ever delegate to them rather than agencies? Without some gains secured by delegating to courts, it is difficult to see why Congress would incur the greater transaction costs of resolving more issues in the legislature to compensate for institutional properties of courts. The literature on delegation to courts has identified several reasons that Congress relies on rights to sue—particularly private rights of action—as an alternative or supplement to bureaucracy. They include, but are not limited to:
- Presidential Subversion. When Congress is concerned that the President will subvert bureaucratic enforcement of congressional preferences, such as under divided government, Congress will be more likely to mobilize private lawsuits, with private rights of action coupled with attorney’s fee awards and economic damages for enforcers, as insurance against executive underenforcement.
- Uncertainty about Endurance of Coalition. The enacting coalition may be concerned that it will not continue in power, and will not be present to engage in congressional oversight of bureaucracy to ensure enforcement of its statutory mandates in the face of bureaucratic resistance or failure. Uncertainty about remaining in power thus encourages Congress to rely on private lawsuits, sufficiently incentivized, to provide for continued “autopilot” enforcement of legislative mandates in the absence of the enacting coalition.
- Bureaucratic Drift. Legislative coalitions sometimes lack faith that bureaucracy will enforce legislation aggressively because they perceive bureaucrats as politically timid, apathetic, careerist, and vulnerable to capture. They therefore advocate in the legislative process for private rights of action in court as insurance against bureaucratic failure.
- Budget Constraints. Lack of adequate tax revenue, or the political costs of raising it, encourages Congress to achieve public policy goals through private lawsuits because it shifts substantial implementation costs away from the state and to private parties. Further, the costs borne by the judiciary are less traceable to particular legislative enactments than appropriations for agencies. This implementation strategy thus can attract a broader support coalition in a tax-averse political environment.
- In the United States’ relatively “antistatist” political culture, reliance on lawsuits often can attract a broader support coalition than is possible with bureaucratic implementation requiring administrative state building.
This list is not exhaustive. The key point is simply that there are a variety of sources of utility to be derived by enacting coalitions from relying on civil litigation in implementation. Under some circumstances, then, it is sensible that they would be willing to incur the costs of legislating with greater substantive specificity because of their choice to delegate to courts, provided that the overall cost-benefit calculus warrants it.
A second question best answered in advance is this: Since the legal territory investigated by this Article is statutory, is Congress not free at any time, under the Article I grant of all legislative powers to Congress, to override any administrative or judicial decision by simply amending the statute? If so, does Congress not ultimately have equal control over courts and agencies? The answer is that while this is formally true, in reality statutory amendment is a weak mechanism of control because it is often extremely costly and frequently impossible. The United States’ institutionally fragmented lawmaking process empowers many veto players, and all will have to regard legislative override as preferable to the status quo in order to coordinate on accomplishing one; the enacting coalition may no longer be intact; its preferences and priorities may have changed; the transaction and opportunity costs of passing new legislation, especially in this fragmented environment, are very high; and there is uncertainty about the legislative outcome if the legislative process is opened up. In this lawmaking system, as Moe puts it, “Whatever is formalized will tend to endure.” Consequently, at the time of drafting a law, an enacting coalition will recognize that it would be a very risky strategy to plan on controlling implementation through legislative amendment. In this institutional environment, tools to control implementing agents without actually passing a new law are enormously valuable.
The argument developed below focuses primarily on such tools of control. It draws on many well-known ideas in the literature on institutional properties of courts and agencies, but the argument is novel in using them to build a theoretical explanation for the degree of substantive policy specificity conditional on the distribution of power between courts and agencies. I argue that legislative coalitions, when according a substantial role to civil actions for enforcement, have incentives to enact more detailed policy substance because:
- Courts are harder for legislators to monitor and communicate with than agencies.
- Legislators lack opportunities to act as intermediaries and wield informal influence on behalf of constituents before courts, which they routinely do before agencies.
- Legislators wield much weaker and less credible sanctioning capacity against courts than agencies.
- Ex ante procedural instruments of control are less effective with courts than agencies.
- The judiciary has more limited policy-making capacity than agencies.
- Existing empirical evidence, both qualitative and quantitative, supports the conclusion that legislative coalitions are aware that courts will be harder to control than agencies postenactment, and have lesser policy-making capacity; that Congress intends to delegate less discretion to courts than to agencies; and that such coalitions design legislation with the goal of influencing implementing agents in the future, with appreciation of the differences between courts and agencies.
It cannot be overemphasized that the argument is purely relative. The argument is not that courts operate free from legislative influence at the implementation stage, or lack policy-making capacity. Nor is it argued that agencies are easily and perfectly controllable by Congress or have limitless policy-making capacity. The argument is that courts, as compared to agencies, have more autonomy from Congress at the implementation stage, and weaker policy-making capacity.
A. Courts are Harder for Congress to Control
1. Monitoring and Communication are Less Effective with Courts
Ex post legislative influence on implementation is, of course, greatly facilitated by legislative knowledge of planned and actual implementation activity, and by an opportunity to communicate legislator preferences to implementing agents before actions are taken. Legislative monitoring and communication is less effective with courts than agencies.
Agencies. Congress uses a variety of methods to monitor agency implementation. In legislation empowering agencies, and on an ad hoc basis, Congress imposes information collection, accounting, record keeping, and reporting requirements on agencies, producing “oceans of data and reports from offices within agencies about ongoing programs.” Congress also maintains its own bureaucratic monitoring capacity to provide independent and detailed information about ongoing agency activities and programs, most notably in the Congressional Budget Office and the General Accounting Office, as well as independent offices of Inspectors General within federal departments and agencies. Members of Congress also rely on “fire alarm” oversight, creating rules, procedures, and practices that allow constituents to examine agency decisions prospectively and seek aid from legislators before such decisions are made.
Most congressional monitoring of agencies takes the form of informal communication through what Aberbach calls the “committee intelligence system.” Legislative committees have ongoing relationships with staff and leaders of agencies under their jurisdiction. Committee members and their staff have dense networks of contacts both within and outside agencies, which they continuously draw on to stay abreast of ongoing agency activities. As Melnick explains, “Agency officials work with Congress day after day, revising legislation, negotiating yearly appropriations, vetting nominees for political posts, and listening to members’ complaints about the consequences of agency decisions.” The continuous contact and regular stream of communication fosters familiarity to the point that some have characterized congressional committees and agencies under their jurisdiction as being “in bed” together, or as “sweetheart alliances.”
Courts. Legislative monitoring and communication with federal judges is far weaker. As compared to a centralized agency with its nerve center in Washington, the judiciary’s highly decentralized structure makes it more difficult for legislators to even observe what issues are being litigated through lower federal courts across the country. Staff of congressional committees with jurisdiction over relevant policy find it more difficult to stay abreast of important judicial decisions from lower federal courts, and when they do hear of cases via constituent complaints, it is typically only after decisions have been rendered. As one committee staffer put it, “Court decisions are not visible. Therefore, they are not easy targets. Agencies are highly visible. We fight with agencies all the time.”
Further, in a decentralized judiciary, legislators and their staff do not have repeated interactions and open lines of communication with the specific federal judge, or a panel of them, that will render a decision of particular concern. They lack dense networks of contacts embedded in the federal judiciary. In fact, now-Court of Appeals judge Robert Katzmann concludes, based on a book-length empirical study of communications between Congress and the federal courts, that there is a “communications gulf” between them. Even if legislators are aware of and concerned about pending or impending issues before federal courts, they are not able to access key decision-makers. Both legislators and judges are acutely aware that they could be legally or ethically compromised by communications about pending cases. As contrasted with the intimate familiarity arising from daily legislative-bureaucratic contact, legislative-judicial communications (to the extent that they occur) are characterized by “awkward unease.” If legislators wish to be heard by the court, they can request permission to file an amicus brief.
2. Informal Interventions are Less Effective with Courts
Agencies. Dense communication networks linking committees and agencies facilitate not just information collection by Congress, but also an opportunity for legislators to communicate their preferences to agencies. When legislators or their staff receive constituent complaints, they can intervene through phone calls, letters, or jawboning agency officials, acting as intermediaries in informal negotiations on the constituents’ behalf. As one long serving member of Congress put it, referring to agency oversight, “We find our letters have a special effect on a lot of people.” Congressional committee members and their staff report that efforts to change agency decisions are regularly effective. The use of such informal influence by legislators provides opportunities to claim credit for resolving problems, which is a form of constituency service that serves electoral goals.
Courts. This strategy is not feasible with courts. Congress does not have opportunities for informal negotiation and influence with courts. If, as discussed in the last section, communication with federal judges about pending or impending case raises potentially compromising ethical and legal issues, efforts to negotiate a favorable judicial decision for a constituent is simply not an option.
3. Sanctions are Less Effective with Courts
Agencies. After passage of a law, at the implementation stage, Congress has an “awesome arsenal” of weapons to punish disobedient bureaucrats. Two of the most important are appropriations and investigatory or oversight hearings. Congressional decisions about agency funding through annual appropriations bills provides an especially potent mechanism of influence. The club of budget cuts can be targeted either at an agency in general, or more specifically at particular programs or activities within it, if Congress wishes to issue a more focused sanction. The appropriations authority can be used as a “backdoor legislative veto” because, if an agency takes actions opposed by Congress, a subsequent appropriations measure can prohibit the expenditure of any funds to support the specific administrative activity in question.
Congress can also use public (and sometimes televised) hearings and investigations to “subject recalcitrant bureaucrats to public humiliation that devastates their careers,” a weapon that can be directed at specific administrators that Congress wishes to target. As two US Senators characterize this power, oversight hearings allow them to “drag realities out into the sunlight and demand a full accounting from those who are permitted to hold and exercise power.” Such public ventilation of attacks and allegations of wrongdoing can have adverse consequences for both agencies and specific administrators.
One notable feature of Congress’s sanctioning power is that it “can be used with the discrimination of a stiletto or the explosive power of a bomb, and this flexibility rescues the legislature from the immobilism of . . . too devastating a weapon to employ for any but the most drastic confrontations.” That is, Congress can calibrate its sanctions against bureaucracy to its purpose.
Courts. Federal judges are far more difficult to sanction because (1) they have life tenure and salary protection, (2) sanctions against them are costlier and less effective because of the generality of their jurisdiction, and (3) it is politically riskier to attack federal courts than bureaucrats because of their higher levels of public legitimacy.
a. The Insulation of Life Tenure and Salary Protection
Perhaps most obviously, life-tenured federal judges enjoy a level of security from job loss unique in American government, and constitutional insulation from reduction in salary. For all practical purposes federal judges cannot be fired and their salaries cannot be reduced. The very purpose of this design in the American constitution was to insulate them from political attack. Based on congressional dissatisfaction with judicial decisions, federal judges cannot be called into hearings for public humiliation that devastates their careers. As compared to bureaucrats, federal judges have a higher level of discretion to make policy choices without fear of direct personal attacks on their job security, income, or future professional opportunities.
b. The Insulation of General Jurisdiction
While Congress is extremely limited in its ability to target and punish individual judges, it certainly has tools to attack the federal judiciary, as an institution, without amending a law. While judicial salaries are expressly protected from being reduced, Congress could let inflation erode them. Congress could refuse to confirm new judges as vacancies in already-authorized judgeships arise, driving up caseload pressures. Most significantly, through annual appropriations, Congress could slash funding for the judicial infrastructure. In order to function effectively, the federal judiciary needs money for buildings, law clerks and other staff, and technological infrastructure, and appropriation of such funds is annually within Congress’s discretion. Congress could devastate the federal judiciary as we know it without passing a new law.
However, the kind of general attacks on the federal judiciary that could be accomplished without passing new law would come at a very high cost compared to more focused attacks on specific agencies, administrators, or programs. When it comes to legislative sanctions against the federal judiciary, short of passing legislation, Congress has only “explosive bombs” that are undiscriminating in their devastation. It lacks stilettos. This is due to the breadth of federal court jurisdiction, and the critical governmental, social, and economic functions performed by the federal judiciary.
As compared with agencies’ more specific policy jurisdictions, federal courts’ jurisdiction spans broadly across all questions of federal law (and beyond it). Effective attacks on the federal judiciary would thus diminish its capacity for all purposes, not just for purposes of engaging in the offending conduct. In the field of constitutional law, the federal judiciary’s functions are core to the integrity of the American state, such as protecting individual constitutional rights against state and federal governmental encroachments, resolving separation of powers conflicts, conflicts among the states, and conflicts between federal and state governments. In the field of statutory policy, it includes frontline implementation responsibilities across many areas of regulatory policy, including civil rights, environmental, labor, antitrust, intellectual property, and all manner of financial regulation, among others. Under the Administrative Procedures Act and Fifth Amendment constitutional due process, the federal judiciary plays a critical role supervising federal bureaucracy, not to mention state bureaucracy under Fourteenth Amendment due process.
The generality and importance of federal judicial functions is such that damage to the federal judiciary as an institution would reverberate widely across social and economic sectors and interests. This raises both the policy and electoral costs of material legislative attacks on the judiciary. On the policy side, damaging the judiciary because of dissatisfaction with how it has treated one issue or policy domain will limit its institutional capacity with respect to many others, in which its role may be vitally important and desirable to legislators. On the electoral side, it is difficult to imagine a member of Congress whose constituents would not be harmed by serious damage to the judiciary, even while some others might be benefited. Accordingly, the scenarios in which a legislators’ net policy preferences or electoral interests are served by such a general attack on the federal judiciary, as compared to a focused attack on a particular agency, administrator, or program, are vastly fewer. The generality of federal court jurisdiction is a great source of insulation from attacks by the political branches.
c. The Insulation of Judicial Legitimacy
Another factor rendering legislative attacks on courts costlier than attacks on bureaucracy is that of judicial legitimacy. Scholars have theorized that this judicial legitimacy arises from qualities of judicial process and reasoning conveying an impression to the public of objectivity, neutrality, and legality, a phenomenon some have referred to as the “myth of legality.” Whatever the causes, since the General Social Survey began asking respondents in 1973 about their confidence in the Supreme Court and Congress, the Court has ranked higher, and the disparity has widened materially over time.
The judiciary’s higher level of public legitimacy increases the probability that material legislative attacks on it will provoke public censure against those who level them, and therefore makes such attacks costlier and less likely. Numerous scholars have observed that attacks by the legislature on the judiciary can be politically costly. This contrasts with legislative attacks on bureaucracy. Bureaucracy does not possess the same degree of presumptive legitimacy in the public eye. In contrast with courts, legislators are more likely to perceive raking bureaucracy and bureaucrats over the coals to be popular with voters, and thus to serve electoral goals. Based on extensive interviews of legislators and their staff concerning their willingness to attack courts versus agencies, Miller concludes that because of the public perception of courts as “mostly non-political” in nature, a congressional committee is “less likely to be as aggressive against the courts as it is against federal agencies.” Obviously, at times legislators see political advantage in attacking courts—particularly rhetorical attacks that pose no real threat. The claim here is only that courts’ distinctive legitimacy in the public eye can make serious attacks on them costlier as compared to serious attacks on bureaucracy.
4. Procedural Controls are Less Effective with Courts
Agencies. In addition to monitoring, informal interventions, and sanctions, agencies can be influenced though legislative designs which use “administrative procedures as instruments of political control,” as McNollgast put it. Legislators use procedural rules, enforceable through judicial review, that give interests that were part of the enacting coalition a privileged capacity to influence agency decisions. Through the Administrative Procedures Act, organic statutes creating agencies, and particular substantive statutes, Congress can require that agencies, for example, give public notice prior to making important decisions, receive comments, hold public hearings, and allow participation in the decision-making process by interested parties. Additionally, Congress can engage in “deck-stacking” in favor of preferred interests with procedural rules such as those governing evidence, proof, and appeals.
It is a hallmark of American administrative governance that there is massive heterogeneity in the nature and extent of procedural control of bureaucracy across agencies, policy areas, and issues. This tremendous heterogeneity in administrative procedure provides an expansive menu of options for those seeking to use them as instruments of political control. Consider the important example of rulemaking. The power to promulgate quasi-legislative rules, filling in the meaning of a statute, is among the most important powers that can be delegated to administrators. They use it to make binding prospective legal rules.
Rulemaking is an area characterized by relatively robust procedural control under the Administrative Procedures Act (and judicial interpretations of it). However, Congress at times imposes important additional procedural controls in specific contexts. Through statutory requirements governing some kinds of important rulemaking, Congress has, at different times and in different statutes, conditioned agency rulemaking powers by requiring that rules cannot be made unless:
- a different, institutionally separate agency first identifies an issue as a legitimate subject of rulemaking;
- the agency first conducts a cost-benefit analysis if compliance will require significant expenditures by state or local governments;
- the agency first prepares an environmental impact statement;
- the industry to be regulated is first given an opportunity to propose a rule, which must be adopted by the agency if “feasible”;
- the agency first convenes a review panel to evaluate, in consultation with small business, the effects of a proposed rule on small business, and prepares a report on the subject to be filed in the public rulemaking record;
- the agency first presents a proposed rule to Congress or some subunit within it for possible veto.
Courts. The manipulation of procedure offers less opportunities for political control in federal courts than in agencies. This is so for two reasons. First, Congress’s procedural strategies will be limited to those that are compatible with the core structure of a civil action. Second, the strategy of procedure as a means of political control of an agent, to be effective and efficient, requires an apparatus to enforce the procedural rules that is external to Congress, and is independent of the agent being governed by the procedural rules. Such an apparatus is lacking with respect to procedural control of the judiciary.
a. Less Flexible Menu of Procedural Controls
If procedural variation is a hallmark of the American administrative state, then, by comparison, procedural uniformity is a dominant feature of federal court procedure since the Rules Enabling Act of 1938. “Trans-substantive” application of the Federal Rules of Civil Procedure—their consistent operation across all substantive causes of action—has been a core animating idea of federal procedure since the Enabling Act. Of course, Congress can and has departed from this norm by fashioning unique statute-level procedural rules targeted to specific types of causes of action, such as heightened pleading in securities cases and administrative exhaustion in prisoner cases, with the goal of influencing substantive policy. Moreover, outside the compass of the Federal Rules, Congress also relies on statute-level rules governing such issues as attorney fee shifting, damages, evidence, and proof to guide implementation in courts. This type of variation in rules governing litigation can be feasibly interwoven with the core structure of a civil action, and Congress has clearly deployed such rules as instruments of political control—trying to shape substantive outcomes—when it delegates to courts.
However, some administrative procedures that provide potent means of political control do not have judicial analogues because they cannot be joined with the basic structure of a civil action. The most important is procedural control of substantive rulemaking. When statutes are centrally implemented in courts, judges are effectively delegated power to make quasi-legislative rules, which they engage in under the rubric of “statutory interpretation.” When Congress delegates the equivalent of substantive rulemaking authority to courts, it has fewer opportunities to control its exercise than would be available with an administrative delegation.
In civil litigation, when parties present issues entitled to resolution, judges act on them. Within the existing core framework for a federal civil action, Congress cannot mandate that a federal judge, before rendering a decision that resolves competing legal claims, report to Congress on the results of a cost-benefit analysis; or allow regulated parties to propose opinions, which would become binding if “feasible”; or constitute a review panel to receive input from small business representatives on their views on a draft opinion before it is signed; or require that an opinion first be submitted for review by a congressional committee, where it might be vetoed, before it is issued. Congress has conditioned administrative rulemaking in these ways. Congress certainly can condition how federal courts make law to some extent as well, but it has more options for procedural control of the fundamental power of rulemaking by agencies than by courts.
b. Lack of External and Independent Enforcement Apparatus
McNollgast’s landmark account of procedural rules as instruments of political control hinges critically on the existence of an apparatus external to both Congress and the administrative state that will enforce the procedural rules when violated by bureaucrats. This control strategy creates a decentralized system for producing bureaucratic compliance with the enacting coalition’s preferences, administered by (1) interested parties who prosecute actions against bureaucrats, together with (2) federal judges as decision-makers in these adjudications. For the strategy to work, McNollgast explains, the “judicial remedy must be highly likely”; “courts . . . play a key role in assuring political control”; and “without them political actors could not rely on decentralized enforcement.” This is a system of judicial supervision of bureaucracy where bureaucratic compliance is motivated by the highly credible threat of judicially imposed remedies. From Congress’s point of view it is efficient because it shifts the cost of monitoring and enforcement onto interested parties and courts.
An analogous strategy of efficient procedural control is not available for Congress to maintain judicial fidelity to the enacting coalition’s preferences. With respect to oversight of the judiciary, Congress can certainly use procedure to mobilize and advantage elements of the enacting coalition who wish to challenge judicial decisions by appealing them up the judicial hierarchy. However, there is no decision-maker external to both Congress and the judiciary to which appeal of judicial behavior can be taken. Thus, Congress cannot delegate the role of supervising the policy decisions of federal judges, as it does the policy decisions of federal bureaucrats. Congress must be the supervisor.
B. Courts Have Lesser Policy-Making Capacity
Generalist judges have limited policy expertise in any given area and are largely dependent on interested parties to the litigation, within an adversarial context, to present relevant information. Agencies are specialists with deep knowledge of their field and substantial independent research capacity, including staffs of scientists, policy analysts, economists, and lawyers. They can and regularly do range widely outside agency staff and stakeholding parties in their collection of information. Greater agency expertise, analytic competence, and information gathering capacity—relative to courts—are probably the most frequently given reasons for preferring bureaucrats over courts in policy implementation.
Further, the radically decentralized nature of litigation and courts, as an institutional infrastructure, renders it a weaker machinery to articulate a coherent and integrated regulatory scheme as compared to bureaucracy. Administrative law scholars have observed that agencies are capable of creating a relatively more centralized, unified, and integrated regulatory scheme, orchestrated by an administrator at the top of a hierarchical agency with powers of national scope. In comparison, providing a larger role for litigation and courts in implementation is associated with policy that is more often piecemeal, inconsistent, and at times contradictory.
This point should not be overstated. Some have argued that courts possess distinctive forms of expertise, such as their analysis of law and procedure, and their capacity to fact-find in adjudications. Others have emphasized that bureaucracy can be quite fragmented and decentralized. Still, in relative terms, it seems very plausible that as between courts and agencies, legislative coalitions are more likely to regard agencies as offering greater expertise and regulatory policy-making capacity. Empirical evidence that this is so will be discussed in Part II.D.
This has implications for Congress’s incentives when it accords a substantial role to civil litigation in implementation. A core assumption in much scholarship on delegation is that delegation to agencies, while entailing a cost of reduced legislative control of policy, serves legislative interests in conserving the resources that would be necessary to write high-quality specific legislation. Delegation to agencies allows Congress to instead write less specific legislation and leverage bureaucratic expertise and capacity to fill out statutory detail. It follows from this logic that when the delegate has less expertise to leverage, legislators have less incentive to delegate through lower degrees of specificity, and more incentive to invest in doing the work necessary to write more specific legislation. In Epstein and O’Halloran’s conceptualization of delegation as a “make or buy” decision, courts have less policy-making capacity to sell, and thus when delegating to courts, Congress must make more policy itself.
Higher levels of substantive statutory specificity when delegating to courts can provide more legislative structure, definition, and consistency to the regulatory scheme. This can compensate for the limited capacity of generalist judges presiding over lawsuits in a decentralized judiciary to render high quality, consistent, and coherent regulatory policy. Some may question the assumptions that legislative coalitions actually care about statutory detail, policy quality, and controlling downstream implementation, or pay attention to differences between courts and agencies when drafting policy substance. Parts II.D and E show that these assumptions are well grounded in a substantial body of empirical scholarship.
C. Delegation to Courts and Substantive Rulemaking Delegations
As discussed in the introduction, the notion of a choice “between” courts and agencies is a stylized theoretical formulation. While those who employ it surely recognize the existence and importance of hybrid regimes, they appear to contemplate that the delegation of more power to courts entails the delegation of less power to agencies. David Engstrom has observed that, in contrast with the notion of a dichotomous choice “between” courts and agencies, the “regulatory reality” is often a “complex ecolog[y]” of public and private enforcement. Indeed, the dichotomous choice formulation is not only stylized, but it also obscures the possibility of a positive association between delegation to courts and delegation to agencies. The theoretical logic developed in this Part thus far points directly to such a positive association. Before explaining why, it is useful present some descriptive information from this project’s data.
The data from this project, discussed in Part III, examines a large body of regulatory legislation and, with respect to discrete regulatory commands, identifies what judicial and administrative implementation provisions Congress employed to carry the commands into effect. It allows one to observe, at a simple descriptive level, the frequency with which rights to sue are deployed independently of administrative authority, versus deployment in conjunction with that authority. The data shows that when Congress provides a de novo civil action (public or private) to enforce regulatory law, 79 percent of the time the same regulatory substance in the statute is also governed by administrative rulemaking, administrative adjudications, or administrative sanctioning authority. This suggests that when civil litigation is present to enforce regulatory law, the regime is very likely to be hybrid. On the other hand, 21 percent of the time that a civil action is provided, none of these forms of administrative power are deployed to govern the same regulatory substance. If one focuses only on private rights of action, 83 percent of the time they are coupled with at least one of these basic forms of administrative power to govern the same regulatory substance, and 17 percent of the time they are accompanied by none of them. While it is certainly true that Congress at times relies on de novo civil actions for all frontline implementation, the dominant approach is that when Congress relies on civil actions, they are combined with administrative power.
If legislative coalitions are concerned, when relying on litigation regimes, about courts’ relative lack of expertise, they have an option in addition to forging policy in Congress. Legislative coalitions can also leverage administrative expertise in conjunction with rights to sue. Congress can join de novo suits with administrative rulemaking delegations, achieving enforcement via civil actions of rules crafted by experts, and introducing more expertise into the elaboration of statutory meaning than would exist with civil actions in the absence of rulemaking.
The Truth in Lending Act of 1968 provides a clear illustration. The law’s reliance on a private right of action arose specifically from the desire among key legislators to avoid “expansion of the federal bureaucracy,” and to find an alternative to “spawning a large administrative enforcement effort.” Having first made this decision, legislative committees then endeavored to draft substantive regulatory policy concerning the precise methods for computing interest rates, and the specific form of required disclosures, but found articulating necessary substantive rules to be “technical and complex,” and a conceptually “daunting task.” If formulating specific rules concerning methods of interest rate computation and disclosure was technical, complex, and daunting for legislative committees—aided by experts—obviously it was not something that could be left to the discretion of courts. As a result, Congress mandated that the Federal Reserve Board make the key substantive rules, to be enforced by civil actions.
Further, if Congress has concerns about limits on its capacity for postenactment influence on judicial interpretation of statutes, relative to postenactment influence on administrative interpretation, it can shift interpretive authority within a litigation regime toward agencies by coupling it with a substantive rulemaking delegation. Because Congress has greater control over agency rulemaking than over judicial interpretation of statutes, such rulemaking delegations within a litigation regime will enlarge Congress’s scope of ongoing influence over elaboration of a statute’s substantive meaning while it is enforced through civil actions. This theory is consistent with Bressman and Gluck’s empirical finding that congressional staff see themselves as working in dialogue with agencies to develop statutory meaning far more than with courts. Such dialogue—operating as an avenue of ongoing legislative influence over statutory meaning—can be fostered by a substantive rulemaking delegation, and will be made more attractive by a litigation regime.
If this dynamic is at play, it is sensible to expect that it will be focused on mandatory and specific rulemaking delegations. Discretionary rulemaking delegations, by their express terms, may or may not be exercised by administrators. General rulemaking delegations, by definition, do not delineate with particularity the subject matter covered by them, leaving administrators broad discretion to determine the subject matter that they will target with rules. In contrast, with a mandatory and specific rulemaking delegation, a Congress seeking ongoing influence over regulatory substance in a litigation regime can proactively and preemptively maximize the agency’s interpretive role on desired issues, and include substantive parameters that limit and shape what rules can be made.
Therefore, the institutional logic of the theory laid out in this Part suggests that, ceteris paribus, when Congress deploys litigation regimes it will have incentives to delegate more, rather than less, rulemaking authority to expert bureaucrats, and that such rulemaking delegations will be more likely to be specific and mandatory. In contrast with the conventional characterization of the choice “between” courts and agencies—which implies that delegation of more power to courts is associated with delegation of less power to agencies—this Article’s theory predicts that the choice of litigation regime will give rise to a perceived need to leverage more administrative rulemaking.
D. Evidence of Legislative Coalitions’ Intent When Delegating
The argument advanced in this Part is underpinned by empirical assumptions that legislative coalitions (1) know that federal courts are harder to control than federal agencies, (2) regard them as having lesser policy-making capacity, (3) intend to delegate less policy-making discretion to them, and (4) actually care about the content of public policy enacted in statutes and how it will be implemented after passage. These considerations were relatively absent in the “institutional neutrality” and “courts as dumping grounds” theories sketched above. Those theories tilted strongly toward viewing legislators as seeking to avoid the transaction costs of achieving specificity, and seeking to avoid the political costs of responsibility for specific policies. They also tended to regard the credit for solving public problems as readily available to legislators who support vague legislation that passes difficult choices on to delegates. Legislators did not garner material utility or credit by determining and controlling policy in detail; they did not fear punishment for failing to do so; they had no concern about being held responsible for the bad effects of policies in the course of implementation. This view of Congress is difficult to square with existing evidence.
1. Legislative Coalitions Know that Courts are Harder to Control
Legislative coalitions are aware that courts will be harder to control in implementation than agencies. A particularly informative source on this point is Mark Miller’s work based on interviews of members of Congress and their committee staff about the likelihood of committee responses to disfavored agency or court decisions. Miller interviewed forty-two members of Congress and thirty-five key staffers spanning three committees. One of his goals was to understand the comparative probability that committee members would seek to change decisions by courts versus agencies with which they disagreed.
One prevalent theme in Miller’s interviews is that courts are perceived as far more difficult to control through means short of legislative amendment. As a member of Congress explained: “This committee does not often attempt to overturn even unpopular court decisions. . . . We defer to the courts usually. But committee reaction to bad agency decisions is much faster and agency misinterpretations of Congressional intent are much more easily fixed.” Another member, sitting on a different committee, echoed the view that congressional committees regularly seek to influence agency decisions—but not court decisions—to adjust policy to legislative preferences. “We must constantly negotiate with administrative agencies, especially in times of split party government. But you never negotiate with the courts.”
A staffer explained that agency decisions are “fixed” with threats and coercion, which do not work on courts: “Congress is better able to coerce agencies than the courts to correct policy problems. Congress can use its influence on the executive branch through letter writing, telephone calls, and budget threats. I am unaware of the use of any threats by Congress against the judicial branch or any level of courts.” Another staffer echoed the same point, stating that Congress’s power is greater to overturn agency decisions than court decisions, and that committees are “willing and able to change agency policy decisions” because “[a] simple phone call or an appropriations rider can change an agency decision.” Yet another staffer, from a different committee, stated: “To overturn agency decisions, all we need to do is ask . . . agency staff. But we can’t ask judges how to overturn court decisions.”
Averaging across the three committees, Miller reports that 90 percent of subjects responded that Congress “has the final say in conflicts with the agencies,” and 49 percent said that this was true in the case of conflicts with courts. Of course, self-reported characterizations of power and efficacy of this sort are of uncertain value as descriptions of reality. But they certainly suggest that members of Congress and their committee staff perceive far greater congressional influence over agencies than courts in the course of policy implementation. From the full set of interviews, Miller concludes: “Congressional committees seem quite willing in general to run roughshod over agencies,” while it is rare that they “attack the courts directly.”
As a result of Congress’s greater capacity to influence agency policy-making absent new legislation, a great deal of committee time is devoted to this goal. One Democratic member explained that there is “day to day political friction with the agencies,” and “[i]f you disagree with the Republican Administration, then you attack the agencies.” A staffer observed that “[m]uch staff time is spent convincing agencies to change their decisions on their own, without needing new legislation.” Miller concludes from his interviews that, while members of Congress regard legislative efforts to change court decisions as “unusual events that only occur under special circumstances,” efforts to change “agency decisions is seen as a normal part of the everyday business of Congress.”
Miller’s interviews with legislators and their staff produced an account that strikingly echoes key aspects of the theory set out earlier in this Part: Congress (through its committees) routinely influences policy-making by administrators through informal interventions, threats, and sanctions, without need for statutory amendment. Congressional committees perceive that it is far more difficult to influence courts. One important lesson from Miller’s interviews is that one need not believe that legislative coalitions think through the details of the institutional arguments set out earlier in this Part. Rather, they internalize basic lessons from repeated experience confronting decisions by courts and agencies with which they disagree, with far greater success in changing the latter. It would be remarkable if this experience did not influence how Congress legislates policy substance when delegating to courts.
2. Legislative Coalitions Intend to Delegate Less Discretion to Courts
Recent work by Bressman and Gluck, based on survey questions and interviews of 137 congressional staff with drafting responsibility, illuminates how statute drafters view delegations to courts versus agencies. Congressional staff drafters were asked whether ambiguous language in statutes signaled an intent to delegate authority to fill in gaps by implementers. Ninety-one percent responded that ambiguity signals an intent that agencies fill in gaps. Only 39 percent responded that ambiguity signals an intent for courts to fill in gaps. On the question of why implementers would want agencies to fill in gaps, over 90 percent responded that they intended agencies to draw upon their expertise to fill in statutory meaning. The survey instrument provided eight reasons for having agencies fill in statutory gaps, and expertise was the second most frequently identified reason. Agency resolution of “implementation details” ranked first.
One of Bressman and Gluck’s key findings is that congressional staff see themselves as working in dialogue with agencies to develop statutory meaning far more than with courts. One theme repeated by congressional staff drafters was that courts are not a desirable source of statutory interpretation. Twenty-three percent of respondents “volunteered that drafters affirmatively prefer that courts not interpret their statutes at all.” Thirty percent “specifically accused the courts of inconsistency or being result-oriented.” Though the survey does not indicate why respondents so regard courts, the theory set forth earlier in this Part, coupled with Miller’s work, suggest one possible explanation: courts’ greater independence from congressional control may allow them to be more “results oriented.” Agencies recognize that committees are willing and able to “run roughshod” over them, as Miller put it. This contributes to agency interpretations that hew closer to committee preferences. It may also be that when courts or agencies issue disfavored interpretations, committees’ weaker capacity to bring courts back in line with committee preferences produces more frequent committee frustration with courts, leading them to regard courts as more “results oriented” than agencies.
Bressman and Gluck’s survey did not ask congressional staff about their perception of the difficulty of postenactment control of courts versus agencies, or the comparative policy-making capacity of courts versus agencies. However, the survey certainly suggests that agencies’ relative expertise is a reason that Congress legislates less specifically, purposefully leaving more gaps in statutes relying more heavily on agency delegations. Congressional staff’s lower likelihood of believing that drafters intended for courts to fill in gaps, and their greater dissatisfaction with courts’ interpretations, would also create incentives for them to legislate more specifically, leaving fewer gaps when delegating to courts.
E. Legislative Coalitions Care about Policy and Seek to Control Downstream Implementation
Legislative coalitions care about controlling the details of policy substance and influencing implementation. Legislators in fact can derive utility from more specific resolution of policy for several reasons. Some members of Congress, and their policy-specialist staff on committees who help draft legislation, may sincerely hold policy preferences that lead them to want to resolve policy issues in legislation, particularly where competing policy options do not have varying political costs. Further, substantive policy specificity, rather than vagueness, can promote credit claiming and blame avoidance, and thus can be a vehicle to pursue political support and reelection.
Highly attentive interest groups, mobilized around the legislative process, tend to advocate for specific policy content and often possess detailed knowledge about the substantive policy positions supported by members of Congress. Indeed, such groups are frequently part of coalitions with legislators that collaborate in formulating and drafting legislative proposals. Disappointing such groups can carry significant political and electoral costs. Legislators are also mindful that, down the road, critics and electoral challengers can effectively publicize and politicize poor policy content of statutes that a legislator supported, and the actual deleterious policy effects that ensued, which motivates them to care about and attend to statutory detail.
This is an empirically grounded view of legislative coalitions. It is supported by an extensive body of empirical scholarship, both quantitative and qualitative, showing that when such coalitions write important laws they make design choices (1) intended to mitigate subversion by implementers, (2) informed by comparisons between courts and agencies, and (3) intended to control future statutory interpretation. First, studies have found that under divided government, where legislative majorities are concerned about executive resistance at the implementation stage, Congress writes more detailed laws, includes more formal structural constraints on administrative action (e.g., appropriations limits, reporting requirements), and includes more structural limitations on presidential influence upon agencies (e.g., qualifications on who the president can appoint, fixing the duration of their service). Recognizing the potential for subversion at the implementation stage, legislative coalitions seek to mitigate the threat at the drafting stage.
Second, when legislative coalitions design implementation, they think strategically about institutional differences between courts and agencies as they bear on ultimate policy outcomes. They are influenced by judgments such as whether courts or agencies are likely to yield more aggressive enforcement, whether courts or agencies will produce more favorable substantive legal interpretations, and whether courts or agencies are more susceptible to capture. Their design choices are influenced by their expectations on such issues.
Third, legislative coalitions act strategically at the drafting stage to control future statutory interpretation by both courts and agencies. Because interpreters often rely on legislative-historical evidence to elucidate the meaning of statutes, enacting coalitions strategically fashion legislative-historical records, such as committee reports, colloquies in hearings, and floor statements, with the goal of influencing interpreting agents’ later construction of the statute. As Rodriguez explains: “Courts pay attention to legislative history. And legislators looking out for their own interests pay attention to what courts pay attention to.” Because of this recognition, “legislators constantly manipulate the documentary record of the statute’s history with the aim of providing statutory interpreters with a basis for a certain construction or constructions.”
Taken together, this body of empirical scholarship is in strong tension with a vision of legislators—contemplated by the “institutional neutrality” and “courts as dumping grounds” perspectives—who see their interests served by passing laws with high-sounding titles while avoiding the costs associated with achieving specificity, and distancing themselves from specific policy choices. It makes abundantly clear that when legislative coalitions, which include highly attentive and knowledgeable interest groups, write important laws, they frequently do so with attention to differences between courts and agencies, and with concern about achieving control over future policy implementation through statutory content and design. The next Part turns to new evidence assembled to test the theory that Congress will invest more effort to develop substantive law, state it in greater detail, and leverage more administrative rulemaking, when deploying litigation regimes.
III. Empirical Analysis
The empirical goal of this project is to examine the relationship between the legislative choice to accord a substantial enforcement role to civil actions, and the legislative determination of, and control over, policy substance in a statute. Measuring variation in the latter, and doing so in a body of statutes large enough for systematic hypothesis testing, presents substantial measurement challenges. In addition, the complexity of the relationships involved, in combination with data limitations, preclude decisive identification of a causal mechanism with an empirical test. The empirical strategy adopted here is one of triangulation. I identify three distinct observable implications of the theory set out in Part II and empirically probe all of them in order to gain traction on the question investigated.
The first is that litigation regimes will be associated, on average, with more attention and effort focused on developing policy substance in the legislative process. If legislative coalitions are actually trying to resolve more policy substance in Congress, this will entail more work in the legislative process. This effort should be evident in the activity of congressional committees as they develop the bill, demonstrated by the amount of testimonial and documentary evidence that was elicited from witnesses in committee hearings. Measurement here will focus on committee hearings on bills prior to passage. The second observable implication of the theory is that litigation regimes will be associated, on average, with more detailed elaboration of policy substance in the statute. As contrasted with measurement keyed to the legislative process, measurement here will focus on characteristics of its end product: the statute actually passed. In particular, the measure will be based on the aggregate volume of language in parts of the statute that make substantive regulatory law. The third observable implication is that litigation regimes will be associated, on average, with more frequent delegation of substantive rulemaking power to agencies, which simultaneously leverages more expertise to make rules enforceable in court, and enlarges the interpretive role of agencies, over which Congress has greater capacity for ongoing influence relative to courts.
Each measure is necessarily indirect. However, if analysis of these three distinct observable implications of the theory—each measured using very different materials—yield consistent results, the results will be mutually reinforcing. Mutually reinforcing results will bolster confidence that the measures tap the underlying concept of interest: legislative determination of, and efforts to exercise control over, policy substance in a statute.
A. The Body of Laws
I start with the 366 federal statutes, passed between 1947 and 2008, that were classified by David Mayhew as “significant legislation.” Coders read all of the Mayhew laws in order to identify those containing domestic regulatory commands. “Regulatory,” as used here, refers to “any governmental effort to control behavior by other entities,” including business firms, government at any level, or individuals. A regulatory “command” is a mandatory proscription of actions that the legislation seeks to prevent (e.g., an employer must not discriminate based on race), or a mandatory requirement that the regulated population engage in required conduct (e.g., a lender must disclose required information to a borrower). Using this broad conceptualization of regulatory commands, I included all laws that contained any civil regulatory commands directed at behavior within the United States. This Article studies legislative choices concerning the implementation of these regulatory commands.
Of Mayhew’s 366 significant statutes, 217 contained domestic regulatory commands. These laws include the major statutory landmarks and pillars of the modern American regulatory state. The average page length of the 217 laws was 111 pages, and thus the body of federal statutory law in the data spanned approximately 24,087 pages of the Statutes at Large. Each law was read in full in order to code the variables described below, which demanded identification of detailed substantive information about the policy content of each statute, as well as detailed information about the provisions governing implementation of the statute’s regulatory commands.
B. Implementation Regimes
Working with this body of regulatory law, coders identified each distinct implementation regime within each law. It is critical to the research design that the unit of analysis is the implementation regime, and not the law itself. An implementation regime is a set of implementation provisions governing a set of regulatory commands (or sometimes only one).
One law can have multiple implementation regimes. Consider the Civil Rights Act of 1964. Of the law’s eleven titles, one contained regulatory commands against transgressing voting rights (Title I), one contained regulatory commands against race discrimination by any government program receiving federal funds (Title VI), and one contained regulatory commands against job discrimination (Title VII). The substantive prohibitions of each of these titles were separate and distinct from one another, and so were the implementation provisions associated with each.
Title I’s prohibitions commanding nondiscrimination in voting were to be implemented through lawsuits by the Attorney General only¾the title contained no private right of action. Title VI’s prohibitions commanding non-discrimination in programs receiving federal funds were to be enforced by agencies that were endowed with rulemaking power, held administrative adjudications on alleged violations, and had sanctioning authority, including cut off of federal funds from discriminatory programs. Title VII’s prohibitions commanding nondiscrimination in employment were to be implemented through a combination of private lawsuits and Attorney General lawsuits in federal district court. With respect to each of these three parts of the law, Congress made a distinct choice and expressly provided a different implementation regime. One was centered on government lawsuits, one on rulemaking and administrative adjudications, and one on a combination of private and government lawsuits. Each of these distinct implementation regimes covers only the regulatory commands under that regime.
The key to the unit of analysis is to identify, through careful content analysis of each law, all substantive regulatory commands in a law and group them under the implementation provisions that govern them. Long and complex regulatory statutes can have many implementation regimes, whereas short and simple statutes can have only one. The 217 Mayhew laws identified with regulatory commands yielded 916 implementation regimes, with a mean number of 4.2 regimes per law. Of the 916 implementation regimes, 55 occur in one-regime laws. Because the model specifications used in the empirical analysis below cannot use information from one-regime laws (for reasons to be discussed), the data examined in the models is limited to the 862 regimes that occur in the 163 multiregime laws. All of the statistical models presented below focus on these 862 regimes.
C. Measuring Reliance on Civil Litigation and Other Implementation Provisions
Scholars have recognized that private versus government prosecution of lawsuits deliver different enforcement characteristics to an implementation regime. They have argued that private enforcers, as compared to government prosecution, can muster a larger volume of suits, leverage more private and insider information, mobilize more efficiently and rapidly, and are less susceptible to capture. However, private actions will tend to focus heavily on profitable cases and may neglect social welfare enhancing ones that are difficult, costly, risky to litigate, or for other reasons have low expected values.
By comparison, public prosecutors can bring important cases that serve public goals even when they are not economically profitable; strategically select cases that are effective vehicles to develop socially desirable doctrine; and make large investments necessary to prosecute expensive, complex, protracted cases, some of which may lead to subsequent private “coattail” actions. Because public and private enforcers draw on different sources of information, possess different kinds of resources, and otherwise have different case selection biases, the combination of public and private rights of action will mean that “fewer good cases are missed.” Scholars have observed, in practice, that litigation under implementation regimes relying on both public and private lawsuits often reflects a division of labor in which each type of enforcer specializes in the kinds of actions that match their comparative advantage. Congressional reliance on a combination of public and private lawsuits can be characterized as a more complete approach to implementation through civil actions than either public or private lawsuits alone.
One distinctive feature of private rights of action identified above—that they can muster a larger volume of suits than public prosecutions—deserves special emphasis. When private rights of action are coupled with public ones, private actions will often comprise a very large share of total litigation. This has significant implications for postenactment control of policy-making and the limited policy-making capacity of the judiciary, as they bear on Congress’s resolution of policy substance. As observed in Part I, scholars have noted that delegation of prosecutorial power to private attorneys and litigants raises the risk that it will be exercised in the service of private, particularistic, and often economic interests, which may conflict with public interests. When this feature of private suits is coupled with the prospect that their volume will far outstrip public ones, it amplifies the control problem concerning the prosecutorial role in implementation.
This is true, too, with respect to the judicial role in implementation. As noted in Part II.C, the empirical reality of statutorily provided civil actions is that a large majority of the time they are joined to administrative delegations. In these hybrid regimes, courts and agencies interact in elaborating statutory meaning. Private rights of action will typically generate significantly more opportunities for statutory interpretation by courts than public rights. This will enlarge courts’ relative role in hybrid regimes, and with it the potential problems of policy control and policy quality associated with delegation to courts. Thus, from the standpoint of the theory advanced in Part II, reliance on private suits will be associated with higher incentives for resolving more policy issues in Congress, and with specifying policy in greater detail.
Litigation regime is coded 1 when a regime contains both private and government rights of action, identifying regimes in which Congress created a complete litigation regime. Private suit only is coded 1 if there is a private right of action, but not a government right of action. Government suit only is coded 1 if there is a government right of action, but not a private right of action. Litigation regime is the equivalent of an interaction of private suit and government suit. The frequency of private rights of action, government rights of action, and a combination of the two (litigation regimes) in the data is displayed in Table 1.
Table 1: Frequency of Civil Actions
|Type||% of Regimes||% of Regulatory Commands Covered|
|Litigation Regime (Gov. & Priv.)||14%||19%|
It bears noting that when private suits are included in a regime, government suits are also used 73 percent of the time. However, when government suits are included in a regime, private suits are also used only 29 percent of the time. This means that litigation regime includes a large majority of the regimes that have private suits, and a comparatively small percentage of regimes with government suits. It is crucial to be clear on what is contained in litigation regime because this will prove to be very important in the statistical models.
Effectively estimating the relationship between reliance on civil litigation and the dependent variables studied here requires accounting for the administrative aspects of implementation regimes as well. With respect to each regime, dummy variables were coded to measure whether it relied on (1) general rulemaking, (2) specific rulemaking, (3) administrative adjudication, or (4) administrative sanctions or orders. The distinction between general and specific rulemaking delegations is important to some of the analysis below, and thus it is necessary to state it clearly. In general rulemaking delegations, Congress vests administrators with broad and weakly guided discretion to make law. Typical examples of such delegations are the Omnibus Crime Control and Safe Streets Act of 1968, in which Congress provided that the “Secretary may prescribe such rules and regulations as he deems reasonably necessary to carry out the provisions of this chapter,” and the Clean Water Act of 1972, where Congress stated that the “Administrator is authorized to prescribe such regulations as are necessary to carry out his functions under this Act.” Such general delegations of rulemaking power appear to be what delegation scholars have in mind when they associate administrative delegations with vague statutes in which Congress does little work and passes on the task of resolving policy substance to administrators.
In contrast, specific rulemaking delegations are more focused delegations that identify specific issues for the rulemaker’s attention and effort. They are far more often mandatory than are general rulemaking delegations. In the data presented below, 78 percent of specific rulemaking delegations are mandatory, as compared to only 32 percent of general delegations. In addition to the general rulemaking delegation in the Clean Water Act of 1972, Congress further mandated specific rules. It provided that the Administrator:
shall, by regulation . . . determine . . . those quantities of oil and any hazardous substance the discharge of which, at such times, locations, circumstance, and conditions, will be harmful to the public health or welfare of the United States, including, but not limited to, fish, shellfish, wildlife, and public and private property, shorelines, and beaches, except that in the case of the discharge of oil into or upon the waters of the contiguous zone, only those discharges which threaten the fishery resources of the contiguous zone or threaten to pollute or contribute to the pollution of the territory or the territorial sea of the United States may be determined to be harmful.
Unlike discretionary delegations to make rules the administrator “deems reasonably necessary to carry out the provisions” of a statute, in specific rulemaking delegations the legislative coalition, on average, has done more of the work required to identify specific issues that will need to be resolved in implementation, frequently including substantive parameters that limit and shape what rules can be made. In the specific rulemaking delegation just quoted, Congress not only mandated rulemaking on quantities of oil discharges to be banned by the Act, but specified certain injures (to shell fish and private property, for example) as cognizable harms to the public health or welfare, while specifying a much more limited range of cognizable harms in the contiguous zone. In specific, as compared to general rulemaking delegations, Congress is often more actively seeking to manage policy substance.
D. Hearings Data and Measuring Congressional Attention
The first hypothesis to be evaluated is that when Congress accords a substantial enforcement role to civil litigation, it will focus more attention, time, and effort on developing policy substance. Committee hearings are primary sites where Congress invests effort in developing policy substance. This does not mean that committee hearings undertake rigorous policy analysis as they unfold. Rather, they are often highly scripted, with extensive prior communications and collaboration between committee staff and witnesses. They are a venue in which committee members may seek to build a strong case for a desired proposal—a case directed to the public, interest groups, and the floor of the parent chamber. In addition to acquiring policy-relevant information, hearings are an opportunity for committee members to acquire political information, such as the likely electoral repercussion of pursuing a bill. However, scholars who acknowledge these features of hearings recognize that they exist alongside, and are entwined with, actual policy learning. Notwithstanding their political dimension, hearings reflect engagement with important sources of information, expertise, and input into substantive policy-making. They are a visible stage on which we can observe sources of information and assistance tapped by legislators and their staff in the course of developing a bill, and what issues are the primary targets of their attention.
For each law in the data, committee hearings in which it was considered were identified. From the pool of hearings for each law, for all laws in which more than twenty witnesses testified, twenty witnesses were randomly drawn. When twenty or fewer witnesses testified in total, all were coded. This yielded a total of 2,916 witness appearances in the 163 multiregime laws, covering 862 regimes. The average length of a witness’s testimony (oral, written, and Question and Answer (Q & A)) was 11 pages. The testimony of all witnesses in the data covered over 32,000 pages in committee hearing volumes. The frequency of different witness types is listed in Table 2.
Coders read each witness’s testimony in full, applying a coding protocol that requires content analysis to identify key information. The following variables were coded for each witness’s testimony. With respect to each regime in the law, (1) whether the witness’s direct testimony related to any regulatory command in that regime, (2) the total number of questions asked by legislators to the witness relating to any regulatory command in that regime, (3) the total number of different legislators who asked the witness questions relating to any command in that regime, and (4) the total number of pages of documents submitted by the witness relating to any command in that regime.
Each of these measures provides only a limited indication of legislative attention and effort. A variable that combines information from each is superior to any one measure standing alone. For each regime-witness combination, I computed the mean of the standardized measures for each of the four items listed above. The dependent variable, legislative attention, is the mean value of the index, for each regime, based on all witnesses coded for the law. Statistical tests discussed in Appendix A confirm the excellent validity and internal consistency of this scale. For ease of interpretation, the scale was shifted up so that zero is the lowest value.
Table 2: Frequency of Witness Types
|Type||% of Total Witnesses*|
|State or City Official||11%|
|Specific Business Representative||15%|
*The column sums to over 100 percent because some witnesses identified multiple titles or affiliations.
Empirical analysis in Appendix C shows that, at the regime level, legislative attention is strongly associated with the word-count measure of specificity (discussed in Part III.G). Thus, this measure of legislative effort to develop policy in committee moves in tandem with how specifically Congress actually articulates policy in the statute. This bolsters the face validity of legislative attention.
The mean composite legislative attention scores for regimes with no lawsuits, government suits, private suits, and litigation regimes are displayed in Figure 1. At the descriptive bivariate level, consistent with the theory presented in Part II, regimes lacking any lawsuits have the lowest scores, followed by government suits, private suits, and litigation regimes. The value for litigation regimes is nearly quadruple the value when there are no lawsuits.
Figure 1: Legislative Attention by Type of Civil Actions Provided
Law-Level Fixed Effects. A key challenge of modeling legislative attention (and substantive specificity) is accounting for the many variables that may influence it. For example, these variables can include the nature of the policy problem at issue, its degree of complexity, the level of ambitiousness with which Congress is seeking to regulate it, the extent of public attention focused on the law during the legislative process, policy conflict over it, and long-run time trends in lawmaking processes and characteristics of legislation. Many variables have the potential to be associated with the level of legislative attention manifest in the committee hearing process, as well as substantive specificity. Adequate measurement of all such variables presents a large challenge.
In order to address this challenge, law fixed effects are employed. This approach leverages only information within laws to estimate the effects of independent variables that vary across their regimes. Law fixed effects (by one estimation method) include dummy variables for each law in the regression. The coefficients for each law dummy soak up the effects of any omitted variables that have consistent effects across regimes of the law, meaning variables that would take the same value for each regime within the same law. Law fixed effects allows for assessment of whether reliance on litigation regimes is associated with higher levels of legislative attention and substantive specificity, and a higher probability of an administrative rulemaking delegation, as compared to other regimes in the same law that do not rely on litigation regimes.
Policy conflict. It is critical to incorporate a measure of policy conflict into the model for two reasons. First, a key part of the conceptualization of legislative attention (and substantive specificity) is that, in part, it reflects efforts to determine policy substance in the law in order to constrain implementers’ downstream range of options. Legislative coalitions will make more such efforts in the face of greater uncertainty about how implementers will exercise authority. Policy conflict in the legislative process is one indication of the likelihood of conflict at the implementation stage, increasing uncertainty around implementation, and correspondingly increasing incentives for legislative coalitions to resolve more issues in the legislature. The more legislative coalitions fear subversion of their policies by opponents at the implementation stage, the more they will seek to pin down the details of the policy they enact. This logic suggests that policy conflict in the legislative process will be associated with higher levels of both legislative attention and substantive specificity.
It also suggests the further hypothesis that the effects of litigation regime on legislative attention will be conditional on policy conflict. The logic of this hypothesis is as follows. Congress focuses more legislative attention on parts of laws governed by litigation regimes in part because it will be more difficult to control litigants and courts at the implementation stage. Policy conflict at the drafting stage serves as a signal of the degree of opposition to expect at the implementation stage. For this reason, the hypothesized effect of litigation regimes on legislative attention may grow in the presence of greater policy conflict.
Law fixed effects control for policy conflict at the law level. However, some regimes in the same law may be the locus of more policy conflict than others. It is therefore necessary to model policy conflict at the regime level. Every witness who testified on a regime was coded, at the regime level, for whether she opposed some regulatory command within it. For each regime, opposition was then constructed as the proportion of total witnesses to testify on the regime who opposed some regulatory commands within it. Opposition is also interacted with litigation regime in order to assess, in one model specification, the conditional hypothesis just discussed: that concern about controlling future policy substance will be especially great when courts will be central implementers, and resistance at the implementation stage is anticipated due to higher levels of opposition to the policy commands in hearings.
Political-institutional environment and law fixed effects. Law fixed effects control for political and institutional variables whose effects would be consistent across regimes in the same law. Both theoretical and empirical work suggests that the political and institutional environment in which a law is passed may be important in explaining statutory substance, including but not limited to specificity. Therefore, it may also be associated with the legislative process that produced that statutory substance. The key point here is simply that law fixed effects control for law-level political and institutional variables, such as divided government, divided control of Congress, party polarization, electoral uncertainty, and interest group environment. Each of these variables would take the same value for each regime in a law, and accordingly their effects are absorbed by the law dummy.
Policy area and law fixed effects. Legislative attention in hearings to a regime’s substantive regulatory policy (and substantive specificity) will likely be associated with policy area. Law fixed effects also function as a control for policy area, effectively providing highly specific policy codes—one for each law. This feature of the model is critical. For example, in the Civil Rights Act of 1964, the level of specificity of Title VII (with a litigation regime) is judged in comparison to the specificity of other regimes in the same act without a litigation regime. It is not judged in comparison to environmental, securities, or consumer laws. Indeed, it is not even judged in comparison to other civil rights laws. For omnibus laws combining regulatory commands spanning multiple policy areas, a law fixed effect is an inadequate policy control, and thus I explicitly include policy area dummy variables for omnibus laws.
Time trends and law fixed effects. Legislative attention in hearings (and substantive specificity) may vary, in part, as a function of long-run time trends. Many have observed, for instance, that federal statutes have grown much longer during the period studied. Other important time trends may relate to legal doctrine, such as the growth and then decline of federal courts’ propensity to liberally construe statutes to imply private rights of action, or post-Chevron changes in judicial deference to agency interpretations. Such time trends may be correlated with legislative attention and/or substantive specificity. By using law fixed effects, and thereby leveraging only variation across regimes in the same law (all of which were, necessarily, enacted in the same year), such time trends are controlled for and their confounding threat is neutralized. For example, regardless of the state of doctrine on implied private rights of action or deference to agency decisions, it would be the same across all titles of the Civil Rights Act of 1964, and thus it does not explain variation in specificity across them.
Scope. It is necessary to control for the magnitude or scope of the regulatory interventions undertaken in a law, because more extensive regulatory interventions may require more legislative attention in hearings and more words to articulate policy. Law fixed effects control for this at the law level. However, some regimes in the same law will have broader scope than others. To measure the extensiveness of a statute’s intervention, scholars have developed a protocol for counting a statute’s “major provisions.” I adapt this approach to the task of counting regulatory commands in each regime. The core idea is to count each separate regulatory command, resulting in a variable measuring the sum of regulatory commands, or the discrete requirements and prohibitions imposed on regulated entities. The protocol for coding regulatory commands is discussed further in Appendix B.
Salience. Legislative attention in hearings (and substantive specificity) may be partly a function of salience. Higher levels of public and interest group attention to a law may lead congressional committees to focus on it more in hearings. Law fixed effects controls for this at the law-level. However, some regimes in the same law may be more salient than others. Howell et al. and Baumgartner and Jones measure the salience of statutes based on the extent of Congressional Quarterly (CQ) coverage. Salience is a regime-level count of the number of lines in CQ statute summaries discussing and characterizing each regime’s regulatory commands, and it serves as a regime-level salience measure.
F. Hearings Models
Models A, B, and C in Table 3, with the regime-level mean value of legislative attention as the dependent variable, are estimated with ordinary least squares regression, with fixed effects for the law, and with robust standard errors clustered on year. When interpreting the coefficients on the mutually exclusive dummy variables litigation regime, private suit only, and government suit only, the reference category is regimes with no lawsuits. Model A is a sparse model with only the implementation characteristics as independent variables, and Model B adds the control variables.
Litigation regime is statistically significant with a large effect in both models. In assessing magnitude, I focus on the model with controls (Model B), where the effect is materially smaller. The presence of a litigation regime is associated with growth in the scale of .15. To put this magnitude in perspective, the mean value of the scale when there are no government or private lawsuits is .12, such that litigation regime is associated with growth of 125 percent relative to that level. When Congress relies on litigation regimes, it focuses a lot more attention on that part of the law in the committee hearings.
Private suit only and government suit only are insignificant. Note that in the data, when Congress relied on private rights of action, a substantial majority of the time (73 percent) it concurrently relied on government rights of action, such that a large majority of private suits are captured in litigation regime. The presence of one element of litigation is not associated with more attention and effort by congressional committees on policy substance. The effect emerges only when Congress makes a clear choice to make civil litigation central to the design by relying on a complete litigation regime. Congress does so a large majority of the time (73 percent) that it relies on private lawsuits.
To examine this further, I drop litigation regime—an interaction between private suit and government suit—to observe only the direct effects of private suit and government suit. The results are presented in Appendix D. What we observe is that private suit is primary in driving the effect of litigation regime. In the absence of litigation regime, private suit is statistically significant and positive, with a substantively large effect. In contrast, government suit is insignificant, and its coefficient is only about one-third the size of private suit. I regard the model with the interaction (litigation regime) as the theoretically stronger model, and it shows that government suit is important to legislative attention conditional upon private suit. However, the alternative specification without the interaction reveals that private suit has a strong independent effect and is the primary driver.
This result is entirely sensible. As discussed in Parts I.B and III.C, as compared to public suits, scholarship in the field recognizes that private suits pose a distinctive control problem on the prosecutorial dimension. Indeed, they have been the primary target of the democracy and public policy critiques. Further, their association with a materially higher volume of suits can greatly elevate the role of courts in implementation, increasing problems of control and policy quality that incentivize resolving more issues in Congress.
With respect to the insignificance of government suit only, Table 1 shows that government suits govern 58 percent of the commands in the data. Relative to private suits, they are widespread and routine. By themselves (without private enforcement), they are not associated with elevated legislative attention. With respect to the insignificance of private suit only, it is important to remember that this variable captures a very small number of regimes (48 out of 916, or 5 percent of regimes) because such a large share of private rights of action are coupled with government rights of action. Empirical examination of these private suit only regimes (reported Appendix E) reveals that, as compared to litigation regimes, Congress was substantially less likely to provide successful plaintiffs with attorney fee awards, or with multiple damages, suggesting lack of intent to actually mobilize private enforcers. And these regimes, on average, were less salient (measured by salience) and were subject to less political conflict (measured by opposition). They also delegated less administrative implementation authority, reflecting less provision for implementation and enforcement on the administrative side as well as the litigation side. All of the results described in this paragraph are statistically significant, and they are presented in more detail in Appendix E. Private suit only, therefore, is a rare event in the data that occurred in less salient and less conflictual regimes in which Congress was less likely to actually incentivize the use of the private right of action, and which provided for weaker enforcement overall. It appears that a private right of action, alone and without incentives to use it, sometimes reflects congressional indifference. In contrast, when Congress intends to make civil actions and courts central to the implementation design, it joins a private right with a government right of action, enacting a litigation regime. It does this a large majority of the time (73 percent) that it relies on private enforcement.
Table 3: OLS Models of Legislative Attention with Law Fixed Effects
|Model A||Model B||Model C||Model D|
|Sparse||Controls||Controls & Interaction||Expert Witnesses|
|Private Suit Only||-.05||.004||.003||-.06|
|Government Suit Only||.003||.01||.02||.02|
Omnibus policy dummies
|Robust standard errors, clustered on year, in parentheses|
|***p<.01; **p<.05; *p<.1|
Returning to Table 3, Model C interacts litigation regime with opposition, and the interaction is significant. With the interaction in the model, litigation regime now captures only the effect when opposition is at its mean. At that level of opposition, litigation regime is associated with growth in legislative attention of .11, about doubling its mean value when there are no lawsuits (.12). When opposition is increased one standard deviation (.19) above its mean, litigation regime is associated with a growth of .29 in legislative attention, which is about two-and-a-half times legislative attention’s mean value with no lawsuits. Congress focuses significantly more attention on policy substance when relying on litigation regimes, and this effect grows larger in the presence of higher levels of policy conflict. The alternative specification presented in Appendix D, which drops litigation regime and incorporates only the direct effects of government suit and private suit, shows that the relationship between private suit and opposition drives this interaction effect. Private suit’s positive association with specificity grows in the face of increasing levels of policy conflict.
A challenge to my interpretation of the model is possible here. A rival theory might be that the conditional relationship just described does not actually reflect more work on policy substance when relying on litigation regimes in the face of higher levels of policy conflict. Instead, legislative attention is just picking up interest group conflict, for example, with business interests squaring off against unions and non-governmental organizations over regulatory policy. Such interest group conflict over policy substance draws in more witnesses, generates more legislator questioning, and elicits more documentary submissions. Further, recall that the “courts as dumping grounds” perspective predicts that higher levels of policy conflict in Congress will be associated with more frequent reliance on courts for implementation in order to achieve a greater “responsibility shift.” Thus, interest group conflict is driving delegation to courts, and driving up legislative attention, while there is no real direct relationship between the two.
This possibility can be probed by leveraging information on the identity of the witnesses. The title and institutional affiliation of every witness was coded. Model D reruns the empirical mode, now including in the analysis only witnesses of a type traditionally associated with policy expertise: federal bureaucrats, state bureaucrats, and persons associated with research institutions (overwhelmingly universities), where the witnesses’ title or position indicated expertise in the subject matter of the law. While such witnesses are not necessarily detached from interest group conflict, they are among the witnesses most likely to be called for their policy expertise. In this expert-only model, the effect of litigation regime is markedly larger. When opposition is at its mean, litigation regime is associated with growth in legislative attention of .17, or a 142 percent increase over its mean value where there are no lawsuits (.12). When opposition is one standard deviation (.19) above its mean, the effect of litigation regime is .37, more than tripling legislative attention relative to when there are no lawsuits.
In a contrasting model (not displayed), I included only witnesses of a type traditionally associated with interest group conflict over regulatory policy: representatives of businesses or business associations, NGOs, and unions. In this model, the main effect of litigation regime becomes insignificant. Its interaction with opposition is significant. When opposition is one standard deviation (.19) above its mean, the effect of litigation regime is .23 (as compared to .37 in the expert-only model). Growth in legislative attention when Congress relies on litigation regimes in the context of more policy conflict is much larger with respect to utilization of bureaucratic and research institution expertise than it is with respect to traditional interest group combatants. This supports the inference that it involves actual policy work in the committee, rather than simply reflecting interest group conflict.
Finally, it is natural to wonder whether some distinctive characteristics of policy problems that require fact adjudications for resolution leads to reliance on litigation regimes, and is also associated with greater effort in developing policy substance (legislative attention), even though there is no direct relationship between the two. We can rule this possibility out. Government suit only and private suit only are not associated with more legislative attention in hearings. Administrative adjudication is also clearly insignificant in every model in Table 3. Litigation regime’s positive and significant association with legislative attention is not present in other circumstances in which Congress provides for fact adjudications.
In sum, Congress focuses significantly more attention on policy substance when relying on litigation regimes. Private rights of action are critical in driving this effect. The effect grows in the presence of higher levels of policy conflict in the hearings. The expert-only model suggests that the growth in legislative attention in the face of policy conflict is associated with actual policy work, as distinguished from merely registering interest group conflict.
G. Coding Statutory Specificity
In order to test the specificity hypothesis, the substantive statutory content associated with each implementation regime was coded for degree of specificity, measured as a word count. Here, I follow the most common empirical strategy of measuring statutory specificity, deployed in several articles by law professors Robert Cooter and Tom Ginsburg, by Tom Ginsburg in independent work, and by political scientists John Huber and Charles Shipan in their noted book Deliberate Discretion?: The Institutional Foundations of Bureaucratic Autonomy. These scholars use word and page counts to measure specificity, where specificity is conceptualized in part as a strategy to control implementing actors, be they courts or agencies. While they use total word and page counts in laws, this Article uses a much more targeted measurement strategy. The specificity variable is constructed as a word count with respect to the portions of each statute, governed by each implementation regime, that lays out the substantive regulatory policy under that regime specifying what conduct is prohibited or mandated.
In regulatory laws such as environmental, civil rights, or antitrust laws, substantive specificity concerns the portions of statutes that actually identify what behaviors, under what conditions, and by whom, shall constitute illegal pollution, discrimination, or anticompetitive actions. Portions of a statute included in the specificity measure were those stating (1) prohibitions and elaborations on them; (2) requirements and elaborations on them; (3) exemptions affirmatively exempting some group or class from the regulatory commands; (4) rules of inclusion affirmatively stating that the regulatory commands apply to some group or class; (5) rules of construction providing express guidance on interpretation, such as explaining the regulatory commands’ relationship to other laws; and (6) definitions of words contained in the forgoing parts of the statute. Further detail on coding specificity is contained in Appendix B.
Word counts may seem abstract, and it is easy to imagine circumstances in which more words will not actually be more constraining on implementers, and may cause confusion that increases interpretive discretion. In this regard, it is important to stress that scholars using word counts as a measure of constraint do not posit an iron law whereby more words are always more constraining. Rather, the claim is that over a large body of statutes, on average and ceteris paribus, increasingly detailed policy instructions will be associated with increasing word counts.
The data analyzed here offer several opportunities to gauge the face validity of the word-count measure in ways that past studies were not able to. One approach is to assess whether and to what extent the word counts are correlated with statutory attributes that are associated, intuitively, with actual resolution by Congress of more issues. As discussed below in Part III.E, the regulatory commands variable is a regime-level count of discrete prohibitions and requirements directed at the regulated entity. If word counts plausibly measure statutory detail, the two variables will be highly correlated. Further, the data include a set of dummy variables indicating whether each regime contained: (1) definitions, (2) exemptions, (3) rules of inclusion, and (4) rules of construction, all as defined above.
Each of these four characteristics add layers of detail to regulatory commands. For example, the Clean Water Act of 1972 banned a variety of forms of water pollution, including oil pollution and the dumping of sewage from vessels. The statute contained twenty-two definitions, including defining “barrel” as “42 United States gallons at 60 degrees Fahrenheit,” and defining “sewage” as “human body wastes and the wastes from toilets or other receptacles intended to receive or retain body wastes.” It laid down a variety of regulations governing manufacturers of “marine sanitation device[s],” including recordkeeping and reporting requirements, but it expressly excepted from coverage “construction of a vessel by an individual for his own use.” While excepting such vessels, it expressly stated that its regulation of vessels did include and “apply to vessels owned and operated by the Unites States unless the Secretary of Defense finds that compliance would not be in the interest of national security.” Finally, it offered several rules of construction, one of which provides that
[n]othing in this section shall affect or modify in any way the obligations of any owner or operator of any vessel, or of any owner or operator of any onshore facility or offshore facility to any person or agency under any provision of law for damages to any publicly owned or privately owned property resulting from a discharge of any oil or from the removal of any such oil.
If the statute had not gone beyond its basic regulatory commands to define “barrel” and “sewage” (and twenty other words), expressly except vessels constructed for personal use, expressly include vessels owned by the US (with a possible national security caveat), and made clear that the statute shall not be construed to override certain damages actions, those issues would have been left to an implementer. By resolving all of them, Congress increased the statute’s substantive specificity relative to if it had resolved none of them. In doing so, Congress added words to the statute.
The presence of (1) definitions, (2) exemptions, (3) rules of inclusion, and (4) rules of construction, are strongly associated with word counts. I created an additive index reflecting the number of these items in each regime, ranging from 0 to 4. The association between each value of the index and word counts is presented in Table 4.
Table 4: Specificity Characteristic Index and Word Counts
|Index Value||Words||% of Regimes Covered|
In Appendix C, models of specificity are presented showing that with law fixed effects included (making comparisons only within laws), and regulatory commands and the index of all four characteristics as independent variables, both are statistically significantly and positively associated with regime-level word counts, with substantively large associations. Definitions, exemptions, rules of inclusion, and rules of construction are each separately significant in the same regression if substituted for the index. The word-count measure moves in tandem with attributes of statutes that intuitively suggest resolution of more policy issues, bolstering the face validity of the measure.
A second and very different approach to evaluating the face validity of the word-count measure examines the relationship between word counts and legislative effort, as measured by legislative attention. If word counts are associated with genuine resolution of more policy issues in a statute, one would expect this to be correlated, on average, with higher levels of legislative effort and attention directed to those parts of the statute in committee hearings. This expectation flows directly from the widely held theoretical view that Congress delegates discretion, in part, due to limits on the time and effort that it can devote to making specific policy, as discussed in Part II. The flip side of this is that more specific policy-making by Congress will require that it invest more time and effort. In Appendix C, a regression is presented showing that¾ controlling for the scope of the regulatory intervention with regulatory commands, and including law fixed effects¾legislative attention is statistically significantly associated with word counts, and the magnitude of the association is very large. Across regimes within the same laws, and controlling for the scope of the regulatory intervention, the word-count measure is strongly associated with attention and effort in committee hearings.
The mean specificity for regimes with no lawsuits, government suits, private suits, and litigation regimes are displayed in Figure 2. At the descriptive bivariate level, consistent with the theory presented in Part II, regimes lacking any lawsuits have the lowest word counts, followed by government suits, private suits, and litigation regimes. The value for litigation regimes is nearly triple the value when there are no lawsuits. The bivariate patterns look very similar to the legislative attention scale.
Figure 2: Substantive Specificity by Type of Civil Actions Provided
H. Specificity Model
The specificity model incorporates the same independent variables as the models of legislative attention displayed in Table 3. The logic given for incorporating those variables into a model of legislative attention also applies to a model of legislative substance. The reason is that this Article deploys legislative attention as a measure of Congress’s effort to develop policy substance, and it deploys specificity as a measure of Congress’s ultimate articulation of policy substance. However, one qualification is necessary. Existing theory points in contradictory directions regarding the relationship between policy conflict (opposition) and specificity, and the way in which the relationship between litigation regime and specificity may be affected by policy conflict. It has already been argued (in Part III.E), from the standpoint of this Article’s theory, that policy conflict will create incentives to write more specific legislation in order to insulate the enacting coalition’s policy decisions against opponents in the future, and that the effect of litigation regime on specificity may grow with increasing levels of policy conflict.
However, the dominant prediction in the literature runs in the opposite direction. As discussed earlier (in Part I.C), the single most frequently stated hypothesis about statutory specificity is that policy conflict is associated with statutory vagueness, with vagueness used as a strategy to avoid adverse political repercussions or to achieve passage where specificity would lead to legislative failure. This dominant view flows into the “courts as dumping grounds” hypothesis (discussed in Part I.D) that conditions of policy conflict will be associated with vague delegations to courts. This is exactly the opposite of this Article’s prediction. It is possible to adjudicate between these conflicting theories with the interaction of litigation regime and opposition, which will isolate the effect on specificity when Congress is relying substantially on civil actions in an environment of heightened policy conflict.
Table 5: OLS Models of Legislative Specificity with Law Fixed Effects
|Model A||Model B||Model C|
|Sparse||Controls||Controls & Interaction|
|Private Suit Only||-446||-112||-115|
|Government Suit Only||56||69||87|
|Omnibus Policy Dummies||—||ü||ü|
|Robust standard errors, clustered on year, in parentheses|
|***p<.01; **p<.05; *p<.1|
Models A, B, and C in Table 5 replicate Models A, B, and C in Table 3, but now with specificity as the dependent variable. The performance of the lawsuit variables in the specificity models closely tracks what we observed in the legislative attention models. The presence of litigation regime is statistically significant with a large effect in both the sparse model (A) and the first model with controls (B). In the model with controls, where the effect of litigation regime is materially smaller, it has a coefficient of 744. Relative to the mean level of words when there are no suits (788), this represents about double the specificity. Just as Congress invests more attention in developing policy substance when relying on litigation regimes, it specifies policy substance in about twice as much detail when it relies upon litigation regimes. Both private suit only and government suit only are statistically insignificant. As with the legislative attention models, the effect of litigation regimes on specificity emerges only when Congress makes a clear choice to make civil litigation central to the design by relying on a complete litigation regime, something it does a large majority of the time (73 percent) that it relies on private lawsuits.
I again drop litigation regime (an interaction of private suit and government suit), to observe only the direct effects. This allows one to examine the separate and independent effects of private suit and government suit. The results are presented in Appendix D. We again observe that private suit is primary in driving the effect of litigation regime. In the absence of litigation regime, while both private suit and government suit are statistically significant, private suit achieves a materially higher level of significance (.004 as compared to .058) while comprising materially fewer observations, and has about double the magnitude.
This result, again, is not surprising. As compared to public suits, private suits pose a distinctive control problem on the prosecutorial dimension, and their massively larger volume can greatly elevate the role of courts in implementation. Correspondingly, there is growth in problems of policy control and policy quality that incentivize greater substantive specificity. It is important to stress that the models with litigation regime (Table 5) show that the dominant source of the specificity effect comes from those regimes in which Congress made a clear self-conscious choice to make civil actions central to implementation by relying on both public and private suits.
Regarding the insignificance of government suit only, I remind the reader that government suits cover 58 percent of the commands in the data (Table 1). Just as these ordinary and routine provisions, in the absence of private suits, are not associated with greater legislative attention, they are not associated with greater legislative specificity. Regarding the insignificance of private suit only, I remind the reader, as discussed in the context of the committee hearings models and Appendix E, that it is a relatively rare event in the data (5 percent of regimes), occurring in lower-salience, lower-conflict regimes in which Congress had little interest, was less likely to incentivize the use of the private right of action with fees or damages multiples, and provided for less enforcement overall. When Congress intends to make civil actions and courts central to an implementation design, it joins a private right with a government right of action, enacting a litigation regime.
Returning to Table 5, Model C interacts litigation regime with opposition, and the interaction is significant. With the interaction in the model, litigation regime now captures only the effect when opposition is at its mean. At that level of opposition, litigation regime remains significant and is associated with an increase of 643 in specificity, or an 82 percent increase relative to specificity’s mean value in the data when there are no lawsuits (788). When opposition is increased one standard deviation (.19) above its mean, litigation regime is associated with a growth of 1,156 in specificity, for an increase of 147 percent relative to specificity’s mean value in the data where there are no lawsuits (788).
This shows that Congress legislates policy substance in greater detail when relying on litigation regimes, and this effect grows in the presence of higher levels of policy conflict. This result is consistent with the hypothesis that policy conflict may exacerbate legislative concerns about policy control when delegating to courts, heightening efforts to pin down substantive policy in greater detail. The result is the opposite of what is predicted by the “courts as dumping grounds” perspective, the logic of which predicts lower degrees of specificity in the face of delegations to courts under policy controversy. Thus, the data strongly rejects this prediction of the “courts as dumping grounds” perspective.
The alternative specification presented in Appendix D, which drops litigation regime and incorporates only the direct effects of government suit and private suit, shows that this interaction effect is driven by the relationship between private suit and opposition. That is, when litigation regime is dropped from the model, the interaction between private civil actions for implementation, and conflict over policy substance, is associated with Congress legislating with a particularly high level of specificity. The interaction between government suits and opposition is not significant. This parallels what we observed in models with legislative attention as the dependent variable.
It is reasonable to ask whether some unique feature of policy problems that require fact adjudications for resolution leads to reliance on civil litigation, and to higher values of specificity, even though there is no direct relationship between the two. However, private suit only and government suit only are insignificant, and administrative adjudication—another way that Congress provides for fact adjudications—is also insignificant in every model in Table 4. Thus, there exists no general association between legislative provisions for fact adjudications and policy specificity.
A further challenge might be pressed about whether the empirical relationship between specificity and litigation regime actually supports the theory set out in Part II. A rival theory might be that Congress first legislates with higher degrees of specificity for reasons having nothing to do with implementation in courts, and once having done so, has less need to draw on administrative expertise, and therefore is more likely to make the already specifically legislated rights directly enforceable in court. This argument flips the direction of the causal arrow, and might be called the “detail first, then civil litigation” theory. This account would be observationally equivalent with the empirical results, but with the outcome produced by a different causal mechanism than the one hypothesized.
One characteristic of the data provides leverage on this question. All litigation regimes were coded for whether they were enacted contemporaneously with the regulatory commands being implemented, or whether those regulatory commands were added by amendment to an already existing litigation regime. Forty-five percent of the litigation regimes in the data are of the latter type. This scenario provides a clear separation in time, with key implementation decisions made first, and statutory substance written subsequently, with foreknowledge of the litigation regime. If the “detail first, then civil litigation” mechanism is driving the results, we would expect that the effect would be absent, or materially less, when Congress is legislating policy substance into preexisting litigation regimes.
I conducted empirical tests entering the two types of litigation regimes into the model separately: (1) those enacted contemporaneously with the policy substance governed, and (2) those in which amendments added the policy substance into a preexisting litigation regime. The association with specificity is statistically significant and of roughly similar magnitude for both types. These empirical results are presented in Appendix F. They show that Congress is similarly specific when it knows that it is legislating policy substance into preexisting litigation regimes, and when it enacts litigation regimes concurrently with the policy substance. The “detail first, then civil litigation” mechanism is not driving the relationship between litigation regime and specificity.
In sum, Congress lays down regulatory policy substance in greater detail when relying on litigation regimes. Private rights of action are critical in driving this effect. The effect grows in the presence of higher levels of policy conflict in the hearings. The effect is durable when analysis is restricted to circumstances in which legislators have foreknowledge that they are writing policy substance into a litigation regime, showing that it is not driven by a process whereby Congress first legislates in detail, and then subsequently elects to rely on a litigation regime.
I. Litigation Regimes and Substantive Rulemaking Delegations
This section examines the relationship between litigation regimes and substantive rulemaking delegations. Before doing so, it is informative to highlight several results on the rulemaking variables in both the legislative attention and specificity models. General rulemaking has a negative coefficient in every legislative attention model, and is significant in five of six models. Relative to the mean value of legislative attention in regimes without general rulemaking (.21), in models in which general rulemaking is significant, it is associated with a reduction of 19 to 24 percent in legislative attention. General rulemaking also has a negative coefficient in every specificity model, and is significant in five of six models. Relative to the mean value of specificity in regimes without general rulemaking (1,056), in the models in which general rulemaking is significant, it is associated with a reduction of 23 to 26 percent in specificity. These results are consistent with the conventional notion that general rulemaking delegations shift policy work from Congress to the bureaucracy, along with responsibility to specify substantive law.
In contrast, specific rulemaking has a positive coefficient in every legislative attention model, and is significant in four of six models.  Relative to the mean value of legislative attention in regimes without specific rulemaking (.17), in the models with controls in which specific rulemaking is significant, it is associated with growth of 24 to 29 percent in legislative attention. Specific rulemaking also has a positive coefficient in every specificity model, and is significant in all six models. Relative to the mean value of specificity in regimes without specific rulemaking (516), in the models with controls it is associated with an increase of 43 to 53 percent in specificity. In specific rulemaking delegations, relative to general ones, Congress is more actively seeking to manage policy substance by identifying specific issues for rulemaking, usually mandatorily, and often with guiding parameters that constrain rulemakers. The models indicate that as compared to regimes without specific rulemaking delegations, regimes with them involve more work in hearings on substantive policy, and higher levels of statutory specificity.
Part II.C argued that, in light of concerns about policy control and policy quality, when Congress delegates to courts it has incentives to delegate more, rather than less, rulemaking authority to bureaucrats. This prediction contrasts with the conventional characterization of the dichotomous choice “between” courts and agencies. The data presented here allows investigating that possibility. I do so by estimating a regression model with substantive rulemaking delegation as the dependent variable, which equals 1 when the regulatory commands in the regime are covered by a substantive rulemaking delegation, and equals 0 otherwise. Sixty-three percent of the regimes in the data contain a substantive rulemaking delegation. The model contains the following independent variables: litigation regime, private suit only, government suit only, administrative adjudication, administrative sanctions, regulatory commands, salience, opposition, and omnibus policy dummies. This is the same model specification presented in Table 3 (Model B), and Table 5 (Model B), but with substantive rulemaking as the dependent variable. A logistic regression model was estimated with law fixed effects. The results are reported in Appendix G. The key results are as follows.
Litigation regime is statistically significant with a very large magnitude. The presence of a litigation regime is associated with a 42-percentage point increase in the probability of a substantive rulemaking delegation. Interestingly, administrative adjudication is also significantly associated with a substantive rulemaking delegation, but with a marginal effect less than half of litigation regime. When Congress deploys administrative adjudications in an implementation regime, the probability that it will delegate substantive rulemaking power grows by only 19 percentage points. Thus, when Congress relies substantially on adjudication in the civil justice system, as compared to adjudication by bureaucracy, the associated growth in the probability that it will incorporate administrative rulemaking is more than double. Congress sees greater value in administrative rulemaking when relying on litigation regimes as compared to administrative adjudications, which are conducted by experts that Congress can more easily influence.
Next, I parse the effects arising from discretionary versus mandatory rulemaking delegations. In regimes containing a discretionary delegation of rulemaking (which occurs in 38 percent of regimes), the administrator is allowed but not required to make rules. In regimes containing a mandatory delegation of rulemaking (which occurs in 46 percent of regimes), the administrator is required by statute to make rules. The coverage of the two types is overlapping because some regimes contain both.
I examined separate regressions isolating as dependent variables discretionary rulemaking alone (without mandatory rulemaking), and mandatory rulemaking alone (without discretionary rulemaking). Results show that, as between mandatory and discretionary, the association between litigation regimes and substantive rulemaking is driven by mandatory rulemaking delegations. Litigation regime is statistically significant with a large substantive effect in the mandatory model (35 percentage point increase), and clearly insignificant in the discretionary model. The opposite is true with respect to administrative adjudication, which is statistically insignificant in the mandatory rulemaking model, and statistically significant and positive in the discretionary rulemaking model (9 percentage point increase). When Congress relies on litigation regimes, it is much more likely to mandate that administrators make rules. When it relies on administrative adjudications, it is modestly more likely to give administrators discretion to make rules, but not to mandate them. These empirical results are discussed in more detail in Appendix G.
Next, I parse the effects arising from specific versus general rulemaking delegations. Specific rulemaking delegations occur in 54 percent of the regimes, and general rulemaking delegations occur in 22 percent. The coverage of the two types is overlapping because some regimes contain both. I examined separate regressions of specific rulemaking alone (without general rulemaking), and general rulemaking alone (without specific rulemaking). The results show that, as between specific and general, the association between litigation regimes and substantive rulemaking is driven by specific rulemaking delegations, which are frequently associated with substantive parameters that limit and shape what rules can be made. Litigation regime is statistically significant with a positive large effect (37 percentage point increase) when the dependent variable is limited to regimes containing only specific rulemaking delegations. When the dependent variable is limited to regimes containing only general rulemaking delegations, litigation regime is statistically significant with a large negative effect (33 percentage point reduction). These empirical results are discussed in more detail in Appendix G. Litigation regimes are associated with a heightened probability of specific rulemaking delegations, and a reduced probability of general rulemaking delegations. When Congress deploys litigation regimes, administrators are more likely to be assigned mandatory and specific rulemaking responsibilities, and they are less likely to be afforded the broad, discretionary, and unguided power of general rulemaking.
The following picture emerges very clearly from the data. Litigation regime has a statistically significant and substantively large association with the delegation of substantive rulemaking authority. Rulemaking delegations that are mandatory and specific drive this effect, suggesting a higher level of legislative effort to proactively stimulate administrative rulemaking on specific subject matter to be enforced in court with a litigation regime. Litigation regimes are associated with efforts to leverage more administrative expertise. Moreover, an enlarged role for rulemaking expands congressional capacity to influence substantive elaboration of the statute postenactment. These results are consistent with and reinforce the theory and evidence linking litigation regimes to greater congressional effort to develop policy substance, and to specify associated substantive policy in greater detail.
In the 1970s, alongside the growing pervasiveness of private lawsuits implementing federal statutes, critiques emerged that legislative reliance on civil litigation and courts for implementation is undemocratic and produces bad public policy. These two critiques have been repeated for four decades and have become a staple of American public policy debates. But these critiques beg a very important question: Does Congress write statutory substance differently when it delegates to courts? The democracy and public policy critiques assume that the answer is no.
An alternative view, advanced in this Article, is that legislative coalitions in fact recognize the problems of policy control by the elected branches, and relatively weaker judicial policy-making capacity, that underpin the democratic and public policy critiques. They anticipate these issues and mitigate them by resolving more policy substance in Congress, articulating a more detailed regulatory scheme, and leveraging more administrative expertise with mandatory and specific rulemaking delegations. Why would they incur the costs of doing this rather than delegating broad discretionary authority to bureaucracy? Delegating to courts and doing more work in Congress is perfectly consistent with standard accounts of legislators seeking to maximize political support, reelection prospects, or policy benefits. Scholars have identified a variety of reasons that make delegation to litigants and courts, under some circumstances, a desirable choice. Because this implementation strategy can yield utility for legislative coalitions, they will be willing to incur attendant costs, provided that the overall calculus makes such delegations attractive.
The empirical results in this Article show that when Congress deploys litigation regimes for statutory enforcement, it is in fact associated with (1) greater attention and effort in developing policy substance in the legislative process, and (2) more detailed elaboration of policy substance in the statute. In a series of different statistical models, these alternative approaches to measuring legislative resolution of policy substance yield strikingly parallel results. The parallel character of the results makes them mutually reinforcing and bolsters confidence that, while the measures are necessarily indirect, they tap the underlying concept of interest¾the resolution of policy substance by Congress. Moreover, the notion that legislative concerns about expertise and policy control are at play when Congress relies on litigation regimes is further strengthened by the association between litigation regimes and substantive rulemaking delegations that are mandatory and specific. Such delegations can simultaneously mobilize more expertise, include substantive parameters that limit and shape what rules can be made, and increase legislative control over elaboration of statutory meaning through formal and informal agency oversight. Finally, this evidence is fortified by qualitative work showing that legislators and their staff are acutely aware that courts are harder to control than agencies during implementation, regard agencies as more expert, and generally intend to delegate less interpretive discretion to courts than agencies.
The empirical results reject the assumptions of the democracy and public policy critiques of civil litigation for policy implementation, the “institutional neutrality” view, and the “courts as dumping grounds” view. However, it is important not to overclaim. The argument made in Part II has many parts, and the data do not allow me to test any specific one. For example, the models do not distinguish between issues of policy control and issues of policy quality. The observational nature of the data, and the complexity of the causal processes implicated, do not permit a decisive statistical test of a causal mechanism. There may be other, entirely different, theoretical accounts that also explain the patterns in the data. However, I have tested rival theories suggested by existing literature and by many scholars who have provided feedback along the way, and none are supported by the data.
The empirical results are consequential for normative debates about the democratic and public policy consequences of litigation regimes. Greater congressional delegations to litigants, lawyers, and courts, in addition to enlarging their role in policy implementation, are attended by the assertion of substantially greater policy-making power by Congress itself. Conventional characterizations of litigation regimes as dislocating power from the administrative state to courts miss this critical fact. Relative to the state of the world imagined by this conventional perspective, the actual state of the world is one in which federal regulation implemented through litigation regimes is more informed by institutional policy-making capacity in the legislature, more often attended by expert rulemaking, and more tied to democratic governance, than previously thought.
This, of course, does not eliminate important debates about the wisdom of litigation regimes for democracy and public policy. Those debates will and should continue. But they cannot be advanced by the simple comparison of courts and agencies, as has been the approach for forty years. Rather, they must be informed by realistic engagement with actual tradeoffs, which are more complex, difficult, and interesting than previously recognized. Congress’s elevated role as a policy-maker, and its more active and focused use of administrative capacity, when delegating to courts must be integrated into the analysis and debate.
Finally, this Article’s findings are relevant to understanding how divided government and party polarization have shaped the development of the American state over the past half-century. Beginning in the late 1960s, divided party control of the legislative and executive branches became the norm and relations between Congress and the president became more antagonistic. Growing ideological polarization between the parties exacerbated the institutional antagonism arising from divided government. These political-institutional conditions have influenced how Congress structures regulatory power through legislation.
When Congress delegates implementation power to bureaucracy, the conjuncture of divided government and growing party polarization has been associated with more extensive efforts to cabin bureaucratic discretion with statutorily imposed rules, procedures, and constraints. Some scholars have lamented these developments on the ground that they inhibit agencies’ ability to execute their missions effectively and efficiently, weakening the capacity and strength of the federal bureaucracy. On this view, divided government and party polarization, operating together, may be regarded as an engine producing pathology and disfunction in the American administrative state. Other work has shown that divided government and party polarization are also powerfully associated with congressional reliance on private rights of action to enforce statutes. From Congress’s point of view, opportunities and incentives for private enforcement, as an alternative or supplement to bureaucracy, offer valuable enforcement insurance when the executive is distant. From the late 1960s to the mid-1990s, in response to a long succession of statutory invitations by Congress, the population-adjusted rate of private statutory suits in federal court increased by about 1,000 percent.
This Article adds an important, and perhaps ironic, piece to the story of the American regulatory state’s development over the past half century. In the face of an increasingly distant executive, Congress increasingly encumbered administrative power while dramatically escalating mobilization of private litigation as an alternative or supplement. This Article shows that Congress’s turn to litigation had consequences for policy substance. More policy substance was shaped by Congress in parts of statutes governed by litigation regimes. Through this pathway, the conjuncture of divided government and party polarization led the American Congress to exercise more of the power delegated to it under Article I to fashion the rules that govern us in the republic of statutes.
Sean Farhang: Professor of Law, Public Policy, and Political Science Associate Professor of Political Science, U.C. Berkeley.
. Guido Calabresi, A Common Law for the Age of Statutes (1982). For a summary characterization of “the Age of Statutes,” see Robert A. Katzmann, Judging Statutes 7–8 (2014); for an extended characterization, see Abner J. Mikva & Eric Lane, Legislative Process ch.1 (1st ed. 1997).
. See Admin. Off. of the U. S. Cts., Table C-2––U.S. District Courts––Civil Statistical Tables for the Federal Judiciary (2017), http://www.uscourts.gov/sites/default/files/data_tables/jb_c2_0930.2017.pdf [https://perma.cc/6B3X-PCNY]. The growing centrality of statutory law, and the increasing role of litigation and courts to implement it, are evident in other countries as well. R. Daniel Kelemen, Eurolegalism: The Transformation of Law and Regulation in the European Union (2011); Ran Hirschl, The Judicialization of Politics, in The Oxford Handbook of Law and Politics (Gregory A. Caldeira, R. Daniel Kelemen & Keith E.Whittington eds., 2008).
. This critique, of course, has been focused on both judicial review of agency actions, as well as legislative reliance on civil actions for enforcement, and particularly private rights of action. This Article’s focus is civil actions and not judicial review of agency actions, but it draws on the broader normative literature on the distribution of policy-making power between courts and agencies.
. See Eugene Bardach & Robert A. Kagan, Going by the Book: The Problem of Regulatory Unreasonableness (Transaction Publishers 2002); Thomas F. Burke, Lawyers, Lawsuits, and Legal Rights: The Battle Over Litigation in American Society (2002); Farhang, supra note 5; Sean Farhang, Legislative-Executive Conflict and Private Statutory Litigation in the U.S.: Evidence from Labor, Civil Rights, and Environmental Law, 37 L. & Soc. Inquiry 657 (2012); Morris P. Fiorina, Legislative Choice of Regulatory Forms: Legal Process or Administrative Process?, 39 Pub. Choice 33 (1982); Margaret H. Lemos, The Consequences of Congress’s Choice of Delegate: Judicial and Agency Interpretations of Title VII, 63 Vand. L. Rev. 361 (2010); Matthew C. Stephenson, Public Regulation of Private Enforcement: The Case for Expanding the Role of Administrative Agencies, 91 Va. L. Rev. 93 (2005); Matthew C. Stephenson, Legislative Allocation of Delegated Power: Uncertainty, Risk, and the Choice Between Agencies and Courts, 119 Harv. L. Rev. 1035 (2006).
. See Bardach & Kagan, supra note 14; Burke, supra note 14; Farhang, supra note 5; Farhang, supra note 14; Fiorina, supra note 14; Lemos, supra note 14; Stephenson, Legislative Allocation of Delegated Power, supra note 14.
. See, e.g., Lisa S. Bressman, Edward L. Rubin & Kevin M. Stack, The Regulatory State 139–56 (2010); David Epstein & Sharyn O’Halloran, Delegating Powers: A Transaction Cost Politics Approach to Policy Making under Separate Powers (1999); John D. Huber & Charles R. Shipan, Deliberate Discretion?: The Institutional Foundations of Bureaucratic Autonomy (2002); Theodore J. Lowi, The End of Liberalism: Ideology, Policy, and the Crisis of Public Authority (1969); Theodore J. Lowi, The End of Liberalism: The Second Republic of the United States (1979); Jerry L. Mashaw, Greed, Chaos, and Governance: Using Public Choice to Improve Public Law ch. 6 (1997); Rachel VanSickle-Ward, The Devil is in the Details: Understanding the Causes of Policy Specificity and Ambiguity (2014); Mathew D. McCubbins, Roger G. Noll & Barry R. Weingast, Administrative Procedures as Instruments of Political Control, 3 J. L. Econ. & Org. 243 (1987); Stephenson, Legislative Allocation of Delegated Power, supra note 14.
. Richard A. Posner, Legal Formalism, Legal Realism, and the Interpretation of Statutes and the Constitution, 37 Case W. Res. L. Rev. 179 (1987); Melnick, supra note 6; William N. Eskridge Jr., Dynamic Statutory Interpretation (1994); Peter L. Strauss, On Resegregating the Worlds of Statute and Common Law, 9 Sup. Ct. Rev. 429 (1994); Robert A. Katzmann, Making Sense of Congressional Intent: Statutory Interpretation and Welfare Policy, 104 Yale L.J. 2345 (1995); Mashaw, supra note 19, at ch. 6.
. Huber & Shipan, Deliberate Discretion, supra note 19, at 76. See Robert D. Cooter & Tom Ginsburg, Comparative Judicial Discretion: An Empirical Test of Economic Models, 16 Int’l Rev. L. & Econ. 295 (1996); Robert D. Cooter & Tom Ginsburg, Leximetrics: Why the Same Laws are Longer in Some Countries than Others (2003), https://works.bepress.com/robert_cooter/134/download/ [https://perma.cc/JY52-H4ET].
. See McCubbins, Noll & Weingast, Administrative Procedures, supra note 19; Mathew D. McCubbins, Roger G. Noll & Barry R. Weingast, Structure and Process, Politics and Policy: Administrative Arrangements and the Political Control of Agencies, 75 Va. L. Rev. 431 (1989); Terry M. Moe, The Politics of Bureaucratic Structure, in Can the Government Govern? (John E. Chubb & Paul E. Peterson eds., 1989); Robert A. Kagan, Adversarial Legalism: Tamed or Still Wild?, 2 N.Y.U. J. Legis. & Pub. Pol’y 217 (1999); Epstein & O’Halloran, supra note 19; James Q. Wilson, Bureaucracy: What Government Agencies Do and Why They Do It ch. 16 (1989).
. See James B. Thayer, The Origin and Scope of the American Doctrine of Constitutional Law, 7 Harv. L. Rev. 129 (1893); Alexander M. Bickel, The Least Dangerous Branch: The Supreme Court at the Bar of Politics (1962); John Heart Ely, Democracy and Distrust: A Theory of Judicial Review (1980).
. See Martha A. Derthick, Up in Smoke: From Legislation to Litigation in Tobacco Politics 5–8, (3rd ed. 2012); Patrick M. Garry, A Nation of Adversaries: How the Litigation Explosion is Reshaping America (1997); Daniel P. Kessler, Introduction, in Regulation vs. Litigation: Perspectives from Economics and Law 1 (Daniel P. Kessler ed., 2011); Andrew P. Morriss, Bruce Yandle & Andrew Dorchak, Regulation By Litigation 14–15 (2009); Walter K. Olson, The Litigation Explosion: What Happened When America Unleashed the Lawsuit (1991); Walter K. Olson, The Excuse Factory: How Employment Law Is Paralyzing the American Workplace (1997); Jeremy Rabkin, Judicial Compulsions: How Public Law Distorts Public Policy (1989); Ross Sandler & David Schoenbrod, Democracy by Decree: What Happens When Courts Run Government (2003); W. Kip. Viscusi, Overview, in Regulation Through Litigation (W. Kip Viscusi ed., 2002); Nathan Glazer, Towards an Imperial Judiciary?, 41 Pub. Int. 104 (1975).
. See Derthick, supra note 28, at 5–8; Kessler, supra note 28, at 1; Morriss, Yandle & Dorchak, supra note 28, at 23–24; Frank B. Cross, Shattering the Fragile Case for Judicial Review of Rulemaking, 85 Va. L. Rev. 1243, 1290–1306 (1999); McCubbins, Noll & Weingast, Structure and Process, Politics and Policy, supra note 25, at 444–45; Thomas W. Merrill, Judicial Deference to Executive Precedent, 101 Yale L.J. 969, 978–79 (1992); Richard J. Pierce Jr., The Role of the Judiciary in Implementing an Agency Theory of Government, 64 N.Y.U. L. Rev. 1239, 1251 (1989); Laurence H. Silberman, Chevron––The Intersection of Law & Policy, 58 Geo. Wash. L. Rev. 821, 821–24 (1990).
. See Morriss, Yandle & Dorchak, supra note 28, at 15; Derthick, supra note 28, at 5–7; Stephenson, Public Regulation of Private Enforcement, supra note 14, at 116–17; Pamela H. Bucy, Private Justice, 76 S. Cal. L. Rev. 1, 66–67 (2002); Frank B. Cross, Rethinking Environmental Citizen Suits, 8 Temp. Envtl. L. & Tech. J. 55, 68–69 (1989); Joseph A. Grundfest, Disimplying Private Rights of Action Under the Federal Securities Laws: The Commission’s Authority, 107 Harv. L. Rev. 961, 969–71 (1994).
. David B. Spence & Frank Cross, A Public Choice Case for the Administrative State, 89 Geo. L.J. 97, 140 (2000). See Donald L. Horowitz, The Courts and Social Policy (1977); Rabkin, supra note 28; Richard J. Pierce Jr, Agency Authority to Define the Scope of Private Rights of Action, 48 Admin. L. Rev. 1 (1996); Richard B. Stewart & Cass R. Sunstein, Public Programs and Private Rights, 95 Harv. L. Rev. 1193 (1982).
. See Kathleen Bawn, Political Control Versus Expertise: Congressional Choices About Administrative Procedures, 89 Am. Pol. Sci. Rev. 62 (1995); Epstein & O’Halloran, supra note 19; Huber & Shipan, supra note 19, at 1–2; Craig Volden, Delegating Power to Bureaucracies: Evidence from the States, 18 J.L. Econ. & Org. 187 (2002).
. See Ely, supra note 26, at 131–34; Mikva & Lane, supra note 2; Fiorina, supra note 14, at 46–52; VanSickle-Ward, supra note 19, at 14–19, 24–32; Lisa S. Bressman, Chevron’s Mistake, 58 Duke L.J. 549, 568 (2009); Neal Devins & Michael Herz, The Battle That Never Was: Congress, the White House, and Agency Litigation Authority, 61 Law & Contemp. Probs. 205, 215–16 (1998); Joseph A. Grundfest & Adam C. Pritchard, Statutes with Multiple Personality Disorders: The Value of Ambiguity in Statutory Design and Administration, 54 Stan. L. Rev. 627, 628 (2002); Jonathon R. Macey & Geoffrey P. Miller, The Canons of Statutory Construction and Judicial Preferences, 45 Vand. L. Rev. 647, 666 (1992); Daniel B. Rodriguez, Statutory Interpretation and Political Advantage, 12 Int’l Rev. L. & Econ. 217, 218 (1992).
. See George I. Lovell, Legislative Deferrals: Statutory Ambiguity, Judicial Power, and American Democracy (2003); Mashaw, supra note 19, at 155–56; Rodriguez, supra note 38, at 218; Victoria F. Nourse & Jane S. Schacter, The Politics of Legislative Drafting: A Congressional Case Study, 77 N.Y.U. L. Rev. 575, 595 (2002).
. See Huber & Shipan, supra note 19; Epstein & O’Halloran, supra note 19; Fiorina, supra note 14, at 45–46; Bressman, supra note 38, at 568; John D. Huber & Charles R. Shipan, The Costs of Control: Legislators, Agencies, and Transaction Costs, 25 Legis. Stud. Q. 25 (2000); Lemos, supra note 14, at 368; Spence & Cross, supra note 32, at 135–36; Stephenson, Legislative Allocation of Delegated Power, supra note 14, at 1037.
. Epstein & O’Halloran, supra note 19, at 47, 49; see also Ely, supra note 26, at 133; Kenneth Culp Davis, Judicial, Legislative, and Administrative Lawmaking: A Proposed Research Service for the Supreme Court, 71 Minn. L. Rev. 1 (1986); McCubbins, Noll & Weingast, Administrative Procedures, supra note 19, at 261.
. Mark A. Graber, The Nonmajoritarian Difficulty: Legislative Deference to the Judiciary, 7 Stud. Am. Pol. Dev. 35 (1993); Eli M. Salzberger, A Positive Analysis of the Doctrine of Separation of Powers, or: Why Do We Have an Independent Judiciary, 13 Int’l Rev. L. & Econ. 349 (1993); Paul Frymer, Black and Blue: African Americans, the Labor Movement, and the Decline of the Democratic Party 7–17 (2008).
. Graber, supra note 46, at 35–73. Graber states that independent commissions are also institutional sites that will be attractive to Congress when it wishes to evade accountability for controversial policy decisions. See id., at 40–41 (discussing creation of the Interstate Commerce Commission as a choice by Congress to distance itself from controversial public policy). I interpret Graber’s singling out of the most difficult-to-control type of administrative organ—independent commissions—and likening it to federal courts, as implying that ordinary bureaucracy is more susceptible to congressional control, and that therefore delegating to it will yield less of a blame shift than delegating to federal courts or independent commissions.
. Tom Ginsburg, Judicial Review in New Democracies: Constitutional Courts in Asian Cases 32 (2003); Salzberger, supra note 46, at 349, 362–63; Ran Hirschl, The Political Origins of Judicial Empowerment through the Constitutionalization of Rights: Lessons from Four Constitutional Revolutions, 25 L. & Soc. Inquiry 91, 104 (2000); John M. Scheb II and William Lyons, The Myth of Legality and Public Evaluation of the Supreme Court, 81 Soc. Sci. Q. 928, 929 (2000).
. For example, in seeking to explain the delegation choice between courts and agencies, political scientist Morris Fiorina motivates an article by asking: What explains the legislative choice to delegate power to make policy choices to agencies, as opposed to legislators making the policy choices themselves and providing for their direct enforcement in courts? Fiorina, supra note 14, at 45, 53. This question assumes that legislators choose between making policy decisions themselves, to be enforced in court, versus delegating them to bureaucrats. Fiorina provides no explanation for why this would be the case. Terry Moe formulates the same question: “Why does Congress delegate authority to an agency, rather than passing detailed laws enforceable in courts?” Terry Moe, The Positive Theory of Public Bureaucracy, in Perspectives on Public Choice: A Handbook 466 (Dennis Mueller ed., 1997). But like Fiorina, he does not explain why this is the choice.
. See Burke, supra note 14, at 14–15; Farhang, supra note 5; Kagan, supra note 3; Melnick, supra note 6; Farhang, supra note 14; R. Shep Melnick, From Tax and Spend to Mandate and Sue: Liberalism after the Great Society, in The Great Society and the High Tide of Liberalism 387 (Sidney Milkis & Jerome M. Mileur eds., 2005).
. See generally David Vogel, The “New” Social Regulation in Historical and Comparative Perspective, in Regulation in Perspective: Historical Essays 155 (Thomas K. McCraw ed., 1981); Sean Farhang, Regulation, Litigation, and Reform, in The Politics of Major Policy Reform in Postwar America 69–70 (Jeffrey Jenkins & Sidney Milkis eds. 2014).
. See Cross, supra note 29, at 1303–05; Charles R. Shipan, Designing Judicial Review: Interest Groups, Congress and Communications Policy 10 (1997); Kenneth A. Shepsle, Bureaucratic Drift, Coalitional Drift, and Time Consistency: A Comment on Macey, 8 J.L. Econ. & Org. 111, 114–15 (1992); Charles R. Shipan, Interest Groups, Judicial Review, and the Origins of Broadcast Regulation, 49 Admin. L. Rev. 549, 555–56 (1997).
. McCubbins, Noll & Weingast, Administrative Procedures, supra note 19, at 250. See also Joel D. Aberbach, Keeping a Watchful Eye: The Politics of Congressional Oversight 130–44 (1990); Robert H. Davidson & Walter J. Oleszek, Congress and Its Members 357 (6th ed. 2006).
. Mark C. Miller, The View of the Courts from the Hill: How Congressional Committees Differ in their Reaction to Court Decisions 234 (1990) (unpublished Ph.D. dissertation, The Ohio State University, Department of Political Science). This lower visibility of court decisions is likely fostered, in part, by less frequent use of “fire alarms” by constituents because they recognize legislators’ limited capacity, or inability, to informally intervene with courts as they do with agencies, as discussed shortly.
. William N. Eskridge Jr., Philip P. Frickey, Elizabeth Garrett & James J. Brudney, Cases and Materials on Legislation and Regulation: Statutes and the Creation of Public Policy 996–97 (5th ed. 2014); McCubbins, Noll & Weingast, Administrative Procedures, supra note 19, at 262–63; Aberbach, supra note 63, at 130–32.
. Miller, supra note 72, at 234 (Staffer stating: “To overturn agency decisions, all we need to do is ask . . . agency staff. But we can’t ask judges how to overturn court decisions.”), 238 (Member of Congress stating: “We must constantly negotiate with administrative agencies, especially in times of split party government. But you never negotiate with the courts.”).
. Michael W. Kirst, Government without Passing Laws: Congress’ Nonstatutory Techniques for Appropriations Control (1969); McCubbins, Noll & Weingast, Administrative Procedures, supra note 19, at 248; Melnick, supra note 6, at 241–43; Davidson & Oleszek, supra note 63, at 357–58.
. It is important to recall that the focus here is on sanctions that do not involve passing a law. The “court curbing” literature in political science has focused on the introduction of bills that would limit the courts power if passed. Though such bills have very rarely passed, scholars have argued that their introduction may send threatening messages that affect judicial behavior through intimidation, or because they may be read by judges as signals of declining judicial legitimacy in the eyes of the public. Gerald N. Rosenberg, Judicial Independence and the Reality of Political Power, 54 Rev. Pol. 369 (1992); Tom S. Clark, The Limits of Judicial Independence (2011). I do not discount the possibility that the threat of sanctioning legislation may be consequential in these ways. The argument of this section is that actual sanctions are less effective as applied to courts compared to agencies. Much of the argument supports the inference that the threat of sanctions against federal courts will also be less effective than the threat of sanctions against federal agencies.
. Jeffery A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited 94 (2002); Georg Vanberg, Legislative-Judicial Relations: A Game-Theoretic Approach to Constitutional Review, 45 Am. J. Pol. Sci. 346, 347 (2001); Ginsburg, supra note 49, at 32; Ura & Wohlfarth, supra note 94.
. See William N. Eskridge Jr., Abbe R. Gluck & Victoria F. Nourse, Statutes, Regulation, and Interpretation: Legislation and Administration in the Republic of Statutes 20 (2014); Henry J. Friendly, Some Kind of Hearing, 123 U. Pa. L. Rev. 1267 (1975); Ernest Gellhorn & Glen O. Robinson, Perspective on Administrative Law, 75 Colum. L. Rev. 771, 789 (1975); Richard E. Levy & Robert L. Glicksman, Agency-Specific Precedents, 89 Tex. L. Rev. 499, 501 (2011).
. While the legislative veto was ruled unconstitutional in 1983, the tool remains consequential and actually appeared in significant legislation more frequently after it was declared unconstitutional. Epstein & O’Halloran, supra note 19, at 100–01. Davidson and Oleszek suggest that agencies have continued to comply with legislative veto provisions because to do otherwise would risk alienating legislators that wield important continuing influence over the agency’s well-being, such as through appropriations. Davidson & Oleszek, supra note 63, at 355–56.
. Eskridge, Gluck & Nourse, supra note 100, at 20; Robert G. Bone, The Process of Making Process: Court Rulemaking, Democratic Legitimacy, and Procedural Efficacy, 87 Geo. L.J. 887 (1999); Stephen Burbank, The Rules Enabling Act of 1934, 130 U. Pa. L. Rev. 1015 (1982); David Marcus, The Past, Present, and Future of Trans-Substantivity in Federal Civil Procedure, 59 DePaul L. Rev. 371 (2010); Stephen N. Subrin, Federal Rules, Local Rules, and State Rules: Uniformity, Divergence, and Emerging Procedural Patterns, 137 U. Pa. L. Rev. 1999 (1989).
. Farhang, supra note 5, at 26–28; Farhang, supra note 14; Margaret E. Johnson, A Unified Approach to Causation in Disparate Treatment Cases: Using Sexual Harassment by Supervisors as the Causal Nexus for the Discriminatory Motivating Factor in Mixed Motive Cases, 1993 Wis. L. Rev. 231 (1993); Joseph L. Smith, Congress Opens the Courthouse Doors: Statutory Changes to Judicial Review Under the Clean Air Act, 58 Pol. Research Q. 139 (2005).
. Melnick, supra note 6, at 160–72; Salzberger, supra note 46; Lemos, supra note 14, at 370–71; Stephen C. Yeazell, Unspoken Truths and Misaligned Interests: Political Parties and the Two Cultures of Civil Litigation, 60 UCLA L. Rev. 1752, 1779–80 (2013).
. McCubbins, Noll & Weingast, Administrative Procedures, supra note 19. McNollgast’s view appears to assume that courts will enforce legislative procedural rules against bureaucracy consistent with the intentions of enacting legislative coalitions, or at least that they will be more faithful to the enacting coalition than bureaucrats. From their point of view, then, it may be that Congress is less in need of a supervisor when it implements directly in courts because of courts’ fidelity to Congress. However, many have doubted the view of courts as faithful agents of legislatures. See, e.g., Lee Epstein, William M. Landes & Richard A. Posner, The Behavior of Federal Judges: A Theoretical and Empirical Study of Rational Choice (2013); Segal & Spaeth, supra note 95; Salzberger, supra note 46, at 359. Indeed, some recent scholarship suggests that statutory drafters actually regard courts as more ideological than administrators. See Bressman & Gluck, infra note 150, and accompanying text. Thus, it seems a significant difference between courts and agencies, in the context of the present discussion, that Congress lacks an external institution for the enforcement of procedural controls on courts.
. See Bressman, Rubin & Stack, supra note 19, at 60–61; James M. Landis, The Administrative Process 22–26 (1938); Davis, supra note 42; Pierce, supra note 29, at 1251; Thomas W. Merrill & Kristin E. Hickman, Chevron’s Domain, 89 Geo. L.J. 833, 861–62 (2001); Spence & Cross, supra note 32, at 140; Stephenson, Public Regulation of Private Enforcement, supra note 14, at 127–29; Lemos, supra note 14, at 377–78; Cass Sunstein, The Most Knowledgeable Branch, 164 U. Pa. L. Rev. 1607, 1617 (2016).
. In discussing the potential problem of private enforcement’s tendency to push the boundaries of statutory meaning beyond legislative intent, David Enstrom observes that “agencies with rulemaking powers can override deviant statutory interpretations by promulgating contrary regulations,” and that agencies may be better positioned to correct such deviations than legislatures. David Engstrom, Private Enforcement’s Pathways: Lessons from Qui Tam Litigation, 114 Colum. L. Rev. 1913, 1939 (2014). The argument I am making here recognizes the same administrative rulemaking capacity as an alternative to judicial elaboration in the context of a litigation regime. But I further argue that legislative recognition of this opportunity informs statutory design; that Congress may intend from the outset for an agency to play a key role in elaborating statutory meaning to be enforced via lawsuits as a baseline feature of an implementation regime, rather than (or in addition to) correcting deviant judicial interpretations; and that one factor inducing this choice is Congress’s greater ability to influence rule elaboration by agencies as compared to federal judges. Whether this control strategy is sensible would depend on the motivation for reliance on litigation to implement. For example, if the legislative coalition is motivated by desire to ensure robust enforcement in the face of limited agency resources, and where the legislative coalition otherwise regards the agency as trustworthy, it would be sensible. On the other hand, if reliance on private enforcement is motivated by legislative distrust of the executive due to divergence in preferences (such as under divided government), this control strategy would be less attractive to a legislative coalition, and would depend on the relative positions of the agency and the court vis-à-vis Congress. Thus, while this strategy to control substantive statutory meaning while relying on litigation regimes will sometimes be attractive, it will sometimes not be. Further, as Engstrom also notes, this control strategy (like all others) can only be partial in light of agency incentives to deviate from legislative preferences, and procedural encumbrances on rulemaking. Id. at 1939–40.
. While specific and mandatory rulemaking delegations do not guarantee that an agency will heed them, they certainly increase the probability that the agency will make rules on specified subject matter as compared to discretionary and general rulemaking delegations (stating, for example, that an agency “may” make rules that are “reasonably necessary” to carry out a law). Further, the presence of a specific and mandatory rulemaking delegation increases the probability that an agency will be ordered by a court to make rules in response to a lawsuit. Eric Biber, Two Sides of the Same Coin: Judicial Review of Administrative Agency Action and Inaction, 26 Va. Envtl. L.J. 461, 465, 476, 478, 482 (2008).
. An important qualification is necessary here. The implication is not that litigation regimes cannot come at a cost to agency power if they are associated with a formal augmentation of the agency delegation. Even if the choice of a litigation regime is associated with an increased delegation of rulemaking power, this would not suggest that the agency’s relative power is undiminished by the presence of the litigation regime. For example, a private right of action’s capacity to draw a large volume of de novo litigation into federal courts will often lead to much greater judicial influence on elaboration of substantive law and its application to individual cases—even in the presence of a rulemaking delegation—as compared to a scenario in which an agency dominates implementation through the singular administrative power of adjudication, subject only to deferential judicial review under the Administrative Procedure Act. Further, Part III.I shows that while litigation regimes are associated with more frequent rulemaking delegations, the delegations are less discretionary in nature.
. Id. at 775. Interestingly, this contrasts what is probably the more common view among law professors: that administrators are more “political” interpreters than courts. See Landes & Posner, supra note 57; Stephenson, Legislative Allocation of Delegated Power, supra note 14.
. See generally Richard F. Fenno, Jr., Congressmen in Committees (1973) (classic political science study of how committee members’ actual goals and environmental constraints shape committee decisions in the House, and recognizing formulation of good policy as one significant committee member goal); R. Douglas Arnold, The Logic of Congressional Action (1992) (study of how members of Congress weigh how policy positions will affect reelection prospects, emphasizing the importance of policy preferences of significant groups within legislators’ coalitions, and legislators’ own preferences when reelection implications are indeterminate); see also Aberbach, supra note 63, at 110.
. Arnold, supra note 152; Farhang, supra note 5, at ch. 6; Shipan, supra note 61; Thomas W. Gilligan, William Marshall & Barry Weingast, Regulation and the Theory of Legislative Choice: The Interstate Commerce Act of 1887, 32 J.L. Pol. 35 (1989). Indeed, Hall and Deardorff highlight that such groups supply substantial expertise, resources, and labor needed to achieve greater specificity, materially defraying the costs actually borne by legislators to achieve it. Richard L. Hall & Alan V. Deardorff, Lobbying and Legislative Subsidy, 100 Am. Pol. Sci. Rev. 69 (2006).
. Huber & Shipan, supra note 19; see also VanSickle-Ward, supra note 19, at chs. 4–6 (finding several other measures of political-institutional fragmentation and conflict to be associated with specificity).
. Arthur W. Murphy, Old Maxims Never Die: The “Plain-Meaning Rule” and Statutory Interpretation in the “Modern” Federal Courts, 75 Colum. L. Rev. 1299, 1312–13 (1975); Rodriguez, supra note 38, at 221; see Melnick, supra note 6.
. David R. Mayhew, Divided We Govern: Party Control, Lawmaking, and Investigations, 1946–1990 (1991). Mayhew relies primarily on media coverage in the New York Times and the Washington Post to identify highly significant statutes. He has provided updates to his identification of significant legislation since initial publication of Divided We Govern.
. Bucy, supra note 30, at 5; John C. Coffee Jr., Rescuing the Private Attorney General: Why the Model of the Lawyer as Bounty Hunter Is Not Working, 42 Md. L. Rev. 215, 226 (1983); David Engstrom, Agencies as Litigation Gatekeepers, 123 Yale L.J. 616, 621, 632 (2013); Stewart & Sunstein, supra note 32, at 1298.
. Barry Boyer & Errol Meidinger, Privatizing Regulatory Enforcement: A Preliminary Assessment of Citizen Suits Under Federal Environmental Laws, 34 Buff. L. Rev. 833 (1985); Engstrom, supra note 166, at 633, 661–62; Howard M. Erichson, Coattail Class Actions: Reflections on Microsoft, Tobacco, and the Mixing of Public and Private Lawyering in Mass Litigation, 34 U.C. Davis L. Rev. 1 (2000).
. Engstrom, supra note 166, at 658; Tamar Frankel, Implied Rights of Action, 67 Va. L. Rev. 553 (1981); Reza Rajabiun, Private Enforcement and Judicial Discretion in the Evolution of Antitrust in the United States, 8 J. Competition L. & Econ. 187 (2012).
. Derthick, supra note 28, at 5; Morriss, Yandle & Dorchak, supra note 28, at 15; Bucy, supra note 30, at 66–67; Cross, supra note 30, at 68–69; Grundfest, supra note 30, at 969–71; Stephenson, Public Regulation of Private Enforcement, supra note 14, at 116–17.
. These include statutory authorization for an administrator to impose fines; issue cease and desist orders, citations, or subpoenas; undertake inspections, recalls, or seizures; and suspend or revoke a license. Fifty-three percent of regimes provided for some form of administrative sanction or order.
. Davidson & Oleszek, supra note 63; Epstein & O’Halloran, supra note 19; Paul Burstein & C. Elizabeth Hirsh, Interest Organizations, Information, and Policy Innovation in the U.S. Congress, 22 Soc. F. 174, 179 (2007); Davis, supra note 42; Daniel Diermeier & Timothy J. Feddersen, Information and Congressional Hearings, 44 Am. J. Pol. Sci. 51 (2000); Ralph K. Huitt, The Congressional Committee: A Case Study, 48 Am. Pol. Sci. Rev. 340, 342–43 (1954); Kevin M. Leyden, Interest Group Resources and Testimony at Congressional Hearings, 20 Legis. Stud. Q. 431 (1995); Spence & Cross, supra note 32, at 135–36.
. Of course, this sampling strategy does not permit comparisons across laws, since (1) one law with twenty witnesses in the sample may have had twenty witnesses testify in total, and (2) another law with twenty witnesses in the sample may have had two hundred witnesses testify in total. Thus, aggregate testimony in scenarios (1) and (2) cannot be compared based on samples of twenty each. However, as discussed below, the use of law fixed effects restricts comparison in the empirical models to different regimes in the same law, thereby holding constant the total number of witnesses who testified in each law.
. For a good discussion of the fixed effects strategy for dealing with omitted variables, see William H. Greene, Econometric Analysis ch. 13 (5th ed. 2003); Jeffrey M. Wooldridge, Introductory Econometrics: A Modern Approach ch. 14 (2006); Joshua D. Angrist & Jörn-Steffen Pischke, Mostly Harmless Econometrics: An Empiricist’s Companion ch. 5 (2009).
. Epstein & O’Halloran, supra note 19; Farhang, supra note 5; Huber & Shipan, supra note 19; VanSickle-Ward, supra note 19; Cooter & Ginsburg, Comparative Judicial Discretion, supra note 23; Cooter & Ginsburg, Leximetrics, supra note 23; Moe, supra note 25.
. Policy dummies were created for all policy areas comprising more than 5 percent of regimes appearing in omnibus laws. These were agriculture, civil rights, food and drug, labor, public health and safety, and securities and commodities exchange. Remaining policy areas were aggregated into an “other” category.
. It may be problematic to compare CQ line counts over time because there may be temporal variation in the average level of salience of significant laws, and in the formatting of the publication itself. However, as discussed above, the use of law fixed effects restricts comparison in the empirical models to different regimes in the same law, thereby holding constant salience and formatting norms.
. VanSickle-Ward argues that in models of specificity, the effect of political conflict in the legislative process on specificity will be conditional on salience, with political conflict in high-salience laws associated with less specificity, and political conflict in low-salience laws associated with more specificity. The logic of the argument is that in higher salience laws Congress is under pressure to act, and thus has strong incentives to navigate policy conflict with vague and indeterminate language (because more specific bills are harder to pass in a politically divided institutional environment). In lower salience laws, there is less pressure to act, and potential opponents can be accommodated more discretely by adding provisions to the law (such as exemptions for objectors), thereby driving up specificity. VanSickle-Ward, supra note 19, at 9, 28, 32. I examined an interaction between opposition and salience in the main models below (Models B & C in Table 3 (legislative attention), and Models B & C in Table 5 (specificity)). The interaction was highly insignificant in all four models (with p values ranging between .49 and .81), and thus I elected not to include it. VanSickle-Ward’s focus was political conflict at an institutional level (divided government, polarization, interest group environment, etc.), which is controlled for in my models with law fixed effects. Thus, this regime-level interaction of witness opposition and salience is not a direct test of her theory.
. The administrative implementation dummies are not part of an exclusive set of categories, either with respect to one another, or with respect to the lawsuit dummies. Each of those four dummies may equal 0 or 1 in any regime, regardless of which of the four mutually exclusive lawsuit codes the regime takes. For that reason, the administrative dummies do not change the fact that the lawsuit dummies in the model are interpreted with respect to the excluded reference category (no lawsuits).
. Dropping the interaction of opposition and litigation regime, litigation regime is statistically significant in both models, with a coefficient of .21 in the expert-only model, and .14 in the interest group model (p>.01 and >.05, respectively). Even in the absence of the interaction, the association between litigation regimes and legislative attention is stronger with respect to policy experts than traditional interest group witnesses.
. Huber & Shipan, supra note 19. VanSickle-Ward takes a different approach, using a content analysis protocol to code specificity in statutes on a scale ranging from 0 to 3. This approach seems especially strongly suited to VanSickle-Ward’s primary use of it, which was coding statutes passed across the states to address the same policy problem. This allowed VanSickle-Ward to create a coding protocol that identified in advance a set of issues which each law might address, and provided criteria for measuring how specifically each state statute addressed that issue. VanSickle-Ward, supra note 19, at 165–75. This strategy would not be promising for coding statutes spanning all policy areas and covering more than sixty years, where no two statutes address the same issue.
. Huber & Shipan, supra note 19, at 44–56. Huber and Shipan use a content analytic coding scheme in their study of state Medicaid laws to code statutory language as “general policy language” that provided indeterminate guidance, versus specific language that imposed more determinate constraints on the parameters of policy, finding that about 80 percent of statutory language fit into the latter category. Id.
. Clean Water Act of 1972, 86 Stat. 816, 863, 871 (1972). In the absence of this definition, an alternative broader definition of sewage, not limited to human body waste, would have been possible. Sewage, Merriam-Webster Dictionary, https://www.merriam-webster.com/dictionary/sewage [https://perma.cc/8CN3-UJDQ] (last updated Aug. 30, 2018) (defining sewage as “refuse liquids or waste matter usually carried off by sewers”).
. This is not to say, of course, that the “courts as dumping grounds” perspective is not an accurate account of particular lawmaking episodes. Rather, the empirical reality of legislative delegations, when viewed across a large population of important statutes, is the opposite of that predicted by the “courts as dumping grounds” perspective on the issue of specificity. Moreover, this empirical result regarding specificity does not allow evaluation of the other dimension of the “courts as dumping grounds” perspective, which concerns the choice of delegation to courts at all. None of the empirical evidence presented in this Article speaks to that question, which I intend to explore with these data in future work.
. While the theory in Part II does not predict that litigation regimes will be negatively associated with general rulemaking, this result fits the theory. Litigation regimes elevate substantive specificity in the legislation itself, as well as specific and mandatory rulemaking delegations, thereby reducing the need for general rulemaking delegations.
. See Nolan McCarty, Keith T. Poole & Howard Rosenthal, Polarized America: The Dance of Ideology and Unequal Riches (2006); Gary C. Jacobson, Partisan Polarization in Presidential Support: The Electoral Connection, 30 Congress & the Presidency 1 (2003).