I N S T I T U T E Can Courts be Bulwarks of Democracy? Jerey K. Staton, Christopher Reenock, Jordan Holsinger and Staan Lindberg Working Paper SERIES 2018:71 THE VARIETIES OF DEMOCRACY INSTITUTE July 2018 Varieties of Democracy (V–Dem) is a new approach to conceptualization and measure- ment of democracy. It is co-hosted by the University of Gothenburg and University of Notre Dame. With a V–Dem Institute at University of Gothenburg with almost ten staff, and a project team across the world with four Principal Investigators, fifteen Project Managers (PMs), 30+ Regional Managers, 170 Country Coordinators, Research Assistants, and 2,500 Country Experts, the V–Dem project is one of the largest ever social science research-oriented data collection programs. Please address comments and/or queries for information to: V–Dem Institute Department of Political Science University of Gothenburg Spra¨ngkullsgatan 19, PO Box 711 SE 40530 Gothenburg Sweden E-mail: contact@v-dem.net V–Dem Working Papers are available in electronic format at www.v-dem.net. Copyright c©2018 by authors. All rights reserved. Can Courts be Bulwarks of Democracy? Jeffrey K. Staton, Christopher Reenock, Jordan Holsinger and Staffan Lindberg ∗ ∗Jeffrey K. Staton is Professor of Political Science at Emory University. Christopher Reenock is Associate Professor of Political Science at Florida State University. Jordan Holsinger is a Ph.D. candidate in the De- partment of Political Science at Florida State University. Staffan Lindberg is Professor of Political Science at University of Gothenburg. This research project was supported by Riksbankens Jubileumsfond, Grant M13-0559:1, PI: Staffan I. Lindberg, V-Dem Institute, University of Gothenburg, Sweden; by Swedish Re- search Council, PI: Staffan I. Lindberg, V-Dem Institute, University of Gothenburg, Sweden & Jan Teorell, Department of Political Science, Lund University, Sweden; and by Knut & Alice Wallenberg Foundation to Wallenberg Academy Fellow Staffan I. Lindberg, V-Dem Institute, University of Gothenburg, Sweden. We performed simulations and other computational tasks using resources provided by the Notre Dame Center for Research Computing (CRC) through the High Performance Computing section and the Swedish National Infrastructure for Computing (SNIC) at the National Supercomputer Centre in Sweden. We specifically acknowledge the assistance of In-Saeng Suh at CRC and Johan Raber at SNIC in facilitating our use of their respective systems. Abstract Independent judges are thought to promote democratic regime survival by allowing per- ceived violations of rules limiting arbitrary power to be challenged non-violently in a fair setting, governed by transparent rules. Yet, judges are often subjected to public shaming and politically motivated removals. Courts are sometimes packed with partisan allies of the government, their jurisdiction is nearly always subject to political control and their deci- sions can be ignored. For all of these reasons, scholars have identified patterns of prudential decision-making that is sensitive to political interests even on the most well-respected courts in the world. If these forces all operate on judges, what, if any, are the conditions under which judges can be conceived of as defenders of democracy? How could judges subject to political pressures stabilize a democratic regime? This document summarizes a book that addresses these questions. We argue that despite these pressures judges can enhance regime stability by incentivizing prudence on behalf of elites, both those who control that state, i.e., leaders, and those on whose support leaders depend. Empirically, we leverage original data on judicial behavior, judicial institutions, and policy using a sample of all democratic political systems for over 100 years. We re-examine empirical claims of existing models of courts and democracy as well as original claims derived from our own work. The election of Donald J. Trump to the presidency of the United States disrupted the otherwise abstract and politically disconnected world of American political science. Scholars once content to publish in journals hidden behind paywalls began writing opinion pieces in print media sources and granting interviews with broadcasters, radio stations and podcasts, all questioning whether American democracy was indeed in danger. With the assistance of a core of young journalists committed to evidence-based reporting, scholars mobilized to bring decades of research on the nature of authoritarianism to the public discourse.1 As the field promoted past research, it also created new measurement strategies designed to characterize changes in important elements of democratic life, engaging the whole discipline in the process of data production.2 In a New York Times opinion piece in December of 2016, Steven Levitsky and Daniel Ziblatt put the matter bluntly, writing Is our democracy in danger? With the possible exception of the Civil War, American democracy has never collapsed; indeed, no democracy as rich or as established as America’s ever has. Yet past stability is no guarantee of democ- racy’s future survival. We have spent two decades studying the emergence and breakdown of democracy in Europe and Latin America. Our research points to several warning signs.”3 Noting that many Americans place their faith in the state’s system checks and balances, Levitsky and Ziblatt remind us that the ultimate success of formal checks on the politically powerful depend on a variety of informal norms, including the notion of a legitimate oppo- sition, partisan and presidential restraint. To this list, we should add respect for general rule of law values, including a deep societal commitment to an independent judiciary as the arbiter of fundamental constitutional norms. Just eight days after President Trump’s inauguration, the American system of checks and balances was tested. On January 27, 2017, with little input from the Departments of State, Homeland Security or Defense, President Trump issued Executive Order 13769, which im- mediately prohibited entry into the United States nationals from Iran, Iraq, Libya, Somalia, Sudan Syria and Yemen.4 The order also suspended the U.S. Refugee Admissions Program 1See for example the reporting in Amanda Taub’s “The Rise of American Authoritarianism,” https: //www.vox.com/2016/3/1/11127424/trump-authoritarianism 2See for example the work of John Carey, Gretchen Helmke, Brendan Nyhan and Susan Stokes developing Bright Line Watch, http://brightlinewatch.org/about-us-new/. 3https://www.nytimes.com/2016/12/16/opinion/sunday/is-donald-trump-a-threat-to-democracy. html?_r=0 4Executive Order: Protecting the National from Foreign Terrorist Entry into the United States, January 27, 2017. 1 (USRAP) for a period of 120 days, widening the policy’s impact beyond the seven named states. As a consequence of requiring the immediate implementation of the order hundreds of individuals were detained at the nation’s airports, some of whom enjoyed permanent U.S. residency status. By Sunday January 29, the Department of Homeland Security clarified that it would not bar the entry of permanent residents, yet roughly 500,000 residents would nevertheless be subject to extended screening activities. 5President Trump’s order required that upon resumption of the USRAP, the Secretary of Homeland Security would be directed to “make changes, to the extent permitted by law, to prioritize refugee claims made by individuals on the basis of religious-based persecution, provided that the religion of the individual is a minority religion in the individual’s country of nationality.” Critically, in an interview with the Christian Broadcasting Network, President Trump clarified that this element of the order was designed to aid individuals of the Christian faith.6 Almost immediately, dozens of legal challenges to the executive order were launched. Many of the initial challenges took the form of habeas corpus petitions seeking the release of individuals who were detained at the nation’s airports.7 The State of Washington filed for declaratory and injunctive relief in order to protect its “residents, employers and educational institutions,” which it argued would be powerfully harmed by the executive order. On January 28, Judge Ann Donnelly of the U.S. District Court for the Eastern District of New York issued an emergency stay of removal, which arguably halted the continued enforcement of Trump’s order; and on February 3, U.S. District Court Judge for the Western District of Washington James L. Robart issued a temporary restraining order on a nationwide basis enjoining the most important sections of Executive Order 13769. This decision was affirmed by a three judge panel of the Ninth Circuit Court of Appeals on February 9. Revisions to the first executive order ultimately produced more than 50 separate litigations carried out across most of the country’s legal system. Reflecting on Judge Donnelly’s emergency stay, ACLU Executive Director Anthony Romero exclaimed, “What we’ve shown today is that the courts can work. They’re a bulwark in our democracy.”8 Our book considers Romero’s claim. We ask, ‘under what conditions, 5http://www.vox.com/2017/1/28/14425150/green-card-ban-muslim-trump 6http://time.com/4652367/donald-trump-refugee-policy-christians/ 7For example, Aziz v. Trump No. 1:17-cv-00116 (E.D.Va. 2017) involved two Yemini brothers, Tareq Aqel Mohammend Aziz and Ammar Aqel Mohammed Aziz, who were detained by the U.S. Customs and Border Protection at Washington-Dulles Airport pursuant to the executive order despite having been previ- ously granted Lawful Permanent Resident status by the State Department. Similarly, Darweesh v. Trump involved two Iraqi men who were detained at John F. Kennedy International Airport despite having valid U.S. visas. 8http://www.latimes.com/nation/la-na-aclu-profile-20170131-story.html. 2 if any, are courts capable of being defenders of democracy?’ This paper is an overview of our book project’s main contributions and it proceeds in five parts. In Section 2, we develop concepts useful when considering judicial effects on regime survival. In Section 3, we review existing theoretical models of judges as bulwarks, highlighting their contributions and ending with a discussion of several open questions whose answers elude current models. In Section 4, we introduce a model of how courts help manage potential misunderstandings between leaders and supporters within a regime. In Section 5, we summarize the main theoretical and empirical insights from the model and summarize empirical implications that we test in our manuscript. Our primary theoretical claims are that independent courts can strengthen regimes in two ways: (1) by incentivizing leaders to less frequently take actions that are likely to raise questions about whether regime rules have been broken and by incentivizing the opposition to less aggressive police potential violations of regime rules, and (2) in limited circumstances, by improving the quality of communication between leaders and the opposition about the rationale for actions that might appear to violate regime rules. As we develop below, by reducing conflict in this way, courts produce two types of exchanges. Leaders exchange inter-branch conflict for partisan conflict, and all elites exchange greater opportunities for peace for governments that are less willing to act on private information, even when all parties would benefit. Finally, for this mechanism to work, courts must be willing to accept defiance of their orders. Indeed, some degree of non-compliance can be a healthy part of a functioning democratic regime. Defenders of Democracy The notion that judges should act as defenders of democratic systems, their values, and their processes is a common and well-respected position. Writing in the Harvard Law Re- view, former Israeli Supreme Court President Aharon Barak clarifies the breadth of Anthony Romero’s proposition and places it in historical context. He writes: The [role] of the judge in a democracy is to protect the constitution and democ- racy itself. Legal systems with formal constitutions impose this task on judges, but judges also play this role in legal systems with no formal constitution. Israeli judges have regarded it as their role to protect Israeli democracy since the found- ing of the state, even before the adoption of a formal constitution. In England, notwithstanding the absence of a written constitution, judges have protected democratic ideals for many years. Indeed, if we wish to preserve democracy, we cannot take its existence for granted. We must fight for it. This is certainly the 3 case for new democracies, but it is also true of the old and well. Anything can happen. If democracy was perverted and destroyed in the Germany of Kant, Beethoven, and Goethe, it can happen anywhere . . . I do not know whether the supreme court judges in Germany could have prevented Hitler from coming to power in the 1930’s. But I do know that a lesson of the Holocaust and of the Second World War is the need to enact democratic constitutions and ensure that they are put into effect by supreme court judges whose main task is to protect democracy (Barak, 2002). Judge Barak’s position reflects well the consensus that developed in the 20th century among the global legal community, which mobilized around the goal of promoting democ- ratization and human rights. A highly professional and independent judiciary came to be understood as one of the central pillars of rule of law advocacy efforts aimed at supporting new democracies and encouraging reform in authoritarian contexts. The clear recognition of an obvious theoretical tension between majoritarian values and legal limits on authority notwithstanding (e.g. Friedman, 2002), judges exercising various forms of constitutional re- view came to be viewed as key defenders of democratic norms. In the introduction to its 2002 report on the promotion of judicial independence and impartiality, the U.S. Agency for International Development writes Judicial independence is important for precisely the reasons that the judiciary itself is important . . . In democratic, market-based societies, independent and im- partial judiciaries contribute to the equitable and stable balance of power within the government. They protect individual rights and preserve the security of per- son and property. They resolve commercial disputes in a predictable and trans- parent fashion that encourages fair competition and economic growth. They are key to countering public and private corruption, reducing political manipulation, and increasing public confidence in the integrity of government. Even in stable democracies, the influence of the judiciary has increased enor- mously over the past several decades. Legislation protecting social and economic rights has expanded in many countries, and with it the court’s role in protecting those rights. The judiciary has growing responsibility for resolving increasingly complex national and international commercial disputes. As criminal activity has also become more complex and international and a critical problem for ex- panding urban populations, judges play a key role in protecting the security of citizens and nations (Miklaucic, 2002). 4 In times where core democratic norms appear to be threatened, where historical under- standings of the limits of state power are suddenly called into question, there is a undeniable optimism in these perspectives. Norms of legislative and executive constraint may be vio- lated, perhaps discarded entirely, yet as long as the courts of law are open for business and judges are willing to constrain the state, democracies may backslide but they are unlikely to collapse. It is a comforting story. The story is comforting not only because of its clear normative appeal but also because of well-known examples in which courts have either claimed or been explicitly delegated a role for protecting democratic norms. In its 1951 Southwest States Case (1 VBerfGE 14), the Ger- man Federal Constitutional Court was asked to invalidate two federal statutes designed to re- organize three Laender created during the period of allied occupation, Baden-Wu¨rttemberg, Baden and Wu¨rttemberg-Hohenzollern, into the single Land Baden-Wu¨ttemberg. The first statute extended the lives of the Laender parliaments until the reorganization could be com- pleted, thus suspending upcoming elections. The second statute laid of the procedures for the reorganization. In its opinion, the Court wrote An individual constitutional provision cannot be considered as an isolated clause and interpreted alone. A constitution has an inner unity, and the meaning of any one part is linked to that of the other provisions. Taken as a unit, a constitution reflects certain overarching principles and fundamental decisions to which indi- vidual provisions are subordinate . . . Thus this Court agrees with the statement of the Bavarian Constitutional Court, “That a constitutional provision itself may be null and void, is not conceptually impossible just because it is part of the con- stitution. There are constituent principles that are so fundamental and such an extent an expression of a law that precedes even the constitution that they also bind the frame of the constitution, and other constitutional provisions that do not rank so high may be null and void because they contravene these principles.” (as quoted in Jackson and Tushnet, 2006, p. 588). Against the backdrop of the principle that the Basic Law ought to be thought of as a logical whole, the nature of which is itself limited by certain higher principles of law, the Court continued, writing The Basic Law has decided in favor of a democracy as the basis for the gov- ernmental system . . . As prescribed by the Basic Law, democracy requires not only that parliament control the Government, but also that the right to vote of eleible voters is not removed or impaired by unconstitutional means . . . It is true 5 that the democratic principle does not imply that the life of a Landtag must not exceed four years or that it cannot be extended for important reasons. But this principle does require that the term of a Landtag, whose length was set by the people in accepting their constitution, can only be extended through procedures prescribed in that constitution, i.e., only with the consent of the people. (as quoted in Jackson and Tushnet, 2006, p. 588). By extending the Laender parliaments’ life without consent of the voters in the affected areas, the federation had violated the right to vote. Beyond invalidating federal statues in order to preserve fundamental democratic liberties, the Court endorsed a powerful principle restricting amendments to the Basic Law that might contravene fundamental norms of a democratic society. Conceived of in this way, constitutional review serves as a backstop against any legislative effort that might undermine basic democratic principles. An extraordinary variant of this power was conferred upon the South African Consti- tutional Court during its transition to democracy. A key element of the compromise that allowed for the relatively peaceful transition to democracy involved an agreement at the Multi-party Negotiating Process to 34 Constitutional Principles which would guide the draft- ing of South Africa’s new constitution. The Constitutional Court was explicitly delegated the power to certify that the constitution conformed to the 34 principles (see Constitutional Court of South Africa Case 23/96.). By the end of the twentieth century, providing a form of constitutional review had become a common piece of the constitutional architecture of transiting states. Notably, democratic reforms across post-communist Europe were accompanied by forms of constitutional control relying on powerful constitutional courts whose initial appointments were drawn from pools of highly qualified and well-regarded jurists (See the excellent discussion in Schwartz, 2000). Despite clear variation in the level of activism, courts across the region were credited with decisions helping transition from an authoritarian past to a democratic form of government. Indeed, the Constitutional Court of Hungary played such an important role in the provision of social and economic rights during the 1990s that Kim Scheppele called it “arguably more democratic than the Parliament even though the judges are not directly elected” (Scheppele, 2005). So too in Latin America have courts been a part of the story through which democratic norms come to be fully adopted. The Constitutional Court of Colombia is recognized in- ternationally for giving meaning and force to core commitments of the 1991 Constitution, requiring the state to provide the social and economic rights which the highest law demands (e.g. Cepeda-Espinosa, 2004; Uprimny, 2003). The Constitutional Court is also credited for developing a flexible jurisprudence on the natural of military jurisdiction, which has al- 6 lowed for the successful negotiation of civil-military relations during a prolongated period of violent conflict (R´ıos-Figueroa, 2016). Similarly the Constitutional Chamber of the Costa Rican Supreme Court is credited with massively expanding access to justice over social and economic rights claims, especially in the context of health (Wilson and Rodr´ıguez Cordero, 2006). And following a notorious delay, in the late 1990s, the judiciary of Chile eventu- ally began to investigate credible claims of gross human rights abuses under the Pinochet regime, a critical source of accountability in the aftermath of the democratic transition (e.g. Huneeus, 2010; Sikkink, 2011). Across many years and multiple political contexts, courts have been empowered to speak to the nature of a state’s democratic practices. They have developed jurisprudence identi- fying the limits of state authority under democratic constitutions; and, they have provided access to citizens seeking redress for violations of core democratic principles. Defensive Failures Just as there are powerful stories of judges coming to the defense of democratic states, it is clear that courts, even courts that are formally independent and unconnected politically from sitting governments, are far from successful sources of democratic restraint in all cases. Created by the communist regime in 1982, the Polish Constitutional Tribunal (CT) emerged through the democratic transition as an important source of constitutional control. Begin- ning in the middle of the 2000s, the CT would become a locus of conflict in the political competition between the Christian democratic Civic Platform (PO) and the conservative, national Law and Justice Party (PiS). The battle would come to a dramatic head in the weeks following the October 2015 parliamentary election. Following eight years as the dominant coalition partner in Poland’s government, public opinion polling in the summer of 2015 strongly suggested that PO was likely to lose a considerable number of seats in the October parliamentary elections. In June, the PO government enacted a new statute on the CT, which permitted it to replace five judges, all of whom had terms that were set to expire after the pending election. Two terms would expire after the seating of the new parliament. Under the prior institutional framework, the new government would have been empowered to appoint these judges. The five additional PO appointments meant that it had appointed 14 out of 15 CT judges; however, Andrzej Duda, the President of Poland, refused to administer the oath of office for the five new appointees. After taking office following an election that gave it a majority of seats in the Sejm, the PiS amended the PO’s constitutional court act, annulling the appointment of the five PO judges. The amendment created five new positions, limited 7 the term of office for the President of the Tribunal, ended the tenure of the sitting President and Vice President, and stipulated that a judge’s term begins only after the administration of the oath of office before the President of Poland. On December 2, the PiS-controllled Sejm appointed five new judges in direct defiance of a CT order demanding the Sejm abstain for doing so until the constitutionality of the amendment could be reviewed. President Duda administered the oath of office to the new PiS judges. On December 3, a five judge panel of the CT found that the three PO-appointed judges who replaced judges whose terms were expiring prior to the new parliamentary session were validly appointed. On December 9, the CT found multiple aspects of the PiS amendments to the Constitutional Tribunal Act to be unconstitutional, including the termination of the President and the Vice President’s terms. The government rejected the decision, refusing to publish it in the state’s Journal of Laws. By the end of the January, 2016 the government passed a budget a bill cutting the CT’s yearly budget by roughly 10 percent. Ignoring con- cerns expressed by the European Commission that it was undermining judicial independence and democracy, the PiS continued its efforts to reform the judiciary through 2017, amending rules for appointing and removing judges across the system and at all levels (Commission, N.d.). The Polish experience is far from unique. Courts seeking to constrain leaders are often the target of institutional attacks. Judges are removed from their posts. Key institutions of judicial powers are reformed or eliminated altogether. Appointment rules are changed so as to concentrate staffing authority in a single power center (Helmke, 2010; Pe´rez-Lin˜a´n and Castagnola, 2009). Some of the attacks are so serious that they render effectively eliminate the courts as a source of constraint. Indeed, in 2007, Bolivia’s Constitutional Tribunal was rendered inquorate as a consequence of politically motivated impeachments and resignations in the context of major conflict between the judiciary and President Evo Morales (Castagnola and Pe´rez-Lin˜a´n, 2011). The Polish case also reminds us that judicial orders are broadly understood to not be self-enforcing, a challenge that is particularly pressing when the target of an order is the state itself (Becker and Feeley, 1973; Birkby, 1966). Critically, although there many examples of non-compliance in settings not characterized by high levels of the rule of law (Ginsburg and Moustafa, 2008), courts are not always obeyed in rule of law states (Vanberg, 2005; Carrubba, Gabel and Hankla, 2008; Chilton and Versteeg, 2018). The Constitutional Bench of Costa Rica’s Supreme Court confronts a variety of compliance challenges in its amparo jurisdiction (Staton, Gauri and Cullell, 2015). The Netanyahu government’s pattern of evading High Court and administrative court decisions across a very wide set of issue areas is particularly notorious (for Civil Rights in Israel, N.d.). 8 Perhaps of greatest concern, scholars have suggested that in order to avoid conflict and non-compliance, judges often engage in politically deferential patterns of decision making, at least in particularly salient cases, which render the constraints they might place on gov- ernments practically non-binding (e.g., Bill Cha´vez and Weingast, 2011; Rodr´ıguez-Raga, 2011; Carrubba, Gabel and Hankla, 2008). Summarizing these challenges, USAID’s Office of Democracy and Governance writes: [I]n several countries, governments have refused to comply with decisions of the constitutional court (e.g., Slovakia and Belarus) and substantially reduced the court’s power (e.g., Kazakhstan and Russia). This illustrates the dilemma con- stitutional courts often face: Should they make the legally correct decision and face the prospect of non-compliance and attacks on their own powers, or should they make a decision that avoids controversy, protects them, and possibly enables them to have an impact in subsequent cases? Bold moves by constitutional courts can be instrumental in building democracy and respect for the courts themselves. However, the local political environment will determine the ability of the courts to exercise independent authority in these high stakes situations (Democracy and Governance, 2002). These facts raise serious questions about the capacity of courts to serve as defenders of democracy. If judges attempting to hold leaders accountable are often the target of institutional attacks; if court orders can be ignored even in states with seemingly significant commitments to the rule of law; if politically savvy judges avoid conflict precisely when they are needed, how is it that they can serve as bulwarks of democracy? What is being protected and what does it mean to pro- tect? In asking whether a court can come to the defense of democracy we confront an immediate conceptual challenge over what it is exactly that we believe courts are defending. While it is not useful to argue over which definition of democracy, among the many alternatives, is “correct,” it is nevertheless helpful to focus attention on the concept scholars seem to have in mind. As it turns out, a cursory review of the cases in which courts have or have not come to the defense of democracy illustrates that the definitions scholars and journalists adopt likely span the conceptual universe. When the German Federal Constitutional Court strikes down the first reorganization act, it does so in order to protect the democratic right to vote, 9 an essential component of minimalist and procedural concepts of democracy. In contrast, what made the Hungarian Constitutional Court so democratic in the eyes of Scheppele was its commitment to social and economic rights, common components of maximalist and substantive concepts of democracy. For the purposes of evaluating the validity of measures we use in our empirical work, it is no doubt critical to provide a clear definition of democracy; however, at this point, it is simply worth noting that scholars claim that courts might protect democracy, conceived of in a variety of ways. What we mean when we ask whether courts can be bulwarks of democracy is whether courts are capable of promoting respect for what we will call “fundamental regime rules.” However we conceive of democracy, the right to vote and to mobilize politically would appear to be fundamental regime rules. It is hard to imagine a political system that we would call democratic in which these rights were not respected. So too is the right to political speech. But the right to health, to equal protection under the law, or to a federal structure of the state might also qualify as fundamental regime rules in particular contexts. From a contractual conception of democratic regimes, these rules reflect the core commitments competing groups make to each other when coming to a compromise over regimes that transfer power peacefully via elections. Their violation calls into question social understandings about the nature of the regime itself. What we are interested in is whether courts can promote adherence to these rules. We must also ask about the outcomes indicating a break with regime rules. In other words, what counts as a failure or a success in the protection of democracy? A simple answer is that democracy is protected when regime rules are respected or when the failure to respect a rule is corrected via the legal process. Thus, to observe and remedy a violation of the right to vote counts are protection whereas the failure to see such a violation or the inability to correct it would not. A more subtle answer recognizes that democratic values are never fully realized in states (Dahl, 1971). There are periods in which particular regime rules are violated while regimes themselves remain relatively robust. What is particularly troubling in democracy is the collapse of social commitments to the peaceful transition of power via the electoral process. On this account, we might then ask whether courts are capable of ensuring not only particular adherence to regime rules but to the underlying process of peaceful compromise essential to democratic politics. Our view is that these two outcomes are intimately related, but as we develop, courts might be capable of promoting peace even when they are incapable of ensuring absolute adherence to certain regime rules. 10 Existing Theoretical Models Let us suppose that we accept Judge Barak’s normative claims about the role of the judiciary in a democratic society. And suppose further that we accept U.S.A.I.D.’s summary of the variety of ways through which independent judiciaries can enrich and sometimes come to the defense of democracy. In doing so we will have accepted a normative claim on the one hand about the role that we envision for judges in a democratic state and on the other hand we will have accepted a number of pathways through which judges might influence democratic politics. What we will not have done is give a model of how judges do this in practice. It is one thing to envision judges as bulwarks of democracy; and yet it is quite another thing for judges to perform this role in the context of a real political system, rife with both routine incentives and serious dangers that can undermine this role for judges in practice. If we are to take seriously the notion of judges as democratic defenders, we will require an understanding of politics in which judges should be expected to play this role in fact. We need to understand the mechanism or mechanisms through which courts might influence democracy. Theoretical models of law and politics generally take one of three views about that role judges might play as a protector of democratic regimes. The first view understands judicial independence, as well as the effective exercise of judicial powers, to be an outcome of political competition. Under conditions of sufficient political competition, competing parties agree to be bound by the decisions of judges because this is a core part of a long-run compromise over the nature of the political regime (Stephenson, 2003; Hanssen, 2004). A similar view is reflected in the account of judicial independence given by North and Weingast (1989). They envision independent courts as part of the institutional architecture by which political leaders make promises that are credible to potential creditors. Judicial independence from this perspective is a choice that leaders make in order to ensure state solvency or promote economic growth. Likewise, the account of independent courts as a form of political “insurance” describes a political process linking political competition to influential courts (Ginsburg, 2003). It is the fear of losing power, and the hope that courts might protect their interests against potential transgressions in the future, that causes political groups to create judiciaries incentivized and empowered to constrain current leaders. Political competition results in the construction of institutions designed to promote judicial independence. These institutions result in judges who engage in independent decision-making. The key implication of this line of research is that, far from a defender of democracies, judges are an outcome of democratic regimes. A second view of emerges from models that have focused on judicial-government inter- 11 actions, seeking conditions under which courts are willing and able of holding governments accountable. One prominent account is that courts are more likely to attempt to hold lead- ers accountable under conditions of political fragmentation (e.g. R´ıos-Figueroa, 2007). As governments find it increasingly difficult to coordinate on a response to unfavorable judicial decisions, judges are more willing to attempt to hold them to regime rules. Another account focuses on public support for courts, suggesting that courts gain leverage over elected officials only in so far as communities are unwilling to countenance state choices to ignore judicial decisions (e.g. Vanberg, 2005). On this second view, judges are capable of holding leaders to regime rules. To do so; however, courts must be willing to identify regime rule violations and they must be able to make their decisions stick. Critically, this line of research courts suggests that courts may be able to hold officials to regime rules, but they are particularly able to do so when they enjoy public support or when government is constrained, typically through the fragmentation of power. When courts do not find support in this way, they are particularly careful about the decisions they make, often failing to exercise their powers in order to save their posts. A final view derives from research on political regimes and focuses on a pernicious in- formational feature of delegating power to an individual or group. The problem is that once state power is delegated to a group it gains an informational advantage over those out of power. Critically, perceptions that political leaders have violated fundamental limits on their authority are common sources of political instability (Boix and Svolik, 2013; Diamond, Linz and Lipset, 1995; Linz and Stepan, 1996; Linz, 1978; O’Donnell and Schmitter, 1986). Managing beliefs about the extent to which leaders are constrained thus represents a critical governance challenge. Scholars have suggested that courts help groups manage this problem by providing an adjudicative process in cases where leaders are perceived to have violated a rule. On one account, courts are assumed to learn private information about whether lead- ers are acting in good faith when challenged (Carrubba, 2005). A related account relaxes the assumption of private information, instead suggesting that through the briefing process, parties necessary for a regime to survive are able to learn about the preferences and resolve of other parties (Carrubba and Gabel, 2014); and thus learn whether perceived violations will be tolerated or not. In each case, information is revealed and regime conflict is avoided precisely because courts are attentive to real political pressures and seek to minimize cases in which they issue orders that are defied. A third account suggests that through jurisprudence that invites experimentation and dialogue courts can help avoid costly conflicts by promoting learning about the nature of perceived violations (R´ıos-Figueroa, 2016). Each of these ac- counts suggests that courts might promote democratic regimes by influencing informational challenges that undermine democratic compromise. As is true of the second view, though, 12 courts on these accounts are strategically deferential, avoiding conflict whenever possible. Each of these views sheds important light on the capacity of courts to serve as bulwarks of democracy. The first view, which is skeptical of courts’ ability to be protectors of political compromise is no doubt persuasive. Put simply, democratic regimes confront some challenges that are unlikely to be fixed by judges exercising their powers. Communities sometimes choose to undermine judicial independence as a means of gaining political control. Courts are, of course, useful mechanisms of social control and so they are often targeted by groups seeking power free from democratic competition (e.g. Ginsburg and Moustafa, 2008). Still, it is possible that courts might play a role in some cases and in some contexts. In that sense, the second view’s focus on political fragmentation and public support for courts offers a compelling account of the conditions under which we might expect courts to be influential. And critically, the third view’s focus on the potential informational role that courts play both focuses our attention on a key problem of democratic governance as well as pointing to the mechanisms through which judges might matter. Empirical Findings Our book manuscript leverages a variety of new data sources that have become available over the last ten to fifteen years. We leverage data on over 100 years of regimes from the Varieties of Democracy Project (Coppedge and Wilson., 2017.),9 which provides us with measures of concepts (e.g., attacks on judicial institutions or non-compliance) that have never been available for a large sample of states and years. We also use data from the Comparative Law Project, which gives us the opportunity to investigate judicial decision making across 40 countries. In addition to these data sources (and many others), we also use information on constitutional rules from the Comparative Constitutions Project (CCP, (Elkins, Ginsburg and Melton, 2014) ). We use these data for three purposes: (1) to revisit empirical debates from the literature using new data and rigorous research designs for observational data whenever possible, (2) to highlight facts that raise puzzles in light of existing arguments and (3) to evaluate several empirical implications of our own theoretical argument, which we summarize below. In addition, the manuscript provides a number of tests of measurement validity for our key variables. In this section, we highlight a few findings that shed light on existing arguments and a few patterns that suggest puzzles to answer. 9The data cover countries and former colonies from 1900 to 2016 across hundreds of concepts. 13 Results that are consistent with existing theory The first view of the relationship between courts and democracy suggests that courts are not so much defenders of democracy as they are outcomes of democratic processes. Consider the response of judicial independence to democratic transitions. Figure 1 below displays the mean levels de facto judicial independence using the Linzer and Staton (2015) (or LJI) measure in the lead-up to and in the aftermath of a transition to democracy. The shaded area represents one standard deviation around the mean. The plot provides simple but striking support for the first view. After transitioning to democracy, de facto judicial independence experiences rather rapid growth in the first 20 years, continuing to climb until roughly 50-60 years under democracy. Enhanced political competition would appear to be associated with increased de facto judicial independence. 0.0 0.2 0.4 0.6 0.8 1.0 Years Relative to Transition Late nt D e Fa cto Judi cial Inde pen den ce 100 75 50 25 Transition 25 50 75 100 Shaded area represents one standard deviation around series mean. Figure 1: De Facto Judicial Independence pre- and post-Democratic Transition Consider also the relationship between party competition among democratic regimes and de facto judicial independence. Figure 2 displays a simple scatterplot of the difference be- tween the largest government party and opposition party and de facto judicial independence for democracies between 1975-2017. The data on party seats were taken from the Database on Political Indicators (Cruz, Keefer Scartascini 2018). Here we see similar evidence that 14 political competition and mean levels in LJI or de facto judicial independence are asso- ciated in expected ways. There are no cases of relatively high levels of de facto judicial independence in non-competitive regimes. Moreover, democracies with the highest de facto judicial independence hover within 25 pts. or so around party parity between government and opposition parties. Here we find simple but striking support for a key elements of both the first and second views of the role of courts in democracy. Courts would appear to be particularly likely to behave independently when political competition is relatively high and where governments do not enjoy significant control over the levers of power.0.2.4.6.81Latent De Facto Judicial Independence -50 0 50 100Difference between Largest Gov. Party and Largest Opp. Party Figure 2: De Facto Judicial Independence and Party Competition Results that are inconsistent with theory The data we have summarized so far are consistent with the claim that independent courts are both the products of democratic processes and in particular those that avoid the concen- tration of power. Here we summarize three additional results, which raise questions about existing theoretical models. We consider whether there is evidence suggesting that political leaders should attempt to build independent courts as political systems democratize. We then consider whether there is evidence suggesting a positive relationship between the rules that scholars believe should incentivize judicial independence and independence itself. We do so by revisiting the findings reported in Melton and Ginsburg (2014), in the context of a difference-in-differences design meant to more precisely identify the effect of institutional change on de facto judicial independence. 15 Does democratization result in the strengthening of institutions designed to promote judicial independence? The Comparative Constitutions Project has spent the last decade collecting and analyzing constitutional texts for all independent states since 1789. The Project uses the information contained in constitutional texts in conjunction with a survey instrument and a coding team to generate a comprehensive dataset of government institu- tions.10 These publicly available data, first released in 2010, allow scholars to extract any combination of constitutional institutions believed to be linked to judicial independence. In an application of these data on the question of judicial independence, Melton and Ginsburg (2014) use a scale of multiple de jure indicators that cover 192 countries from 1960-2008. We include these indicators and consider their variance around the time at a transition to democracy. Figure 3 below displays the mean levels in de jure judicial independence in the lead-up to and in the aftermath of a transition to democracy. The shaded area represents one stan- dard deviation around the mean. This figure offers a striking comparison to Figure 1, which summarizes the changes in de facto judicial independence following transitions to democ- racy. The data suggest that formal institutions designed to enhance judicial independence are not particularly likely to be strengthened after a transition to democracy, potentially the most dramatic increase in political competition that a country can experience. In fact, de jure institutions, while exhibiting a slight upward trend after transition, remain relatively flat throughout a regime transition, indicating that autocracies and democracies offer sim- ilar commitments via parchments to the judiciary. Even after 50 years of experience with democracy, relatively high variance among democratic states in their de jure institutions persists. We also consider the relationship between de jure and de facto judicial independence over the lifetime of different regime types. Figure 4 below displays measures of de jure and de facto judicial independence on the y- and x-axes, respectively. Blue bubbles are democratic regimes (primarily on the right side of the plot) while red bubbles are autocratic regimes (primarily on the left side of the plot). Bubble size corresponds to the regime’s age, with large bubbles being older. The panels represent two years of data, twenty years apart. The upper panel is from 1990 and the lower panel is from 2010. A reasonable expectation from the literature is that democratic regimes would have a triangular distribution with low de jure democracies having a broad distribution of de facto judicial independence along the x-axis, but a relatively tight clustering of democratic regimes in the upper-right corner of the graph. Consolidated democracies are expected to have several de jure institutions (e.g., fixed tenure, high barriers to removal, etc.) in their constitutions 10See http://comparativeconstitutionsproject.org/ 16 0 1 2 3 4 Years Relative to Transition Num ber of D e Ju re In dep end ence Ind icato rs (M elton and Ginsb urg) 50 25 Transition 25 50 Shaded area represents one standard deviation around series mean. Figure 3: De Jure Judicial Independence pre- and post-Democratic Transition 17 0.0 0.2 0.4 0.6 0.8 1.0 0 1 2 3 4 Latent De Facto Judicial Independence − 1990 Num ber of D e Ju re Inde pen den ce I ndic ator s − 199 0 0.0 0.2 0.4 0.6 0.8 1.0 0 1 2 3 4 Latent De Facto Judicial Independence − 2010 Num ber of D e Ju re Inde pen den ce I ndic ator s − 201 0 Figure 4: Plots of De Facto and De Jure Judicial Independence by Regime Type and Age. Blue Bubbles are Democratic while Red Bubbles are Autocratic. Bubble Size Corresponds to Regime Age. 18 and relatively high de facto independence. This expectation is, however, only partially realized. It is the case that long-lived democracies migrate toward to right portion of the figure, moving from mid- to high-level de facto independence over their lifetime. However, this migration appears to be independent from the formal de jure institutions adopted in their constitutions. There are many consolidated democracies with independent judiciaries that have few traditional markers of formal independence enshrined in their constitutions. On the other hand, younger democracies appear to believe that adopting de jure institutions may be the pathway to establishing an independent court. Most new democracies have several formal institutional protections represented in their constitution despite having relatively less independent courts. Also interesting to note is that autocratic states of all ages appear to adopt de jure institutions committing the state to an independent judiciary despite having extremely low de facto independence, suggesting that their formal institutional commitment is little more than symbolic window dressing. Of course the associations represented in Figure 4 are merely correlations. We also con- sidered whether de jure institutions have a causal effect on de facto judicial independence. We began with the Melton and Ginsburg (2014) dataset and coded all new de jure insti- tutional changes across five de jure indicators as interventions between 1959-2008. These institutions included provisions for lifetime terms, selection procedures, removal conditions, removal procedures and salary insulation. Interventions for each of these institutions were quite rare over the time period with 7, 18, 9, 7 and 6 treatments, respectively. We then used coarsened exact matching to match these intervention cases to control cases that shared economic development, population and religious characteristics. We then analyzed the effect of these institutional interventions on de facto judicial independence with a difference-in- difference analysis with fixed effects for both units and time. The results of this analysis, reported in Table 1 below, suggest that newly adopted de jure institutions have no indepen- dent causal effect on de facto judicial independence. Puzzling Empirical Patterns Consistent with the second view of courts and democracy, the simple patterns we reveal suggest that judicial independence is closely related to the fragmentation of power and political competition. Yet, we do not find support for a key element of models connecting democracy to de facto judicial independence. Democratic transitions do not seem to result in a strengthening of institutions thought to incentivize judicial independence. Also, like Melton and Ginsburg, we find no relationship between de jure and de facto independence in democracies. Both of these findings raise questions about the first two views of courts and 19 Table 1: Difference-in-Difference Estimates for Impact of de jure Institutions on de facto Judicial Independence Effects of De Jure Institutions on De Facto JI Life Term Selection Proc. Removal Condition Removal Proc. Salary Insulation De Jure Institution 0.059 -0.029 -0.009 0.021 0.010 (0.031) (0.0175) (0.015) (0.097) (0.010) Time-Varying Controls GDP per capita (thousands) 0.163** 0.091 0.061 0.072 0.108 (0.076) (0.067) (0.032) (0.039) (0.137) Population -0.048 0.037 -0.236 0.051 0.056 (0.221) (0.121) (0.105) (0.045) (0.148) Constant -0.489 -0.484 -2.045 -0.264 -0.985 (1.647) (1.279) (0.788) (0.635) (1.500) Unit Fixed Effects YES YES YES YES YES Time Fixed Effects YES YES YES YES YES Number of Observations 212 745 253 134 155 Note: De jure institution coefficients are estimated from a difference-in-difference estimator modeled around changes in particular de jure institutions. Standard errors are clustered on country-episodes. Two-tailed tests of statistical significance are displayed: *p < .05, **p < .01. democracy. Critically, no existing theoretical account sits easily with a number of clear empirical pat- terns, both in the data we summarize and well-known stories about courts and democracy. First and perhaps foremost, peak courts broadly understood to be independent of sitting gov- ernments have been central players in the collapse of democratic regimes. Notably, conflicts between the Supreme Court of Chile and Salvador Allende in 1973 and between the Supreme Court of Honduras and Manuel Zelaya in 2009 created legal grounds that coup-plotters would use to rationalize their actions (Ruhl, 2010; Valenzuela, 1978). So, it is certainly possible that courts can be part of a democratic breakdown. Second, courts seeking to constrain leaders are often the target of institutional attacks (Helmke, 2010; Pe´rez-Lin˜a´n and Castagnola, 2009). In 2007, Bolivia’s Constitutional Tri- bunal was rendered inquorate as a consequence of politically motivated impeachments and resignations in the context of major conflict between the judiciary and President Evo Morales (Castagnola and Pe´rez-Lin˜a´n, 2011). Third, judicial orders are broadly understood to not be self-enforcing, a challenge that is particularly pressing when the target of an order is the state itself (Becker and Feeley, 1973; Birkby, 1966). Critically, although there many examples of non-compliance in settings not characterized by high levels of the rule of law (Ginsburg and Moustafa, 2008), courts are not always obeyed in rule of law states (Vanberg, 2005; Carrubba, Gabel and Hankla, 2008). The Constitutional Bench of Costa Rica’s Supreme Court confronts a variety of compliance challenges in its amparo jurisdiction (Staton, Gauri and Cullell, 2015). The Netanyahu 20 government’s pattern of evading High Court and administrative court decisions across a very wide set of issue areas is particularly notorious (for Civil Rights in Israel, N.d.). Perhaps of greatest concern, scholars have suggested that in order to avoid conflict and non-compliance, judges often engage in politically deferential patterns of decision making, at least in particularly salient cases, which render the constraints they might place on gov- ernments practically non-binding (e.g., Bill Cha´vez and Weingast, 2011; Rodr´ıguez-Raga, 2011; Carrubba, Gabel and Hankla, 2008; Chilton and Versteeg, 2018). Summarizing these challenges, USAID’s Office of Democracy and Governance writes: [I]n several countries, governments have refused to comply with decisions of the constitutional court (e.g., Slovakia and Belarus) and substantially reduced the court’s power (e.g., Kazakhstan and Russia). This illustrates the dilemma con- stitutional courts often face: Should they make the legally correct decision and face the prospect of non-compliance and attacks on their own powers, or should they make a decision that avoids controversy, protects them, and possibly enables them to have an impact in subsequent cases? Bold moves by constitutional courts can be instrumental in building democracy and respect for the courts themselves. However, the local political environment will determine the ability of the courts to exercise independent authority in these high stakes situations (Democracy and Governance, 2002). It is hard to square any of the existing views with the frequency of attacks on courts as well as the incidents of non-compliance. If independent and powerful courts are simply the outcome of long-run compromises over democracy itself, it is hard to explain why there is some much intra-regime variation in attacks and non-compliance. And if judges are un- derstood to be politically deferential when threatened, indeed if their ability to help reveal hidden information depends on this kind of behavior, then again it is hard to understand why courts find themselves in conflict so often. Consider Figure 5, which displays an index of two instruments taken from the V-Dem database, judicial attacks and judicial purges. V-Dem’s Judicial Attacks instrument mea- sures the extent to which the government attacked the courts, claiming they were either corrupt, incompetent or politically motivated in their operations. V-Dem’s Judicial Purges instrument measures the extent to which judges had been removed arbitrarily or for political reasons. We constructed an additive index of these two variables and plotted them against de facto judicial independence, or LJI. As the Figure displays, more independent judiciaries are less likely to be the target of government attacks. However, we nevertheless see a decent amount of variation in government attacks against these types of courts. Why would we 21 see intra-regime variation in judicial attacks if independent courts are an outcome of the democratic process? -2024VDEM Judicial Attacks 0 .2 .4 .6 .8 1Latent Judicial Independence (Linzer and Staton, 2015) Figure 5: De Facto Judicial Independence and Party Competition Now consider Figure 6. This figure displays an index of two instruments taken from the V- Dem database, noncompliance with high court decisions, and noncompliance with other court decisions. These V-Dem instruments measure the extent to which the government failed to comply with important decisions by the courts with which is disagreed. We constructed an additive index of these two variables and plotted them against de facto judicial independence, or LJI. Figure 6 shows that noncompliance with court decisions is clearly more of an issue for less independent judiciaries. Yet, similar to the results for judicial attacks seen above, we see that even fairly independent courts have heterogeneous experiences with noncompliance. Why is this? Standard theoretical accounts on the role of democratic courts do not offer particularly satisfying explanations. These patterns raise questions about the mechanism linking courts to regime survival. If judges attempting to hold leaders accountable are often the target of institutional attacks, and if court orders can be ignored even in states with seemingly significant commitments to the rule of law, why would independent judges render regime rules credible? Similarly, if politically savvy judges avoid conflict precisely when they are needed, it is unclear how courts can be bulwarks of democracy. 22 -2-1012VDEM Judicial Noncompliance 0 .2 .4 .6 .8 1Latent Judicial Independence (Linzer and Staton, 2015) Figure 6: De Facto Judicial Independence and Party Competition Modeling the Management of Regime Rules Our goal is to develop an account of courts and democracy that captures salient features of existing models and yet can speak to the puzzling findings reviewed at the end of the last section. We are particularly interested in the second and third views summarized above. We seek a model of courts and democracy in which judges are sensitive to political pressures and in which they might influence communication between political elites. This section summarizes our theoretical account. Specifically, we consider whether and how delegating some form of judicial review to a court can influence the ability of powerful actors in society to manage commitments to regime rules. An essential problem in the management of regime rules involves addressing poten- tial miscommunication between leaders, coalition partners and possibly political opponents about the validity of actions taken in light of the regime’s constraints. This problem is com- plicated by indeterminacies in natural language and the basic human inability to foresee all future contingencies, which imply that even those rules that are formalized in foundational documents are subject to differing interpretations. Many rules are context dependent. Economic crises, domestic disturbances, budget short- falls, or war may require a leader to pursue extraordinary actions, which might not be or- dinarily tolerated.11 Consider the Peruvian constitutional crisis of the early 1990s. After 11Of course, some constitutional arrangements typically anticipate this problem, calibrating the state powers so as to properly meet economic crises, natural disasters or security threats (Gross, 2011); however, 23 nearly a decade of attempting to bring to heel elements of the Shining Path and the Tu- pac Amaru´ Revolutionary Movement (MRTA), newly elected Peruvian President Alberto Fujimori sought to enhance his powers via emergency power legislation. By the President’s account, “his emergency measures were needed both to battle terrorism and to restructure the state and economy. He required an iron hand to reform the judiciary and break the gridlock created by the opposition in Congress” (Cameron, 1998, pp.127). Suspicious that President Fujimori’s plea for greater authority to fight domestic terrorism was in actuality a raw power grab, Congress resisted. It altered the legislation to “subordinate the executive to the rule of law, assert congressional supremacy over law-making, and require the executive to justify its use of emergency powers” (Cameron, 1998, pp.127). Soon after, Fujimori, in conjunction with the intelligence community and the military, initiated his autogulpe. Informational asymmetries between governors and the governed make this problem par- ticularly vexing. Leaders often have more precise information about the possibility for im- plementing policy initiatives. This advantage makes it possible for leaders to be less than truthful about the true reasons behind a change in policy and for that reason skepticism is a sensible reaction to government actions that plausibly violate regime standards. The Israeli Interior Ministry’s rationale for failing to comply with High Court decisions on a variety of policies dealing with the separation barrier, treatment of migrant workers and educational equality between the Arab and Jewish populations highlighted practical difficulties, relied on appeals about impracticalities. Specifically, the ministry claimed that delays in changing their policies were due to the extreme complexity of these cases, some of which entail significant budget ex- pense, some which have implications for third parties, some of which require the establishment of new procedures and various complex administrative actions. Be- cause of their complexity, these court rulings require an extended period during which they can be implemented (for Civil Rights in Israel, N.d.). Was this true? Were the conditions such that these policies could not be amended or was the Interior Ministry simply refusing to do what it surely could have? Disagreements over these kinds of claims are the kinds of the disagreements that commonly lead to regime breakdown in some contexts. Of course, it is unclear that they have significantly raised the probability of regime collapse in Israel. The problems associated with interpreting a leader’s claim inhere nonetheless. even when present, these “states of exception” are themselves open to interpretation. 24 What role for courts? To build our model, it is useful to return to the third view of courts and regimes, which focuses on an informational role for judges. We are led to ask how a judicial process impact the ability of leaders to communicate clearly about their intentions and rationales for policy choices that might appear to be violations of regime rules. One possibility is that the review process reveals special information to judges about the true nature of the political conflict they resolve. Judges might then consider revealing what they learn to uncertain parties. As long as judges are incentivized to truthfully reveal what they learn judicial review might solve the underlying communication problem. Carrubba’s (2005) model of review assumes such a role for a court. As long as judges are reasonably certain that governments will comply with their decisions, they will be willing to reveal the information that they learn via the litigation process. The literature on authoritarian institutions reflects this kind of mechanism. Svolik (2008) and Boix and Svolik (2013) suggest that authoritarian institutions can be a key component of managing coalitions when leaders are better informed than their supporters. By requiring distinct leaders to deliberate and discuss their policy views and understandings of the facts, legislative processes reveal hidden information. Whether or not all information is revealed, it is safe to assume, the argument goes, that legislative processes at least reduce uncertainty. Perhaps the litigation process could serve the same purpose. It is no doubt common to assume that litigation, generally speaking, reveals information about the case facts and parties’ motivations (Bull and Watson, 2004; Clark and Kastellec, 2013). So, perhaps all that is necessary is to assume that some information will be revealed about a leader’s motives. The difficulty with this assumption is that regime misunderstand- ings follow from beliefs that some kinds of leaders have incentives to dissemble. And if that is true, it is unclear why a litigation process that involves a party believed to be less than truthful would necessarily convince another party that what has been presented or said or recorded is in fact true. This is especially true in so far as we are talking about the state itself, and so rules regarding perjury would have to be enforced on itself. This lack of trust is exactly what causes the problem of managing regime understandings and what we are trying to solve. Assuming that courts simply provide this information does not answer why and how they might do it. Carrubba and Gabel (2014) provide a mechanism by which information could be revealed in the process of litigation. They develop a model in which regime rules are managed via litigation before a court. The particular setting that they have in mind is the Court of Justice of the European Union. The problem on which they focus is that in some years, some states will confront local challenges that make it inefficient to comply with long-run 25 agreements to European regulations. Other governments in the system might agree with this state’s interest in violating the treaty, at least in the short-run; however, there is an infor- mational asymmetry. Only the non-compliant state truly knows whether the local challenge it confronts is sufficient to render its long-run commitments inefficient. Carrubba and Gabel suggest that the litigation process provides states with an opportunity to send costly signals about their private information. The key element of the model involves three steps. First, private parties (or perhaps the Commission) file law suits alleging that an EU regulation has been violated. Second, states decide whether to file briefs in support or in opposition to a state’s legal position in a case. Second, in light of the distribution of briefs, states make costly efforts to convince the states that have filed in opposition to their position that non- compliance with an unfavorable CJEU decision ought not to be punished collectively by the states acting together. Only states that confront significant local challenges will be willing to take the effort necessary to convince its treaty partners of the inefficiency of complying with European regulations in the current instance. The first step in this process alerts states to potential violations of law while the second and third steps provide a coordinated process for revealing information about collective understandings of European law as well as one that allows states to reveal private information. Although we see our approach as broadly consonant with Carrubba and Gabel’s account, we depart from their model in a number of ways, which we believe helps shed light on parts of the problem that they do not highlight. In the Carrubba and Gabel model, judicializing the process always results in the revelation of private information held by defendant states. This happens because some types of governments (i.e., those who are being forced to pay very high costs for compliance with European law) will always be wiling to make the effort necessary to convince other states of their position. It is important, however, to consider the possibility that judicial processes do not result in information revelation, precisely a core feature of the politics of managing regime rules is that it is sometimes impossible to discern a leader’s type because the information she reveals is consistent with what all leaders would say. We develop a model in which we can directly seek conditions under which courts are able to incentivize the revelation of private information. This, as it turns out, has a lot to do with how we conceive of judicial preferences, and ultimately judicial independence. Second, we ask whether judicial processes might reduce regime conflict even if they do not materially affect information transmission. We find that they do and yet in doing so, courts produce other challenges for regimes. Third, the costly signal that states send in Carrubba and Gabel’s model is worth considering carefullly. In the model, states simply expend “effort.” The authors suggest that effort can be understood as persuasive in nature. Of course, it is unclear how a well-constructed argument about why a state has violated a regime rule 26 would either be particularly costly to that state or persuasive to the other states or both. Legal argumentation might be profitably understood as cheap talk. If effort can instead be understood as forms of payments to states or otherwise as direct benefits to state interests, then we will have to consider the possibility that the opposing briefs that governments file are designed to prompt targeted states to buy off their support. If this is possible, then the distribution of briefs offers defendant states a far from perfect sense of who really supports them. Finally, effort could be conceptualized as some kind of activity that is simply costly to the state in some way. What would count as such an effort is unclear. In our model, we will be clear about the kind of costly action that a leader might be led to take in a litigation. The only type of action of that sort in our model is non-compliance with a court order. This, as it turns out, proves to be consequential. Model We will consider a model that deals with a fundamental challenge of managing a regime com- promise, which reflects key elements of the Stephenson and Fox (2011) model of pandering. Consider a political regime that consists of a leader, endowed with governing power, and a supporter on whom the leader depends.12 We assume that the leader and the follower have come to some understanding about regime rules, which may include promises to limit terms of office, procedures for changing rules, policy limits or promises to divide regime surplus in particular ways. The players know that the political context in which they operate some- times requires the leader to take actions that temporarily violate prior agreements. In order to maintain the spirit of a regime compromise, leaders are sometimes called upon to take extraordinary actions, which in normal times would surely constitute a violation of regime rules. Indeed, if the political context does not warrant extraordinary policy measures, such measures may be used to shift the balance of power in the favor of the leader. Critically, the supporter cannot necessarily know that a leader who is taking an extraordinary action is doing so appropriately or in a power grab. We first consider several properties of this regime in the absence of a court. We will then introduce a court and consider the possible differences that it makes. 12The leader here may be thought of as a governing party or coalition in a either a democratic or autocratic context. The supporter may be conceptualized as whatever set of individuals on whom the leader depends for continued leadership. This may be thought of as a party, a group of individuals or the leader’s “winning coalition.” The key point is that the leader has immediate and formal control over the instruments of governance. 27 Baseline Model: Timing and Information Following the setup in Stephenson and Fox (2011) we begin by assuming that a political context can be understood by the players to be “normal” or “extraordinary,” as can be the policies leaders adopt and the responses to those policies by followers. Specifically, let X = {n, x} denote the set of possible descriptions of political contexts or actions taken by the leader or supporter. We denote a political context ω ∈ X. We assume that a political context is drawn from a Bernoulli distribution over X, which is known commonly to the players. We let pi reflect the probability of an extraordinary context. We assume further that the leader observes ω and then proposes a policy response, p ∈ X, where p = x reflects an extraordinary policy that might violate regime rules and p = n reflects a policy that is clearly understood not to violate these rules. Given this information structure, the leader may be conceived of as one of two types, t ∈ X. We will refer to the leaders as the extraordinary and normal types. Observing p, the supporter responds to the policy, choosing an r ∈ X. We assume that r = x induces a costly conflict of uncertain outcome with the leader whereas r = n results in continued support of the leader per the regime rules. A mixed strategy for the leader, σl, assigns a probability distribution over X for each state, where σl(ω) indicates Pr(p = x|ω). Similarly, a mixed strategy for the supporter, σs, assigns a probability distribution over X for each policy she observes, and where σs(p) indicates Pr(r = x|p). Preferences The game deals with how to ensure that policy responses match circumstances while also minimizing conflict. To focus on that problem, we normalize the value of the regime agree- ment to 1. A failure to respond to an extraordinary political context reduces the value of the regime to both players. We scale the regime value by γ ∈ (0, 1) in the event that the ω = x 6= p. We assume that conflicts, should they emerge, are resolved probabilisti- cally. Specifically, if r = x, the leader receives v, where v ∼ Be(α, β). In the event that ω = r = x 6= p, the leader and supporter receive γv and γ(1 − v), respectively. Finally, should ω = n = r 6= p, and the normal leader has unnecessarily taken the extraordinary policy, we assume that the value of the regime to the leader (supporter) increases (decreases) by k > 1. This implies the following payoff function for the leader: 28 ul =    1 if ω = p ∧ r = n 1 + k if ω = n = r 6= p γ if ω = x 6= p = r v if r = x ∧ ¬(ω = x 6= p) γv if ω = r = x 6= p. Likewise, the payoff function for the supporter is giving by us =    1 if ω = p ∧ r = n 1− k if ω = n = r 6= p γ if ω = x 6= p = r 1− v if r = x ∧ ¬(ω = x 6= p) γ(1− v) if ω = r = x 6= p. Analysis Our solution concept is Perfect Bayesian equilibrium. We assume that beliefs are derived via passive conjectures at information sets that are not reached. There are two types of equilibria in the model, two of which involve the leader adopting the same policy independent of the political circumstances, and one in which the types partially separate from each other. We begin by considering the possibility for a complete revelation of the political circum- stances. First note that the supporter will react normally to the normal policy, no matter her beliefs. Independently of the political context, an extraordinary response to a normal policy only wastes resources since the leader has not attempted to shift the balance of power. In the worst case scenario, where the state warrants an extraordinary action but the leader acts normally (i.e., ω = x 6= p), the supporter would only compound the problem (i.e., γ(1− E(v)) < γ). The key question for the supporter is how to respond to an extraordinary policy. Suppose that the leader chooses a policy that is matched to the state (p = ω). If this were true, the supporter would infer the leader’s type upon observing the policy. Thus, upon observing the extraordinary policy, the supporter would accept the policy, selecting a normal reaction (r = n). This kind of reaction would provide the normal leader with a strong incentive to adopt an extraordinary policy, and thus the equilibrium would unravel. Now suppose that each type selected a policy mismatched to the state. In this case, the supporter would again correctly infer the leader’s type from the policy. Upon observing the extraordinary policy, she would know that she confronts a normal leader attempting to change the balance of power. For that reason, she would react extraordinarily (setting r = x), and that would incentivize 29 the normal leader to adopt the normal policy. Thus, the baseline model is inconsistent with communication that fully reveals to the supporter the true state of the world. Lemma 1. There is no PBE in which the types fully separate. Political Opportunism and Political Failure There are two types of equilibria in this model, one in which the leader types can be expected to take the same action and another in which the extraordinary leader can be distinguished partially from the normal leader. Consider the first type of equilibrium. In an equilibrium in which both leaders adopt the extraordinary policy, the supporter’s beliefs are defined via Bayes’s rule when she observes the extraordinary policy (p = x). By passive conjectures, she holds the same beliefs when observing an unexpected normal policy. As the prior probability of an extraordinary political event increases, the supporter is naturally more likely to accept an extraordinary policy response. Definition 1. Let p¯i ≡ 1 − αk(α+β) denote the prior probability of an extraordinary set of political circumstances above which the supporter will choose r = n if she observes p = x when she expects the leaders to adopt the same policy. When the probability that the political context is extraordinary is sufficiently high (when pi ≥ p¯i), the supporter will accept an extraordinary policy response, knowing that sometimes she will fall victim to opportunistic behavior by the normal leader – a leader whose extraor- dinary policy is ill suited to the state. As the consequences of an inappropriate use of the extraordinary policy become increasingly problematic (k increases), that is, for large shifts in the nature of the regime that follow from opportunistic leader behavior, it becomes in- creasingly difficult to sustain this kind of behavior. Still, if ever the supporter is expected to behave in this way, both leaders will select the extraordinary policy for all other values of the model’s parameters. In contrast, for low values of pi, the model is consistent with a pooling equilibrium in which both leaders simply adopt the normal policy, independent of the state, expecting the supporter to select r = x if she observes the extraordinary policy. For this kind of profile of strategies to be part of a PBE, the extraordinary leader must be unwilling to engage in conflict, which requires that the consequences of failing to respond to extraordinary circumstances must be less severe than the political conflict that would ensue were he to move forward with an extraordinary policy (i.e., γ ≥ E(v)). Proposition 1. For pi ≥ p¯i, there exists a pooling PBE in which both leaders adopt p = x and the supporter sets r = n in response to all policies. For pi < p¯i and γ ≥ E(v), there exists a pooling equilibrium in which both types adopt p = n and the supporter sets r = n if p = n and r = x if p = x. The supporter’s beliefs are equal to her priors at all information sets. 30 Interpretation These equilibria reflect two distinct, and yet related problems of manag- ing regime rules. The first, investigated carefully by Svolik (2012), involves deterring leaders from reneging on regime rules by taking advantage of uncertainties about the true nature of policy challenges. An equilibrium in which the leaders pool on the extraordinary policy involves such rule violations – they are successful in political contexts in which the likelihood that regime rules ought to be bent is relatively high. In such contexts, opportunistic leaders can take advantage of perceived crises to alter bargains that were previously necessary for compromise. The second problem, addressed by Stephenson and Fox (2011), is that some- times leaders with the best of intentions are deterred from taking extraordinary actions when such actions would be appropriate, i.e., political circumstances that call for extraordinary action are not responded to appropriately. This kind of policy failure emerges in the second equilibrium when it appears that political circumstances do not warrant extraordinary ac- tion, yet in reality they do. Leaders hoping to act in good faith can be deterred from doing so because of the very same uncertainties that can produce opportunistic behavior. Despite the fact that these equilibria present one of two political failures, critically, the players avoid conflict in both cases, albeit for different reasons. Political Conflict The second type of equilibrium will involve conflict between the players, at least on occasion. Suppose that the supporter reacts extraordinarily to an extraordinary policy with positive probability, and that further, the extraordinary leader moves forward with the extraordinary policy, but that the normal leader chooses the extraordinary policy with positive probability. In this kind of equilibrium, the supporter must be selecting the extraordinary response at a sufficient rate to deter the normal leader from always attempting to take advantage of the supporter’s uncertainty (i.e., always setting p = x); and, yet the normal leader cannot be too aggressive so that the supporter would always react extraordinarily (i.e., always setting r = x). Definition 2. Let ρ ≡ kk+1−E(v) denote the value of σs(x) such that the normal leader is indifferent between his two policies; and, let λ ≡ pi(1−pi)(1−E(v)k−1) − pi (1−pi) denote the value of σl(n) such that the supporter is indifferent between her two responses, when the extraordinary type is expected to adopt p = x. It is useful to note that 12 ≤ ρ ≤ 1 and that 0 ≤ λ ≤ 1 as long as pi < 1 − α k(α+β) , so that this kind of equilibrium requires that the prior probability that the state is extraordinary is sufficiently low (i.e., pi < p¯i). 31 Proposition 2. For pi < p¯i and γ < E(v), there exists a semi-separating equilibrium in which the players adopt the following strategies: sl(ω) = { 1 if ω = x λ if ω = n ss(p) = { 0 if p = n ρ if p = x, The supporter believes Pr(ω = x|p = x) = pipi+(1−pi)λ and may have any beliefs having observed p = n. In this case, policy failure is avoided, however, it is exchanged for both allowing violations of regime rules and costly conflict on occasion.13 The probabilities of regime violations and conflict depend on the equilibrium rates with which normal leaders adopt extraordinary policies and supporters refuse to accept them. What is more, a third political problem emerges. Specifically, conflict, when it emerges, can reflect efforts to stop opportunism, but it can also reflect a fundamental miscalculation – where supporters react extraordinarily to a leader’s good faith effort to solve a serious policy problem. Interpretation Figure 7 summarizes features of the three equilibria in the baseline model. The left panel displays the probability with which the normal leader adopts the extraordi- nary policy in any equilibrium. When it is sufficiently likely that the state is extraordinary (above p¯i), the game’s equilibrium will involve both leaders selecting the extraordinary pol- icy (pooling on p = x), where the normal leader always takes advantage of the fortuitous situation. Below p¯i, equilibrium opportunism depends on the way in which the consequences to the extraordinary leader of failing to match the state relate to the expected outcome of a conflict. Should the consequences of a policy failure be insignificant relative to the ex- pected outcome of a regime conflict (i.e., γ > E(v)), both leaders will adopt the normal policy, resulting in a policy failure but one that avoids a costly conflict. However, should the consequences of policy failure be significant relative to the expected outcome of a regime conflict, then the equilibrium will be in mixed strategies. As the figure suggests, the normal 13The same strategy profile can be part of a PBE if γ ≥ E(v), as long as k is sufficiently small (k < 1−γ−E(v)(1−γ) γ−E(v) ). To simply the discussion, we will assume that k is larger than this threshold in cases where it would matter. This assumption does not influence the findings that follow. In the appendix, we will show that as k increases, the probability of conflict at lower levels of pi decreases. This result holds should there also be a mixed strategy equilibrium for γ ≥ E(v), since in such a case, increasing k would make it more difficult to sustain such an equilibrium, and in it’s place would be a pooling equilibrium in which the conflict probability is 0. 32 γ≤E(v) γ>E(v) 0 . 2 . 4 . 6 . 8 1 P r o b a b i l i t y o f O p p o r t u n i s m 0 .2 .4 .6 .8 1 Probability of extraordinary state Equilibrium Opportunism γ≤E(v) γ>E(v) 0 . 2 . 4 . 6 . 8 1 P r o b a b i l i t y o f C o n fl i c t 0 .2 .4 .6 .8 1 Probability of extraordinary state Equilibrium Conflict Equilibrium Probabilities of Opportunism and Conflict Figure 7: The left panel displays the equilibrium probability of opportunism in the model’s three equilibria. The right panel shows the probability of regime conflict. leader’s probability of choosing the extraordinary policy rises in the prior probability that the state is extraordinary, reflecting the fact that it becomes easier to take advantage of the supporter as expectations become increasingly certain that the political circumstances do in fact warrant extraordinary action. The right panel shows the equilibrium probability of conflict across our three cases. Notably, when a conflict is likely to be particularly costly (or when policy failures are in- significant), there will be no conflict in equilibrium. This is reflected by the blue curve, which tracks the probability of conflict in the pooling equilibria. When the policy failure consequences are significant (again relative to expected outcomes of a conflict), in contrast, the probability of conflict rises in the prior probability of extraordinary circumstances, but only up to p¯i, after which it drops to zero, because once people believe that the state of the world is truly extraordinary, supporters will accept extraordinary policies and normal leaders will take advantage of them. Thus, in the baseline model, the probability of conflict is both non-linear in expectations about the seriousness of potential political crises, as well as conditional on the consequences of failing to respond to those consequences. 33 Constitutional Review Model We now consider a model in which a court endowed with constitutional review powers is called upon to evaluate the policy produced by the leader. The model is sparse. For one, we assume that the policy is reviewed if enacted, setting aside concerns over who would have the incentive to challenge a policy in court, as well as questions of jurisdiction or standing, which might limit the ability of potentially aggrieved parties to sue. Likewise, a judicial decision will come in the form of a veto, setting aside concerns over rules that the court might set to govern future policy outcomes , as well as properties of rules, which might complicate the interpretation of the decision. Despite these simplifications, the model is designed to address the core ideas in work on courts and their effects on regime compromise. What is essential here is that the court can issue a decision that, formally at least, sets aside the policy adopted by the leader. Timing and information In this version of the model, we continue to assume that a political context is first drawn from a Bernoulli distribution over X, and that the leader adopts a policy p1 ∈ X. If p1 = n, then the structure is identical to the baseline, i.e., the supporter chooses a reaction r1 ∈ X. Instead, if p1 = x, we assume that the court is called upon to issue a decision. The court selects a decision d ∈ {0, 1}, where 0 indicates a ruling finding that the policy does not violate regime rules and where 1 indicates the opposite. Should the court select 0, as in the baseline, the supporter then selects an r2 ∈ X. Should the court select 1, the leader is called upon to play again, and may select p2 ∈ X. We conceive of p2 = x as a policy reimplementing the extraordinary policy, the consequences of which we parameterize. Finally, the supporter selects an r3 ∈ X should the leader adopt p2 = x. A mixed behavioral strategy for the leader, σl, assigns a probability distribution over X for each state, and for all histories (ω, x, 1). Similarly, a mixed behavioral strategy for the supporter, σs, assigns a probability distribution over X for all histories (ω, n), (ω, x, 0), and (ω, x, 1, p2). Preferences The supporter’s preferences reflect those of the baseline. She wishes the policy to match the state; she loses utility from opportunistic leadership behavior (reflected by k), from a policy failure (reflected by γ) and in the event of conflict (1−v), precisely as in the baseline. None of this is influenced by the court’s decision, so that in particular, the supporter does 34 not pay a cost for the litigation, and does not care whether the leader adopts pi = n if ω = n in response to a court order or when first called upon to set the policy. We make a similar assumption about the leader, however, we also assume that should the leader defy a court order, he pays a cost b > 0. This cost may be interpreted in a variety of ways from the consequences of public protest, electoral losses, a decrease in international investment flowing from increased uncertainty about the inviolability of contracts, etc. The key point is that defying a court order may be more or less costly to the leader. We assume that in the event that p1 6= p2, the leader may accept the court’s decision without cost. Our approach to the court’s preferences reflects elements of the literature in comparative judicial politics. We will assume first that the court wishes to allow extraordinary policies only should the state of the world warrant it. Thus, we assume that the court’s preferences track the supporter’s identically. We do so in order to provide minimal autonomy from the government, which is necessary for any information to be influenced by the court, but the assumption also sets the stage for a court to play the role of political insurance. We also assume that, all else equal, the court would prefer to be obeyed, and thus assume that the court pays a cost c > 0 should the leader defy an order. We do not assume that the court receives a special signal about the true state of the world. Instead, the court will have to infer it as anyone else in society. The court’s utility is characterized as follows. uc(pi, rj;ω) = I1a− I2c for i = 1, 2, j = 1, 2, 3; and, where I1 is an indicator variable that takes the value of 1 when the court has set d = 0 if ω = p1 and 0 otherwise; and, where I2 is an indicator variable that takes the value of 1 if p2 = x and 0 otherwise. Thus, a reflects the value of the making a decision that matches the state and c the cost of being defied, as described above. Analysis As in the baseline model, the supporter will always react normally to the normal policy. We can restrict the leader’s strategies as well. Note that the leader is indifferent between setting p1 = n and setting p1 = x and p2 = n, whatever the supporter’s strategy. For this reason, we will not consider strategies that involve the leader initially adopting the extraordinary policy, knowing that the court will strike it down, a decision that he will ultimately accept.14 As in the baseline, this version of the model is inconsistent with behavior that fully reveals the state of the world to the players who are uncertain about it. Suppose that the leader sets pi = ω. Upon observing the p1 = x, the court would uphold the policy and the 14Such strategies could be part of a PBE, but they would exist in precisely the same region of the parameter space consisting with pooling equilibria in which the leader types both adopt p1 = n. 35 supporter would accept it. This would incentivize the normal leader to set p1 = x, as well. And of course, if the leader set p 6= ω, then upon observing p1 = x, the court would infer that the normal leader had engaged in opportunism, strike down the policy, and the normal leader would accept the decision, knowing that the supporter would set r3 = x should he not back down. Lemma 2. There exists no fully separating equilibrium in the constitutional review game. The equilibria in the constitutional review game turn on the court’s beliefs over the state, which are shared of course by the supporter. For sufficient confidence that the state is extraordinary, the court will declare the extraordinary policy to be an acceptable use of power. Definition 3. Let c¯ ≡ a−c2a denote the value of pi above which the court will select d = 0 if the leader types are expected to adopt the same policy. This threshold depends on the cost the court perceives for being defied. As will be clear, any case in which c > a, and the court cares more about compliance than exercising its judgment sincerely, results in equilibria that reflect the baseline case precisely. Obviously a highly deferential court, one that behaves as if it were not part of the political system, will not influence behavior at all. The interesting cases involve those in which a > c, and thus the court might be willing to exercise its authority to declare policies invalid under regime rules. Equilibria in the Constitutional Review Game The constitutional review model contains three classes of equilibria, which reflect closely the results of the baseline model, as the following proposition suggests. We first state these conditions formally, and then compare the results to the baseline, relying in part on Figure 8 below. Proposition 3. There are two types of pooling equilibria in the constitutional review game, which can be divided into four sub cases: Case 1a: For pi ≥ max{p¯i, c¯}, there exists a pooling PBE in which both leaders adopt p = x, the court selects d = 0, and the supporter always sets r = n. Case 1b: For p¯i ≤ pi < c¯ and b ≤ min{k, 1 − γ}, there exists a pooling PBE in which both leaders adopt p = x, the court selects d = 1, and the supporter always sets r = n. Case 2a: For c¯ ≤ pi < p¯i and γ ≥ E(v), there exists a pooling PBE in which both leaders adopt p = n, the court selects d = 0, and the supporter always sets r = x. 36 Case 2b: For pi < min{p¯i, c¯} and γ ≥ E(v)− b, there exists a pooling PBE in which both leaders adopt p = n, the court selects d = 1, and the supporter always sets r = x. Beliefs for the court and supporters are equal to their priors at all information sets. As in the baseline model, there are also mixed strategy equilibria in which the extraor- dinary leader adopts the extraordinary policy for sure yet the normal leader choses the extraordinary policy with positive probability. Proposition 4. In a mixed strategy equilibrium of the constitutional review game, the ex- traordinary leader always sets pi = x and the normal leader chooses pi = x with probability λ. The court sets d = 0 if either a < c or if a > c and λ ≤ a−c(1−pi)(a+c) . It sets d = 1 otherwise. The supporter chooses r = n if ever p = n, chooses r = x with probability equal to ρ if the court sets d = 0; otherwise, she sets the probability of r = x to ρ2 ≡ k−bk+1−E(v) . The supporter and court believe that Pr(ω = x|p = x) = pipi+(1−pi)λ . The supporter may have any beliefs having observed p = n. Importantly, if ever the players adopt mixed strategies in equilibrium, the normal leader plays the exact same strategy as he did in the baseline. The reason is that if ever the supporter is called upon to react to an extraordinary policy – independently of the court’s decision – her expected utility calculus is identical to the baseline model. Thus, to render the supporter indifferent, the normal leader adopts the same probability distribution over X. In contrast, the mixed strategy that the supporter adopts depends on the court’s decision. Should the court uphold the policy in equilibrium, the supporter would adopt the same mixed strategy; however, should the court be expected to strike the policy down, the normal leader confronts an additional cost to opportunism. This is because he must defy the court order. For that reason, the supporter’s equilibrium probability of an extraordinary action must drop in order to prevent the normal leader from simply adopting the normal policy. Thus we can state the following result. Result 1. In any mixed strategy equilibrium of the constitutional review game, the probability of conflict is weakly lower than it is in the baseline game. Interpretation Figure 8 provides a visual summary of the key differences between the equilibria of the baseline and constitutional review games. The upper left displays regions of the parameter space for which the three classes of equilibria emerge. The x-axis reflects the prior probability that the state is extraordinary, whereas the y-axis reflects the consequences of a failure of the extraordinary leader to choose an appropriate policy. The upper right displays the same information for the constitutional review game, in a case where the costs 37 to the court of being defied are relatively large. The bottom panel also depicts the results of the constitutional review game, but for relatively low costs of defiance. E(v)! Policy ! Failure! (pi=n)! Potential! Conflict! Consequence of! Policy Failure (γ)! Prior Probability of ! Extraordinary State (π)! Baseline Model! Opportunism! (pi=x)! E(v)! Potential! Conflict! Consequence of! Policy Failure (γ)! a− c2a Prior Probability of ! Extraordinary State (π)! E(v)-b! Policy ! Failure! (pi=n)! Judicial Review Model (Large c)! Opportunism! (pi=x)! E(v)! Potential! Conflict! Consequence of! Policy Failure (γ)! Prior Probability of ! Extraordinary State (π)! a− c2aE(v)-b!1-b! Policy !Failure!(pi=n)! Opportunism!(pi=x)!Judicial Review Model (Small c)! Figure 8: Classes of Equilibria in the Baseline and Constitutional Review Games. The upper left panel describes equilibria in the baseline model for varying levels of γ and pi. The upper right panel describes equilibria for the constitutional review model for large values of c. The bottom panel describes equilibria for the constitutional review game with small values of c. The figure displays immediately the consequences of including constitutional review in this interaction, even a form of constitutional review that does not inject the interaction with additional information about the underlying state of the world. Consider the upper right panel. Although the region of Opportunism has not changed, because the court’s threshold for upholding policies is relatively low, what has changed is the region where mixed strategy equilibria were possible. In their place we find pooling equilibria in which policy failures result. The addition of the court increases the cost of adopting the extraordinary policy. Clearly, if the costs of defying a court (b) are close to zero, then the constitutional model 38 reduces to the baseline, as the lower threshold on the y-axis would converge on the higher threshold. Similarly, as we discussed above, if the court’s costs of defiance (c) increase, the first threshold on the x-axis converges on zero, and again, the constitutional review game reduces to the baseline. The bottom panel displays the effects of constitutional review when judges do not perceive considerable costs to being defied. The region in which mixed strategies are possible is further reduced, because the court is willing to strike policies at higher probabilities when the state is extraordinary (i.e., ω = x), and thus the costs of pursuing the extraordinary policy are higher in a larger set of contexts. Similarly, the policy failure region begins to invade the region in which opportunism was possible. Combining this analysis with the Result 1, the role of a court in helping elites manage regime rules becomes clear. First, by raising the costs of taking extraordinary measures, courts endowed with review powers encourage leaders to be careful about how they respond to political challenges. Importantly, however, this effect comes at a price – namely leaders who might have attempted to solve policy challenges with extraordinary measures are more likely to let such opportunities go. This avoids conflict, but it does so by inviting policy failure. The second mechanism by which courts reduce conflict is more subtle. In an equilibrium in which the normal leader sometimes engages in opportunistic behavior, if the court is expected to strike such policies, the equilibrium probability with which the supporter reacts aggressively is lower relative to the game without the court. The reason is that since the costs of opportunism have been increased, the supporter must be less likely to react extraordinarily (set r = x) in order to offset the normal leader’s increased incentive to adopt the normal policy (p = n). Figure 9 shows the practical effects of these dynamics. The left panel shows the equilib- rium probabilities of conflict in the baseline model versus those in the constitutional review model. We consider the case in which γ < E(v), where there is the possibility of mixed strategies (and thus conflict) in both the baseline and constitutional review games. The left panel shows the equilibrium probabilities of conflict for E(v)−b < γ < E(v). We refer to this as middling values of γ – substantively we are considering a case in which the consequences of failing to solve the policy crisis are reasonably high. For values of pi below the court’s threshold for upholding policies, the difference in probabilities is striking. Where the model without review shows an increasing probability of conflict, the model with constitutional reviews shows that the probability of conflict is flat and zero for low values of pi. Above that threshold, at roughly pi = .38 in the example, conflict emerges in the constitutional review game but it is always lower than in the baseline. Likewise, the rate at which the probability increases is lower than it is in the baseline model. Above p¯i conflict is equally likely in both 39 cases. Turning to the right panel, the results are nearly identical. When γ < E(v) − b, and the consequences of failing to respond to a political crisis are highly significant, con- flict emerges with positive probability in both game types; however, the rate at which this probability increases remains lower in the constitutional review game. We can thus state the following result. Review No Review 0 . 2 . 4 . 6 . 8 1 P r o b a b i l i t y o f C o n fl i c t 0 .2 .4 .6 .8 1 Probability of extraordinary state Middle γ Review No Review 0 . 2 . 4 . 6 . 8 1 P r o b a b i l i t y o f C o n fl i c t 0 .2 .4 .6 .8 1 Probability of extraordinary state Low γ Probabilty of Conflict (Baseline vs. Const. Review) Figure 9: Equilibrium Conflict Probabilities in Baseline and Constitutional Review Games. Black curves indicate probabilities in the baseline model whereas the dashed red curves indicate probabilities in the constitutional revue game. The left panel shows the results for middling values of γ. The right panel shows results for high values of gamma. In each panel, the probability of conflict in the constitutional review game is never larger than it is in the baseline game, for all values of pi. Result 2. The probability of conflict is weakly lower in the constitutional review game than in the baseline game. Summary The baseline model identified conditions under which we might expect opportunism, policy failure and political conflict as elites attempt to manage regime rules. The constitutional review model suggests that conflict is less likely with constitutional review carried out by a minimally independent court, even one that does not enjoy special informational advantages. To do so, courts first must be willing to use their authority and they must accept that doing so will result in conflict with leading officials. Non-compliance must be a part of the process by 40 which courts help groups manage regime compromises. Constitutional review also is likely to reduce opportunism. However, both of these effects, decreases in conflict and opportunism, come at the expense of an increased likelihood of policy failures. By encouraging leaders to limit their power, tensions are relieved, yet leaders will sometimes fail to respond to policy challenges when their supporters would like them to do so.15 A Model with a Stronger Version of Autonomy The court in our model is autonomous in the sense that it does not necessarily share the preferences of the leader – it simply wants the policy to match the state. Of course, modeled this way attaches the court to the supporter, which is itself another way to lack autonomy. That is, the court asked whether there is evidence of compelling contextual reasons for an extraordinary policy – this is exactly what the supporter wants to know. Set up in this way, constitutional review nevertheless reduces conflict, but it does so without powerfully resolving the elites’ information problems. An alternative approach to autonomy might do more. Consider the court’s payoff function. Suppose that we let I1 take on the value of 1 only if p1 = ω and the court judges the means to be appropriate to the ends. That is, the court is no longer simply interested in knowing whether the facts warranted an extraordinary action. Instead, it wants to know whether the means are appropriate, say given the norms and traditions of a democratic society. Consider a policy in which the court does not find the means appropriate even if ω = x. Here, the court would strike down the policy for all beliefs in the state, so long as a > c. In this case, there exists a fully separating equilibrium, one in which only the extraordinary leader sets p = x. For this to work, the extraordinary leader must be prepared to defy the court, while the normal leader is not, so that k ≤ b < 1 − γ. The costs of a backlash must not be too high so that the extraordinary leader would accept the failure but they must not be low enough to induces the normal leader to act opportunistically. In this equilibrium, the supporter would learn the true state of the world. The court would still be defied if ω = x, but the benefit to society of defiance is clear communication of the elite’s understandings of regime rules. Conflict would be avoided because of the accurate communication of the leader’s true motives. The bottom line is that a court that is sufficiently independent from important political interests in society can powerfully resolve the regime’s communication problems. Such a court would do so by exercising judgment, even knowing that it’s decisions are likely to be met with resistance or outright defiance. 15In the appendix, we consider a final version of the model, in which our concept of judicial autonomy is somewhat stronger than it has been in the constitutional review model. 41 General Implications of the Analysis We began our theoretical discussion by highlighting features of existing accounts that we be- lieved were worth considering further. This section summarizes our view of how the model helps clarify the role that courts might play in models in helping elites manage communi- cation about potential violations of regime rules. We also discuss additional implications of the analysis. We arrange our discussion around several questions. • Under what conditions do courts help leaders reveal private information? – Carrubba and Gabel provide a persuasive account of how judicial processes might incentivize the revelation of private information, yet we might suspect that judicial processes do not always have such an effect. In our model, courts change the nature of communication between elites under three conditions. Judges must have preferences that are disconnected from both the leadership and the opposition. Simply rendering courts independent of the government is not enough for the judicial process we consider to help reveal information. Second, judges must be willing must be willing to be defied and governments must be willing to defy them. In these circumstances, judicial review can improve communication about the rationales for taking actions that appear to violate regime rules. It is worth asking whether this scenario is likely to emerge in many states during many years. Appointment processes commonly connect judges to particular political projects. Alternation of power can produce diverse judiciaries but it is unclear how judges can ever be fully disconnected from the central cleavages that structure the politics of their states. And of course as we have argued, courts can influence regime stability even if they are unable to meaningfully influence communication among elites, a point to which we now turn. • If judges do not influence communication between elites can they influence regime stability? – We find that they can. Even in cases where courts do not materially influence the exchange of information, they influence regime conflict in two ways. They provide incentives to all types of leaders to be more prudent, to take aggressive actions less often. Simultaneously, the in the event that opposition leaders observe an action that is a potential violation of regime rules, courts cause these leaders to police this kind of activity less aggressively. By providing leaders with a clear and costly pathway to the production of extraordinary policies (defiance of the judiciary), judicial processes lower the temperature of intra-elite conflict. 42 • What are the broader implications of this kind of judicial effect on regime conflict? We see several. – The first implication is that courts translate potential conflict in political society into inter-institutional conflict. Importantly, though, autonomous judges cannot fully resolve political instability. We should expect to observe regime conflicts even when judges are independent. – Second, non-compliance with judicial orders can be part of a robust political sys- tem. Clearly, there are well-understood normative rationales for expecting leaders to comply with judicial orders. And broad disrespect for judicial authority could have negative consequences for a wide array of economic and political activities. That said, non-compliance can also also be part of a process by which leaders communicate about their resolve and the seriousness of the policy problems they confront. This kind of communication can strengthen regimes. – For this type of mechanism to work, judges must not be “single-minded pursuers of compliance,” an assumption that seems to pervade much of the political science literature on judicial politics. While it is unlikely that judges prefer a world in which orders are routinely ignored, there is abundant evidence that judges understand that legal systems sometimes require flexibility. It is useful to recall that non-compliance need not manifest as a categorical refusal to implement a judicial order forever and under all circumstances. Some forms of non-compliance emerge in contexts where judges and administrators negotiate the best way for the state to fulfill its duties - a process that plays out over time. In these contexts, judges understand that their orders will not be implemented. The Colombian Constitutional Court’s effort to force the Colombian state to address a massive failure to protect victims of the internal conflict is particularly instructive in this regard (Cepeda, 2009). So too is the experience of international judges, especially judges on regional human rights courts (European Court of Human Rights or the Inter-American Court of Human Rights) where non-compliance is understood to be a massive problem. A participant at a Brandeis University conference of international jurists, commenting on the the eventual implementation of decisions regarding gay rights suggested, “There is wisdom in waiting, as events occur later, and decisions that are not enforced become enforced” (The International Center for Ethics and Life, N.d.). Although the problems international judges confront are often thought to be distinct from those confronted by domestic judges, in many respects they are quite similar. It may be that in that similarity may 43 lie useful insight into how legal systems might best wrestle with the tensions that follow from pursuing many rule of law values. Empirical Implications The theoretical model suggests a number of empirical implications. Here we highlight a number of salient claims, beginning with claims regarding the actions of leaders. The primary way through which the independent courts influence regime conflict is by incentivizing prudence on the part of leaders. The flip side of this coin is that leaders with private beliefs that are suggestive of a serious policy problem will be less likely to take necessary actions in the presence of independent courts than they would be in their absence. This to say that the price of a reduction in regime conflict is in part what we are calling policy failure. As an empirical matter, we believe that it is difficult, perhaps completely infeasible to identify measures (even proxy measures) of policy failure itself. We can measure actions that might be perceived as extraordinary. Specifically, we will consider the decision of a state to declare a state of emergency. • Leaders should be less likely to declare a state of emergency in the presence of an independent court. We are interested in identifying other types of activities that could be considered “extraor- dinary” actions on behalf of leaders and are open to suggestions at this stage of our analysis. Finally, we intend to conduct case studies of leadership decision-making processes in the context of threats to the state. We anticipate the following. • Leaders confronting independent courts will be likely to debate what can be done to address perceived crises without violating the law. We are considering a comparison of responses to terrorism in Turkey prior to and following the recent democratic backslide. We are also considering an analysis of George W. Bush’s response to the 9/11 attacks and to the Trump administrations efforts to impose its travel restrictions in light of significant judicial pushback. We also consider implications of the model for regime conflict itself. We have already conducted a series of analyses of democratic regime survival as well as coup attempts (Staton, Reenock and Lindberg, 2015). We consider the following expectations there. • Democratic states should be particularly vulnerable to breakdown (coup attempts) when judiciaries are significantly attacked by politicians. Attacks on court’s institutions 44 and personnel incentive the kind of strategic behavior that renders courts irrelevant to intra-elite conflict. • Democratic states need not be vulnerable to breakdown (coup attempts) to simple non- compliance. Indeed, non-compliance is part of the process by which the mechanism we describe works. • Democratic states should be made less vulnerable to breakdown in the presence of strong civil societies, which empower courts and incentivize them to be willing to render independent decisions. We also can consider implications of the model with respect to responses to aggressive state decisions. As we know, there will be no conflict in cases in which actors believe that there are good reasons for leaders to behave aggressively. Thus, we anticipate the following. • Conditional on a leader taking an aggressive action, conflict is most likely (least likely) when conditions observable by all members of society suggest that there is (not) a crisis. We are still contemplating how to carry out this analysis exactly, but we have been thinking about responses to different types of emergency declarations following different types of crisis. Admittedly, we are still working on this one! Conclusion The book has a concluding chapter. 45 References Barak, Aharon. 2002. “A Judge on Judging: the Role of a Supreme Court in a Democracy.” Harv. L. Rev. 116:19. Becker, Theodore Lewis and Malcolm Feeley. 1973. The Impact of Supreme Court Decisions: Empirical Studies. 2d Ed. Oxford University Press. Bill Cha´vez, Rebecca, John A. Ferejohn and Barry A. Weingast. 2011. A Theory of the Polit- ically Independent Judiciary: A Comparative Study of the United States and Argentina. In Courts in Latin America, ed. Gretchen Helmke and Julio R´ıos-Figueroa. Cambridge University Press. Birkby, Robert H. 1966. “The Supreme Court and the Bible belt: Tennessee reaction to the” Schempp” decision.” Midwest Journal of Political Science pp. 304–319. Boix, Carles and Milan Svolik. 2013. “The Foundations of Limited Authoritarian Govern- ment: Institutions, Commitment, and Power-Sharing in Dictatorships.” Journal of Politics 75(02):300–316. Bull, Jesse and Joel Watson. 2004. “Evidence disclosure and verifiability.” Journal of Eco- nomic Theory 118(1):1–31. Cameron, Maxwell A. 1998. “Self-Coups: Peru, Guatemala, and Russia.” Journal of Democ- racy 9(1):125–139. Carrubba, Clifford J. 2005. “Courts and Compliance in International Regulatory Regimes.” Journal of Politics 67(3):669–689. Carrubba, Clifford J., Matthew Gabel and Charles Hankla. 2008. “Judicial Behavior under Political Constraints: Evidence from the European Court of Justice.” American Political Science Review 435–452(4):109. Carrubba, Clifford J and Matthew J Gabel. 2014. International courts and the performance of international agreements: A general theory with evidence from the European Union. Cambridge University Press. Castagnola, Andrea and An´ıbal Pe´rez-Lin˜a´n. 2011. The Rise (and Fall) of Judicial Review. Cambridge University Press p. 278. 46 Cepeda-Espinosa, Manuel Jose´. 2004. “Judicial activism in a violent context: The origin, role, and impact of the Colombian Constitutional Court.” Wash. U. Global Stud. L. Rev. 3:529. Cepeda, Manuel Jose´. 2009. “The constitutional protection of IDPs in Colombia.” Judicial Protection of Internally Diplaced Persons: The Colombian Experience . Chilton, Adam S and Mila Versteeg. 2018. “Courts’ Limited Ability to Protect Constitutional Rights.” U. Chi. L. Rev. 85:293. Clark, Tom S and Jonathan P Kastellec. 2013. “The Supreme Court and percolation in the lower courts: an optimal stopping model.” The Journal of Politics 75(01):150–168. Commission, Vencie. N.d. “Opinion on the Act of the Constitutional Tribuna.”. Coppedge, Michael, John Gerring Staffan I. Lindberg Svend-Erik Skaaning Jan Teorell David Altman Michael Bernhard M. Steven Fish Adam Glynn Allen Hicken Carl Hen- rik Knutsen Joshua Krusell Anna Lu¨hrmann Kyle L. Marquardt Kelly McMann Valeriya Mechkova Moa Olin Pamela Paxton Daniel Pemstein Josefine Pernes Constanza Sanhueza Petrarca Johannes von Ro¨mer Laura Saxer Brigitte Seim Rachel Sigman Jeffrey Staton Na- talia Stepanova and Steven Wilson. 2017. “V-Dem [Country-Year/Country-Date] Dataset v7.” Varieties of Democracy (V-Dem) Project . Dahl, Robert. 1971. Polyarchy: Participation and Opposition. New Haven: Yale University Press. Democracy, Office and Governance. 2002. “Guidance for Promoting Judicial Independence and Impartiality, Revised Edition.”. Diamond, Larry, Juan J. Linz and Seymour Martin Lipset. 1995. Politics in Developing Countries. Baltimore: Johns Hopkins University. Elkins, Zachary, Benjamin Ginsburg and James Melton. 2014. “The Comparative Constitu- tions Project.”. URL: http://www.comparativeconstitutionsproject/ for Civil Rights in Israel, The Association. N.d. “Project Democracy: Fighting for the Ground Rules.”. Friedman, Barry. 2002. “The birth of an academic obsession: The history of the counterma- joritarian difficulty, part five.” Yale Law Journal pp. 153–259. 47 Ginsburg, Tom. 2003. Judicial Review in New Democracies: Constitutional Courts in Asian Cases. New York: Cambridge University Press. Ginsburg, Tom and Tamir Moustafa. 2008. Rule by law: the politics of courts in authoritarian regimes. Cambridge University Press Cambridge. Gross, Oren. 2011. Constitutions and Emergency Regimes. In Comparative Constitutional Law, ed. Tom Ginsburg and Rosalind Dixon. Research Handbooks in Comparative Law Edward Elgar Publishing pp. 334–355. Hanssen, F Andrew. 2004. “Is there a politically optimal level of judicial independence?” American Economic Review 94(3):712–729. Helmke, Gretchen. 2010. “The origins of institutional crises in Latin America.” American Journal of Political Science 54(3):737–750. Huneeus, Alexandra. 2010. “Judging from a guilty conscience: The Chilean judiciary’s human rights turn.” Law & Social Inquiry 35(1):99–135. Jackson, Vicki and Mark Tushnet. 2006. Comparative Constitutional Law. Second ed. Foun- dation Press. Linz, Juan J. 1978. Crisis, Breakdown and Reequilibration. In The Breakdown of Democratic Regimes, ed. Juan J. Linz and Alfred Stepan. Baltimore: Johns Hopkins University. Linz, Juan J. and Alfred C. Stepan. 1996. Problems of Democratic Consolidation. Baltimore, MD: Johns Hopkins University Press. Linzer, Drew A. and Jeffrey K. Staton. 2015. “A Global Measure of Judicial Independence, 1948-2012.” Journal of Law and Courts 3(2):223–256. Melton, James and Tom Ginsburg. 2014. “Does De Jure Judicial Independence Really Matter?: A Reevaluation of Explanations for Judicial Independence.” Journal of Law and Courts 2(2):187–217. URL: https://doi.org/10.1086/676999 Miklaucic, Michael. 2002. Guidance for Promoting Judicial Independence and Impartiality, Revised Edition. Technical report U.S. Agency for International Development. North, Douglass and Barry Weingast. 1989. “Constitutions and Commitment: The Evolution of Institutions Governing Public Choice in 17th Century England.” Journal of Economic History 49(4):803–832. 48 O’Donnell, Guillermo and Philippe C. Schmitter. 1986. Transitions from Authoritatian Rule: Tentative Conclusions about Uncertain Democracies. Baltimore: Johns Hopkins. Pe´rez-Lin˜a´n, An´ıbal and Andrea Castagnola. 2009. “Presidential Control of High Courts in Latin America: A Long-term View (1904-2006).” Journal of Politics in Latin America pp. 87–114. R´ıos-Figueroa, J. 2007. “Fragmentation of Power and the Emergence of an Effective Judiciary in Mexico, 1994–2002.” Latin American Politics & Society 49:31–57. R´ıos-Figueroa, Julio. 2016. Constitutional Courts as Mediators: Armed Conflict, Civil- military Relations, and the Rule of Law in Latin America. Cambridge University Press. Rodr´ıguez-Raga, Juan Carlos. 2011. Strategic Deference in the Colombian Constitutional Court. In Courts in Latin America, ed. Gretchen Helmke and Julio R´ıos Figueroa. Cam- bridge University Press. Ruhl, Mark. 2010. “Honduras Unravels.” Journal of Democracy 21(2):93–107. Scheppele, Kim Lane. 2005. “Democracy by judiciary. Or, why courts can be more democratic than parliaments.” Rethinking the rule of law after communism pp. 53–54. Schwartz, Herman. 2000. The struggle for constitutional justice in post-communist Europe. University of Chicago Press. Sikkink, Kathryn. 2011. The Justice Cascade: How Human Rights Prosecutions Are Chang- ing World Politics (The Norton Series in World Politics). WW Norton & Company. Staton, Jeffrey K, Christopher Reenock and Staffan Lindberg. 2015. “Bending but not Breaking: Rule of Law Tensions and Regime Survival.”. Staton, Jeffrey K., Varun Gauri and Jorge Vargas Cullell. 2015. “The Costa Rican Supreme Court’s Compliance Monitoring System.” Journal of Politics 77(3). Stephenson, Matthew C. 2003. ““When the devil turns. . . ”: The political foundations of independent judicial review.” The Journal of Legal Studies 32(1):59–89. Stephenson, Matthew Caleb and Justin Fox. 2011. “Judicial Review as a Response to Polit- ical Posturing.”. Svolik, Milan. 2008. “Authoritarian Reversals and Democratic Consolidation.” American Political Science Review 102(02):153–168. 49 Svolik, Milan. 2012. The Politics of Authoritarian Rule. Cambridge: Cambridge University Press. The International Center for Ethics, Justice and Public Life. N.d. “The International Rule of Law: Coordination and Collaboration in Global Justice.”. Uprimny, Rodrigo. 2003. “The constitutional court and control of presidential extraordinary powers in Colombia.” Democratization 10(4):46–69. Valenzuela, Arturo. 1978. Chile. In The Breakdown of Democratic Regimes, ed. Juan J. Linz and Alfred Stepan. Baltimore: Johns Hopkins University. Vanberg, Georg. 2005. The Politics of Constitutional Review in Germany. New York: Cam- bridge University Press. Wilson, Bruce M and Juan Carlos Rodr´ıguez Cordero. 2006. “Legal opportunity structures and social movements: The effects of institutional change on Costa Rican politics.” Com- parative Political Studies 39(3):325–351. 50