%20-%20https://sketch.com%20--%3e%3ctitle%3e3ECBBF50-FD08-412A-943D-2CFC5E025A16%3c/title%3e%3cdesc%3eCreated%20with%20sketchtool.%3c/desc%3e%3cg%20id='Stylesheet'%20stroke='none'%20stroke-width='1'%20fill='none'%20fill-rule='evenodd'%3e%3cg%20transform='translate(-926.000000,%20-668.000000)'%20id='BUTTONS'%3e%3cg%20transform='translate(100.000000,%20597.000000)'%3e%3cg%20id='BTN-/-RIGHTS'%20transform='translate(816.000000,%2065.000000)'%3e%3cg%20id='ICON-/-Rights'%20transform='translate(10.000000,%206.000000)'%3e%3cpath%20d='M13.5752841,0%20L1.34943182,0%20C0.59375,0%200,0.586666667%200,1.33333333%20L0,22.6666667%20C0,23.4133333%200.620738636,24%201.34943182,24%20L17.6505682,24%20C18.40625,24%2019,23.3866667%2019,22.6666667%20L19,5.36%20L13.5752841,0%20Z%20M13.5752841,1.89333333%20L17.0838068,5.36%20L13.5752841,5.36%20L13.5752841,1.89333333%20Z%20M1.34943182,22.6666667%20L1.34943182,1.33333333%20L12.1988636,1.33333333%20L12.1988636,5.36%20C12.1988636,6.10666667%2012.7926136,6.69333333%2013.5482955,6.69333333%20L17.6235795,6.69333333%20L17.6505682,22.6666667%20L1.34943182,22.6666667%20Z'%20id='Shape'%20fill='%23595959'%20fill-rule='nonzero'%3e%3c/path%3e%3ccircle%20id='Oval'%20stroke='%23595959'%20stroke-width='1.5'%20fill='%23595959'%20cx='9.5'%20cy='13.5355339'%20r='5.5'%3e%3c/circle%3e%3cpolyline%20id='Path-2'%20stroke='%23FFFFFF'%20stroke-width='1.5'%20transform='translate(9.500000,%2013.035534)%20rotate(45.000000)%20translate(-9.500000,%20-13.035534)%20'%20points='8%2016.0355339%2011%2016.0355339%2011%2010.0355339'%3e%3c/polyline%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
I. Introduction
Much scholarship investigates how the executive and legislative branches of government constrain the judiciary. Relatedly, scholars investigate courts’ roles as veto players under different institutional arrangements and specifically the extent to which they moderate or fuel conflict between the executive, legislature, and other actors. Constitutional courts, which often are a fourth branch of government and reside outside of the regular judicial branch, can play a moderating role and contribute to interbranch stability—they have upheld and even defined the separation of powers among political institutions. The conflict-resolution role of constitutional courts is especially crucial during regime transitions and in young democracies where a strong commitment to constitutionalism is yet to emerge and the powers of different actors are unclear and conflicting. Yet, it is by no means assured that constitutional courts will contribute to regime stability and integrity, whether due to their limited powers, lack of autonomy, or other factors. If constitutional justices are beholden to the elected branches, the regular judicial branch, or other powerful actors, constitutional courts may turn into surrogates of one actor as opposed to successful and impartial mediators (Brown and Wise Reference Brown and Wise2004).
Constitutional courts in the former communist countries have generally been a success story (Schwartz Reference Schwartz2000). They have made decisions with important financial, budgetary, and political consequences; they have frequently made decisions counter to the interests of the ruling parties and have been considered by many the primary defenders of the constitution. Yet some have been dominated by the interests of powerful executives or have fomented conflicts between the different branches of government. With democratic backsliding in countries such as Russia and populism sweeping Central and Eastern Europe, constitutional courts have become the target of illiberal parties. Some activist courts in countries until recently considered the most advanced democracies in the region, such as Hungary and Poland, have been packed with loyalists, their independence drastically reduced. Given these developments, the determination of which institutional safeguards are conducive to the autonomy of high courts and regime stability is a salient question. Such an investigation sits squarely within the literature on the costs and benefits of delegative democracy (O’Donnell Reference O’Donnell1994) and on supervisory agencies, such as prosecutors and ombudsmen, created when other institutions are not functioning properly (Moreno, Crisp, and Shugart Reference Moreno, Crisp, Shugart, Mainwaring and Welna2003). Such supervisory institutions may hold other institutions accountable, but they also must be supervised themselves.
For several reasons, this study focuses on the Bulgarian Constitutional Court (BCC), a fourth branch of the government considered outside the regular judiciary. The BCC, along with similar courts in Romania, Slovakia, and Slovenia, has become an important political veto player (Ginsburg and Bugaric Reference Ginsburg and Bugaric2016), is considered by many an important defender of democracy (see Ganev Reference Ganev2003), and has remained a well-respected institution. While it is a powerful court (see Schwartz Reference Schwartz2000; Sadurski Reference Sadurski2010), the BCC has not been the victim of constitutional crises that have curtailed other Eastern European courts’ powers. The Bulgarian case merits further study as, institutionally, Bulgaria has a mix of institutions borrowed from established Western European democracies and the Soviet Union. For instance, Bulgaria, similar to most Southern European states, introduced in 1991 a Supreme Judicial Council (SJC) to select, promote, and dismiss judges in the regular judiciary. The SJC is partially composed of judges elected by their peers and serves as the governing body between the government and the ordinary courts, established to insulate the judiciary from parliamentary majorities.
However, Bulgaria’s transition to democracy and related institutional reforms allowed the continuance (with only modest reforms) of some institutions inherited from the communist period, namely, the procurator general (PG) with strong prerogatives, following the traditional Soviet-style procuratura model. The Venice Commission, an advisory body to the Council of Europe and an internationally recognized authority in the field of constitutional law, concluded that the “over-powerful” Bulgarian procuracy may threaten judicial independence (Venice Commission 2008). The Bulgarian public views PGs as central figures in the administration of justice, and the media reports on them almost daily. There is a widespread perception that the office of the PG has been used for economic repression and political blackmail, which is reflected in the low approval ratings of all PGs (Svobodna Evropa 2019). Given such dynamics, it is important to investigate whether the PG can subvert judicial decisions including those before the BCC.
Naturally, after the collapse of communism, the countries of Central and Eastern Europe looked west for successful institutional and policy precedents to emulate. However, the interactions between traditional and nontraditional institutions have created some unexpected consequences and at times compromised the independence of the BCC. The activism of the BCC has been compared to that of the Hungarian Constitutional Court, which until constitutional reforms in 2011 was considered one of the most independent in the former communist countries (Schönfelder Reference Schönfelder2005; Sheppele Reference Sheppele2011). Many Western observers praised the BCC for its strong reputation after the democratic transition (Schrameyer Reference Schrameyer1993; Schwartz Reference Schwartz2000). From 1991 to 1997, when the country was ruled by mostly left-wing governments, the BCC, dominated by judges from the center right, blocked many attempts of the government to reverse market-oriented reforms (Smilov Reference Smilov2009). Critics, however, have at times accused the court of instigating conflict between the judiciary and the executive and siding with the ordinary judiciary on key cases to increase its own influence (Popova Reference Popova2010). A number of its decisions have been controversial, and nongovernmental organizations (NGOs) and several constitutional justices have indicated that politics influence case decisions.Footnote 1 Some experts have argued that the “super independence” of the Bulgarian judiciary and the BCC have made them “self-serving and corporatist” (Smilov Reference Smilov2009, 5).
While the overall impact of the BCC on Bulgarian democracy and stability and an evaluation of the court’s independence is beyond the scope of this research, we focus on the following questions: Do certain political actors, and in particular the PG, influence the voting decisions of constitutional justices? What other political factors drive BCC judges’ constitutional review decisions? By answering these questions, the study has implications for interbranch relations and the role of PGs in constitutional review in transitional contexts. Whether the Bulgarian constitutional justices are susceptible to political influence from powerful political actors, and in particular the PG, has important implications. It sheds light on judicial autonomy, which is crucial for the stability of Bulgarian democracy but also helps us evaluate whether the Kelsenian model of constitutional review, vested in a constitutional court separate from the judiciary, has served its intended purpose—to act as a firewall, insulating the regular judiciary from political questions. While constitutional courts may mediate institutional conflicts, if institutions are fragile and lack a history of autonomy, which is often the case during times of transition, the executive, the legislature, or other important actors, such as the PG, might attempt to influence the decisions of the court.
In this article, we examine the relevant institutional incentives in the Bulgarian context, review the comparative work on judicial decision-making, and derive testable implications about the impact of the Bulgarian PG and other interested parties on the voting decisions of Bulgarian constitutional justices. We test our theory employing original data consisting of 2,324 individual judges’ votes on 206 constitutional review decisions before the BCC between 1992 and 2012 and a model that takes into account the interrelatedness of the decisions of the PG and the justices. We find that, as predicted, the powerful Bulgarian PG affects judicial voting, as do judges’ alignments with the executive and the opinions of interested parties.
II. The Role of Political Actors on Constitutional Court Judges’ Decision-Making
Judicial scholars argue that judges are affected by the strength of other political actors, such as the executive and congress and powerful prosecutors as well as nongovernmental groups. A significant body of literature examines the influence of the executive branch on high courts in the United States and in comparative contexts (see Helmke Reference Helmke2002; Iaryczower, Spiller, and Tommasi Reference Iaryczower, Spiller and Tommasi2002). In Bulgaria, the executive and the powerful PG are particularly influential. Although the power of the PG is similar to that of the US solicitor general, who argues before the US Supreme Court, the influence of Bulgaria’s PGs comes from their unique position in the Bulgarian political system. The Bulgarian PG is an independent agent but is located within and a part of the regular judicial branch, formally independent of the executive branch with expansive powers. The PG also heads the National Investigative Service and “can transfer to it, investigations at its discretion” (Open Society Institute of Sofia 2019, 37).
To clarify the strong prerogatives of the Bulgarian PG and other political actors as well as the incentives faced by BCC justices, we provide the political and institutional context. The 1991 Bulgarian Constitution endowed the judiciary and also the BCC with extensive powers. Like many constitutional courts, the BCC has a posteriori abstract constitutional review, among other powers listed in the constitution (Art. 149; Smilov Reference Smilov2009, 10). Similar to many European countries, Bulgaria adopted a mixed system for appointing its judges, whereby different branches of government each unilaterally appoint a fixed number of justices. The National Assembly, the president, and an assembly of the Supreme Court of Cassation (SCC) and the Supreme Administrative Court (SAC), the country’s top appellate and administrative courts, choose four judges apiece. This distribution of appointments across branches is thought to enhance judicial independence and make “it virtually impossible for a single political force to colonize the Court” (Ganev Reference Ganev2003, 607). Judges on the BCC must have high professional standing and moral integrity and must have practiced law for 15 years (Const., Art. 147[3]).
Unlike the US Supreme Court, the BCC does not conduct constitutional review through cases and controversies whereby injured parties appeal from lower-court decisions. Rather, it reviews the constitutionality of laws in the abstract referred to it by specific political actors, namely, the president, no less than one-fifth of all members of the National Assembly, the Council of Ministers, the SCC and SAC, the PG, the ombudsman (since 2006), and the Supreme Bar Council (since 2015). While only important political actors may bring cases to the court’s attention, the decision-making process by the BCC is enhanced by allowing the participation of amicus curiae or “interested parties” briefs (provided by a wider range of actors, including NGOs), which the court may request. Amicus curiae are not unique to Bulgaria but exist in many European high courts (Stone Sweet Reference Stone Sweet2000; Ganev Reference Ganev2003; Garoupa, Gomez-Pomar, and Grembi Reference Garoupa, Gomez-Pomar and Grembi2013).Footnote 2 The PG can refer any case to the court or opine its position as an interested party, but the PG cannot do both on the same case, and its appearance on a case is not mandatory.
Several institutional safeguards insulate BCC judges from political pressure. The court has 12 judges who serve nine-year nonrenewable terms. Judges’ terms are staggered such that one-third of the court is replaced every three years. The court makes decisions using a majority voting rule requiring that more than half of the judges (i.e., 7 of 12 judges) agree. For most decisions, nine judges must be present for a vote to meet the quorum requirement. Abstentions are not allowed. Judges who disagree on abstract review cases issue “dissenting opinions,” which are in the public record.
BCC judges are largely self-disciplined. Few circumstances require a judge to resign, and the BCC itself decides under what conditions. A pretrial criminal investigation can be initiated against a constitutional judge, but only after two-thirds of the judges vote in favor of lifting immunity after the PG provides the BCC with sufficient evidence of a premediated crime (Const. Court Act, Art. 9). The court is financially independent from the executive with its own independent budget, which it proposes and only the National Assembly can amend. After service on the bench, many judges go on to prestigious careers in academia, the regular judiciary, politics, or international organizations. See supplemental material (SM) 1 for information on the BCC judges. Next, we discuss the influence of other political actors on the BCC.
The Supreme Judicial Council
To protect the independence of the ordinary judiciary, Bulgaria, like most other Southeast European states, established the SJC as the coordinating body between the government and the ordinary courts under its post-democratic-transition constitution. The SJC appoints, promotes, transfers, demotes, and removes judges, procurators, and investigating magistrates within the regular judicial branch.Footnote 3 Although formally independent, the SJC is often at the center of political conflicts. Every ruling party in postcommunist Bulgaria has tried to control the SJC, and according to many observers, including prominent members of the judiciary, the SJC is not insulated from outside influence. In several cases, new governments have changed laws in order to allow the replacement of members of the SJC with those friendly to the new government, with the expectation that these newly appointed members would return the favor later (Center for the Study of Democracy 2010, 214).
The method of selection of the SJC has been criticized multiple times by the Venice Commission as it undermines the SJC members’ independence (see Venice Commission 2008, 2009). The SJC has 25 members, with the two supreme courts and the PG as members ex officio. The National Assembly and judiciary each elect 11 members. Before the 2015 judicial reforms, 6 of the 11 members from the judicial quota were elected by judges, four by procurators, and one by the investigators.Footnote 4 All decisions of the SJC, before and after the reforms, were made under simple majority rule with a quorum requirement of one-half of the members. The joint staffing at the level of the SJC enhanced the control of the PG over the judiciary. In effect, procurators and judges on the SJC jointly appointed and removed themselves, which is certainly an unusual feature of the organ by comparative standards. The joint appointment process was criticized as inappropriate as it threatens the independence of both regular judges and procurators (Penev Reference Penev2014, 206), and its reform was recommended by the Venice Commission (2008, 2009).
While a 2015 judicial reform split the SJC into a judicial collegium and procuratorial collegium to prevent joint staffing decisions by judges and prosecutors, it de facto increased the control of the PG over the procuratorial collegium and increased the parliamentary parties’ pressure on the members of the judicial collegium. Because after the reforms only six members in the judicial collegium were selected by the ordinary judiciary and the heads of the two supreme courts, the members selected by judges remained a minority even in the judicial collegium (six members) relative to the members selected by the parliament (eight members). By contrast, after the reforms, in the procuratorial collegium the number of members selected by prosecutors and investigators (five members) is equal to the members selected by the parliament. This turns the PG into the decisive voter in the procuratorial collegium, who can dominate staffing and other important decisions. Due to the PG’s de facto majority, the SJC has largely been dominated and continues to be dominated by the PG, a position with very strong prerogatives in the Bulgarian political system, established below, including the ability to request that the SJC remove any prosecutorial or regular judicial officer it proposes.
Although established in 1991 and reformed in the 2000s and again in 2015, which is beyond the scope of the data analysis, SJC reforms reveal the shortcomings of the SJC since its creation. The Venice Commission (2008, para. 34) criticized the SJC, not because it was one body composed of both prosecutors and judges per se but because it did not appear “to maintain the distinction as to function and power between public prosecutor and judge.” Specifically, the Venice Commission (2008, paras. 25, 39) criticized the pre-2015 setup of the SJC because “the prosecution should have no involvement whatsoever in the ultimate administration of justice, nor in the appointment and functioning of judges, and the operation of the court system,” and likewise allowing judges to choose prosecutors inhibits this latter groups independence as well.
The Procurator General: “Only God Is above Me”Footnote 5
There is little empirical work outside of the US example that explains how prosecutors influence judicial decision-making on constitutional matters (see Voigt and Wulf [Reference Voigt and Wulf2019] on criminal matters). The Council of Europe’s Consultative Council of European Prosecutors (CCPE) has closely scrutinized PGs in Europe due to their role in noncriminal cases, including their authority to enter into cases involving constitutional issues, possibly violating the principles of separation of powers or acting in a way that is not “on behalf of society and in the public interest” (CCPE 2008). The CCPE has further scrutinized procurators who are located within the judiciary with complete autonomy and legal and nonlegal functions, such as the Bulgarian PG. In contrast to some other Western systems, the Bulgarian Constitution placed the PG within the judiciary to avoid political influence from the executive or the legislature.
Prerogatives of the Bulgarian PG
After its democratic transition, Bulgaria introduced some radical political reforms but also preserved some Soviet-type institutions. In so doing, it made changes to the prerogatives and placement of the PG, such as locating it within the judicial branch (Trendafilova-Batchvarova Reference Trendafilova-Batchvarova1997, 134), but the PG continued its strong supervisorial role over the judicial system and criminal investigations and generally the execution of punishment and coercive measures.Footnote 6 The 1991 Constitution retained the right of prosecutors to order special investigative measures such as the authority to arrest, search and seize homes, and issue warrants (Trendafilova-Batchvarova Reference Trendafilova-Batchvarova1997, 144). Bulgaria kept the Soviet-style prosecutor as it was seen as independent from other political actors, a defender of citizens’ rights, and a necessity in a new democracy and because it had a significant role in the criminal justice system, which had procedures that themselves were not amended until after the mid-2000s (144).
Since Bulgaria’s democratic transition, the PG is appointed by the president, who is the head of state and has important prerogatives but is not the chief executive in Bulgaria’s parliamentary system, having mostly a representative role. The PG serves for seven years, a mandate longer than most other political actors. The PG may refer constitutional cases to the BCC or enter cases as amicus curiae—not unique from a comparative perspective. A 2006 constitutional amendment attempted to make the PG more accountable, requiring the PG to submit an annual report to the National Assembly and the SJC. However, such changes did not increase the procurator’s accountability, as noted by the European Court of Human Rights in Kolevi v. Bulgaria (Decision of the European Court of Human Rights, No. 1108/02, para. 135 [2009]).
Bulgarian procurators have the right to obtain information and documents, carry out inspections, summon citizens or legal entities without warrants, release detained individuals without court intervention, and allow or block certain parts of an investigation (Judiciary System Act, Art. 145). Constitutional amendments in 2003 allow prosecutorial liability for nonofficial or official behavior constituting an intentional crime of a public nature. Authorization from the SJC, over which the PG (as shown above) has tremendous influence, is required for charging or detaining a prosecutor himself, unless the prosecutor is caught in the act of committing the crime. Essentially, the PG is immune from criminal prosecution and virtually irremovable (Const., Art. 69, 70, 132).Footnote 7
Empirical Implications
The PG is not controlled by any of the branches of government, and it is unclear whether any political figure can hold the PG accountable. This, however, does not entirely free the PG from informal political pressure from the other branches of government, and specifically the executive. PGs have generally been perceived as agents of the executive—the preferred choice of the party in power and possibly other smaller, pivotal parties.Footnote 8 The politicization of the PG selection process and its susceptibility to influence from the executive was publicly acknowledged by Prime Minister Boyko Borisov, who shared his frustration with having a PG “selected” by one party while having his deputy “selected” by another. Bulgarian PGs, however, have not always been compliant agents of the executive. While the preferences of the executive are an important determinant of PG and judicial behavior that merits further investigation, the preferences of the PG and those of the executive or the party in power should not be conflated.
According to a number of observers, once elected, omnipotent PGs have turned against governments that supported their candidacy (Hein Reference Hein2006; Kalaydzhieva Reference Kalaydzhieva2015). The PG has become a political force in its own right, often confronting politicians through criminal investigations that may damage their careers even if no conviction follows.Footnote 9 The PG’s strong prerogatives have allowed former PGs to shield narrow interests and criminal groups, which has been alleged by a number of high-level procurators, officials, and politicians (Mediapool 2002). Given the hierarchical structure of the procuracy and its extensive powers, it is possible for rent-seekers to influence judicial decision-making through the PG. For example, the first PG, Ivan Tatarchev (1992–99), allegedly connected to organized criminal networks, allowed chronic delays of criminal cases. Bulgaria’s second PG, Nikola Filchev (1999–2006), protected family members and blocked salient criminal cases (Hristov Reference Hristov2002; Center for the Study of Democracy 2010).
As far as the BCC, the PG influences its justices’ votes in a number of ways. First, as mentioned above, the PG can initiate a pretrial investigation against any judge, which could potentially prompt lifting the judge’s immunity. The decision to bring charges or to initiate a pretrial investigation is not subject to judicial review, and the PG does not need to have strong evidence against the targeted judge—that will be required later when the case is sent to a trial court with a bill of indictment. Before the preliminary investigation is completed, there is no actual decision by the PG to prosecute. For this reason, disclosing the names of people subject to preliminary investigation constitutes a violation of their rights and violates the law. Despite that, this information has been leaked by the procuracy multiple times, tarnishing the reputation of judges, members of parliament, and politicians and possibly hurting their career prospects based on often little to no evidence. There is a widespread perception among experts and the public that the “bringing of charges” and the preliminary investigations are the most significant instrument of unchecked prosecutorial power used against politicians, members of the government, and magistrates.Footnote 10 Unlike judges from the judiciary, members of parliament, or other magistrates, BCC judges targeted by the PG’s pretrial investigations against them have not become public information.Footnote 11 Despite this, the ability of the PG to initiate investigations, and importantly to initiate pretrial investigations or to bring charges, can be a powerful deterrent and discourage constitutional justices from alienating the powerful PG.
Second, the PG may potentially exert pressure on constitutional justices in a less direct way. The inclusion of the PG as an ex officio member of the SJC may deter judges and undermine their impartiality (Venice Commission 2017). While procurators and investigators on the SJC are dependent on the procuracy and tend to vote unanimously, judges—being a more heterogeneous group due to their diverse appointers—tend to split their vote, which undermines their influence on the SJC (see Kapital, December 13, 2003). As a result, the PG has been able to dominate voting on the SJC, especially related to decisions to suspend judges or lift their immunity.
This influence of the PG matters because a number of former BCC judges continue their judicial careers by joining the judiciary or international courts. Another 25%–30% of judges continue with careers in the National Assembly or politics more generally (Open Society Institute 2005). Given the long tenure of the PG and its control over the SJC, the governing body of the judiciary, BCC judges are incentivized to avoid alienating the PG, especially if they plan on continuing with careers in the judiciary or politics, which many do, as documented in SM1.2. Such incentives are strengthened by the relatively short staggered terms of BCC judges and the fact that if they move to the regular judiciary, the PG can target them with an investigation and possible suspension, of which there are precedents by the PG against regular judges and a significant number of politicians (see n. 11). Given the arguments above, we expect:
Hypothesis 1. Judges are more likely to vote for declaring a law unconstitutional if the procurator general refers a case to the court or files a brief claiming that a law is unconstitutional.
Other Political Actors
Other political actors may influence judges’ individual decisions as well. Specifically, judges’ alignment with the political preferences of the governing coalition that enacts the law under review as well as the opinions of interested parties should influence judges’ decisions. Judges may defer to legislators with whom they are politically aligned, as established on several high courts (see Segal and Spaeth Reference Segal and Spaeth2002; Amaral-Garcia, Garoupa, and Grembi Reference Amaral-Garcia, Garoupa and Grembi2009; Garoupa et al. Reference Garoupa, Gomez-Pomar and Grembi2013; Pellegrina and Garoupa Reference Pellegrina and Garoupa2013), for two main reasons. First, judges may defer to political allies in the legislature because they are concerned that those actors may retaliate against them or the court in which they serve (McCubbins, Noll, and Weingast Reference McCubbins, Noll and Weingast1995)—a reasonable concern, as witnessed by the treatment of judges recently in Poland and Hungary, whose powers and positions have been drastically altered when they ruled against the governing coalition (Human Rights Watch 2013; Santora Reference Santora2018). Second, judges may favor the policy passed by political party allies in the governing coalition because they agree with that or to demonstrate loyalty to influential party members.Footnote 12 Political actors may help judges who have been party allies to secure both political and legal positions after their relatively short terms on the BCC (see Garoupa and Ginsburg Reference Garoupa and Ginsburg2015; SM1.2 for examples of some judges’ post-BCC posts). Given this, we predict:
Hypothesis 2. Judges aligned with parties in the governing coalition, which enacted the law under review, are less likely to vote to declare a law unconstitutional than nonaligned judges.
Just as judges may vote in a way that favors powerful government actors or those with whom they are politically aligned, judges also will be attentive to the interests of other actors such as interested parties who closely monitor the work of the BCC. Judges will listen to such interests especially when they raise concern about the constitutionality of a law being reviewed by the court. The presence of interested party briefs indicating that the law is unconstitutional informs the court that it “has outside support by a political actor who has an interest in monitoring the response to the courts’ decision if the court notes a constitutional violation” (Vanberg Reference Vanberg2005, 103). Further, these same public actors are in a position to draw attention to a court’s decision to strike down a law if the government “attempts to evade” such a court ruling, and in this way these public actors can serve as “allies” to the court when it strikes down a statute. A growing proportion of interested parties opining unconstitutionality allow the court to gauge the strength of its allies. Additionally, an increase in the percentage of briefs suggesting that a law should be stricken or upheld might provide a crude measure of public opinion on an issue (Kearney and Merrill Reference Kearney and Merrill2000). Because amicus curiae briefs target particular issues, they allow justices to make precise calculations about public opinion, unlike more general public opinion polls (Collins Reference Collins2004). Given the above arguments, we predict:
Hypothesis 3. Judges’ propensity to declare laws as unconstitutional will increase as the average number of interested parties who opine unconstitutionality increases.
III. Data and Research Approach
The data consist of 2,324 individual judges’ votes on 206 constitutional review decisions before the BCC from 1992 to 2012. A decision represents the BCC’s opinion on the constitutionality of an entire law or part of a law.Footnote 13 Most decisions analyzed here deal with the constitutional review of laws that the parliament in power at the time of review enacted. The dependent variable is a binary variable indicating whether a judge voted to find a law or part of a law unconstitutional.
The data on judicial decisions came from the BCC’s website, which makes all the court’s decisions available in Bulgarian, and were assembled with the help of several Bulgarian research assistants. The data include most of the constitutional review cases in our time period, excluding the cases that lacked information on variables of interest—the opinions of the interested parties as well as the decisions taken during interim governments. Interim governments served very short time periods and generally did not enact major policy changes. The BCC heard a very small number of cases under interim governments, and in these periods there are no reliable measures for the government composition supporting the prime minister. We also exclude cases existing under the other powers of the court, such as 68 interpretative decisions that seek to interpret the normative sense of constitutional provisions and do not seek to review the constitutionality of laws enacted by the National Assembly.
The analysis includes both judge- and case-level variables. Judge-level variables include judges’ alignment with the governing coalition, sex, prior political background, and tenure. Aligned Judge indicates whether the judge is politically aligned with the governing coalition at the time of constitutional review. Because, as stated in the previous section, sometimes the PG might act as an agent of the executive, this variable also controls for the influence of the executive and the party in power and allows us to disentangle the effects of the PG that are not due to executive pressure. For judges’ ideological associations, we independently researched the background of each judge from newspaper and scholarly articles. A full description of judges’ attributes and how their alignment was coded are found in SM1.1 and SM1.3.
Female Judge indicates the sex of the judge. Several studies of US federal courts indicate that female judges decide issues differently from male judges (see Boyd, Epstein, and Martin [Reference Boyd, Epstein and Martin2010] for this proposition and references to works with similar findings); however, this has not been established universally (George Reference George2001). For the time period of our analysis, there have been 10 female constitutional court judges on the BCC out of a total of 42 judges, necessitating the inclusion of this variable.
Judge Tenure is the number of years a judge has been on the BCC at the time of the decision, created by subtracting the year the judge was appointed to the court from the year of the decision. Judges who have been on the court only a short time are more likely to have been chosen by the sitting president or National Assembly and more willing to show loyalty to that appointer. Further, judges may exhibit freshman effects, which may make them hesitant to disagree with the majority, as have been found on the US courts of appeals (see Hettinger, Lindquist, and Martinek Reference Hettinger, Lindquist and Martinek2003) but have not been found to block voting on the US Supreme Court (see Bowen and Scheb Reference Bowen and Scheb1993). Judges nearing the end of their terms may be more eager to use their votes as a signal that will positively influence their careers after the bench (Magaloni Reference Magaloni, Mainwaring and Welna2003).
The variable Judge Prior Political Background refers to the judges’ prior service in politics. Judges coming from politics might be motivated by political concerns. By contrast, former professional judges and law professors may internalize the norm of judicial independence (Sternberg, Shikano, and Sieberer Reference Sternberg, Shikano and Sieberer2016) and may be more likely to insulate themselves from “popular sentiments” (Sisk, Heise, and Morris Reference Sisk, Heise and Morris1998, 1477). Prior careers in a legal environment, rather than politics, may encourage “organizational loyalty” to the constitutional court, making judges view disagreement among judges as “irresponsible” and undermining judicial ethics (Ulmer Reference Ulmer1970, 582).
While judges’ attributes are important determinants for judges’ individual decisions, case-level variables, denoting the position of the PG and that of interested parties, also influence judicial choice. The information about the PG’s role, if any, in the case as well as the actors filing amicus briefs and their positions regarding the constitutionality of the law were all found within the case decision from the BCC website. To determine whether justices are likely to follow the PG, the analysis includes two variables: PG Refer, which is coded 1 if the PG referred the case to the court and 0 otherwise; and PG as IP Unconstitutional, which is coded 1 if the PG appeared on a case as an interested party and submitted a brief opining unconstitutionality. When PG as IP Unconstitutional is coded 0, this means that the PG explicitly opined that the law was constitutional or that the PG does not appear as an interested party on the case.Footnote 14 The PG either refers a case or appears as an interested party and submits a brief. These two actions are mutually exclusive.
The BCC offers another methodological advantage in that it reports on the briefs submitted by amicus curiae (i.e., specific governmental bodies and agencies as well as NGOs or, in certain cases, individuals) for each case it reviews. Judges’ decisions might also follow the positions of actors opining that a law is unconstitutional that they reveal in interested party briefs. An increase of such briefs indicates the strength of the court’s allies if it decides to strike down a law. Further, these same allies have an interest in calling attention to a court’s decision if actors in government attempt to evade the court’s decision. The participation of amicus curiae or “interested parties” through the filing of briefs is not procedurally unique to Bulgaria, but many European courts choose not to reveal them. Amicus curiae that suggest that a law is unconstitutional provide a measure for public support for the court if it also opines a law is unconstitutional. As such, judges are likely to account for the proportion of interested party briefs that support a finding of unconstitutionality (see Collins Reference Collins2004; Vanberg Reference Vanberg2005). The analysis includes a variable, Percent IPs Unconstitutional, which is the proportion of interested party briefs in favor of unconstitutionality, calculated as the sum of interested party briefs in favor of unconstitutionality divided by the number of all interested party briefs, excluding the opinion of the PG if one was submitted.Footnote 15
The analysis includes other important contextual variables affecting judges’ decision-making that were created by reviewing outside sources. PG Tenure refers to the number of years the PG has been in office at the time of the BCC’s decision. This variable is included because judges and courts may follow the PG, as do US solicitor generals, because of their experience as litigators and repeat players (see McGuire Reference McGuire1995; Black and Owens Reference Black and Owens2012). As the number of cases the BCC hears per year is relatively constant and the PG’s term is fixed, the number of years the PG has served provides a proxy for its experience (see Wohlfarth Reference Wohlfarth2009).
Additionally, decisions of judges and high courts have been shown to fluctuate with the political composition of the polity itself, such that judges act more assertively or independently under divided or fragmented government and more deferentially under unified government (Gely and Spiller Reference Gely and Spiller1990; McCubbins et al. Reference McCubbins, Noll and Weingast1995). While this finding has been sustained when evaluating decisions on a number of courts including in nondemocracies (see Epperly Reference Epperly2017), it is not always influential (see Herron and Randazzo Reference Herron and Randazzo2003; Hilbink Reference Hilbink2012). To control for the political strength and unity of the government, the analysis includes the variable Government Strength, which captures the percentage of legislative seats the party or parties in the governing coalition have in Bulgaria’s unicameral parliament (see SM2.1 for this coding and its sources).
Some of the regressions also include two case-level variables. Complex Case refers to whether the case deals with a particularly complex policy area in which special legal acumen may be necessary. As laws increase in complexity, judges on the BCC may rely more on opinions of experts outside of the court. Following Vanberg (Reference Vanberg2005, 104), a case is considered complex if it involves “economic regulation, state-mandated social insurance (unemployment, health, and retirement insurance), civil servant compensation, taxation, federal budget issues and party finance.” Complex cases involve laws that are very detailed and require a high level of knowledge to understand. Another case variable, Separation of Powers, refers to whether a case involved an issue that effectively aggrandizes or diminishes the powers of one branch of government over the other, which according to Shapiro (Reference Shapiro, Gloppen, Gargarella and Skaar2004) is an important area of review for constitutional courts. These laws usually involve reforms to the structure and powers of government institutions. SM2.2 provides examples of both types of cases.
Finally, the composition of the entire court (which varies by case due to quorum rules) and the number of judges aligned with the governing coalition influence both the judges’ and PG’s decision-making calculi. As noted by Rogers (Reference Rogers2001) and Brouard and Hönnige (Reference Brouard and Hönnige2017), an important determinant of whether courts strike down a law as unconstitutional is the political composition of the full court in relation to the legislature. For individual judges, who may be torn between following the PG or the government in power when they disagree, knowing the political orientation of other judges with whom they vote as well as the strength of the full court’s political alignment with the legislature whose laws it reviews are important considerations. As shown by literature on panel effects (Fischman Reference Fischman2015), individual judges tend to change their votes in agreement with judges with other political orientations as the latter group increases in size. Thus, judges should be more likely to defer to the legislature when the number of aligned judges on the court increases.
Likewise, the political orientation of the entire court is a consideration for the PG (see Wohlfarth Reference Wohlfarth2009). In deciding whether to opine a law as unconstitutional, the PG may anticipate and be constrained by the court’s likely position on a case of constitutional adjudication based on its political composition. In its own decisions, the PG will factor in whether the court will be more or less likely to defer to the legislature.
To capture the full court’s political orientation, the variable Percent Court Aligned with Governing Coalition measures the percentage of judges who are aligned with parties that are within the governing coalition. The percentage aligned took the number of judges who were aligned with the government coalition and divided it by the number of judges on each case. Appendix SM3.1 provides summary statistics for all the variables.
Disentangling the independent political influence of the PG requires additional analysis because if the results indicate that judges vote in line with the preferences of the PG, this influence may not necessarily be political; it could be purely judicial. That is, the PG could be choosing to opine unconstitutionality when the case against a law is legally meritorious, in much the same way that US Supreme Court justices follow solicitor generals because their positions have legal merit due to the their professionalism and the quality of their legal opinions (Jenkins Reference Jenkins1983; see Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Black and Owens Reference Black and Owens2012).Footnote 16 Solicitor generals also have a built-in advantage due to their ability to select cases that they know they will win (Jenkins Reference Jenkins1983; Black and Owens Reference Black and Owens2012).
Further, the latent factor, legal merit, might be driving the decisions of both the PG and the justices. Therefore, failure to account for the endogenous nature of the decision of the PG as well as for the effect of unobserved factors such as legal merit would overestimate the influence of the PG. Although factors such as the legal merit of the case might affect the decisions of both the PG and the justices, their relative influence might vary by decision maker because the justices and the PG might have different opinions about legal merit or different goals or views on the function of the laws. While we do not have a direct measure of legal merit, we employ statistical methods that take into account latent factors that affect the decisions of both the PG and the justices.Footnote 17
A model of dichotomous choice examines the effect of relevant factors on the decision of the justices to rule in favor of unconstitutionality and the decision of the PG to opine unconstitutionality. Given the interrelatedness of the decisions of the PG and the judges, these decisions should not be modeled as independent but rather as related. Estimating the two models separately ignores the correlation between the error terms in the two models, which would result in inefficient estimates. A bivariate probit model would address this and account for unobserved factors such as legal merit that influence both decision makers. A single-equation model would fail to capture the total effects of factors that shape both decisions.Footnote 18 As noted by Kimball (Reference Kimball2006, 323), single equations with their own dependent variables fail “to capture the total effects of the factors that shape both outcomes. Econometrically, this misspecification is a form of omitted variable bias, which violates the assumption of zero correlation between the independent variables and error terms. Subsequently, the estimated error term absorbs the effect of the excluded variable and results in parameter estimates that are biased and inconsistent” (see also Hartzell, Hoddie, and Bauer Reference Hartzell, Hoddie and Bauer2010).
The bivariate probit model assumes that the random error terms in the equations are correlated and that the covariance of the random error terms is a constant, ρ, rather than zero (as in univariate probit models). If legal merit or other factors that influence the decision of the PG to opine unconstitutionality are not randomly distributed across the population of cases, then the error terms in the two equations will be correlated, suggesting that the same unobservables that affect the PG’s position also influence the voting decisions of the justices. If a correlation between the error terms is identified, the bivariate probit model corrects for the correlation, eliminating the effects that the nonrandom selection of cases have on the voting decisions of the constitutional justices. The remaining difference in the probability that justices vote for unconstitutionality in cases where the PG opines unconstitutionality and those where the PG does not can be attributed to the expressed opinion of the PG in favor of unconstitutionality.
There are other dependencies in the data that, if neglected, might lead to incorrect estimates of standard errors—the same justices voting on different cases and the votes being nested within cases. The data reveal that within-judge dependency is weak (intraclass correlation coefficient of .03) but within-case dependency is significant (intraclass correlation coefficient of .61), suggesting that we cannot ignore within-case dependency. Accounting for possible intracase correlation, standard errors are clustered by case in our main models and some of our robustness checks. Failure to cluster may lead to underestimated standard errors and incorrect inferences.
IV. Results
Case characteristics might constrain justices voting on the same case. While we take into account some case characteristics such as complexity, some of the relevant characteristics are unobserved. To take into account such unobserved characteristics of the cases, we employ a bivariate probit model. Table 1 presents our main bivariate probit model with case-level clustering of standard errors to account for the dependence between votes on the same case. In table 1, three separate probit regressions are analyzed. The first regression analyzes factors related to judges themselves, the PG, and the position of interested parties. The second regression adds covariates about the strength of the government and features of the laws reviewed, and the final regression adds a variable regarding the courts’ political alignment with the governing coalition. The table includes coefficients for each variable and the standard errors. In the text, when describing the impact of a variable on judges’ votes, we refer to marginal effects listed in terms of percentages. The marginal effects indicate the effect on the dependent variable moving the independent variable in question from 0 to 1 if a binary variable and from the minimum to the maximum value if a continuous variable is being analyzed.Footnote 19
Determinants of Judges’ Decisions Finding Laws Unconstitutional (Bivariate Probit)
Note.—Robust standard errors in parentheses. N = 2,324; PG = procurator general; IP = interested party.
a Excludes the position of the PG as an IP.
* p < .10.
** p < .05.
*** p < .01.
As suggested above, a bivariate probit model is more appropriate for this analysis as it accounts for the interrelatedness of the two decision makers (the judges and the PG) and allows for the possibility that both decisions are affected by latent factors, such as legal merit. The Akaike information criterion (AIC) and Bayesian information criterion (BIC) for the bivariate probit regressions, found in table 1, indicate that the first regression is the simplest and best fitting one. The results of the second part of the first regression (table 1, col. 2) are depicted in figure 1. Although the AIC is slightly smaller for regressions 2 and 3, the BIC is much lower for the first regression. Across regressions, the estimated correlation coefficient, or ρ, is large and significant at the .05 level in regression 1 (the best fitting model) and at the .10 level for the remaining regressions. This significant and large ρ justifies the use of the bivariate probit model and indicates that the decisions of judges and the PG are indeed correlated. The results of the second stage reflect the selection-corrected effect of the PG positions on justices’ votes.
Average marginal effects with 95% confidence intervals for the second equation of the bivariate probit (table 1, col. 2). Dependent variable is individual judges’ decisions of unconstitutionality: a, individual-level determinants; b, case-level determinants.
Failing to use a bivariate probit regression would result in incorrect inferences regarding the PG’s influence on the judges’ decisions. Here, the single-equation model would have underestimated the influence of the PG on judges’ decisions. This interdependence between the decisions of the judges’ and prosecutors has seldom been tested (see Bailey, Kamoie, and Maltzman Reference Bailey, Kamoie and Maltzman2005; Wohlfarth Reference Wohlfarth2009; Black and Owens Reference Black and Owens2012). However, Zorn (Reference Zorn2002, 163) argues that modeling the solicitor general’s decisions more specifically is crucial for understanding the actual mechanism by which the solicitor general influences the US Supreme Court. We address this important issue in the Bulgarian context. Focusing on the first part of the bivariate probit regression in table 1 (corresponding to col. 1), the PG is influenced by case factors, such as the complexity of the case and the issue area. Further, the PG is concerned with the strength of the government it is opposing when opining a law is unconstitutional, a factor that does not influence individual judges when it appears later in regressions 2 and 3 (i.e., cols. 4 and 6, respectively). The results also show that the PG responds to the position of interested parties.
As to individual factors influencing judges’ decisions of unconstitutionality in this second part of the bivariate probit regression (see table 1, col. 2, and fig. 1), these are substantially consistent with the single-equation probit models in SM4 as to coefficient sign and significance. Aligned judges are 14% less likely to find laws unconstitutional, but judges’ prior political background is not significant in the bivariate model. As to the influence of the PG as an interested party and other amicus curiae, these are both significant in all the bivariate regressions, and the inclusion of these variables improves the fit of this model as confirmed by a Wald test.Footnote 20 Judges are 56% more likely to find a law unconstitutional when the PG opines unconstitutionality.Footnote 21
In addition to the main results, we conducted robustness checks found in the appendix that are consistent with the main results in table 1. First, the results for probit models using fixed effects at the judge level and additional regressions with alternative variables for the PG’s positions appear in table SM5.1.
Results for a nonrecursive bivariate probit model in which a direct relationship between the behavior of the justices and the PG is not posited appear in table SM5.2. In this nonrecursive model, the dependent variable for the first equation is still whether the PG opines unconstitutionality, but this variable does not appear as an independent variable in the second equation regarding judges’ decisions. The sign and significance of the variables of interest are consistent with the results from the recursive bivariate probit presented in table 1. Wald tests, however, indicate that the recursive bivariate is the better model.
Finally, to further address the interdependence of the judges and PG’s decisions, we conduct an extended regression model (ERM) for which the dependent variable is the decision at the judge level, and the decision of the PG is modeled as an endogenous covariate (see table SM5.3).Footnote 22 In these regressions, a significant correlation between the judges’ decisions and those of the PG indicates that the PG’s decisions are indeed endogenous. In all three models, the correlation coefficient is substantively large and significant at the .01 level. The substantive results from the ERM are identical in coefficient sign and significance to the bivariate results.
V. Implications
The results demonstrate that if Bulgaria wants stronger and more independent judges, then Bulgarian justices need to be further insulated from the political influence of the powerful PG. The Bulgarian Constitution provides for strong independence and autonomy of the Bulgarian judiciary, PG, and constitutional court. While this combination is generally highly valued in Western legal practice, when combined with a lack of accountability and checks from other branches of government this independence may make these institutions susceptible to political pressure (Hein Reference Hein2006). Representatives of the Council of Europe and the European Commission have continuously cautioned Bulgaria that the PG’s excessive prerogatives create opportunities for political abuse and have called for their revision. Even the former PG and current BCC chair, Boris Velchev, admitted that the system allows for political abuse (Nikolov Reference Nikolov2009). Despite this, the existing institutions still allow for dictatorship by the PG. The preservation of the Soviet procuracy model and the absence of formal political control over the PG has made it susceptible to nontransparent influences from politicians, interest groups, and even organized crime.
Although beyond the scope of our analysis, the judicial reforms of 2015 preserved the prerogatives of the PG and its influence on the appointment and discipline of judges, counter to the recommendations of the Venice Commission. These reforms did split the SJC into two chambers—judicial and procuratorial—which was encouraged by the Venice Commission but preserved the abundant influence of the PG. The 2015 reforms do not seem to have alleviated the powerful and unaccountable influence of the PG. Further reform is necessary to weaken the influence of the PG on judicial decision-making.
VI. Conclusion
The broad purpose of this study was to determine to what extent the BCC may operate autonomously and whether other political actors, such as the PG, influence judges in their constitutional review decisions. As with the US solicitor general, the results showed that BCC judges have a high propensity for following the opinions of the PG. While the literature on US prosecutors explains that their influence on judges’ decisions is due to legal acumen, status as a repeat player with resources, or ideological alignment with judges, our results seem to indicate that judges weigh heavily the preferences of the Bulgarian PG, most likely due to its stature as a political actor and power to affect judges’ reputations. The power afforded the Bulgarian PG, styled after the Russian version, appears to have created incentives for judges to often follow its lead. As repeatedly suggested by the Venice Commission (2008, 2017), such extensive noncriminal powers and the placement of the PG within the judiciary may harm the independence of constitutional judges.
The results also indicate that judges themselves exhibit some bias toward governments with whom they are politically aligned and that judges not aligned with the government are more sensitive to the opinions and recommendations of the PG. Our goal here is not to diminish the importance of the court or its contribution. The court has generally served its role as a protector of the constitution, and many of its decisions have contributed to the stability of Bulgarian democracy. Many constitutional justices have demonstrated their impartiality, and the results show that judges consider the view of interested parties, including NGOs, as expressed in amicus curiae briefs. However, Bulgaria’s institutional arrangements still make constitutional justices vulnerable to political influence and thus weaken this court as an institution.
On a broader level, the analysis also shows the dangers to emerging democracies that perhaps reform too quickly, combining old institutional structures (such as a procuracy, unaccountable to any other actor but able to exert influence on the judiciary) with new ones, namely, a powerful and autonomous constitutional court. Many young democracies in Central and Eastern Europe and Latin America have structures similar to Bulgaria. Some of these countries created constitutional courts after democratic transitions or constituent assemblies and created powerful prosecutor generals, which have historical antecedents similar to the Russian-style procurator model or share some of the other attributes of the Bulgarian PG as to powers, location in government, and ability to intervene in constitutional controversies, not simply criminal matters. For example, Ukraine’s prosecutor general is similar to Bulgaria’s (Venice Commission 2008) and is seen as quite powerful, as alluded to by that country’s president when discussing the use of the Ukrainian prosecutor to pursue investigations against the son of President Donald Trump’s political rival (Savage, Schmidt, and Barnes Reference Savage, Schmidt and Barnes2019). Further, in North Macedonia the powers and lack of confidence in prosecutors and special prosecutors have become quite apparent since the political upheaval and high-level corruption in 2015 (see Dimitrieva and Tiede Reference Dimitrieva and Tiede2020). Colombia’s procurador general, who has expansive powers and is obligated to give an opinion in every constitutional case before this country’s constitutional court, is also viewed as overly political and influential.
While such prosecutors as well as constitutional courts should provide a further check on elected branches that overstep their power, their very existence may suggest that other government institutions suffer from an “accountability deficit” (Moreno et al. Reference Moreno, Crisp, Shugart, Mainwaring and Welna2003), one that is not easily remedied if the PG is unchecked and constitutional courts are influenced by outside political actors. Bulgaria has provided a cautionary tale as to how a government deals with pressures to democratize or meet EU conditionality while at the same time acknowledging its own historic legal institutions.
'/%3e%3cpath%20d='M17%207.971L15.711%206.677%2010.912%2011.495%2010.912%200%209.088%200%209.088%2011.495%204.289%206.677%203%207.971%2010%2015z'%20transform='translate(8%208)'/%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
