Lifting the veil of secrecy – dissenting opinions in the subnational constitutional courts of Germany

ABSTRACT

Dissent is an integral feature of decision-making in collegial courts. However, unless procedural rules provide for the publication of a dissenting opinion and judges make use of this opportunity, courts appear to outsiders as impersonal, monolithic institutions. The following contribution explores dissenting opinions in German subnational constitutional courts (Landesverfassungsgerichte). Today, most of these courts provide for open dissenting opinions. Based on an empirical analysis of 1,115 cases decided by the constitutional courts of ten states (Länder) over a ten-year period (2009–2018), I find that dissent is driven predominantly by decision-level factors. Notably, the likelihood of dissent increases where a decision taps into the relationship between federal and state constitutional law and where more than one dissenting opinion is published. By contrast, I find no evidence for ideology or career background to impinge on the decision of a judge to author a dissenting opinion.

KEYWORDS:

• Dissenting opinions
• rational dissent theory
• subnational constitutional courts
• ex ante model
• federalism
• Germany

Introduction

Dissent is an inescapable fact of judicial decision-making in collegial courts. It has always existed. However, it makes a tremendous difference whether or not dissent is allowed to be disclosed to outside audiences. In the U.S. Supreme Court, dissenting opinions have a long-standing tradition as a legal institution, dating back to the era of Chief Justice Marshall (Bader Ginsburg 2010; Kelemen 2018, 58–62). By contrast, dissenting opinions have for long been alien to courts in the continental European legal sphere. Following the civil law tradition, court rulings have been regarded as unanimous and impersonal decisions that ought to reflect the application of doctrinal methodology and place a premium on providing legal certainty.

Dissenting opinions lift the veil of secrecy surrounding judicial deliberations and decision-making. As such, they expose not only the dissenter, but the entire court. They debunk the myth of pure objectivity and uniformity of judicial decisions. What is more, dissenting votes cast doubt on the infallibility of the majority, signalling ‘that, in the dissenters’ view, the Court's opinion is not just wrong, but grievously misguided’ (Bader Ginsburg 2010, 2) and that the court could have – in principle – decided differently on the merits of a case. As such, dissenting opinions are testament to different legal opinions that reflect the pluralism inherent to a democratic society (Lietzmann 2006, 270). Today, the traditional incompatibility of dissenting opinions in civil law systems no longer holds. Twenty-one Member States of the European Union (EU) provide for dissenting opinions in at least some courts, usually the court of highest instance. Accordingly, there has been a recent surge in scholarly interest in exploring judicial dissent beyond the U.S. context, in part motivated by estimating judges’ ideal points, which is not limited to constitutional courts in continental Europe (Bricker 2017; Kelemen 2018; Dalla Pellegrina, Garoupa, and Gómez-Pomar 2017; Hanretty 2012), but extends to the constitutional courts of Turkey (Abad Andrade 2020) and the Supreme Courts of Argentina (Muro et al. 2020), Brazil (Desposato, Ingram, and Lannes 2015), Canada (Songer, Szmer, and Johnson 2011) and the Philippines (Dalla Pellegrina, Escresa, and Garoupa 2014).

The German Federal Constitutional Court (FCC) set the example for many constitutional courts throughout Europe. Following a legislative amendment, and inspired by the experience of the U.S. Supreme Court, dissenting opinions have been allowed at the FCC since 1971 (Eggeling 2011, 79–91). However, it deserves mentioning that the debate of the late 1960s concerned the introduction of dissenting opinions at the constitutional courts (in the plural), i.e. at both the federal and state level. However, it was not until 1982 that the state of Hamburg first introduced the option for justices¹ to submit a signed dissenting opinion for the public record. At present, ten out of 16 subnational constitutional courts permit open dissenting opinions (Thierse 2020).²

This article provides the first empirical analysis of dissenting behaviour at these courts. Drawing on original data covering more than 1,100 decisions handed down by ten subnational courts in the period 2009–2018, I explore factors driving judicial dissent at both the decision-level and the level of individual justices. Methodologically, it exploits the fact that – given the embeddedness in a federal polity – a host of extraneous variables can be controlled for (Reutter 2021b). At the same time, the study contributes to the comparative analysis of decision-making within constitutional courts by analyzing dissent in collegial courts that differ in some crucial regards from both the FCC, and arguably all constitutional courts at the national level.

The remainder of the article is structured as follows: The next section describes the organization and jurisdiction of subnational constitutional courts and gives an overview of the provisions regarding dissenting opinions. Thereafter, I review theoretical arguments for the incentive to dissent. I evaluate the applicability of rational dissent theory for subnational constitutional courts and develop testable hypotheses for explaining judicial dissent. Subsequently, I present the data and the empirical analysis. The final section concludes and considers potential research questions beyond the scope of this article.

Constitutional courts at the subnational level

Germany is an exceptional case in that it combines democracy, federalism, civil law tradition and constitutional review at both the federal and state (Land) level (Reutter 2021b, 3). The German states are political systems in their own right. By virtue of this quality, they possess the right to adopt constitutions of their own and establish subnational courts that are mandated and obligated to enforce these constitutions. This competence has been confirmed by the FCC's doctrine of ‘separate but interdependent constitutional spaces’ (Höreth 2020, 53). Thus, the first constitutional courts were established in Bavaria (1947), Hesse (1948), Bremen (1949), Rhineland-Palatine (1949) and the former South-Western provinces of Württemberg-Baden, Württemberg-Hohenzollern and Baden even before the Basic Law was adopted in October 1949. Like the FCC, which followed in 1951, these subnational constitutional courts are not super-appellate bodies, but specialized constitutional tribunals that are restricted to adjudicating on matters where constitutional principles are at stake.

Their jurisdiction differs across states and in most cases is not entirely congruent with that of the FCC. All subnational constitutional courts possess the competence to adjudicate on disputes between governmental bodies (Organstreitverfahren), evaluate the constitutionality of norms and acts of public authority and invalidate unconstitutional acts, including parliamentary statutes. As is the case at the federal level, the right of initiative for constitutional review rests with ordinary courts (concrete review) and political bodies (abstract review).³ Important differences exist with regard to the constitutional complaint procedure, which grants private individuals the right to seek redress for the purported infringement of fundamental rights by an act of public authority. Bremen, Hamburg, Lower Saxony and Schleswig-Holstein have thus far not introduced constitutional complaints.⁴ In turn, subnational constitutional courts wield some competences that are pertinent to the subnational polities and therefore unknown before the FCC: This regards, first, municipal constitutional complaints which can be brought by local authorities that claim an infringement of their constitutional right to self-administration. Second, all subnational constitutional courts possess jurisdiction to resolve disputes over procedures of direct democracy (Reutter 2021b, 5), which is only feebly institutionalized at the federal level.

The most striking difference compared to the FCC and – to the author's best knowledge – all national constitutional courts, is the fact that the subnational constitutional courts are composed of justices serving in a part-time, honorary capacity next to their professional activity as judge at an ordinary court, professor of law at a university, attorney in a law firm or employment in a non-judicial job (Reutter 2021a, 5).⁵ The procedures governing the nomination and election of justices vary considerably across states. However, a formal (confirmatory) election by state parliaments is the norm. The right to propose candidates may extend to state governments, parliaments and state courts. While eight out of 16 states provide for nominations only from within parliament, two (Bavaria and Hesse) grant a right of nomination to all three branches, i.e. state governments, state parliaments and the courts.⁶ For the nomination by parliaments, one can distinguish between nomination by party groups, special committees, groups of members, and the Bureau or Council of Elders. The majority requirements for the election in parliament range from simple majority (Bavaria and Hamburg) to a two-thirds majority of all members (Brandenburg, Northrhine-Westphalia, Saarland, Saxony-Anhalt, Saxony, and Thuringia) (Reutter 2021a, 2020a).⁷ The terms of office range from four-year legislative period (Bremen) with unlimited renewability to a twelve-year, non-renawble stint (Mecklenburg-Western Pomerania) (Reutter 2020a, 226).

Legal provisions regarding dissenting opinions in subnational constitutional courts

In most states, the provisions regulating dissenting opinions at subnational constitutional courts are modelled after those of the FCC. The Eastern German Länder of Berlin, Brandenburg, Saxony-Anhalt and Thuringia have adopted the provisions of the pertinent paragraph of the procedural act for the FCC almost literally. The latest state to introduce the option of publishing a dissenting opinion as part of an encompassing constitutional amendment is Northrhine-Westphalia in 2017 (Thierse and Hohl 2017). Saxony is the only state to explicitly preclude dissenting opinions and the publication of the voting outcome in the procedural act on the constitutional court (Thierse 2020, 182). Despite the fact that a majority of German states have explicitly regulated the option of dissenting opinions in procedural acts of their constitutional courts, the use of this legal institute is still to be seen as an exception to the general rule of the secrecy of deliberations. In a number of procedural acts on subnational constitutional courts, the obligation of judges to remain silent on the process of deliberations is stipulated explicitly (Thierse 2020, 183).

More detailed provisions on the submission of a dissenting opinions are included in the rules of procedure of the courts. Justices must generally announce their intention to publish a dissenting opinion. In most cases, a dissenter has to announce his or her intention to dissent as early as possible, but no later than before his or her colleagues sign the decision. The provision to announce the intention to dissent ‘as soon as the state of deliberations permit’ is very obviously aimed at launching additional discussions so that a consensual decision can be reached (Abad Andrade 2020, 109–10). The standard deadline for submitting the dissenting opinion is three weeks after the decision has been drafted. As is the case at the FCC, a number of states permit a dissenting judge to expound the content of his or her dissenting opinion after the court has delivered its decision. This is true of Bremen, Mecklenburg-Western Pomerania, Schleswig-Holsteins, Saxony-Anhalt and Thuringia. At any rate, the dissenting opinion is not part of the court decision, but an adjunct, a personal and individual perspective on the merits of case.

Theory and hypotheses

Rational dissent theory (Epstein, Landes, and Posner 2011) serves as the workhorse for most empirical studies of judicial dissent. The basic assumption is that judges weigh the costs of dissenting from the majority opinion against the potential benefits. Drafting a dissenting opinion entails effort costs for the dissenter, but also imposes collegiality costs on the majority, which may feel compelled to produce a more carefully reasoned and encompassing legal decision that addresses and preempts potential disagreement. However, a dissenting opinion also promises reputational benefits for the dissenter to the extent that the dissenting opinion receives recognition in subsequent jurisprudence, the legal community and the wider public. Besides reputational gains, dissenting opinions also reflect differing ideological convictions and policy motivations of judges (Segal and Spaeth 2002). They can be authored for very different reasons, either to defend the legal (and political) status quo or to voice publicly visible criticism against a jurisprudence that fails to stay abreast altered social or political circumstances. Accordingly, dissenting opinions are used to issue a warning against an overzealous adaption of jurisprudence to new social realities or as a signpost for future legal development and the invitation to future review (Thierse 2020, 199).

The institutional context of subnational constitutional courts seems to mitigate the positive and neutralize the negative impacts of dissenting opinions. Thus, while the lower majority thresholds for the election of justices and the possibility of renewable terms could imply a greater risk of partisan behaviour compared to the FCC, the utility of using dissenting opinions opportunistically to secure reelection is offset by the fact that the office of justice at a subnational constitutional court is a part-time and honorary activity (Geck 1983, 360–61). Decisions issued by subnational constitutional courts receive only scant attention in the media, which constrains the opportunities for individual judges to raise their profile by publicly dissenting from the majority opinion. This is all the more true to the extent that the media reports only about the existence, but not about the exact content of dissenting opinions (Eggeling 2011, 232).

In addition, the internal organization of subnational constitutional courts seems conducive to consensus-oriented interactions between judges. Like most constitutional courts throughout Europe – including the FCC – subnational constitutional courts are built on an ex ante model of judicial deliberations and opinion-writing (Cohen 2014). Before the full panel decides on a case, pre-deliberations take place and a draft decision is drawn up by a rapporteur. Where an oral hearing takes place, the rapporteur prepares a written vote that may include a proposal for the structuring of the oral hearing. Usually, one rapporteur is selected to shepherd a case through the deliberations, although a tandem of a rapporteur and co-rapporteur is the norm for more complex and comprehensive cases. Given the strong agenda-setting prerogatives and informational advantage of rapporteurs, co-panelist judges may refrain from dissenting simply because they lack the factual and legal knowledge (Bricker 2017, 173–74; Meyer and Hönnige 2017, 229).

There is sufficient evidence that where dissenting opinions are possible, constitutional and high courts in civil law systems display markedly lower dissent rates than the U.S. Supreme Court or the State Supreme Courts (Bricker 2017, 175; Kelemen 2018, 92–93). However, institutional context und internal organization only explain lower dissent rates, i.e. dissent at the aggregate level. To explain the occurrence of dissenting opinions, it is necessary to consider individual- and decision-level factors.

Professional background of justices

While disagreement is inevitable in judicial decision-making, open dissent is a voluntary and purposive act that likely reflects the personality and individual traits of a judge. The professional background of a justice can be expected to be of particular importance for the context of subnational constitutional courts, where paths to the judicial bench and therefore the overall composition are even more varied than in national constitutional courts, let alone the U.S. Supreme Court, ‘which is almost exclusively comprised of career judges’ (Bricker 2017, 175). Career academics stand out from their colleagues with a professional background as career judge. While the latter are trained into applying the law in real-world cases, socialized into working in collegial courts that are (still) governed by the principle of secret deliberations and accustomed to a culture of consensual decision-making, the former are used to working in solitude in offices to craft and apply legal theories. Unlike professional judges, law professors routinely become involved in public debates with outspoken positions that comment on and influence case law and legal development (Bricker 2017, 175). In particular, dissenting opinions in subnational constitutional courts have been characterized as substitute for commentaries or scientific publications on the state constitutions, which are in short supply (Eggeling 2011, 233; Gärditz 2013, 481). Therefore, I expect career academics to be more inclined to authoring dissenting opinions (H1).

Ideological composition of the panel

The ideological positions and normative convictions of judges may enter the calculus of judicial decision-making in several ways. First, in keeping with the attitudinal (or policy-based) model of judicial behaviour, judges are ‘single minded seekers of legal policy’ (George and Epstein 1992, 325). They use jurisprudence as a vehicle to realize their policy preferences and ‘etch their political values into law’ (Epstein and Knight 2013, 14), without taking into account the preferences and potential (re-)actions of their colleagues, other political bodies or the public. In this perspective, dissent is the expression of a deviation from policy preferences reflected in the majority opinion. In a different reading, the ideological position of a judge can become a factor of judicial dissent when this serves to signal commitment to a legal interpretation that has lost out in judicial deliberations, but corresponds to the policy agenda of a party. Whether judges are conceptualized as ‘lawmakers’ interested solely in realizing their personal policy goals or as agents of political parties, I expect justices with ideological positions far away from the panel median to be most inclined to engaging in dissent (H2). While this is an individual-level effect, a decision-level factor that may be driving dissenting opinions is the overall ideological heterogeneity of the panel. All else equal, ideologically more diverse panels should be associated with a greater probability of dissent (H3). This is because not only the ideological positions, but also the intensity of policy preferences among judges on a particular outcome will diverge (Epstein, Landes, and Posner 2011, 108).

Collegiality and opportunity costs of dissenting

Recent scholarship has stressed the relevance of individual motivations other than the pursuit of policy goals (Epstein and Knight 2013). Among these, the job satisfaction derived from maintaining good collegial relations appears as one factor most applicable to the context of German subnational constitutional courts. Judicial dissent can jeopardize good collegial relations by requiring the rapporteur(s) and the majority to invest more time and resources into explaining and justifying their legal reasoning. However, the marginal collegiality costs are bound to decrease for each additional dissent: If two justices dissent instead of one, the majority will likely not need to invest twice the time to consider the dissenters’ arguments or draft a majority opinion that has twice the word count. Conversely, the opportunity costs for dissenting will decrease with every additional dissent: If one justice has indicated her willingness to dissent, other colleagues may feel less convinced by the majority decision and also find it easier to make their disagreement public because they are not singled out as dissenters. Therefore, dissenting opinions are hypothesized to be more likely when there are additional judges that dissent (H4).

Type of proceedings

Organstreit proceedings involve public bodies in a contradictory procedure; one body (or a component part) vindicates an infringement of constitutional rights. As is the case at the national level, Organstreit proceedings have been found to be an instrument of the opposition in state parliaments (Flick 2009; Meyer and Hönnige 2017, 232–33) (Reutter 2017, 96; Thierse and Hohl 2017, 264). They are a clear example of political actors seeking to influence the constitutional court to weaken political opponents (Ewert and Hein 2016). As a result, we should expect party-political conflict to play a greater role in judicial deliberations and increase the likelihood of disagreement in Organstreit proceedings (H5a).

Abstract review procedures are known to be a preferred tool of the parliamentary opposition to qualify governmental decisions (Vanberg 1998; Hönnige 2007). Abstract review is the most evident example of external influence on a court to make use of its veto power. Like the FCC, subnational constitutional courts possess the competence to strike down legislation that infringes constitutional norms. Abstract review has concrete policy implications, and this contrasts squarely with Organstreit proceedings, which can only ascertain that an act (or failure to act) breaches constitutional norms. Accordingly, we should expect abstract review procedures to be most conducive to spurring dissent (H5b).

Constitutional complaints have been characterized as a ‘wailing wall’ for citizens (Blankenburg 1998). They grant individuals access to the constitutional court. At the same time, there are high obstacles to overcome. Crucially, the procedure is a subsidiary and extraordinary legal remedy, and the routine admissibility requirement is that a plaintiff has exhausted all legal remedies at the ordinary appeal stages. Since constitutional courts require proof of a personal violation of fundamental rights and are a non-contradictory procedure, they do not (easily) lend themselves to politicization (Ewert and Hein 2016). What is more, like constitutional complaints brought before the FCC, most complaints fail tests for admissibility (Reutter 2020b, 170). As a result, we can expect constitutional courts in most cases to be unspectacular, run-of-the-mill decisions that produce little to no substantive disagreement. This effect will likely outweigh the impact of the relatively few cases that do make it on the docket because of an acknowledged potential for a development of constitutional law. Accordingly, it is assumed that dissenting opinions are less likely to be authored in constitutional complaint procedures (H5c).

Judicial strength of a decision

Besides the type of procedure, a second factor at the decision-level is the strength of a decision. Following the conceptualization offered by Pócza, Dobos, and Gyulai 2017, 1560–33) and Reutter (2021b, 3), strength is taken here to refer to the extent to which court rulings constrain the range of decision alternatives for policy-makers. In other words, strong court decisions are an indicator of judicialization (Hirschl 2008; Landfried 1994; Stone Sweet 2000). A prime example of a strong decision is a ruling in which a constitutional court invalidates a statute as null and void and strikes it from the statute book. It is plausible that these are precisely the decisions that will spark the most severe controversy among judges. After all, judges are perfectly aware that the annulment of a statute as unconstitutional is the most incisive interference in democratic law-making. As noted by Visser (2015, 282), deference to the decisions taken by a democratically elected and accountable legislature are the rule rather than exception. Therefore, we should assume that at least some judges will speak in favour of exercising judicial restraint or seeking less intrusive solutions, e.g. the annulment of only parts of the statute or a constitutionally compatible interpretation where rulings have a strong bearing on policy choice. We can thus hypothesize that dissenting opinions are more likely to be authored for stronger decisions (H6).

Constitutional review in a federal polity

Germany is a federal polity in which constitutional law is premised on both the Basic Law and the state constitutions (and increasingly on European Union law) (Möstl 2005). The doctrine of separate and autonomous constitutional spheres of the state (Länder) and federal (Bund) levels of government is to some degree at odds with the normative and structural binding power exerted by the Basic Law (GG), i.e. the federal constitution. Not only is the exercise of public authority at the subnational level bound by federal constitutional law, but also the effectiveness of subnational constitutional law is limited by key principles enshrined in the Basic Law. These include, above all, the principle of homogeneity (Art. 28 GG), the supremacy of federal law over subnational (constitutional) law (Art. 31 GG) and the division of legislative competences between (Art. 70–74 GG) (Höreth 2020). They thus operate at the crossroads of subnational and federal constitutional law. Subnational constitutional courts routinely receive cases in which plaintiffs claim an infringement of legislative competences or fundamental rights enshrined in the Basic Law, and most courts today claim the competence to evaluate the compatibility of subnational acts of government with provisions of the Basic Law (Alexy 2020, 53). Evidence from case studies suggests that the autonomy of the states to regulate and legislate and the boundaries imposed by provisions of the Basic Law or FCC case law figure prominently in dissenting opinions across courts (Thierse 2020). In general, overt disagreement among judges could be more prevalent where judicial review of subnational acts of public authority taps into the relationship between the Basic Law and subnational constitutional law (H7).

Research design

I investigate the determinants of dissenting opinions at the subnational constitutional courts of all ten states that provide for open dissenting opinions at their constitutional courts throughout the period of investigation (2009-2018): Berlin, Brandenburg, Bremen, Hamburg, Hesse, Lower Saxony, Mecklenburg-Western Pomerania, Saxony-Anhalt, Schleswig-Holstein and Thuringia. My period of analysis has not been covered by existing studies on dissent at the subnational level in Germany (Eggeling 2011; Thierse 2020) and allows me to include Schleswig-Holstein, which introduced its own constitutional court only in 2008 (Flick Witzig 2017). In compiling my dataset, I drew on several sources. First, I consulted the archives of the respective constitutional courts, which are either available on the courts’ websites or integrated into legal databases hosted by the administrations of the state governments. Second, I perused the official collections of decisions published jointly by all subnational constitutional courts covered in my analysis (with the exception of Hamburg). Third, I triangulated the information with the entries in the Juris database.⁸ Finally, I relied on a dataset and coding scheme provided by Reutter that builds on the decisions contained in the official collections.⁹

My dataset contains information on a total of 1,115 court decisions, 946 (85%) of which are not included in the Reutter dataset. I included all final decisions on main proceedings, as well as applications for interim measures. Other procedural motions such as pleas of remonstrance, bias petitions or applications for review of the judgment were excluded from the analysis. The 1,115 decisions are associated with a total of N = 8,821 individual voting decisions by judges which form the units of analysis. Because of missing information for two judges from Hesse on the nominating party or institution, 10 observations are dropped from the multivariate analysis.

Table 1 reveals that the courts differ vastly in terms of caseload and also display variation across types of procedure. Thus, whereas the constitutional court of Brandenburg handled 597 proceedings (excluding procedural motions) in the period 2009–2018, the constitutional court of Thuringia – which is most similar in terms of jurisdictional competence and internal organization (Lembcke and Güpner 2018, 103) – decided on merely 58 cases. At the same time, dissent rates differ markedly across subnational constitutional courts. As can be gleaned from Table 2, dissenting opinions also vary by type of proceedings. Thus, while constitutional complaints account for the bulk of cases in the dataset, more than 99% of these cases were decided without a dissenting opinion. By contrast, abstract review proceedings, while accounting for a marginal share of just about 2% of all cases analysed, are associated with a relatively high degree of dissent.

Table 1. Constitutional court decisions by procedure type.

Land	Number of decisions	Decisions with at least one dissenting opinion (%)	Constitutional complaints	Abstract review	Organstreit proceedings	Other
Berlin	236	3 (1.3)	194	2	13	27
Brandenburg	597	1 (0.2)	554	2	13	28
Bremen	13	2 (15.3)	0	6	3	4
Hamburg	23	1 (4.3)	0	0	7	16
Hesse	33	3 (9.1)	17	0	5	11
Lower Saxony	18	0 (0.0)	0*	4	5	9
Mecklenburg-Western Pomerania	55	3 (5.5)	6	0	35	14
Saxony-Anhalt	57	2 (3.5)	10	2	5	40
Schleswig-Holstein	25	3 (10.0)	0*	3	4	18
Thuringia	58	17 (29.3)	21	7	9	21
	1,115

Table 2. Dissenting opinions by type of proceedings.

Type of proceedings	Total number of decisions (%)	Decisions with dissent	Percentage of decisions with dissent
Organstreit proceedings	99 (8.9)	7	7.1
Abstract review	26 (2.3)	9	34.6
Constitutional complaint	802 (71.9)	7	0.9
Concrete review	10 (0.9)	0	0.0
Scrutiny of elections and the status of members of parliament	32 (2.9)	4	12.5
Scrutiny of procedures related to direct democracy	13 (1.2)	2	15.4
Municipal constitutional complaint	89 (8.0)	4	4.4
Other (e.g. applications for interim measures)	44 (3.9)	2	4.5
	1,115

For a multivariate analysis of judicial dissent, I combined decision-level data with data at the level of individual justices, which was also obtained in parts from a dataset compiled by Reutter. Missing data were gathered by conducting desk research using publicly available sources such as plenary protocols, official journals, almanacs and media reports.¹⁰ Additional information was requested from the press offices of the constitutional courts, where necessary. The dataset contains information on 190 individual justices, 147 of whom served as regular members of the constitutional court in the period of investigation.

The unit of analysis is judge per decision, as every constitutional judge in principle has the opportunity to publish a dissenting opinion on any decision on the court's agenda. The multivariate analysis is conducted on 8,811 observations for which there is complete information at both the decision and judge-level. The dichotomous outcome variable Dissent assumes the value 1 in case an individual justice has delivered a dissenting opinion. As Table 1 illustrates, dissenting opinions are a rare event. Of all 1,115 decisions analyzed, only 35 were decided with at least one dissenting opinion, a share of just above 3%. This figure is markedly lower than the dissent rate of 11% reported for the FCC in the period 2003–2010 and outshined by the Czech and Slovakian constitutional courts, that exhibit dissent rates of 29 and 25%, respectively (Bricker 2017, 174).

There are two groups of explanatory variables: At the level of individual justices, I distinguish between four different professional backgrounds. Professional judge denotes justices who serve as judges in their regular office and is used as the reference category. Law Professor marks those justices who are tenured professors of law at a university. Attorney refers to justices who are practicing attorneys in their full-time occupation. Lay judge is a category reserved for justices occupied in a field outside of the legal realm. Distance to panel median is a variable that measures the ideological distance from the median judge on the left-right dimension. For every justice in my dataset, I combined information on the nominating party with information on that party's left-right position. The latter information was retrieved from Bräuninger et al. (2020) and is based on manifesto data at the state party level. For each justice, I used the legislative term that covered the respective confirmatory vote. Data on the nominating party or institution come from Reutter (2021) and were supplemented by information I gathered using publicly searchable databases, archives and media reports. For ex officio members and constitutional justices nominated by state governments, I used the governmental parties’ positions, weighted by their share in a fictitious electoral coalition. For justices nominated by several parties or formally nominated by a designated electoral committee, I similarly used the respective parties’ left-right position, weighted by seat share in the electoral coalition.

Using the same data on party ideology, I constructed a variable named Ideological dispersion of panel, which is calculated as the standard deviation of the ideological positions of nominating parties. Since the decision texts routinely indicate the exact composition of the panel, I was able to construct decision-variant measures of this variable. I differentiate between three specific proceedings that are expected to vary systematically with the probability of judicial dissent: Abstract review, Organstreit proceedings and Constitutional complaints. All remaining proceedings are grouped in a residual category which serves as the reference category. The variable Strength of decision captures the degree of policy intrusion of a decision. It is based on an additive index first proposed by Pócza, Dobos, and Gyulai (2017) and adapted by Reutter (2021b). The index ranges from 0 to 9, with the minimum value 0 indicating either the rejection of an application or the declaration of the claim as justified, but without consequences for lawmakers, and the maximum value 9 indicating the formulation of legally binding requirements derived from an interpretation of constitutional norms. For decisions with more than one dissenting opinion, the number of additional dissenting opinions is rendered by the variable Additional dissents. Furthermore, the dummy Reference GG captures whether the decision concerns the relationship between subnational constitutional law and the Basic Law.

Control variables are included at both the individual and decision level. The dummy President indicates that a judge served as President of the court. Seniority measures the tenure of a justice and is calculated as the time span (in years) between the first date in office and the date of the decision. Electoral majority refers to the share of votes for the respective justice in the parliamentary confirmation vote (based on the total number of votes cast).¹¹ Male accounts for the fact that justices are still predominantly male. At the level of decisions, I account for the variation between judgments (Urteil), which are preceded by an oral hearing, and resolutions (Beschluss). Summary statistics for all variables included in the regression models are reported in the Appendix.

Due to the hierarchical structure of the data, where individual judges are nested in decisions, and decisions nested within constitutional courts, and in order to account for unobserved heterogeneity, I fit a mixed-effects logistic regression model. The analysis proceeds stepwise. Model 1 is an ‘empty’ model that simply reports the variance components. Model 2 is a random-intercept model that includes predictors at the individual level (level 1). Model 3 adds predictors at the decision-level (level 2). Finally, Model 4 adds a random coefficient for Additional Dissent at level 3 to ascertain whether the effect of the covariate varies across courts.

Analysis and discussion

The variance-components model (Model 1) indicates substantial clustering in the data and justifies to assume dependence in the individual justices’ decisions. Substantively, an intra-class correlation of 0.79 at the decision-level and 0.29 at the court-level signifies that roughly 29% of the variance in the outcome variable is explained by differences between courts, while 79% of the variance lies between decisions within courts. This implies that only 21% of the variance is explained by individual-level characteristics (i.e. between judges within decisions). While the intra-class correlations do not change substantively, Model 2 reveals that contrary to our theoretical expectations the professional background of justices does not have a systematic impact on the proclivity to dissent. Neither category emerges as significant compared to the reference category (professional judge). These effects remain intact in Model 3. In light of the data, H1 must therefore be rejected. As can be seen from Table 3, adding predictors at level 2 does not only reduce the variance at the decision-level, as can be expected, but also leads to a decrease in the between-court variance. This suggest that there are court-level differences in the distribution of predictors that are related to dissent. The intra-class correlation of 0.685 indicates that conditional on the fixed-effects covariates, the combined random effects at the decision and court level are estimated to explain about 69% of the residual variance of the outcome variable.

Table 3. Determinants of judicial dissent - Results from mixed effects logistic regression models.

		Model 1	Model 2	Model 3	Model 4	Model 5
Predictors	hypothesis
Individual level
Law Professor	H1		0.118	0.131	0.097	0.122
			(0.37)	(0.37)	(0.37)	(0.37)
Attorney			0.406	0.326	0.295	0.290
			(0.51)	(0.50)	(0.50)	(0.50)
Lay judge			0.206	0.102	0.057	0.099
			(0.87)	(0.84)	(0.84)	(0.83)
Distance to panel median	H2		−0.627**	−0.577**	−0.556*	−0.582**
			(0.21)	(0.22)	(0.22)	(0.22)
President			−1.858*	−1.864*	−1.873*	−1.864*
			(0.78)	(0.78)	(0.78)	(0.78)
Seniority (years)			−0.023	−0.020	−0.038	−0.021
			(0.04)	(0.04)	(0.04)	(0.04)
Electoral majority			0.003	0.000	−0.003	−0.002
			(0.02)	(0.02)	(0.02)	(0.02)
Male			−0.004	0.021	−0.006	0.017
			(0.34)	(0.34)	(0.34)	(0.34)
Decision level
Ideological dispersion of panel	H3			0.180	−0.272	0.076
				(0.47)	(0.51)	(0.47)
Additional dissents	H4			2.542***	2.707***	5.007
				(0.70)	(0.72)	(2.75)
Organstreit	H5a			−0.079	−0.289	−0.120
				(0.76)	(0.77)	(0.73)
Abstract review	H5b			−0.046	−0.407	0.107
				(1.02)	(1.07)	(0.98)
Constitutional complaint	H5c			−1.000	−0.859	−1.150
				(0.77)	(0.79)	(0.77)
Strength of decision	H6			−0.351	−0.285	−0.331
				(0.18)	(0.19)	(0.17)
Reference GG	H7			1.773**	1.718**	1.713**
				(0.62)	(0.62)	(0.60)
Judgment				1.789*	1.739*	1.778*
				(0.73)	(0.74)	(0.70)
Constant		−7.779***	−7.188***	−7.652***	−7.631**	−7.126**
		(1.04)	(2.18)	(2.27)	(2.49)	(2.26)
Random effects
Intercept variance (court-level)		4.523	4.857	2.063	2.420	2.219
		(2.42)	(2.68)	(1.42)	(1.68)	(1.50)
Slope variance (court-level)						10.788
						(14.41)
Intercept variance (decision-level)		7.786	8.098**	5.075**	4.280**	4.461**
		(2.41)	(2.56)	(1.67)	(1.44)	(1.54)
Intraclass correlation (court-level)		0.290 (0.10)	0.299 (0.11)	0.198 (0.11)	0.242 (0.13)	0.223 (0.12)
Intra-class correlation (decision-level)		0.789 (0.05)	0.797 (0.05)	0.685 (0.07)	0.671 (0.08)	0.670 (0.08)
N (level 1)		8,821	8,811	8,811	8,064	8,811
N (level 2)		1,115	1,115	1,115	1,115	1,115
N (level 3)		10	10	10	10	10
Log-likelihood (df)		−280.341 (3)	−268.736 (11)	−245.529 (19)	−234.776 (27)	−244.326 (20)
AIC		566.683	559.472	529.057	523.552	528.651

Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. Likelihood-ratio χ² (1 df)    2.41 (p = 0.121) for the assumption that Model 3 is nested in Model 5.

Also contradicting H2, judges far away from the median in terms of left-right are not found to be more likely to dissent. Instead, they have a significantly lower probability of dissenting from the majority decision. Contrary to H3, ideological heterogeneity of the panel does not have a systematic effect on the likelihood of dissent. As expected by H4, dissenting opinions become more likely when a judge is not the only dissenter. The variable Additional dissent has a strong effect and is highly significant (p < 0.001). Turning to H5, neither Organstreit proceedings nor abstract review are associated with a greater likelihood of dissenting opinions. Constitutional complaints are associated with a lower probability of dissent, yet this effect fails to reach statistical significance. H6 is falsified in light of the data. Stronger decisions are associated with a significantly lower, not higher probability of dissenting behaviour. In turn, H7 finds support: Where a decision implies a consideration of the relationship between provisions of the Basic Law and subnational constitutional law, the probability of dissent increases significantly (p < 0.01).

Turning to the control variables, at the individual level, presidents are significantly less likely to table dissenting opinions, which could be explained by a special responsibility of these justices to remain neutral and seek consensus. The other control variables do not display any significant effect. At the decision-level, I find judgments, i.e. decisions pronounced after an oral hearing, to be positively associated with dissenting opinions (p < 0.05).

To offer an illustration of the substantive findings, Figure 1 graphs predicted probabilities for various values of the variable Additional dissents, contrasting decisions where the relationship between subnational constitutional law and the Basic Law was at stake and those where there was no reference to the Basic Law. All else equal, the predicted probability of dissent is almost not distinguishable in a scenario with a single dissenter. It rises steeply to a difference of over 21 percentage points in a scenario with three additional dissenters. However, the confidence intervals become exceedingly large for higher values of Additional dissents, which suggests that the effect of the covariate Reference GG is estimated with less precision.

Figure 1. Marginal effects of additional dissents on dissent probability.

The above results are based on the random-intercept model reported under Model 3. Model 4 adds year-fixed effects. I also fit a model with a random coefficient for Additional dissents at the court level (Model 5). Since the subnational constitutional courts differ in size,¹² dissenting opinions could impose differential collegiality costs on the majority and differential opportunity costs on the dissenters. For instance, the intention of other judges to dissent could have a more modest effect on the probability to publish a dissenting opinion in courts with a larger panel size. The results indicate that the variances at both the decision and court level do not change substantially compared to the random-intercept model. A likelihood ratio test fails to reject the simpler random-intercept model in favour of the random-coefficient model, supporting the assumption that there are no differential effects of additional dissent on the probability of dissenting opinions across courts. The substantive results are robust to other estimation techniques, the exclusion of constitutional courts that do not feature individual constitutional complaints, and the restriction to single-authored dissenting opinions. The reader is referred to the Appendix which reports the results of a series of robustness checks.

Judicial dissent in subnational constitutional courts: Conclusions and outlook

Although the legal institute of dissenting opinions has become an established feature of German subnational constitutional review over the course of the past 30 years, our empirical knowledge of what drives judicial dissent has been limited. Existing research remains confined to anecdotal evidence and descriptive inference. This is the first study to assess judicial dissent in the subnational courts of Germany based on a large-N analysis. As such, it contributes to a growing body of research on judicial politics and judicial behaviour beyond the iconic U.S. Supreme Court.

What are the implications for the empirical analysis of other constitutional courts, and how far do the results travel? Some might argue that there is hardly any comparative merit in analyzing dissent in German subnational constitutional courts for an understanding of decision-making in national-level constitutional courts, given that the former are staffed with justices acting in an honorary capacity, manage a modest caseload and process cases with a limited scope for constitutional interpretation for the federal level, let alone beyond Germany. These concerns can be alleviated. First, subnational constitutional courts are genuine constitutional courts, established to uphold the supremacy of constitutional law over ordinary law. In the past years, the subnational constitutional courts have stepped out into the limelight of public awareness with politically controversial decisions regarding the expansion of suffrage to minors, EU citizens and non-EU aliens, gender equality and affirmative action, policy initiatives using procedures of direct democracy, and the banning of headscarves for female teachers (Thierse 2020; Reutter 2022). Most importantly, subnational constitutional courts are collegial courts. As such, theories of judicial behaviour should also be applicable to them. In this vein, some results confirm previous findings from other institutional contexts. Thus, as also found by Muro et al. (2020) for the Supreme Court of Argentina, additional dissents are associated with a significantly higher probability of dissent. This is compatible with the explanation that the opportunity costs of exposing oneself as a dissenter are lower when other justices also signal their disagreement with the majority decision. Furthermore, dissent was found to be more likely in cases where the relationship between subnational and federal constitutional law is at stake. This finding is in line with the observation that more complex cases, i.e. those containing a greater number of references to constitutional provisions or that tap into multiple issues, lead to a greater likelihood of dissent (Bricker 2017; Songer, Szmer, and Johnson 2011). It also corroborates the assumption that interaction between courts across levels of government may become a driver of dissent (Peterson 1981, 418). Stronger decisions, i.e. those that constrain political discretion, were not found to be associated with a greater probability of dissenting opinions. While this refutes theoretical assumptions, the result is consistent with recent research on dissent in the Argentinian Supreme Court (Muro et al. 2020).

Altogether, decision-level factors emerge as far more relevant predictors of dissenting opinions at German subnational constitutional courts compared to individual-level factors. Notably, neither career background nor ideological position were found to have an impact on the inclination to dissent, a finding at odds with previous research from other national-level constitutional courts. At the same time, this speaks to the conceptualization of subnational constitutional courts as decision-making bodies that, in spite of partisan influence on the appointment and (re-)election of judges, manage to produce decisions that transcend party-political preferences or professional attitudes (Reutter 2021b, 14, 2021a). The overall consensual nature of subnational constitutional courts is also reflected in the finding that neither the type of proceedings nor the strength of the decision exerts a significant effect on the probability of dissent. Despite the institutionalized opportunity for justices to publish a dissenting opinion, an overwhelming share of decisions by subnational constitutional courts are issued unanimously. This can be attributed, first, to the low overall number of cases before subnational constitutional courts that lend themselves to politicization and judicialization. For instance, while the judicially ‘strongest’ decisions in the sample attain an index value of 8 on a 9-point scale, the mean value is just under 1. Therefore, an overwhelming proportion of subnational constitutional court decisions will simply not offer any reputational gains to be reaped from publishing a dissenting opinion. Second, the high share of unanimous decisions evidences the efficiency of the ex ante model of judicial deliberations, which delegates substantial agenda-setting powers to the rapporteur. Holding deliberations that are based on a draft report seem to work towards promoting (ostensibly) consensual decisions because informational advantages and agenda control by the rapporteur, and – to a lesser degree – the president, raise the costs of dissent for fellow colleagues (Bricker 2017, 174; Dalla Pellegrina and Garoupa 2013, 577). Third, procedural requirements such as announcing the intention to dissent as early as possible in the deliberations apparently contribute to fostering consensus. Thus, the rapporteur and the president may take issue with reservations and address point of critique by potential dissenters.

Given that determinants of judicial dissent tested in other contexts proved of only limited significance for the case of German subnational constitutional courts, what is the way forward? The theoretical underpinnings of the analysis would benefit from probing deeper into the substance of the rulings and dissenting opinions. After all, not all dissents are created equal, as they may be authored for very different reasons. It is beyond the scope of this article to explore the substance and various motivations for dissenting opinions. However, taking court decisions on the extension of voting rights to EU foreigners and non-EU aliens at the state and municipal level as an example, dissenting opinions have been shown to both signal ambition for a new legal interpretation and to criticize the departure from established legal doctrine (Thierse 2020). Qualitative research that builds on a more fine-grained analysis of the decisions and the dissenting opinions of the 35 cases identified in this sample certainly has its merits. It could be meaningfully combined with legal linguistics (Goźdź-Roszkowski 2020) and draw on encompassing text data gathered as part of the research for the present article. More generally, future research should dedicate more attention to the relevance of court interactions and the embeddedness of constitutional courts in judicial network structures as determinants of dissent. In this regard, the study of German subnational constitutionals courts as actors in a federated constitutional network has the potential to inspire socio-legal studies of dissent among constitutional courts that form part of the European constitutional network. The history of European integration attests to the fact that constitutional law and constitutional orders are not static, but adapt to new legal circumstances and social challenges. It is therefore conceivable that dissenting opinions are indicative of a more general phenomenon, reflecting disagreement about judicial competences, jurisdictional boundaries and the balance between democratic self-determination and legal integration in multi-level polities.

Acknowledgments

I am indebted to Werner Reutter for sharing his data on subnational constitutional courts and for providing invaluable feedback and encouragement. I would also like to thank Uwe Kranenpohl and Oliver Lembcke for helpful comments. Alina Schulte provided excellent research assistance for this article. An earlier draft of this article was presented at the 28th scientific conference of the German Political Science Association (GPSA/DVPW), which was held online from September 14-16, 2021.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Central Research Development Fund of the University of Bremen under Grant ZF01/2020/FB08/Thierse_Stefan.

Notes

1 In keeping with the terminology known from the U.S. Supreme Court and following Reutter (2021), I employ the term justice for judges at constitutional courts.

2 At the Bavarian constitutional court, justices may write a dissenting opinion, but only in an anonymized form.

3 In the exclusive case of Bavaria, private citizens can, by means of the so-called popular action (Popularklage), initiate an abstract review.

4 Given the congruence of fundamental rights at the subnational and federal level, constitutional complaints before the FCC remain the ‘reservoir’ avenue for the redress of rights violations in states without a separate constitutional complaint procedure.

5 Brandenburg and Berlin provide for the possibility of appointing full-time constitutional justices. It should be noted that constitutional justices at the Austrian Constitutional Court also exercise their office alongside their regular employment as professional judge, attorney at law, prosecutor or law professor (Holzinger 2010). However, Austrian constitutional justices receive a remuneration for their subsidiary office that by far exceeds the rather symbolic allowances that justices at German subnational courts are entitled to.

6 In Bavaria, the right of proposal rests with the president of the Constitutional Court, in Hesse, it is the president of the Higher Regional Court which wield this right (Reutter 2021a, 8).

7 In Hesse, a two-thirds majority applies for the election of the five professional judges, while the seven non-judge members are elected by simple majority.

8 Juris (juris.de) is a commercial provider of legal information that features the most encompassing database on the jurisprudence of courts of all jurisdictions and at all levels of government in Germany.

9 The data on the decisions and the justices are retrievable under https://www.sowi.hu-berlin.de/de/forschung/projekte/LVerfGe/Statistiken (last accessed on January 14, 2022).

10 The Reutter dataset excludes six professional judges at the constitutional court of Hesse which are elected by a designated parliamentary committee. Also, information on constitutional justices from Bremen for the period 2011–2015 had to be completed for this analysis.

11 Where justices serve as ex officio members of the constitutional court, which within my sample is the case only for the president of the constitutional court of Bremen, I imputed the value 100.

12 For instance, while the panels in Bremen, Mecklenburg-Western Pomerania, Saxony-Anhalt or Schleswig-Holstein are composed of 7 justices, the constitutional court of Hesse features panels with 11 justices.