Introduction
Different theories have been proposed to explain judicial behaviour over the last several decades. Those theories can be broadly categorised into three different approaches. Legalistic theories suggest that constitutional judges simply interpret and apply the Constitution with a conformist view of precedent, within a given set of legal norms, and in adherence to the law (hence, individual judicial characteristics have no significant role in explaining court decisions). From a completely different perspective, the attitudinal model sees judicial preferences, with special emphasis on ideology or sincere political viewpoints, as the main determinant.Footnote 1 Finally, strategic theories recognise the importance of judicial preferences, but argue that they are implemented while taking political and institutional realities into account (for example, judges will consider the possible reaction of whomever appointed them, allowing policy goals to interact with disposition preferencesFootnote 2 ). These theories were, however, developed to explain judicial behaviour in the United States, initially at the Supreme Court and, more recently, in the federal courts.
In the European tradition, legal scholars have remarked that the German model of a specialised constitutional court has been more successful in avoiding the contamination of legal decisions and judgments by political determinants. They recognise that the appointment mechanism to these courts is political in nature, but that the resulting internal balance of the court decision-making is overwhelmingly politically neutral.Footnote 3 The apparent depoliticisation of this model, in contrast with the US institutional arrangements, seems to be confirmed by the leading advocates of such design.Footnote 4 However, recent studies in Europe have shown that the patterns of judicial behaviour are not immune to ideology. Still, these patterns somehow play out differently than at American federal courts.Footnote 5 In this respect, there seems to be a gap between recent empirical work and more traditional legal accounts in European constitutional law.
Our work is part of the literature that proposes that European constitutional judges likewise respond to their personal preferences and, at the same time, make their decisions within an institutional context that induces certain strategies.Footnote 6 Given the political nature of constitutional judges’ selection and appointment, it is almost natural to expect them to exhibit the same political preferences as the party that selects them. At the same time, important political and institutional constraints can incentivise consensus in courts where constitutional judges are willing to abdicate their ideologically-preferred outcomes to forge significant majorities that enhance judicial reputation.
In this article, we look at the particular case of Belgium as an excellent example of the European tradition. Until 1970, Belgium was a unitary state. Several major stages of state reform have turned Belgium into a complex federal state composed of language areas, communities and regions. There are four language areas: the Dutch, the French and the German language areas, plus the bilingual Brussels-Capital area. These areas have no powers of their own but merely serve to delimit the competency area for communities and regions. There are three communities: the Flemish-, the French- and the German-speaking Community. They hold competencies in policy areas more oriented to personal welfare (e.g. education, culture, health care and social policy). There are also three regions: the Flemish, the Walloon and the Brussels-Capital region. These are granted powers in areas more directly related to the territory itself (e.g. the economy, employment, agriculture, housing, energy, transport and the environment). The union maintains competency over various federal policies.Footnote 7
Belgium’s federalisation process has led to a complex institutional framework. At the federal level, the main institutions are the King, the federal Parliament (House of Representatives and Senate) and the federal government. Each community and region has its own parliamentary assembly, government and civil service. In 1980, however, Flemish politicians decided to officially merge the Flemish region (and its institutions) with the (institutions of the) Flemish Community, resulting in one Flemish parliament, one Flemish government and one Flemish administration. As the regions and communities are created by the state under the terms of constitutional law, there is no historical background to help establish consensus over their competences and powers. Frequently, the Belgian Constitutional Court entertains conflicts of jurisdiction between the different communities and regions or between these communities and regions and the central government.Footnote 8
Whilst the US Supreme Court has received much empirical attention from legal scholars and political scientists, the empirical debate on other constitutional courts is largely an emerging literature.Footnote 9 However, somewhat remarkably, the Belgian Constitutional Court has not been studied in depth by empirical scholars.Footnote 10 Our article aims to fill that gap.
The implications arising from our study of comparative judicial politics are significant. First, we document the extent to which American-based theories of judicial behaviour are relevant to foreign jurisdictions. Second, we find that the presumption that European courts are politically neutral is empirically misguided. Third, we show that judicial review of federalism is contaminated by ideological and strategic aspects that cannot be explained by purely legalistic accounts.
The paper is structured as follows. An overview of the Belgian Constitutional Court is presented in the first section, including procedure and operation. The hypotheses are discussed in the second section. Regression analysis is presented in the third section. The results are discussed in the fourth section, together with a few examples and legal implications. This is followed by a short conclusion. Please note that some supplementary materials are available online.
The Belgian Constitutional CourtFootnote 11
From the outset, in 1984, the Belgian Constitutional Court was officially inaugurated as the ‘Court of Arbitration’,Footnote 12 delivering its first judgment in April 1985. Its original mission was to supervise compliance with the constitutional division of powers between the state, the communities and the regions. In 1988 and 2003, the powers of the Constitutional Court were significantly extended. They now also include the supervision of adherence to the fundamental rights and liberties guaranteed in Section II of the Constitution (Articles 8 to 32) and of Articles 170 (principle of legality in tax-related matters), 172 (equality in tax-related matters) and 191 (protection of foreign citizens). In 2007, its name was changed to ‘Constitutional Court’.Footnote 13
The Belgian Constitutional Court consists of 12 judges appointed for life (but subject to mandatory retirement at 70) by the King from a list of two candidates proposed respectively by each of the two chambers of Parliament by a majority of at least two-thirds of the voting members.Footnote 14 Six judges must belong to the Dutch language group, six to the French language group. One of the judges must have an adequate knowledge of German. Each linguistic group is composed of three judges with at least five years of experience as a member of parliament, and three judges with a legal background (university professor of law at a Belgian university, magistrate with the Court of Cassation or the Council of State, legal secretary with the Constitutional Court). The judges of each linguistic group elect a President, who presides over the Court for a term of one year, in rotation with the other President.Footnote 15
Normally, cases are heard by a bench of seven judges.Footnote 16 According to Article 59 of the Special Statute of 6 January 1989, presidents participate in all cases, this in order to guarantee continuity and consistency in the case law of the court. One disadvantage of this rule is that the court cannot decide two cases simultaneously.Footnote 17 The other five judges are appointed in accordance with a complex system of alternation which guarantees that each bench has at least three judges from each linguistic group, at least two judges with prior legal qualifications, and two former members of parliament. All justices have to participate in a variety of cases, therefore excluding specialisation. The first judge of each linguistic group, appointed for the case at hand, acts as reporting judge.Footnote 18 Decisions are taken by a simple majority vote. Since under normal circumstances the number of deciding justices is uneven (7), it is unavoidable that one language group will be in the majority. However, since Article 54 of the Statute of 6 January 1989 states that the presidency of the court rotates annually between the two presidents, the language majority also changes each year. This system prevents one language group from abusing its majority position, since the next year it will be in the minority.Footnote 19 By calling for an uneven number of justices, the legislator wished to make sure that the court would be able to reach decisions. A justice is moreover forced to vote in favour or against a particular proposal. Refusal to take a decision is punishable under Article 258 of the Penal Code: justices are hence not allowed to abstain from taking a decision.Footnote 20 Article 57 of the Special Statute of 6 January 1989 states explicitly that Article 258 of the Penal Code (concerning a refusal to grant justice) is also applicable to the justices of the Belgian Constitutional Court.
Although a bench of seven judges is the rule, the Presidents may decide to submit a case to the Belgian Constitutional Court in full session. They can each do so whenever they deem it necessary, and they are obliged to do so when two of the seven judges who make up the (ordinary) bench so request. This option is rarely invoked.Footnote 21 At least ten judges, and in any case an equal number of Dutch-speaking and French-speaking judges, must be present for the full Court to rule. The presiding judge casts a deciding vote in the event of a tie. Obviously, the right to submit a case to the Court in full session combined with casting the deciding vote gives the presiding judge special authority.Footnote 22
Several mechanisms have been implemented with the intent of avoiding overt party alignment. These include the writing of single opinions, secret voting (making it impossible to track individual votes), and opinions written in such a way as to make the decision appear fairly consensual, since no concurring or dissenting opinion is allowed (in contrast to the US practice of issuing separate opinions). The precise method of decision-making within the court (in addition to the fact that decisions are taken by a simple majority) remains unknown. Some information on the decision-making process is revealed in case law and in interviews with presidents of the court. Former President Alex Arts said, in an interview:Footnote 23
There are within the Constitutional Court communitarian (language) contradictions, differing opinions and ideological differences of opinion. There are no overt tensions: we do everything we can to prevent those tensions. We do everything we can to maintain amiability within the court. Everyone realises that this is necessary. No one benefits from open conflicts, since we will still need each other the next day.
Formally, we are required to come to a decision, and in case of a full meeting, seven votes are sufficient. In the event six votes are cast in favour and six votes against, the vote of the functioning president is decisive.
The votes in our country are cast in secret and hence I cannot tell you how often there is no consensus as we decide by majority. But you may assume that consensus is the rule and that decisions taken by simple majority are the exception confirming the rule. My good colleague, the late Louis-Paul Suetens, once told me when I just started here ‘Here no one wins or loses’.
The proportionality test is essential since the reasoning which constitutes the basis of our decisions is never purely legal; there is also a societal opinion behind it. Our decisions combine legal insights and common sense. We try to be ‘wise’. When we test whether an equal or unequal treatment by the legislator is justified, we try to strike a balance between law and justice.
Legal doctrine is critical of the way in which decisions are made by the Belgian Constitutional Court. The criticism is more particularly focused on the non-transparent character of the decision-making process and on the fact that there are no dissenting opinions, making it difficult to discern the true motives behind a decision of the Constitutional Court. For example, Moonen, in several publications, has pleaded in favour of more transparency concerning the decision-making process of the Belgian Constitutional Court. He mentioned, inter alia, ‘To the extent that the limited amount of information available allows us to draw any conclusions, research concerning the Constitutional Court shows that, in the decision-making process, a complex series of factors are taken into account’.Footnote 24 He has also argued that ‘The court itself could provide more clarity in its decisions concerning the choices available, the path chosen, and especially why one solution was chosen over another’.Footnote 25
Former President Alen of the court is, however, strongly opposed to the issuing of dissenting or concurring opinions. He held ‘That would mean the end of the court, I think. Especially in our country, where we have a communitarian divide and it is continually necessary to seek consensus between the different linguistic communities. Public individual opinions would make such consensus impossible or at least very difficult to achieve’.Footnote 26 More recently, De JagereFootnote 27 has empirically examined third-party intervention in procedures before the Belgian Constitutional Court and argues that this ‘inclusiveness’ is relatively limited. Furthermore, she suggests including the opinions of third parties in the decision-making process in order to enhance the democratic nature of decisions. MaesFootnote 28 has also criticised current decision-making by the Constitutional Court, but from a different angle. She pleads inter alia in favour of involving amici curiae in order to increase the quality of the decision-making and she defends publishing dissenting opinions which would contribute to an open debate on constitutional decision-making.
In this article, we focus on conflicts between regions, communities and the central government as regulated by the Belgian Constitution. Each entity may submit a petition to the Belgian Constitutional Court within six months of publication of legislation that it alleges exceeds the competence of the central government or of a particular region or community. The Constitutional Court is the final arbiter concerning the allocation of competences. If a law is declared unconstitutional, it is then entirely or partially annulled.Footnote 29 Judgments annulling a disputed regulation have absolute binding force from the moment they are published in the Moniteur belge. In principle, the annulment has retroactive effect, which means that the annulled statute must be deemed to never have existed.Footnote 30 The decision is not open to appeal.
The Belgian Constitutional Court handles several other types of cases, not only the ones examined here. Aside from the authorities designated by statute (the Council of Ministers, etc.), an action for annulment may also be brought by any (natural or legal) person who has a justifiable interest.Footnote 31 Also, any tribunal may refer preliminary questions to the Constitutional Court.Footnote 32 In most of these cases, however, neither the plaintiff nor the defendant has a clear political identification. We have chosen to concentrate on disputes between the federal government and regional governments as well as on disputes where opposing parties are both regional governments so that the existence of political conflict or alignment can be easily assessed and will not require additional subjective consideration by the researchers.
Several Belgian scholars have conducted empirical research with respect to the cases decided by the Belgian Constitutional Court.Footnote 33 Those studies provide an analysis of the number of cases (a random sample of the entire population) and the actors involved, but do not correlate those results to the political background of the judges as we do in this article. In fact, those studies focus on reasoning and justification, deliberation, and constitutional dialogue, concluding that the Belgian Constitutional Court vacillates between deference to the legislator and remedying legislation (with preponderant use of the equality clause). Patterns of partisan behaviour are not directly addressed by those empirical studies.
The hypotheses and dataset
Relevant actors at the Belgian Constitutional Court
Our article focuses on cases litigated at the Belgian Constitutional Court where one side is the federal government and the other side is a regional government, as well as on cases where opposing parties are both regional governments. As in any federal arrangement, each party naturally wants to maximise its jurisdiction. At some point, enhancing the political competence of one party necessarily reduces the other party’s influence, be it in the federal government or in a regional government.
The legislation reviewed by the Belgian Constitutional Court reflects a conflict of competence that was unable to be solved politically. One side has a political interest in enacting legislation that is challenged by the other side since a settlement could not be negotiated. In that respect, Belgian federal and regional legislation follows standard political cycles. Federal and regional institutions are frequently dominated by non-identical political majorities, and conflicts that cannot be reconciled without the intervention of the Constitutional Court naturally emerge between jurisdictions.Footnote 34
When reviewing conflicts of jurisdiction, constitutional judges are inevitably confronted with two competing political interests. The political nature and the political implications of constitutional review in this context are understandable and inevitable.
We have gathered information about cases which allows us to identify whether or not the Belgian Constitutional Court decision is favourable to one party in particular. However, as explained previously, we cannot directly observe how each constitutional judge votes in any particular decision (in the best civil law tradition, individual votes are not publicised).Footnote 35 For each particular decision, we know the composition of the Constitutional Court, that is, the constitutional judges that actually participated in the decision (hereafter, we shall refer to the composition of the Constitutional Court at each decision as the Panel of judges).Footnote 36 We also know the proportion of judges in the Panel belonging to each political coalition. For each decision, the three most critical judges are the President and the Rapporteurs (the constitutional judges who draft the text of the decision, that is, the reporting judges). These judges play strategic roles in amassing a majority of votes to pass a particular opinion. They provide the focal point that, presumably, significantly influences the final outcome.Footnote 37
Unlike at the US Supreme Court, the President of the Belgian Constitutional Court is elected by his fellow judges. More precisely, the judges of each linguistic group elect a President who presides over the Court for a term of one year, in rotation with the other President.Footnote 38 The President of the Panel is, in general, the President of the Court. In the absence of the President of the Court, the longest-serving judge or, if necessary, the oldest judge belonging to the same language group, serves as President of the Panel. When a case is handled in a language that does not correspond with the language group the President of the Court belongs to, he or she is obliged to delegate his powers to the other President. (The language to be used is always the language of the Petitioner.) In our sample (1985-2012), we have 15 different Presidents of the Court and an equal number of Presidents of the Panel.
The two Rapporteurs are appointed by means of an identical system of rotation. Specifically, the first judge of each linguistic group - appointed for the case at hand - acts as reporting judge. Unlike at the US Supreme Court, the Rapporteur is chosen before the initial meeting of the Panel and is therefore a member of the judicial majority. At the same time, and unlike in some other countries (e.g. Italy), the Rapporteur is not picked by the President based upon a criterion of expertise.
Hypotheses
At this stage, we can make two contrasting hypotheses. The first hypothesis is simply that the likelihood of a Petitioner succeeding is not influenced by the President being affiliated with the Petitioner’s coalition, by the Rapporteurs being affiliated with the Petitioner’s coalition, or by the proportion of judges in the Panel affiliated with the Petitioner’s coalition. The alternative (second) hypothesis is that the likelihood of a Petitioner succeeding is influenced by the President being affiliated with the Petitioner’s coalition, by the Rapporteurs being affiliated with the Petitioner’s coalition, or by the proportion of judges in the Panel affiliated with the Petitioner’s coalition.
The literature on comparative judicial politics supports the first hypothesis. By virtue of institutional arrangements, constitutional judges in European courts are largely kept insulated from political interests. This is also consistent with legalistic accounts, according to which there should be no correlation between the likelihood of a Petitioner succeeding and measures of political preference.
Two additional reasons could support this hypothesis: first, Shapiro’s conjecture that central governments usually win in cases against regional governments, since civil law courts are less influential and excessively dependent upon the federal government.Footnote 39 In other words, political alignment is not systematic because the particular influence of the federal government vis-à-vis regional governments dominates constitutional adjudication in this context (independent of any particular party interest). Second, the economic theory of judicial independence sees constitutional review as an instrument for achieving an effective balance of federal and regional powers. In order to maximise influence and perform their role as referees, constitutional judges are not systematically aligned with particular political interests, given the need to foster a perception of neutrality in reference to the litigation in the Court.Footnote 40
The alternative hypothesis is in line with standard political science theories, namely the attitudinal model and the strategic model.Footnote 41 By these two accounts, a correlation should exist between the likelihood of a Petitioner succeeding and measures of political preference. In particular, when policy preferences differ between a Petitioner and the Court, we should expect decisions against the wishes of a Petitioner.
Belgian constitutional judges are appointed by means of a complex institutional arrangement that is, however, greatly influenced by national political parties. Those making the appointments will select individuals who are ideologically close to their policy preferences (preferences are expected to be aligned between ‘appointer’ and appointee; therefore, in a model of sincere voting such as the attitudinal model, we should expect the appointee to reflect the preferences of the ‘appointer’). At the same time, in a context of limited tenure (life tenure subject to mandatory retirement at the age of 70) and with an eye to future appointments (such as positions in future administrations, or other political sinecures), constitutional judges seriously weigh the political repercussions of Court decisions for their ‘appointers’ (the appointee should reflect, for strategic reasons, the preferences of the ‘appointer’ ).Footnote 42 Therefore, we expect consistent alignment in the way constitutional judges vote and the interests of their ‘appointers’ (due to both selection and incentives). This reasoning applies to all constitutional judges, including the President and the Rapporteurs. Due to a lack of individual observations, here we investigate the application of this reasoning to the President, the Rapporteurs and the share of judges in the Panel. However, we should take care to emphasise the pivotal role played by the President and the Rapporteurs, as explained in previous sections, which, on the whole, strengthens our empirical strategy.Footnote 43
The two (mutually exclusive) hypotheses are tested on a panel of data recording all decisions of the Belgian Constitutional Court during the period 1985-2012 concerning disputes where one side is the federal government and the other side is a regional government, as well as cases where opposing parties are both regional governments. (This is the entire population of cases, and not a random sample as in previous studies on the Constitutional Court).Footnote 44 We expect political preferences to be extremely relevant in the area of judicial review of federalism. In the 28-year period examined, there were 136 such cases: 96 cases (71%) where one side is the federal government and the other side is a regional government,Footnote 45 and 40 cases (29%) where both sides are regional governments.
Dataset
Data has been drawn from the Belgian Constitutional Court website, which is a fully comprehensive collection of all the decisions taken by the Court each year.Footnote 46 Other publicly available sources, such as the websites of the several parliamentsFootnote 47 and the most important Belgian press outlets (such as De Standaard, De Tijd, Le Soir, La Libre Belgique) have also been used to associate judges with their political preferences.Footnote 48 For each petition, we collected information regarding both Plaintiffs and Defendants (federal government or regions), the declaration of the constitutionality of the law at stake, the date the petition was filed, the date of the decision, and the composition of the Panel of judges,Footnote 49 noting personal characteristics of its members.
A combination of factors characterises the disputes brought to the attention of the Belgian Constitutional Court. First, constitutionality can be questioned either by a region on a federal law, by the federal government on a regional law, or by a region on the law of another region.Footnote 50 In line with this, we constructed a set of binary variables each taking value one when either the federal government or a given region are plaintiffs, and zero otherwise. Similarly, we also generated a parallel set of dummies indicating whether the federal government or a given region are Defendants. Similarly, we built a dummy variable taking value one when the Petitioner succeeds in a dispute (Petitioner succeeds) and zero otherwise. Success of the Petitioner occurs when the latter questions a Defendant’s law and at least one article of that law has been declared unconstitutional.Footnote 51 , Footnote 52 From Table 1, it turns out that the Petitioner succeeds in 70% of cases. However, breaking down the rate of success by the type of Petitioner, one observes that the federal government succeeds less frequently (63%) than all regions except for Flanders (60%). The French Community succeeds 85% of the time and the Brussels region 71%, whereas Wallonia always succeeds.
Table 1 Summary statistics
Observations: 136.
(†) Computed according to the preferences of the median justice in the panel.
(††) No. of parties present in both coalitions/ total No. of parties (pooling the two coalitions)
(†††) Days, initial date: Jan. 1, 1900.
Second, in some cases a petition involves more than one decision because: (1) the same actor raises different concerns that each require a separate decision; or (2) the same petition involves several actors. To reflect this, in the process of data collection we: (a) recorded a declaration of unconstitutionality each time at least one concern was at stake (i.e. an article of law) did not comply with the Constitution (70% of the sampled cases); and (b) when more than one actor took part in a petition, the decision of the Court was entered more than once. The choice for (b), in particular, is important in order to allow the assignment of a political affiliation to different actors involved in the same petition.Footnote 53
Third, we collected information on personal characteristics of the judges, such as Date of appointment to the Court, Age at the time of appointment (54 on average, see Table 1), Age at the time of decision (65 on average, see Table 1), and Gender (93% of the Panel justices are male, and only in 4% of the cases are the Presidents women, as emerges from Table 1).Footnote 54 Such features are aimed at capturing the effect of possible career concerns. For example, individuals in different age groups may have different career concerns (considering, also, that judges must retire at the age of 70).Footnote 55 This could push them to either show a greater degree of independence or, conversely, to favour a particular political party or coalition, depending upon whether the judge intends to take either a politically or a non-politically related job immediately upon leaving the bench. The gender variable controls for similar effects. In the same vein, we also accounted for whether an individual Judge has ever become President (58% of the Panel justices, slightly higher for the First Rapporteur, i.e. 68%), has a Legal background (i.e. was a law professor, 52% of the Panel justices, 60% for the President), or belongs to the French quota (52% of the Panel justices). Average values of all these variables have also been computed on the Panel of judges deciding each dispute.
Fourth, judges’ political preferences have been assigned as follows. If a judge is a former member of parliament, a preference is associated according to the political party he belonged to. Obviously, no subjective assessment is required for these judges. We followed a different approach when alleging, and later aligning, preferences for judges with a legal background. For these judges, we used elements imputing a specific political orientation expressed prior to the judge’s appointment to the Court (e.g. advisor for a political party; candidate for the elections for a political party; chef de cabinet).Footnote 56 For all judges with a legal background, we also allowed for political neutrality in the interest of empirical robustness (see the following section).
Fifth, using information on the outcomes of political and regional elections, we built dummy variables taking value one if there was political alignment between members of the Panel and the Petitioner, both at the moment of the sentence (Date of decision) and at the date the petition is filed (Date of filing).Footnote 57 This allows the testing of two possibilities: the first possibility is that political alignment between the Petitioner and actors in the Court reflects ideological preferences in terms of policy outcome (in this case the political alignment at the date of filing should be relevant in terms of the decision). The second possibility is that political alignment may result from dispositional concerns in terms of actual policy implications (in this case the political alignment at the decision date should be relevant). An additional caveat is that, since judges are allegedly affiliated with one party, whereas both the federal and the regional governments are coalitions, we assumed that there is political alignment between a given judge of the Panel and the Petitioner if the latter’s coalition includes the party the judge is affiliated with. For each Panel, we considered the alleged political preference of the President, the First Judge Rapporteur and the median judge in the Panel. Then, in line with this definition, we computed the political alignment of these actors with both the Petitioner and the Defendant. From the descriptive statistics in Table 1 it emerges that in 60% of the cases there is political alignment between the Petitioner and the President and in 68% of the cases there is political alignment between the Petitioner and the First Rapporteur, whereas in almost half of the cases the median justice of the Panel is aligned with the Petitioner (46%). As for the Defendant, alignment with the President occurs 46% of the time, whereas in 60% of the cases there is alignment with the First Rapporteur. In 60% of the cases the median justice of the Panel is aligned with the Defendant.
In Figures 1-3 we combined the Petitioner’s success rate with its political alignment with members of the Court. First, by comparing the cumulative height of group 1’s bars (success) with the cumulative height of group 2’s bars (failures) in both Figures 1 and 2, it becomes clear that the Petitioner tends to win more frequently, regardless of any kind of political alignment. For our purposes, the most relevant feature is that the number of successes of the Petitioner appears to be considerably higher when there is political alignment with either the President of the Panel or with the First Rapporteur (first bar, Figures 1 and 2) compared to the case of no alignment (second bar, Figures 1 and 2). In addition, the equal height of the third and the fourth bars in Figure 1 confirms the previous evidence, suggesting that the Petitioner’s number of failures is independent of political alignment with the President. Instead, the third and the fourth bars in Figure 2 indicate that there is a higher chance of failure for the Petitioner when there is alignment with the Rapporteur. Overall, in the case of the Petitioner’s alignment with the Rapporteur, preliminary evidence is not as straightforward as in the case of alignment with the President. In Figure 3, the Petitioner’s successes (straight line) and failures (dashed line) are reported as a function of the percentage of judges in the Panel being aligned to the Petitioner. It is clear from the graph that that the number of the Petitioner’s successes increases more rapidly than the number of failures as the proportion of members in the Panel having the same political preferences grows.
Figure 1 Petitioner success and failure according to political alignment with the Court President Group 1: number of Petitioners’ successes. Group 2: number of Petitioners’ failures. Vertical lines refer to cases in which the President’s alleged political preferences are aligned with those of the Petitioner. Horizontal lines refer to cases in which the President’s alleged political preferences are not aligned with those of the Petitioner.
Figure 2 Petitioner success and failure according to political alignment with the First Rapporteur Group 1: number of Petitioners’ successes. Group 2: number of Petitioners’ failures. Vertical lines refer to cases in which the First Rapporteur’s alleged political preferences are aligned with those of the Petitioner. Horizontal lines refer to cases in which the First Rapporteur’s alleged political preferences are not aligned with those of the Petitioner.
Figure 3 Petitioners’ successes and failures according to the political preferences of Panel judges Horizontal axis: percentage of judges in the Panel having the same political preferences as the Petitioner. Vertical axis: number of Petitioners’ successes (continuous line) and failures (dashed line). Note that the figures on the vertical axis are to be read as ‘Number of cases favourable (or unfavourable) to the Petitioner when at least 0.1, 0.2, 0.3, etc. judges in the Panel are aligned with its political preferences’.
Next, we created proxies of the quality of the Petitioner’s issue. The first reflects the political alignment between the Petitioner and the Defendant and is computed as the ratio between the number of parties common to the two coalitions and the total number of parties involved in the dispute (i.e., considering the two coalitions pooled together) (Political alignment between Petitioner and Defendant). Since we implicitly assume that disputes are political in nature, this is aimed at controlling for a presumed different quality of the plaintiff’s argument in case the Petitioner and the Defendant belong to the same coalition. This measure has also been interacted with partisan alignment variables to account for the possible smoothing effect that a Defendant with preferences identical to those of both the Court and the Petitioner may have on the rate of success of the latter. From descriptive statistics in Table 1, it turns out that alignment between the Petitioner and the Defendant occurs in 28% of cases. The second variable follows the discussion by Dalla Pellegrina and Garoupa (2013) in that it measures the time gap between the date in which a petition has been filed to the decision date (Time from filing to decision) (on average 1.2 years, Table 1). This can also be interpreted as an additional measure of the quality of the Petitioner’s issue, being higher the longer it takes to reach a decision. The last variable used to measure the strength of the Petitioners’ arguments is the number of parties involved in the coalition that initiated the legal challenge (Petitioner: Number of parties in coalition). On the one hand, the larger the coalition, the less likely it is that parties will agree to challenge the actions of their counterparts on reasons of policy. On the other hand, however, if the Petitioner does opt to challenge, we expect that there should be good legal/ non-policy related reasons for doing so (i.e. the issue is of higher quality).
Finally, we considered Petitioner and Defendant fixed-effects, and their interaction. Fixed-effects for actors of the Court have also been included. All fixed-effects are intended to capture time-invariant specificities, such as the fact that Petitioners may differ in terms of the strength of their own representatives. Further details regarding the descriptive statistics can be found in Table 1.
REGRESSION ANALYSIS
We estimate the following equation:
$$\eqalignno { \Pr \left( {Petitioner\,succeeds_{{it}} \,{\equals}\,1} \right)\,{\equals}\, &\Phi ( {\alpha _{1} Alignment\,between\,judges\,and\,Petitioner_{{it}}} \cr & {\plus}\alpha _{2} X_{{it}} {\plus}\mu _{j} {\plus}\delta _{h})$$
where each observation refers to the i-th decision of the Court occurring at time t, involving Petitioner j and Judge h (either President, First Rapporteur or the entire Panel). In particular, µ
j
and δ
h
are, respectively, Petitioner and Court members’ fixed-effects. The dependent binary variable, Petitioner succeeds, is equal to one if the Petitioner succeeds in decision i either decided or filed at time t. We assume that
$$\Phi $$
has a normal distribution.
Our main purpose is to investigate whether the rate of success of the Petitioner is influenced by the alleged political position of constitutional judges (Alignment between judges and the Petitioner). Therefore, positive and significant parameters associated to the variables measuring political alignment between the Petitioner and members of the Court (α 1 ) should point in favour of the attitudinal and strategic models, while a more legalistic view would be sustained in case α 1 turns out not to be significant.
We account for features of the decision (X it ) which may influence the probability that the Petitioner succeeds, such as whether litigation involves the government rather than two individual regions, the identity of the parties involved, the number of parties in the coalitions that initiated the legal challenge, filing and decision dates, and time elapsed from filing to decision. We further evaluate the contribution of previously mentioned individual characteristics of Judge-President, Judges-Rapporteurs, and the Panel of judges taking part to each individual decision of the Court.
In addition, the lack of independence within decisions arising from the same petitions engenders the need for clustering the standard errors of the estimated parameters at the petition level. In our specific context, this should produce unbiased estimates of standard errors,Footnote 58 having also good asymptotic properties.Footnote 59
Equation (1) is estimated through a probit regression. Results are reported in Tables 2 and 3, where columns differ depending on the sub-sets of covariates included.Footnote 60 In particular, in Table 2 we account for characteristics of the case and of the parties involved, whereas in Table 3 we consider personal characteristics of the judges and judge fixed-effects. Due to the presence of a relatively high number of dummy variables, we performed regressions in which the political alignment between members of the Court and, respectively, the President, the Rapporteur, and members of the Panel are included separately. Note that in both Tables 2 and 3 we consider the political alignment between the government and Court actors at the decision date, whereas in further analysis (Tables 4 and 5) we study the same estimates considering alignment at the date of filing. We measure the goodness-of-fit through the ratio of the explained to unexplained variance using Count R-squared.Footnote 61
Table 2 Probability that the Petitioner wins a dispute
Probit estimates. Marginal effects are reported. Standard Errors clustered at the petition level in parenthesis. * significant at 10%; ** significant at 5%; *** significant at 1%
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is pronounced (decision date). All estimates include Petition fixed-effects.
Petitioner Wallonia, Defendant government*Petitioner Wallonia, Case involves two regions*Petitioner Wallonia, Case involves two regions*Petitioner Brussels: dropped because the Petitioner wins 100% of the time.
(†) Computed according to the preferences of the median justice in the panel.
(††) No. of parties present in both coalitions/ total No. of parties (pooling the two coalitions).
Table 3 Probability that the Petitioner wins a dispute as a function of judges’ personal characteristics
Probit estimates. Marginal effects are reported. Standard Errors clustered at the petition level in parenthesis. * significant at 10%; ** significant at 5%; *** significant at 1%
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is pronounced (decision date). All estimates include Petition fixed-effects.
(†) Columns (g)-(h): Computed according to the preferences of the median justice in the panel.
(††) Column (g): Share of members.
(†††) Column (g): mean values computed on the judges taking part to each Panel.
Table 4 Probability that the Petitioner wins a dispute – Political alignment at the date of filing Dependent variable: Petitioner succeeds
Probit estimates. Marginal effects are reported. Standard Errors clustered at the petition level in parenthesis. * significant at 10%; ** significant at 5%; *** significant at 1%
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is filed (filing date). All estimates include Petition fixed-effects.
Petitioner Wallonia, Defendant government*Petitioner Wallonia, Case involves two regions*Petitioner Wallonia, Case involves two regions*Petitioner Brussels: dropped because the Petitioner wins 100% of the time.
(†) Computed according to the preferences of the median justice in the panel.
(††) No. of parties present in both coalitions/ total No. of parties (pooling the two coalitions).
Table 5 Probability that the Petitioner wins a dispute as a function of judges’ personal characteristics – Political alignment at the date of filing Dependent variable: Petitioner succeeds
Probit estimates. Marginal effects are reported. Standard Errors clustered at the petition level in parenthesis. * significant at 10%; ** significant at 5%; *** significant at 1%
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is filed (filing date). All estimates include Petition fixed-effects.
(†) Columns (g)-(h): Computed according to the preferences of the median justice in the panel.
(††) Column (g): Share of members.
(†††) Column (g): mean values computed on the judges taking part to each Panel.
We further check for empirical robustness by concentrating on judges with a legal background. As discussed previously, since these judges are appointed on the basis of their legal rather than their political backgrounds,Footnote 62 their political positions cannot easily be deduced, or, even if it is possible, the results are questionable. A way of dealing with this problem is to adopt a criterion based on random imputation of political preferences for all those judges.Footnote 63 Random numbers from 1 to 7 (each one corresponding to a given political party) are assigned by extracting from a uniform distribution.Footnote 64 We performed 300 random draws and an equivalent number of regressions per each specification corresponding to columns (a)-(i) and (a)-(h) of Tables 2 and 3 respectively. Then, we computed the average values of the parameters estimated from each block of 300 regressions along with their standard errors. Results are reported in Tables 4 and 5. Finally, we excluded from the sample all cases involving other acts regulating the same issue (Tables 6 and 7).
Table 6 Probability that the Petitioner wins a dispute Dependent variable: Petitioner succeeds - Random assignment of political preferences for judges with legal background
Probit estimates. Average marginal effects from 300 iterations are reported. Average Standard Errors from 300 iterations in parenthesis. Standard Errors are clustered at the petition level. Number of iterations /300 providing significant parameters at 10% or better in brackets.
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is pronounced (decision date). All estimates include Petition fixed-effects.
Petitioner Wallonia, Defendant government*Petitioner Wallonia, Case involves two regions*Petitioner Wallonia, Case involves two regions*Petitioner Brussels: dropped because the Petitioner wins 100% of the times.
Table 7 Probability that the Petitioner wins a dispute as a function of judges’ personal characteristics Dependent variable: Petitioner succeeds - Random assignment of political preferences for judges with legal background
Probit estimates. Average marginal effects from 300 iterations are reported.
Average Standard Errors from 300 iterations in parenthesis. Standard Errors are clustered at the petition level.
Number of iterations /300 providing significant parameters at 10% or better in brackets.
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is pronounced (decision date). All estimates include Petition fixed-effects.
(†) Columns (g)-(h): Computed according to the preferences of the median justice in the panel.
(††) Column (g): Share of members.
(†††) Column (g): mean values computed on the judges taking part to each Panel.
Discussion
The Empirical Importance of Political Preferences
The econometric results seem to provide evidence of a significant positive relationship between the President’s alleged political preference for the Petitioner’s coalition and the success of the Petitioner (Tables 2 and 3). In particular, the baseline specification of the model (see Table 2, column (a1)) predicts that a discrete change from 0 to 1 of the dummy variable associated to the political alignment between the President and the Petitioner increases the probability that the latter succeeds by 24.2%. This outcome is in line with the specifications that control for the identity of both Petitioners and Defendants (the estimated Petitioner’s rate of success is 24.1%, Table 2, columns (b)-(c)), whereas including the interaction of the President’s preferences for the Petitioner’s coalition with the alignment between Petitioner and Defendant increases the odds of success of the Petitioner by 60% (Table 2, column (a2)).Footnote 65
According to the output of Table 2, there also seems to be a positive relationship between the success of the Petitioner and the political preference of the median judge in the Panel for the Petitioner’s coalition, although this appears less systematic as it involves only the specification in which we account for the alignment between the Petitioner and the Defendant. In this case, estimates indicate that the likelihood of success of the Petitioner when aligned with the median judge increases by 77.8% (Table 2, column (g2)). The same holds true when considering the alignment of the Petitioner’s preferences with those of the First Rapporteur, which raises the rate of success of the Petitioner by 23.5% (Table 2, column (d2)).
In terms of sign and significance, the results reported in Table 2 are confirmed when referring to the model that accounts for specific characteristics of the judges (Table 3). First, the mean value of the parameters in columns (a)-(c) in Table 3 indicates that political alignment between the Petitioner and the President of the Panel significantly increases the rate of success of the former by 21.8 %. A weaker positive influence of the alleged alignment between the median judge in the Panel and the Petitioner on the odds of success of the latter is also confirmed. More in detail, it turns out that a discrete change from 0 to 1 of the dummy variable associated to the alignment between these two actors increases the probability that the Petitioner wins by 23.9% (Table 3, column (h)). On the other hand, there is no evidence of significant connections between the alleged affiliation of the First Rapporteur and that of the Petitioner’s coalition.
There is, instead, weaker evidence for the fact that political alignment between the Petitioner and actors of the Court at the date of filing, rather than at the decision date, affects the decision of the Court itself, especially as long as the median judge in the Panel and the First Rapporteur are concerned (see Tables 4 and 5). In line with the discussion in previous sections, this suggests that the likelihood of success of the Petitioner mostly depends on dispositional concerns in terms of actual policy implications and also, but to a lesser extent, on ideological components related to the expected policy outcome.
In Table 3 (column (g)) one can observe that panel judges with a legal background have a critical role in increasing the rate of success of the Petitioner. Such an outcome has led us to further investigate this feature modifying the mechanism of assigning political preferences to these judges. The results of the analysis conducted under random assignment of political preferences to judges with a legal background (see previous section for methodological details) provide intriguing evidence. Looking at the results in Tables 4 and 5, one can observe that, when disregarding the preferences of judges with a legal background, the estimated parameters related to the alignment between the Petitioner and Court members (where significant) are slightly lower in magnitude compared to those obtained in the main analysis of Tables 2 and 3. While mostly relevant, the number of regressions that yield significant parameters (indicated in square brackets in Tables 4 and 5) is relatively small. In other words, it seems that the presence of judges with a legal background tends to strengthen the effects of the political alignment between the Petitioner and the Court, an outcome that is particularly evident with respect to the President. This tends to support the evidence in favour of the higher bias that judges with a legal background show in favour of the Petitioner. In fact, it is possible that lawyers and professors are less neutral because: (i) they are not professional career judges, and therefore their appointments could be more politicised; and (ii) they may be more concerned about (political) future appointments since they do not have a judicial career to return to.
In line with the discussion above, we also performed estimates excluding cases involving other acts regulating the same issue so as to avoid alleging political alliances in the event of the Petitioner’s failure, since another norm is actually at stake. As reported in Tables 8 and 9, we find that the parameters of interest, although less significant, are in line with the estimates in the main analysis.
Table 8 Probability that the Petitioner wins a dispute Dependent variable: Petitioner succeeds – Excluding cases involving other acts regulating the same issue
Probit estimates. Marginal effects are reported. Standard Errors clustered at the petition level in parenthesis. * significant at 10%; ** significant at 5%; *** significant at 1%
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is pronounced (decision date). All estimates include Petition fixed-effects.
Petitioner Wallonia, Defendant government*Petitioner Wallonia, Case involves two regions*Petitioner Wallonia, Case involves two regions*Petitioner Brussels: dropped because the Petitioner wins 100% of the time.
(†) Computed according to the preferences of the median justice in the panel.
(††) No. of parties present in both coalitions/ total No. of parties (pooling the two coalitions).
Table 9 Probability that the Petitioner wins a dispute as a function of judges’ personal characteristics Dependent variable: Petitioner succeeds – Excluding cases involving other acts regulating the same issue
Probit estimates. Marginal effects are reported. Standard Errors clustered at the petition level in parenthesis.
* significant at 10%; ** significant at 5%; *** significant at 1%
All estimates refer to political alignment between Petitioner/Defendant and judges of the Court at the time the sentence is pronounced (decision date). All estimates include Petition fixed-effects.
(†) Columns (g)-(h): Computed according to the preferences of the median justice in the panel.
(††) Column (g): Share of members.
(†††) Column (g): mean values computed on the judges taking part to each Panel.
Overall, our results seem to provide evidence that when the President is allegedly affiliated with the Petitioner’s coalition, and (to a lesser extent) when the likelihood that the median judge in the Panel supports the Petitioner increases, the odds of success of the Petitioner go up in a statistically significant way. We tried to investigate whether this outcome could be driven by Petitioners’ strategic behaviour. In fact, the Plaintiff may be more inclined to choose to make a petition to the Court when it is more likely to be supported by Court members.Footnote 66 Relating the political preferences of both the Plaintiff and the Defendant to the alleged political affiliation of the President of the Panel, we find some evidence of such behaviour. More precisely, according to the descriptive statistics, it turns out that in 60% of cases the Petitioner is allegedly aligned with the President’s political preferences at the decision date, but only 46% of Defendants (Table 10). The percentages appear even more polarised when considering the subset of disputes in which a region is the Plaintiff (51% versus 32%) and disputes that involve only regions (55% versus 13%). Similar indications hold in relation to the alignment between the First Rapporteur and either the Petitioner or the Defendant. No clear-cut suggestions emerge with respect to the Panel of judges. Analogous figures are observed once political alignments at the date of filing are considered (Table 11).
Table 10 Actors involved in disputes decided by the Constitutional Court: Political alignment at the decision date
Table 11 Actors involved in disputes decided by the Constitutional Court: Political alignment at the date of filing
There is also notable evidence arising from the significant parameters associated to some covariates. First, according to the output in Table 2, it turns out that when the political preferences of the Petitioner and the Defendant are aligned, the success rate of the former is 75% lower than otherwise (mean value of the significant parameters associated to the variable Alignment between Petitioner and Defendant, all columns). The estimated parameters in Table 3 are lesser in magnitude, indicating that the success rate of the Petitioner is 35% lower (mean value of the significant parameters, columns (a)-(c)) when there is alignment with the Defendant compared to the situations of conflict. This is somehow in contrast with the view that the quality of the plaintiff’s argument is higher when the Petitioner and the Defendant belong to the same coalition, although it is consistent with the observation that the Court is more reluctant to favour the Petitioner when complex questions are involved (consequently, litigation involving aligned political coalitions on both sides is likely to indicate legal complexity). Another possible interpretation is that there is less conflict in cases where the Petitioner and the Defendant are politically aligned, so any potentially exploitable link between the Petitioner and the actors in the Court becomes less relevant in terms of increasing the former’s odds of success.
Second, there is strong evidence that the government is less likely to be favoured when it is the Plaintiff in a dispute (Table 2, columns (a1)-(a2), (d1)-(d2), (g1)-(g2)), suggesting that the government is at some disadvantage in terms of influence on the Court as compared to the regions. According to Table 2, the mean value of the significant parameters associated with the variable Petitioner: government indicates that the government tends to record a 48.9% lower likelihood of success compared to regions. Estimates in Table 3 report a similar average value of the significant parameters (42.7%).
Third, the negative and significant parameters associated to the date of decision suggest that, in more recent disputes, the rate of success of the Petitioner has experienced a general reduction. Instead, the time elapsing from the date of filing to the date of decision, which we included as a proxy of the quality of the Petitioner’s issue, seems to have no effect on its probability of success.
Fourth, interesting results emerge from interacting Petitioners’ and Defendants’ fixed-effects. As shown by columns (a1)-(a2), (d1)-(d2), (g1)-(g2) of Table 2, Wallonia (the residual category) tends to win more frequently as compared to both the government and other regions when regions are plaintiffs. In addition, it turns out that when raising an issue against the Brussels capital region, the government has systematically fewer chances to win as compared to when they face other regions or communities (Table 2, columns (b), (e) and (h)). However, in the opposite scenario (Brussels questions a law of the Government), this does not occur. Rather, in such situations Flanders and the French Community seem to be more frequently penalised (columns (c), (f) and (i), Table 2)), whereas when a dispute involves two regions, Wallonia and Brussels are systematically more successful.Footnote 67
Additional interesting results stem from the regression analysis that accounts for judges’ personal characteristics. In particular, judges who were relatively older at their date of appointment tend to support the Petitioner more frequently, as the positive sign of the parameters associated to the variable Age at appointment indicates, especially as far as Presidents are concerned (Table 3, column (b)). If the date of appointment correctly reflects a judges’ expertise, this outcome should indicate that more experienced Presidents are likely to retire after their time on the bench, hence they may have less interest in asserting independence. This result is coherent with the negative parameter of the variable Date of appointment, since more recently appointed Presidents (Table 3, column (b)) are more likely to care about appearing impartial, and thus are less inclined to accord favourable treatment to the Petitioner in response to the systematically higher rate of success (as emerges from Table 1). The Age at the Decision variable is never significant. Besides the aforementioned effects of judges with a legal background, First Rapporteurs from the French quota are also more likely to express bias in favour of the Petitioner (Table 3, column (f)). This may – at least partially – explain the high rate of success of Wallonia in disputes before the Constitutional Court. Finally, all count R2 reported at the bottom of Tables 2 and 3 indicate that the specified models provide a relatively high goodness of fit.
Examples
Following upon our empirical analysis, we document our discussion with examples in line with our results. For example, in the Tobacco Advertisement case (CC Be. 102/1999), the Belgian federal law of 10 December 1997 abolishing tobacco advertisements was challenged before the Belgian Constitutional Court by (inter alia) the Walloon Executive. At the time, the two Socialist and the two Catholic parties (from both sides of the language border) were in power. The specific statute, however, was passed by a majority of Flemish parties in the federal parliament. The law was highly politicised. It affected not only the tobacco industry, but also industry-sponsored sporting events such as the Francorchamps Formula I races. This put the economic interests of the Walloon Region at stake. The Court, after reviewing the law with an eye to competing allocating rules and fundamental rights, upheld it, except for one ‘detail’. The ban on tobacco advertisements did not apply to internationally organised events that took place before 31 July 2003, which was the date the court expected the introduction of a Europe-wide ban based on a European Directive. Some legal commentators considered this a ‘political verdict’,Footnote 68 and some politicians stated that the Court had jeopardised its status as an impartial decision-maker.Footnote 69 The case was decided by 10 judges, with five from the Dutch language group and five from the French language group. Both the presiding judge and the First Rapporteur were from the French language group.
Another case deals with the competences in the domain of the export of nuclear material and equipment (CC Be. 168/2011). In 2003, the competence for dual-use items (peaceful and military) was transferred to the regions, simultaneously with the control of the trade in military equipment. The regional authorities are thus responsible for issuing permits for the import, export and transit of dual-use items and technologies in Belgium. Note, however, that export control policy for dual-use items is an exclusive competence of the European Union. Only the EU has legislative powers, and the regions are thus only responsible for the implementation of European legislation. Consequently, their scope for taking decisions is limited. It is pertinent to this case that, before the European legislation was introduced, nuclear material was already controlled by the Belgian Act of 9 February 1981 on the export conditions for nuclear material and equipment and technical information. This act conformed to international agreements on the non-proliferation of nuclear weapons. Based on the Act, the trade in nuclear materials requires additional approval by the Federal Minister of Energy (in addition to a licence issued by the regional authorities in accordance with the EU Regulation). The Federal Minister bases his decision on recommendations from the Advisory Committee on the Non-proliferation of Nuclear Weapons. The political parties in Flanders striving for Flemish independence (NV-A) and for more competences for the region (CD&V) found that the federal Committee was too strict and that it hindered Flemish economic interests. When, in 2010, an amendment to the Act of 1981 was adopted introducing even stricter rules, the Flemish Executive tabled an appeal against it. The Court did not annul the amendment. Both the presiding judge and the First Rapporteur were associated with a Flemish party (the liberals) which at the time was not part of the Flemish government, but was part of the federal government.
Policy and legal implications
How can these results be interpreted in light of the role adhered to by the Belgian Constitutional Court and within the historical context of its creation? As mentioned previously, the decision to have a Constitutional Court consisting partially of professional lawyers (judges and law professors) and partially of politicians was a deliberate choice made by the legislator in order to guarantee the representation of various groups in society in the Court.Footnote 70 For the creators of the Constitutional Court, it was important that the Court be made up not only of non-elected judges who lacked democratic legitimacy, but also to have elected representatives of the people amongst its membership. This balance between theory and practice was also meant to make constitutional review more acceptable to the legislator.Footnote 71
Moreover, the two-thirds majority required for nominations tends to make the Court as representative as possible of the differing political and ideological ideas in Belgian society. Hence, this reflects the separation of powers and allays fears of a gouvernement des juges. Footnote 72 A consequence of the structure is that the nomination process is inherently political. Some have criticised the involvement of politicians in the Court since, for politician-judges, being a lawyer is not even a requirement. More particularly, the presence of those ‘political’ judges in the Court was criticised because of their presumed lack of independence and impartiality.Footnote 73 Even today, the political character of the nominations continues to be criticised.Footnote 74
Interestingly, successive Presidents of the Constitutional Court have not themselves denied the political nature of the nomination process, and to some extent of the decision making.Footnote 75 The proportionality tests applied by the Court are never merely based on legal arguments. Former President Alex Arts held that ‘The reasons that constitute the basis of our decisions can never be merely legal - they are also based on social opinions. Our decisions combine legal knowledge with common sense. We try to be “wise”’. Yet another former President holds that the judges are aware of political realities and certainly take them into account.Footnote 76
However, although the judges do recognise the political nature of the nominations and of some decisions, they are of the opinion that they are able to make completely independent and impartial decisions. After nomination, so it is thought, the links with politics are cut, leaving judges able to decide with complete independence.Footnote 77 The judges hence recognise differences of opinion, but argue that they strive for consensus, especially since they are engaged in a game involving repeat players.Footnote 78 This indicates that both legislator and judges are aware of the political nature of the nominations, but consider this fair as it guarantees the representation of various opinions in society. Moreover, it was also thought that judges would be able to act with complete impartiality and independence once they were nominated. Our data indicate a different outcome.
There has been some debate on reform with respect to the nomination process. Those reforms, however, are not radical in the sense that they would, for example, eradicate the nomination of former politicians, but are rather suggestions to organise hearings where the Senate could question potential candidates.Footnote 79 Others fear that this would be a mere formality since the candidates would probably never reveal their true opinions and preferences at hearing.Footnote 80 For the recent nomination of the (female) judge Riet Leysen, the Senate organised a hearing and in the process caused political turmoil.Footnote 81 It is, moreover, doubtful whether a different design would necessarily generate different patterns of judicial behaviour in Belgium. After all, our data does not show less bias on the part of ‘professional’ judges than from the ‘political’ ones. On the contrary, we indicated that, in some cases, there is even greater bias on the part of ‘professional’ judges. If the nomination system were therefore to be reformed so that only professional lawyers (and not politicians) could be members of the Court, this would not necessarily change the biases we found. After all, statements made by various stakeholders in Belgium indicate that political affiliations are common knowledge, also of judges, and that there is, moreover, a systematic interwoveness between high magistrates and the political class in Belgium.
CONCLUSIONS
In this article, we have tested the extent to which political variables can explain the behaviour of constitutional judges in Belgium when dealing with conflicts between the central government and regions, and between regions themselves. Rules of procedure prevent us from observing individual votes - a common problem in many empirical studies. Important constitutional courts such as the Italian, the Austrian and the French bodies do not formally register individual votes or dissents. Therefore, we have analysed the decisions of the Court, controlling for the political allegiances of the four political actors we are able to observe directly for each decision: the President, the two Rapporteurs (the two judges who are chosen to draft the opinion of the Court), and the panel of judges involved.
Our empirical results seem to confirm that when the President and, to a lesser extent, the First Rapporteur (who is chosen by a process largely different from the American experience, which therefore does not pose an endogeneity problem) are allegedly affiliated with the Petitioner’s coalition, and the share of the panel of judges supporting the Petitioner’s coalition increases, the odds of success of the latter go up. The result is particularly robust when tested with alternative specifications. Therefore, our empirical results seem to be inconsistent with the view - generally held by legal scholars - that ideology or political variables play no role in explaining behaviour at the Belgian Constitutional Court. Decision-making in the Court is therefore inevitably political in nature. This is also inconsistent with the view that judges have of themselves as being politically neutral and impartial after being nominated.
Still, the empirical outcome does not indicate a sharp politicisation or party agenda as documented by empirical studies concerning the American reality. In fact, our results are easily comparable to previous studies on Italy and France (where direct observation of individual votes is unfeasible) or Germany, Portugal and Spain (where individual votes are observable). Unlike many empirical studies in the United States, we do not find that constitutional judges in Belgium are overwhelmingly reactive to political considerations.
With respect to the more general question of explaining judicial behaviour, our article provides evidence going against the standard formalist accounts. Political preferences seem to matter significantly. As in most civil law countries, and in contrast to the US, there is little correlation between judicial philosophy and political ideology. Therefore, it cannot be held that constitutional judges are pure formalists, yet divided with respect to judicial philosophy. At the same time, it should be emphasised that the evidence also shows that it is not only political preferences that matter in constitutional review.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1574019617000050
