Despite an increase in the degree of political contestation, the European Union (EU) is continually accused of having a democratic deficit. On the one hand, we find an increasing number of roll-call votes and dissenting opinions within the Council of Ministers. Moreover, within the European Parliament, the competition among party groups is becoming more similar to what we are used to from national parliaments. On the other hand, the increasingly democratic politics inside the European Parliament and EU Council remain unconnected to the views of the public (Follesdal and Hix Reference Follesdal and Hix2006: 553). The decisions taken by the Council of Ministers in Brussels are widely considered to be non-transparent. This information asymmetry has enhanced governments’ discretion vis-à-vis their national parliaments.
Today, the directly elected European Parliament acts as a co-equal legislator under the ordinary legislative procedure. Yet, the empowerment of the European Parliament is hardly considered a sufficient remedy for the potential agency drift implied by European integration. European election campaigns are often motivated by domestic party competition, members of the European Parliament (MEPs) remain widely unknown, the powerful European political groups undermine MEPs’ discretion, and – similar to a presidential system – the European Parliament does not elect a European government (Hix and Marsh Reference Hix and Marsh2011; Hix et al. Reference Hix, Noury and Roland2006).
An alternative means of increasing transparency and reinforcing democratic accountability in EU politics is the timely and effective scrutiny of EU policy proposals by national parliaments. Since the 1980s all national parliaments have introduced powerful European affairs committees, which, in cooperation with sector-specific committees, are tasked with scrutinizing European politics (for an overview, see Bergman et al. Reference Bergman, Müller, Strom and Blomgren2003). Although the design and effectiveness of these scrutiny mechanisms differ, their application is often driven by domestic party competition (Auel and Benz Reference Auel and Benz2005). Moreover, some governments initiate scrutiny in order to improve their bargaining position by revealing a domestic constraint to a negotiation partner in Brussels (Finke and Dannwolf Reference Finke and Dannwolf2013; Holzhacker Reference Holzhacker2002).
All of these approaches imply that parliamentary scrutiny reduces existing information asymmetries, thereby affecting legislative behaviour in the Council of Ministers and in the European Parliament. Otherwise, parliamentary scrutiny would be ineffective and therefore inappropriate for reinforcing democratic representation and accountability in the EU. This article is the first to test the effects of parliamentary scrutiny on voting behaviour in the European Parliament. Specifically, I consider whether parliamentary scrutiny caused German and Czech MEPs to defect from the voting instructions issued by their group leadership. My results strongly suggest that in both cases parliamentary scrutiny increases the likelihood that MEPs from governing parties will vote against the instructions of their group leadership. These results emphasize the findings of earlier studies, according to which MEPs from governing parties are in close contact with their national party leaders and sensitive to the strategy that their government is pursuing in the Council (Hix et al. Reference Hix, Noury and Roland2006: 509). In other words, scrutiny seems to reinforce governments’ control over ‘their’ MEPs. By contrast, the effect of parliamentary scrutiny on MEPs whose parties are in the domestic opposition is significantly weaker and depends on a dossier's political salience.
This article proceeds by deriving its main arguments from the existing literature on voting behaviour in the European Parliament and on the explanations for parliamentary scrutiny. Subsequently, I justify my case selection, present my measure of parliamentary scrutiny and explain the voting behaviour of German and Czech MEPs during the sixth European Parliament. I conclude with a brief discussion of my findings.
The Effect of Parliamentary Scrutiny on Legislative Behaviour
In order to evaluate parliamentary scrutiny we could focus either on its effect on the resulting EU policy or on its effect on legislative behaviour. Following the first approach, Schneider et al. (Reference Schneider, Finke and Bailer2010) first find evidence that the relative seat share of the opposition parties in the European Affairs Committees affects the government's bargaining power when negotiating in the Council of Ministers.Footnote 1 However, EU policies are the result of a complex inter-institutional decision-making process. Accordingly, it is very difficult to isolate the effect of parliamentary scrutiny being initiated in a single national parliament. Here, I solve this identification problem by evaluating the effect of parliamentary scrutiny on legislative behaviour directly. Specifically, I raise the question of whether MEPs vote differently on bills that are under parliamentary scrutiny by their domestic legislatures. Below, I first summarize the state of research on voting behaviour in the European Parliament before discussing the potential effect of domestic-level scrutiny.
Voting Behaviour in the European Parliament
The study of legislative voting behaviour has a long tradition (see, for example, Carey Reference Carey2009; Cox and McCubbins Reference Cox and McCubbins2007; Lindstädt et al. 2011; Poole and Rosenthal Reference Poole and Rosenthal1991). All of these studies share the assumption that voting behaviour is not exclusively motivated by ideology, but that it is also influenced by legislators’ worries about their chances for re-election. With a view to reaching the European Parliament, MEPs’ office-seeking intentions are usually considered a means to maximize either their policy gains or their electoral success (see, for example, Finke Reference Finke2012; Meserve et al. Reference Meserve, Pemstein and Bernhard2009). Accordingly, MEPs must please their electorate and their party, which usually allocates the resources inside parliament, such as committee memberships, speaking time, attractive rapporteurships and future careers in the party leadership. Even more importantly, list-based electoral systems enable parties to determine directly the electoral fate of their MPs. Therefore, domestic MPs are conceived as agents of two principals: the party and the voters (Carey and Shugart Reference Carey and Shugart1995). The tension between these two principals is stronger in electoral systems that encourage MPs to seek a personal vote, compared with party-centred systems, in which parties have greater control over an individual MP's electoral fate (Carey and Shugart Reference Carey and Shugart1995). Other factors that have an impact on the degree of party group cohesion are the institutional and financial resources of political parties (Carey and Shugart Reference Carey and Shugart1995), the capability of party leadership (Bailer et al. Reference Bailer, Schulz and Selb2009), MPs’ individual ambitions and career paths (Meserve et al. Reference Meserve, Pemstein and Bernhard2009; Lindstädt et al. Reference Lindstädt, Slapin and van der Wielen2012) and the effects of the electoral cycle (Lindstädt et al. Reference Lindstädt, Slapin and van der Wielen2011).
The European Parliament differs from national parliaments in adding a third principal. European political groups are associations of politically like-minded national parties. They are generally perceived as being very powerful inside the European Parliament, but they hold no powers over an MEP's electoral fate. Instead, MEPs are elected by list-based proportional representation systems in their respective member states, where national parties are frequently able to predetermine the list of candidates.
On the empirical side, numerous studies focus on the varying cohesiveness of the European political groups (Faas Reference Faas2003; Hix et al. Reference Hix, Noury and Roland2006; Lindstädt et al. Reference Lindstädt, Slapin and van der Wielen2012; Meserve et al. Reference Meserve, Pemstein and Bernhard2009) and examine their role in coalition formation in the European Parliament (Finke Reference Finke2012; Hix et al. Reference Hix, Roland and Noury2007; Kreppel and Tsebelis Reference Kreppel and Tsebelis1999). According to Hix (Reference Hix2004), the capacity of national parties to sanction deviating behaviour mainly depends on the design of the electoral system. The power of the national party increases with its ability to determine the list of candidates for European elections. Meserve et al. (Reference Meserve, Pemstein and Bernhard2009) find evidence that an MEP's likelihood of defecting from the political group depends on individual ambitions. Whereas some MEPs aim at staying at the European level, others intend to return to domestic politics (Scarrow Reference Scarrow1997). The authors find a curvilinear effect of age on the likelihood of defection, according to which middle-aged MEPs are most compliant with their European political group.
More importantly for the context of this article, Hix et al. (Reference Hix, Noury and Roland2006) discover systematic differences between the voting behaviour of MEPs whose national parties are represented in the Council and that of MEPs whose parties are not represented in the Council. Many proposals that come before the European Parliament are coordinated with the Council during informal talks; therefore, these proposals have the governing parties’ approval. Hence, MEPs from governing parties may find it harder to support substantial amendments, at least if they intend to return to domestic politics at some point in their careers.
Recently, empirical studies began to focus on vote-level characteristics. For example, earlier studies did not distinguish between legislative proposals as compared to own-initiative reports, let alone between the various legislative procedures. Hoyland (Reference Hoyland2010) reveals that MEPs vote differently on legislative proposals to the way the vote on non-legislative resolutions. Likewise, Meserve et al. (Reference Meserve, Pemstein and Bernhard2009) find that MEPs with national ambitions are more likely to defect from their political groups on proposals that imply an extended delegation of power to the EU.
Overall, existing research finds European political groups to be surprisingly cohesive, despite their limited powers over MEPs’ electoral fate (Hix et al. Reference Hix, Roland and Noury2007). A straightforward explanation for this finding is that national parties often pay limited attention to the proceedings inside the European Parliament. Next, I discuss the effect of parliamentary scrutiny.
Voting Behaviour under Scrutiny
Here, I am interested in the effect of another vote-level characteristic: namely, parliamentary scrutiny. I compiled data which indicate for every legislative proposal issued by the European Commission between 2005 and 2009 whether it was referred for scrutiny to the European affairs committees of the German and Czech parliaments. The question is whether, and if so how, domestic-level parliamentary scrutiny affects the likelihood that MEPs will defect from the voting instructions issued by their group leadership. To answer this question I start by discussing the motivation underlying parliamentary scrutiny of EU law proposals. What are the effects that parties hope for when engaging in costly and time-consuming scrutiny?
For over a decade, the literature on parliamentary scrutiny has been dominated by studies evaluating its effectiveness and evolution from an international comparative perspective (Bergman Reference Bergman1997; Bergman et al. Reference Bergman, Müller, Strom and Blomgren2003; Raunio Reference Raunio2005; Saalfeld Reference Saalfeld2005). More recently, research increasingly points to the fact that a comparative assessment of formal powers, rules and resources does not suffice to evaluate the role of national parliaments in EU politics (Auel and Benz Reference Auel and Benz2005; Holzhacker Reference Holzhacker2002). Most of these studies either highlight the potential inter-institutional balance (Bergman et al. Reference Bergman, Müller, Strom and Blomgren2003; Raunio Reference Raunio2005) or portray parliamentary scrutiny as a tool used by the opposition to keep its government in check (Martin and Vanberg 2004). In the case of European Union politics, effective control of the government is all the more important because the leading minister holds a distinct information advantage with respect to the proceedings inside the Council of Ministers. More recent theoretical approaches focus on the strategic interactions between government and opposition as well as among coalition partners. Based on a typology developed by Döring (Reference Döring1995), Holzhacker (Reference Holzhacker2002) highlights the importance of the strategic interaction between party groups and the leading ministers. The leading minister might not pay attention to party groups at all (non-party mode), might interact with members of other party groups (interparty mode), or might act across party boundaries (cross-party mode). According to Auel and Benz (Reference Auel and Benz2005: 389), parliamentarians find themselves in a dilemma between strictly scrutinizing the government and optimizing the policy outcome in the Council of Ministers. By tying the hands of the responsible minister too closely, domestic MPs from the governing parties risk a worse bargaining outcome (Auel and Benz Reference Auel and Benz2005: 373). To overcome this dilemma, they cooperate informally with the responsible ministries, thereby withholding information from the opposition and avoiding formal scrutiny. By contrast, opposition parties can make use of informal contacts to European actors (Auel and Benz Reference Auel and Benz2005: 388). Holzhacker (Reference Holzhacker2002: 470) finds that German and Dutch opposition parties tend to initiate scrutiny over issues that are salient for the public.
More recently, Finke and Dannwolf (2012) present evidence that parliamentary scrutiny of EU law proposals is motivated by the underlying information asymmetries within the two-level game typical of EU politics. Following the literature on two-level games, they argue that governments that are represented in the Council of Ministers can improve their international bargaining power by strategically revealing a credible domestic constraint (Hug and König Reference Hug and König2002; Schneider et al. Reference Schneider, Finke and Bailer2010). According to this logic, the leading minister has an incentive to reveal the position of her coalition partner if this move credibly shrinks her discretion for granting concessions in Council negotiations.
Almost all of these approaches imply that parliamentary scrutiny reduces existing information asymmetries. In other words, parliamentary scrutiny should increase the transparency of EU politics by drawing attention to the policy proposal at hand. It has been outlined above that the most important domestic principal of an MEP is his national party. Once parliamentary scrutiny has been initiated, the party leadership will be acutely aware of its MEPs’ voting behaviour – not least because an ongoing scrutiny procedure may draw the interest of domestic stakeholders to the issue at hand. Therefore, MEPs will find it increasingly difficult to violate the interests of their national party and their voters by following adverse voting instructions from their European political group. Hence, I expect that parliamentary scrutiny has a positive effect on an MEP's likelihood of defection.
Hypothesis 1: MEPs are more likely to defect when voting on a law proposal under scrutiny by their domestic parliament.
Please note that in H1 parliamentary scrutiny can be conceived of as a proxy for the political salience of a legislative proposal. Following Finke and Dannwolf (2012), I consider scrutiny to be a function of both the political salience of a dossier and the strategic considerations on behalf of the initiating parties. Accordingly, the dossier's political salience is one of my most prominent control variables. However, my next argument, which highlights the difference between MEPs from governing and opposition parties, is based on the strategic considerations underlying parliamentary scrutiny.
The existing literature on voting cohesion in the European Parliament suggests that MEPs from governing parties are more likely to defect from their group's voting instructions (Hix et al. Reference Hix, Noury and Roland2006). Specifically, they act as a crucial link supporting inter-institutional agreements. Based on the results from earlier studies, Finke and Dannwolf (2012) identify three potential reasons for initiating parliamentary scrutiny. First, the leading minister calls for scrutiny in order to improve his bargaining position in the Council. This strategy would be severely foiled by fellow party members voting against the government's strategy inside the European Parliament. Second, a third party (usually the opposition) calls for scrutiny to reveal the leading minister's domestic weakness. This strategy may be attenuated by fellow party members in the European Parliament voting explicitly against any mediocre compromise, thereby reinforcing the leading minister's position in the Council. Third, the opposition aims at blaming the government for supporting an unpopular EU policy. Because domestic stakeholders (in particular the voters) cannot observe the legislative behaviour in the Council, they will assume that the MEPs’ actions represent the government's position. Therefore, the government will find it more difficult to claim innocence if its own MEPs voted in favour of an unpopular political outcome. All three arguments suggest that the observable voting behaviour in the European Parliament affects a government's credibility vis-à-vis its international bargaining partners or its domestic stakeholders. In contrast, the voting behaviour of MEPs from opposition parties does not imply similar ulterior motives. Therefore, I expect parliamentary scrutiny to have a stronger effect on the voting behaviour of MEPs from governing parties as compared to MEPs from opposition parties.
Hypothesis 2: MEPs whose national parties are represented in government are more likely to defect when voting on a law proposal under scrutiny by their domestic parliament.
Finally, the effect of parliamentary scrutiny on MEPs’ voting behaviour depends on a law proposal's political salience. For highly salient dossiers, domestic-level scrutiny may – as an intended or unintended consequence – draw the attention of MEPs’ domestic-level stakeholders, such as trade unions or interest groups. The most salient dossiers, such as the EU chemicals regulation REACH or the Services Directive may even draw the attention of the general public. Although MEPs from opposition parties hardly care about the strategic motivations underlying parliamentary scrutiny, they must be concerned about their voting behaviour being observed and evaluated by important domestic stakeholders. Therefore, I argue that the effect of scrutiny on MEPs from opposition parties depends on a report's political salience.
Hypothesis 3: MEPs whose national parties are not represented in government are more likely to defect when voting on a highly salient law proposal under scrutiny by their domestic parliament.
The remainder of this article tests my hypotheses using data on the voting behaviour of German and Czech MEPs during the sixth European Parliament. I start by introducing the research design and the data collection. Thereafter, I present a statistical test of my argument using regression models.
DATA
For an empirical test of my hypotheses I use a new data set that includes all EU legislative proposals between 2006 and 2008.Footnote 2 Unfortunately, gathering comparable information on parliamentary scrutiny in all 27 member states is a very difficult, if not impossible, undertaking. Here, my selection of two parliaments follows a dissimilar-cases design – that is, I test the robustness of my substantial findings against very different domestic contexts. Specifically, I gathered information on the scrutiny activities during the grand coalition in Germany (2006–8) and the Czech minority government (2007–9). The differences between the two coalitions are significant: one was an oversized and stable grand coalition between two integration-friendly parties in the largest and oldest EU member state; the other was an unstable minority coalition headed by the starkly integration-sceptic Civil Democratic Party (ODS) in one of the smaller and younger EU member states. Nevertheless, the provisions for parliamentary scrutiny are very similar, granting individual political groups the power to refer documents to the committees (COSAC 2007: 15). Whereas this dissimilar-cases design ensures a certain level of robustness against different political contexts, the generalizability of my findings is limited. Specifically, the results are not easily generalized to either majoritarian or presidential systems. Furthermore, my findings cannot be easily generalized to political systems which allow parliaments to issue more or less binding negotiation mandates. This is the case in all three Nordic member states, as well as in Austria, Hungary, Slovakia, Estonia, Latvia, Poland (only Sejm) and Lithuania (COSAC 2007).
The grand coalition between the Social Democratic Party (SPD) and the Christian Democratic Union/Christian Social Union (CDU/CSU) in Germany held 466 of the 614 seats in the sixteenth German Bundestag and 72 seats in the European Parliament. At the time, the opposition consisted of the three smaller parties, the liberal Free Democratic Party (FDP), the Green Party and the socialist The Left (Die Linke). In total, the German opposition parties held 27 seats in the European Parliament. Both government parties are renowned for their pro-EU integration positions. Of the opposition parties, only Die Linke holds a rather integration-sceptic position.
The Czech minority government was formed by the Civil Democratic Party (centre-right), the Christian and Democratic Union (KDU) and the Green Party (SZ) between January 2007 and March 2009. In this coalition the Civil Democratic Party (81 seats) was the senior, the Christian and Democratic Union (13 seats) and the Greens (6 seats) the junior partners. Accordingly, the minority government held only 100 of the 200 seats in the Czech Chamber of Deputies (Poslanecka Snemovna). In contrast to the German situation, this government was inaugurated in January 2007 after a period of deadlock that followed the general elections in June 2006. After a vote of no-confidence, Prime Minister Mirek Topolanek resigned in May 2009 (Linek and Lacina Reference Linek and Lacina2010). Considering the composition of the coalition, the smaller parties were overrepresented in attaining five (Christian and Democratic Union) and two (Greens) of the 15 ministries. The Czech government parties held 14 seats in the European Parliament, compared to 10 seats for the opposition.
My crucial explanatory variable is whether scrutiny was initiated in either of the two chambers. Despite all the differences listed above, the scrutiny procedures in the two countries’ national parliaments are very similar. Following all four chambers’ rules of procedure, the committee chair sets the agenda for each meeting (Deutscher Bundestag Reference Deutscher1980: §61; Poslanecka Snemovna 1995: §36). Nevertheless, in both parliaments political groups possess powerful instruments that allow them to influence the meeting agenda. In the German Bundestag it takes 30 per cent of the deputies, in the Czech lower house only 20 per cent, to prompt a debate in plenary or to call in a committee meeting. Moreover, both lower chambers grant weekly question times to the opposition, during which they can broach any issue they choose. Party group influence on the committee's agenda has been institutionalized by meetings of all party group coordinators before each committee meeting. Finally, each party group chairs at least one committee in which it can initiate scrutiny measures.
The Czech Senate can only delay legislation, while the Bundesrat is an equal co-legislator in certain areas. The difference between the two countries is reflected in my empirical findings. I find only 2 out of 480 cases in which the Czech Senate (but not the lower house) initiated scrutiny. By contrast, the Bundesrat does so regularly.
To construct my variable on parliamentary scrutiny, I resorted to the IPEX database,Footnote 3 which provides scrutiny information at the level of EU documents such as the Commission's law proposals. For every proposal, IPEX includes an entry indicating the history of scrutiny activities in the lower and upper chamber of all member states. In using the IPEX website, I had to overcome several obstacles with respect to the completeness, reliability and comparability of the data. First, the IPEX database includes reliable information for a limited number of countries and over a limited time period. Second, different parliamentary administrations apply different rules for notifying ‘scrutiny’. To ensure the comparability of the applied scrutiny definition in the German and Czech parliaments, we contacted the national IPEX correspondents who are responsible for maintaining the database. In both countries each of the four chambers notifies ‘scrutiny in progress’ if an EU document has been referred to one of the committees. Based on this information, I constructed a dichotomous variable for each chamber. Third, in order to establish the reliability of the data set, I cross-checked the data with national archives. In the Czech case I drew random samples from the IPEX data and cross-checked the information with the protocols provided by the European Affairs Committee of the Chamber of Deputies.Footnote 4 For the German Bundestag I even reproduced the data set by directly analysing the official documents.Footnote 5 As a result of these efforts, I am certain that my most important independent variable is complete, reliable and comparable across the two countries.
During my period of observation, between 40 per cent and 60 per cent of Commission proposals were scheduled for discussion in the committees of either Bundestag or Bundesrat each year. In the Czech Republic the numbers were similar. But, whereas the Czech Senate restricts its scrutiny activity to cases in which the Chamber of Deputies initiates scrutiny, the Bundesrat must be considered a powerful and independent arena for parliamentary scrutiny.
For my dependent variable, I rely on the voting records of all roll-call votes cast during the sixth electoral period in the context of the co-decision procedure. The data have been provided to me by the European Parliament. Here, I assume that a group's political position is identical to the observed voting behaviour of the majority of its members (Bailer et al. Reference Bailer, Schulz and Selb2009; Hix et al. Reference Hix, Noury and Roland2006).Footnote 6
During the three core years of the German grand coalition (2006–8) the European Commission initiated a total of 342 law proposals under the co-decision procedure; I observe roll-call votes for 197 of these procedures, with an average of 3.06 roll-call votes per procedure (=603 observed votes). During the period of the Czech minority government between January 2007 and April 2009 the European Commission initiated a total of 269 law proposals under the co-decision procedure; I observe roll-call votes for 167 of these procedures, with an average of 3.37 roll-call votes per procedure (=562 observed votes). Unsurprisingly, the percentage of observed voting defection on roll-call votes is significantly higher among Czech MEPs (9.9 per cent) than among German MEPs (5.3 per cent). Voting defection is especially low for MEPs from the two German governing parties (4.0 per cent).
The operationalization of my hypotheses requires a measure that approximates the political salience that a party attaches to an EU law proposal. Lowe et al. (Reference Lowe, Benoit, Mikhaylov and Laver2011) suggest the following scale for measuring the importance of a policy dimension j: sj = log((mj + 1)/M), where mj is the number of sentences the party manifesto devotes to issue j and M indicates the length of the entire manifesto. I follow this approach, but assume that the political salience depends on two dimensions. First, it depends on the importance of the corresponding policy area for the domestic party competition. For this purpose I resort to the Comparative Manifesto Project (CMP) data set and assign suitable quasi-sentences to each of the 20 policy areas.Footnote 7 Second, proposals within the same policy area differ vastly with respect to their importance. I approximate this proposal-specific degree of politicization by the length of the corresponding European Parliament report ri. Proposals which have been initiated at approximately the same moment in time compete for the available budget of political attention R. Important proposals draw the attention of many actors – that is, they provoke many amendments and opinions by several committees. Applying Lowe et al.'s (2011) proposal, I arrive at the following two-dimensional measure of a proposal's political salience: si = log((mj * ri + 1)/(M * R)). Accordingly, a party attaches a low political salience to a dossier if the corresponding European Parliament report is short or if the corresponding policy area is irrelevant to the domestic party competition.
My control variables follow the existing studies on voting defection in the European Parliament. At the level of individuals I control for the known curvilinear effect of age. Furthermore, I control for experience and seniority by adding dummy variables for whether or not MEPs have been members of the national parliament or government and a variable for their length of membership in the European Parliament. Finally, I control for MEPs’ gender. At the vote level I control for the lopsidedness by a variable measuring the percentage of ‘Yes’ votes relative to the necessary quorum (simple or absolute majority). Furthermore, I control for the nature of the legal instrument. In addition, I add two variables indicating whether or not the amendment has been authored by an MEP's own group as well as whether or not the roll-call vote has been sponsored by an MEP's own group. Finally, all models feature fixed effects for policy areas approximated by the leading committee. Due to a potential endogeneity issue I refrain from using NOMINATE scores as controls for the ideological position or the ideological proximity of an individual MEP to his group leadership. However, I did so in an alternative model specification without any effect on my substantial results.Footnote 8Table 1 provides the summary statistics for both of my data sets.
Table 1 Descriptive Statistics
With respect to statistical modelling, I follow Meserve et al. (Reference Meserve, Pemstein and Bernhard2009: 10–11), who find a substantial variation in defection tendencies among MEPs, even after explicitly modelling determinants of political ambitions, experience and superiority. In addition, I expect to find systematic variation in defection rates across ranked-choice votes (RCVs). To model both sources of variation appropriately, Merserve et al. make a convincing argument for the necessity of crossing random effects for individual MEPs and votes.
For each of the two countries I estimate a series of three models, beginning with a baseline model that does not include my measure for parliamentary scrutiny. The results are displayed in Table 2. Comparing the results obtained for Czech MEPs to the ones obtained for German MEPs, I find considerable differences that reflect my most-dissimilar-cases design. In fact, the only significant commonality seems to be the positive effect of a dossier's political salience on the likelihood of defection. German MEPs are more likely to defect if the proposal at stake is a directive (as opposed to decision or regulation). I find the same curvilinear effect for age as Meserve et al. (Reference Meserve, Pemstein and Bernhard2009) for the Czech MEPs (middle-aged MEPs defect the least), but I find no such effect for German MEPs. German MEPs are less likely to defect on lopsided votes, whereas I find no such effect for Czech MEPs. Most of these differences can be explained by either the relative power of the German parties within their respective political groups or the lower level of experience among Czech MEPs, given that all of them had newly joined the European Parliament in the sixth European Parliament (see Lindstädt et al. Reference Lindstädt, Slapin and van der Wielen2012).
Table 2 Logistic Regression on Y = MEPs' Voting Defection (crossed random effects model)
Notes: . = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001. All models include fixed effects for policy area (approximated by leading committee) and a constant term (not shown). MEPs’ ideology is operationalized by their NOMINATE estimates for votes cast between May 2004 and December 2005.
Despite all these differences, I find very similar results for both groups of MEPs with respect to my first two hypotheses. Model 2 introduces my variable for parliamentary scrutiny in both national parliaments’ upper and lower chambers. It turns out that parliamentary scrutiny in the Czech Chamber of Deputies has a very strong positive effect on the likelihood that MEPs will defect. Due to perfect correlation I cannot enter the scrutiny activities of the Czech Senate into the same model, a finding which resembles the upper house's weakness in the Czech political system. By contrast, scrutiny in the German Bundesrat is less correlated to the activities observed for the Bundestag. In fact, scrutiny in the Bundesrat has a very strong positive effect on the likelihood of defection. For the German MEPs the share of defection rises from approximately 2.7 per cent (no scrutiny) to around 6.1 per cent (scrutiny); for the Czech MEPs the share of defection rises from 4.5 per cent (no scrutiny) to approximately 11.6 per cent (scrutiny).
My final model includes the interaction effect necessary to test my Hypotheses 2 and 3. To illustrate my results I graph the marginal effect of parliamentary scrutiny on the likelihood to defect for different levels of political salience (Figures 1a and 1b). First, I consider only the interaction between ‘parliamentary scrutiny in the lower chamber’ and ‘membership in a governing party’. The results strongly support my second hypothesis. In both countries the effect of parliamentary scrutiny is significantly stronger for MEPs from governing parties (bottom graphs in Figures 1a and 1b) than for MEPs from opposition parties (top graphs in Figures 1a and 1b). For MEPs from the German governing parties, scrutiny increases the likelihood for defection by approximately 9.5 per cent to a total of approximately 13.5 per cent. For MEPs from the Czech minority government I observe an increase by 39 per cent, which increases their overall likelihood to defect to approximately 57 per cent. In other words, under scrutiny, members from the Czech governing parties are more likely to defect than to abide by the line of their European political group.
Figure 1a The Effect of Parliamentary Scrutiny in the Chamber of Deputies on the Likelihood of Czech MEPs Defecting, Evaluated at Different Levels of Political Salience. Top: MEPs from Opposition Parties. Bottom: MEPs from Governing Parties
Note: Dashed lines indicate 99 per cent confidence intervals.
Figure 1b Marginal Effect of Parliamentary Scrutiny in the Bundestag on the Likelihood of German MEPs Defecting, Evaluated at Different Levels of Political Salience. Top: MEPs from Opposition Parties. Bottom: MEPs from Governing Parties
Note: Dashed lines indicate 99 per cent confidence intervals.
My third hypothesis is tested by a two-way interaction effect between ‘parliamentary scrutiny in the lower chamber’, ‘membership in a governing party’ and the dossier's ‘political salience’. Given the inherent complexity of two-way interactions, the substantial results are best presented by plotting the marginal effects on the predicted probabilities. The top graph of Figure 1a depicts the marginal effect of scrutiny on the likelihood of MEPs from the Czech opposition to defect on their European political group for different levels of political salience. The results do not support my third hypothesis. I find that MEPs from the Czech opposition reveal a higher likelihood to defect on their European group, regardless of whether or not the dossier is under scrutiny in the Chamber of Deputies. At the 99 per cent confidence level scrutiny does not reveal a significant effect on MEPs from opposition parties for any value of political salience.
By contrast, my results for the German sample are more supportive of my third hypothesis. Here, the effect of parliamentary scrutiny on the defection of MEPs from opposition parties depends strongly on a dossier's political salience. Surprisingly, the marginal effect of scrutiny turns out negatively for dossiers of low political salience. Currently, we lack any convincing explanation for this increased loyalty. However, scrutiny reveals a very strong positive effect on MEPs’ likelihood of defection in case of high-salience dossiers. In fact, for the most salient cases, the likelihood of defection increases by up to 30 per cent. By contrast, for those cases not subjected to parliamentary scrutiny, the likelihood of defection decreases in political salience. As a consequence, I find a significant and positive effect of scrutiny for the most salient political dossiers.
In sum, the results generally support my argument that scrutiny increases MEPs’ likelihood of defecting on their party leadership. Specifically, if the law proposal comes under scrutiny by national parliaments, MEPs from both German and Czech governing parties are more likely to defect from their political group. For governing parties, this effect does not depend on a dossier's political salience. This finding seems to support the argument that governments use parliamentary scrutiny to reinforce their strategic bargaining position in the Council. Where the previous literature found that the behaviour of MEPs from governing parties responds to their government's strategy (Hix et al. 2005), my findings indicate that this nexus is reinforced by domestic-level parliamentary scrutiny. The effect of parliamentary scrutiny on MEPs from opposition parties differs. In the Czech case, I find no significant effect of parliamentary scrutiny on the voting defection of MEPs from opposition parties. In respect of MEPs from the German opposition the effect of scrutiny depends on the dossier's political salience.
CONCLUSION
Given that European political groups hold virtually no powers over MEPs’ re-election chances, the existing literature finds a surprisingly high level of voting cohesion. In fact, voting defection in the European Parliament occurs no less frequently than voting defection in national parliaments. In this article, I have asked in how far domestic-level parliamentary scrutiny affects voting behaviour in the European Parliament. To my knowledge this is the first article to test the effects of parliamentary scrutiny on voting behaviour in the European Parliament. I argue that scrutiny draws the attention of national party leadership to the issue at hand. Further, it forces national parties to take an explicit position on an issue – that is, they can no longer turn a blind eye to the behaviour of their fellow party members in the European Parliament. As a consequence, it should be increasingly difficult for European political groups to discipline their MEPs successfully if there are conflicting political interests.
However, the causal relationship between scrutiny and the likelihood of defection is complex. Most importantly, it depends on whether or not an MEP's national party is in opposition or in government. Domestic-level scrutiny has an even stronger positive effect on the likelihood of MEPs defecting if their parties are represented in the European Council of Ministers. This finding presents further support for the strategic interactions between governments and parliaments at both national and European levels. Previous empirical studies of parliamentary scrutiny found that governments initiate scrutiny to reinforce their strategic bargaining position in the Council (Finke and Dannwolf Reference Finke and Dannwolf2013; Holzhacker Reference Holzhacker2002). The previous literature on voting behaviour found that MEPs from governing parties are responsive to their government's strategy in the Council (Hix et al. Reference Hix, Noury and Roland2006). My findings bring both strands of literature together and suggest that MEPs’ responsiveness to their government's strategic position is reinforced by domestic-level parliamentary scrutiny. From a normative perspective, this finding suggests that strengthening parliamentary scrutiny mechanisms may have unintended consequences. Instead of strengthening parliamentary control of the government, scrutiny seems to reinforce governments’ control over ‘their’ MEPs.
In respect of MEPs from opposition parties, my evidence is mixed. Scrutiny has no significant effect on the voting behaviour of MEPs from Czech opposition parties. By contrast, scrutiny causes a significant increase in the likelihood of defection for MEPs from the German opposition parties. One possible explanation may be the fact that Czech MEPs reveal a significantly higher level of defection in the first place. Future work is needed to disentangle the empirical effects of salience and scrutiny on voting behaviour, the latter being a consequence of the former.
In any case, this article has considered two parliaments that, although they represent very different member states, sponsor highly similar scrutiny procedures. Although this case selection ensures the robustness of my results with respect to domestic politics, it also defines the border to generalizability. We do not know to what extent differently designed scrutiny mechanisms produce a different effect on MEPs’ voting behaviour. Similarly, it may very well be that crucial features of a member state's political system, such as minority- and single-party government or unicameralism, alter the effect of domestic-level parliamentary scrutiny. Hence, the next step in this line of research is to extend the data set to include parliaments with different scrutiny mechanisms and which work in different political systems.
APPENDIX
Measurement of Proposal-Specific Political Salience (si)
Note: Every proposal i is assigned to a policy area j.
M * R defines the complete set of political text relevant to all policy areas j and all proposals i. The dark hatched area of mj *rj defines the subset of political text relevant to proposal i. Following the importance scale suggested by Lowe et al. (Reference Lowe, Benoit, Mikhaylov and Laver2011) the salience of proposal i is therefore:
$${\rm {s}_i}\, = \,{\it log} (({{m}_j}\,\ast\,{}{{r}_i}\, + \,1)/(M\,\ast\,R))$$
The total length of all European Parliament reports refers to a fixed time period during which proposals compete for attention. Here I use the reports on all proposals that have been drafted up to six months earlier or later than proposal i.
Acknowledgements
The author would like to thank Tanja Dannwolf, Bjorn Hoyland, the participants of the ‘Economics and Politics Seminar’ at Heidelberg University and two anonymous referees for their comments on earlier versions of this article.
