|Why Do MEPs Defect? An Analysis of Party Group Cohesion in the 5th European Parliament(*)|
|European Integration online Papers (EIoP) Vol. 6
(2002) N° 2;
|Date of Publication in : 12.3.2002|
|| Abstract |
Back to homepage |
| This paper's comments page | Send your comment! to this paper |
If one applies Mezeys (1979) typology for parliaments to the European Parliament, one can easily see that it has moved from a minimal legislature with hardly any influence on the policy output to a reactive legislature with policy-influencing power. In other words, it has a joined a group of parliaments, to which most Western European parliaments belong. At the same time, the European Parliament has acquired additional responsibilities usually held by parliaments apart from that core-defining function (Norton 1990: 1), for example budgetary powers and powers of appointment (Westlake 1998): The EPs formal influence in the selection of the Commission President and the College of Commissioners as a whole is already considerable. Hence, the EP increasingly resembles normal legislatures a fact that is also true for its internal organisation (committees, party groups), as Bowler and Farrell (1995) have shown. Based on the resemblance of traditional legislatures combined with the nonetheless existing peculiarities of a transnational institution, the EP is a promising case for the general study of legislatures and the application of legislative theories. The present study focuses on the voting behaviour of legislators and as a result of that on patterns of group cohesion (and the lack of it) in legislatures. More precisely, the aim of this study is twofold: First of all, it is designed to be the extension of previous studies in reporting the development of patterns of group cohesion in the European Parliament. Secondly (and more important), it is designed to incorporate and operationalise the differing pressures on legislators in a more thorough and rigid way than previous studies have done by drawing on a variety of factors (among them for example methods of candidate selection and electoral systems).
The plan of the study is as follows: In part II, a theoretical framework for the study of legislators will be outlined, which was mainly, albeit not exclusively developed in the context of the US Congress. Based on that, the peculiarities of a transnational parliament like the European Parliament (and their consequences for this study) are described (part III). After that, a short overview of previous studies about the internal politics of the European Parliament will be given, followed by the precise research questions of this study (part IV) and a description of the data and the methods used (part V). Finally, the empirical findings will be presented (in part VI).
In modern democracies, it is obviously no longer the case that citizens meet in a central agora (as Rousseau assumed) to run the political system directly. Instead, contemporary democracy is representative democracy that establishes the link between citizens (or more precise: voters) and the output of the political system through the institution of the political party and especially the party group. (1) Simply stated, modern democracy is as Aldrich put it unworkable save in terms of parties (Aldrich 1995: 3)(2).
Accepting that would normally require a precise definition of what a party is. However, it turns out to be more difficult than expected to define the term in a study that deals with the European Union. One has to decide whether the so-called European parties are to be subsumed under the term or not (see Bardi 1994, Bell/Lord 1998, Johannson 1998, Niedermayer 2001). If one considers Epsteins (1979: 9) definition of a party (any group, however loosely organized, seeking to elect governmental officeholders under a given label, authors emphasis), the question immediately arises whether that can possibly apply to the European parties: After all, candidates in European elections do not use a common European label (Henschel 1994).
These problems, however, can be avoided if one approaches the concept of party from a different angle: Based on earlier work by Key (1964), Mair distinguishes three faces of a party a party in public office (i.e. officeholders), a party on the ground (i.e. members) and a party in central office (i.e. party officials) (Mair 1994). These faces are present in the case of European parties as well, even though the members faces look very different (national parties instead of individuals) and the face of the party officials is rather pale (because they are so weak) (Bardi 1994, Niedermayer 2001). As for the face of the officeholders (at least the legislators in the EP), it looks very much like that of legislators of other national, regional or local parties. In other words, the relationship and the relative strength between these three might be quite different from traditional parties, but legislative parties the main focus of this study are very well comparable.
This leads to the question of why party groups exist. The problem at stake is nicely put by Cox and McCubbins (1993: 108): How can a group of formally equal and self-interested legislators, with demonstrably diverse preferences on many issues, agree on the creation or maintenance of a party?(4) Obviously, legislative parties must serve the legislators interest.
It is a truism, of course, that politicians have preferences. But preferences alone sometimes yield awkward results, a fact political scientists have been aware of ever since the work of Arrow (1953) and McKelvey (1976). Hence, modern political science as reflected in the fundamental equation of politics (Hinich/Munger 1997: 17) regards political outcomes as the product of the interaction of preferences and institutions. Parties and party groups are among the institutions that interact with politicians preferences: Sometimes parties foster politicians goals (and they will happily turn to their party), sometimes they do not. This means that the role of the party itself is influenced by a larger structure. Hence, the role and status of legislative parties (as well as their value for legislators) must be
understood not only in relation to the goals of the actors most consequential for parties, but also in relation to the electoral, legislative, and executive institutions of the government. Fiorina was correct: only given our institutions can we understand political parties (Aldrich 1995: 5).
Consequently, parties and party groups might be of different value to politicians depending on the political system that is analysed! One of the aims of this study is precisely to elaborate the circumstances that influence the value of party groups for legislators. Hix et al. (1999) have recently argued (based on an argument developed by Strom (1990) for the motivations of parties) that legislators interests are threefold: re-election, policy, and office. So, why and when do they need parties to achieve (some of) these goals?
Before we take a closer look at each of the three goals, it is crucial to keep two things in mind: First, the three goals are everything but independent of each other (Laver 1997), and secondly, the relative importance of the three goals is everything but equal. Instead, there is a clear lexicographic order. Re-election is the most important goal: Without re-election, there is neither office (within the legislature) nor policy (influence). And it is much harder to influence policy out of office than it is in office (e.g. by using the strategic advantages of holding an important committee chair). Hence, legislators will whenever they have to give primary attention to re-election, followed by office-seeking and finally policy-seeking.
Why and when are parties and party groups needed to achieve (or secure) re-election? The first reason is that parties are usually perceived as the main actors in the political arena and that as a consequence the individual candidates fate depends very much on the appearance of the party as a whole. As Cox and McCubbins note: substantial components of a partys record affect all its members similarly: for example, all are hurt by a scandal or helped by perceptions of competence, honesty, and integrity (Cox/McCubbins 1993: 112). They also provide empirical support (for the US Congress) for the existence of electoral tides, i.e. connections between the results of individual candidates of the same party.
Such an argument, however, does not necessarily lead to the conclusion that parties can be expected to be cohesive. Obviously, a cohesive party group is more likely to be perceived as strong or efficient. Still, this does not mean that legislative parties will actually be cohesive, as MPs might sometimes face short-term incentives to defect from a party group line. This situation can be compared to a prisoners dilemma: incentives to defect yield an outcome that is non-pareto-optimal: a non-cohesive, ineffective party group that is perceived to be weak.
Axelrod (1981) has argued that the incentives to defect can be dealt with, if one regards the problem as an iterated and not as a one-shot interaction. Including future considerations into the reasoning of legislators makes them realise that they will all be better off, if they work together. However, he assumes that players can recognize another player and remember how the two of them have interacted so far (Axelrod 1981: 308). It seems questionable, though, whether these assumptions are met in the case of legislators without party groups. No legislator can (and wants to) monitor the behaviour of all his fellows. It is thus rational for legislators to form a party group and delegate the function of monitoring to a central leadership.(5) The final piece that is necessary to make the process work is that the party leadership must have means to reward or punish legislators for their behaviour. One area of reward is the distribution of valued offices within a legislature. We are going to turn to that later. With regard to the re-election aspect, however, the party might also be in a position to reward or punish by controlling the use of the (valued) party label (Cox/McCubbins 1994). This, however, depends to a large extent on the interaction of electoral laws and control of candidate selection (Mitchell 2000: 340)(6) that we turn to now.
Concerning candidate selection, practices vary greatly, and the locus of effective control ranges accordingly (Gallagher 1988a: 4). Rahat and Hazan (2001) have recently listed several dimensions for the study of candidate selection. Two of these are crucial for our purposes: the nature of the selectorate and the degree of decentralisation. Katz has provided a theoretical reason the emergence of the cartel party of why the selection process can be expected to become more centralised (Katz 2001), notwithstanding Billes (2001) empirical finding that the process has actually seen a greater involvement of party members in the last 30 years. Be that as it may, there are obviously different levels of punishment potential to discipline legislators. Clearly, legislators can be much easier disciplined, if candidates are chosen by a small inner circle of the party (group) leadership rather than a ballot of all members or by regional party organisations. Hence, a centralised process of candidate selection forces individual legislators to devote greater attention to their party, provided that the links between the party group and the parts of the party that are in control of candidate selection are close. If this is true, the party can ensure the cohesion of a legislative body by weeding out potential troublemakers (Bowler et al. 1999: 6), simply by not allowing them to use the brand name (Aldrich 1995: 49).
The value of that brand name, however, depends (at least partially) on the electoral system. A continuum (with decreasing value of the brand name) exists that ranges from party-centred systems to candidate-centred systems (Mitchell 2000: 341). The dimension is based on what Carey and Shugart have labelled the incentive to cultivate a personal vote (1995: 417). Bowler and Farrell summarise the effects as follows. They depend on the
visibility (ease of monitoring) and the capacity of voters to reward/punish them for their efforts. Two, related and overlapping, aspects of an electoral system that alter this capacity are whether electors vote for candidates or parties, and district magnitude. . . . If voters are not able to vote for individuals but must vote for a party list, there is little point in an elected representative cultivating a personal vote since she/he would be unlikely to benefit electorally. In fact under most forms of party list one would reasonably expect legislators to spend time cultivating a good relationship with the party officials who order candidates on the list, rather than voters. (Bowler/Farrell 1993: 53)
As for district magnitude, the argument is that the larger the district the less easier it is to cultivate a personal vote and the less important is an individuals vote. Thus, the following extreme cases emerge: a party-based system consisting of fixed lists in one nation-wide constituency and a candidate-based system with an open ballot in local constituencies. The latter clearly gives the candidate a much higher degree of independence from his party than the former, hence he can be expected not to give primary attention to his party in the latter case. There is, of course, one last caveat to the arguments given so far: They only hold true if voters as well as parties care about a specific legislature and the behaviour of their legislators within that legislature!
We now turn to the office-seeking goals of legislators, which are assumed to be the second highest in the hierarchy of goals. In this context, the question that needs to be tackled is whether being a member of a legislative party in any way advances the possibilities of achieving more office. What kinds of parliamentary incentives (Bowler et al. 1999) (in contrast to the electoral incentives outlined above) are there that might lead to cohesive party groups?
Laver and Shepsle (1999) have presented a theoretical account based on their portfolio-allocation model (Laver/Shepsle 1990; 1996) of why party groups might emerge and be cohesive. Based on the assumption that legislators have different tastes, they provide a rational choice model that shows that the likelihood of being (or becoming) what they call a strong party(7) increases if legislators co-operate. Consequently they conclude that
there are incentives for independent parties or factions not only to fuse but also to submit to the discipline of other factions with different tastes, ceding strategic autonomy in the expectation of generating more favored government policy outputs that would otherwise arise.(8) The strategic benefits of party discipline thus act as a kind of gravitational force, providing a logic that holds together what on the face of things may appear to be diverse parts of a single party. (Laver/Shepsle 1999: 46)
As they point out, their approach has nothing to do with electoral benefits, but it is only driven by internal proceedings in the legislature. Notwithstanding that, however, there is still one severe problem in our context: The Laver/Shepsle-argument assumes the existence of a government whose survival is determined by a legislature. This is not always the case (as in the US Congress or the European Parliament). It is still possible, though, to apply the underlying idea of the model even to those legislatures. The factor that basically drives the portfolio-allocation model is the disproportionate influence [of ministers, the author] on the substance of cabinet decisions within their jurisdiction (Laver/Shepsle 1999: 30). A very similar function is performed by committee chairs (or rapporteurs) in parliaments without the need to support a government: If they act strategically, they also have a disproportionate influence on parliamentary decisions.
The important question then is how these key posts are distributed in a given legislature. Several methods are available: For example, the distribution could be entirely based on seniority. However, distribution usually follows party lines either proportionally to the size of party groups or with more or less clear advantages for the majority party (as in the US Congress). Within party groups, the same questions arise: How are these valued posts distributed there? Are they in the hands of the leadership of the party groups? That would clearly be the most efficient way to discipline legislators. Empirical evidence for the US Congress suggests that this actually is a very powerful tool in the hands of the leadership: Loyalty to the party leadership is a statistically and substantively important determinant of who gets what assignment (Cox/McCubbins 1993: 186).
To conclude, one can say that office-seeking behaviour gives under certain structural circumstances further incentives to form or join a party group. However, it is very well possible that re-election seeking behaviour and office-seeking behaviour pull MPs in different directions. This might, for example, be the case, if regional parties choose candidates, but central party officials distribute parliamentary posts. Both might have very different policy preferences and expect their MP to act accordingly.
This leads to the last goal policy-seeking. At first glance, it might seem obvious that legislators do have certain policy preferences of their own. However, one could question, whether policy per se does influence the reasoning of candidates at all. Instead, they could adjust their policy positions in order to maximise their chances of re-election and internal promotion (Laver 1997).
Still, one cannot neglect policy for the simple reason that policy positions are the only things that can be easily observed, for example, through roll call votes. Everything has to be deduced from the behaviour that legislators reveal on policy questions. After all, voters (and indirectly parties) do only care about the policy positions that their agents take on. We, therefore, have to analyse the ways in which party groups foster (or hinder) the (instrumental or sincere) policy positions of legislators.
The environment that legislators face when making policy decisions is one of uncertainty. Since the legislature as a whole decides on policy, they can by no means be sure a priori that their individual policy goals will be fulfilled. Weingast assumes that legislators simply cannot afford to take that risk. Hence, he argues that a tendency towards universalism will emerge: A universal coalition will form, which will pass legislation that gives each legislator one policy decision that is crucial to him. This is his so-called universalism theorem (Weingast 1979). However, there might also be another, and in terms of transaction costs cheaper way to achieve that: Legislators could also increase the likelihood of forming certain coalitions, which is the underlying idea of a party. If one regards a party in this way and then takes this idea to its extreme (i.e. that uncertainty is replaced by absolute certainty), it becomes absolutely crucial to gain a majority, since the majority can enact whatever it wants to (Aldrich 1995: 35). Party groups then act as legislative cartels (Cox/McCubbins 1993: 2). However, this also includes the usual problems attached to cartels: For members of a cartel, there is always an incentive to cheat. Some of the mechanisms described above (plus the possibility of institutional rules) might limit those incentives, but what if re-election seeking behaviour provides an incentive to cheat?
Even if parties are composed of legislators with similar tastes (i.e. policy positions) on some dimensions, there are, nonetheless, differences that might for example depend on structural differences in their (electing) constituencies. This is what Fenno (1978) has called Home Style. Denzau et al. have built on that and argued that the home style interferes with the internal considerations of a parliament, since actions of legislators are signals through which constituents . . . monitor legislator performance (Denzau et al. 1985: 1132). If one assumes that constituents care both about legislative results and about legislative behavior (Denzau et al. 1985: 1118), legislators might be forced to display a certain behaviour (congruent to the constituents expectations), even if that leads to unsatisfactory results. For legislative parties, this means that defections of MEPs on policy questions are sometimes unavoidable since MPs have to establish a credible reputation among their key supporters, be they voters or officials of the central party. This once again shows that different pressures on legislators are always possible.
To conclude: If re-election seeking behaviour strongly influences the actions of legislators and if the legislative party is not in control of the re-election process, then cohesiveness of party groups should come as a surprise. And even if party groups are nonetheless cohesive, one must still be careful not to overestimate the role parties play. After all, parties that matter should affect and alter the outcomes a point also made by Krehbiel: if parties are empirically significant, then politics should be significantly different with parties from what it is without them (Krehbiel 1993: 240).
Krehbiel actually doubts that this is the case: Based on theoretical work by Snyder (1991) and Groseclose (1996), who both argue that pivotal members of a legislature as a whole (instead of party group members) are the key to success in legislatures, he has questioned the importance of legislative parties in a series of articles (Krehbiel 1993; 1995; 1999a; 1999b). He argues that legislators with similar preferences usually join similar parties and hence it is impossible to tell whether preferences or parties are the driving forces behind legislative politics. Consequently, he dismisses Patterson and Caldeiras hypothesis (homogeneous parties . . . increase the chances of party voting (Patterson/Caldeira 1988: 129)) or Rohdes concept of conditional party government (Rohde 1991) that requires homogenous preferences within parties and heterogeneous preferences between parties as trivial: Parties are said to be strong exactly when they are superfluous (Krehbiel 1999b: 35). Instead, the real test for party group strength is to analyse the cohesiveness of parties when there is a considerable divergence of preferences within parties and a considerable overlap of preferences between parties. Only then would the results be different depending on whether party-based or preference-based theories are superior.(9)
To sum up: We have to keep in mind that parties can be understood only in relation to the polity, to the government, and its institutions, and to the historical context of the times. ... They are ... endogenous institutions (Aldrich 1995: 19) and the key to understanding them depends on the interaction of the three discussed goals of legislators as well as the institutional environment. Sometimes legislators may be able to win more of what they seek to win, more often, and over a longer period by creating political parties (Aldrich 1995: 28), but sometimes they may not. Most of the work done in this area has analysed the US Congress. The present study seeks to widen the perspective by applying the theory to the European Parliament. Because of the variance inherent in a transnational legislature, this promises to foster the understanding of legislative politics considerably.
What do we have to take into account when applying the theoretical framework just outlined to the European Parliament? First of all, one has to accept the fact that party groups do play a very important role in the European Parliament. The EPs own rules of procedure acknowledge their existence and grant certain rights to them, for example in terms of financial and organisational assistance (Nessler 1997). They also have privileges regarding the workings of the parliament, for example concerning parliamentary questions or the allocation of speaking time. Hence, the rules of procedure provide clear incentives to form party groups.(10) Moreover, they provide clear incentives to form transnational parliamentary groups, since the number of MEPs that is necessary to form a group is lower the higher the number of nationalities among these MEPs is.
What is the result of these incentives? In the current EP, eight transnational party groups used to exist. However, the Technical Group of Independent Members (TGI) has ceased to exist meanwhile. The remaining seven presently comprise 593 of the 626 MEPs. Table 1 gives the name and abbreviation as well as the number of MEPs and the number of nationalities for each party group.
As outlined above, the importance of the party group for re-election depends on the value of its brand name and the interaction of the electoral system with the methods of candidate selection, which is subject to the nature of the election at stake.
Concerning European elections, Reif and Schmitt (1980) have coined the term second-order national elections to describe the specific nature of these elections. What they mean by that is that they are not truly European elections, being based on a European campaign about European issues, but rather national elections based on national campaigns about national issues. In other words, they are not fought on the basis of what the European Parliament does, but on purely national grounds. As a result, one has to conclude that neither has the election result any consequences for the working of the European Parliament nor vice versa. If one further takes into account the low turnout in and the striking indifference (Teasdale 1999: 435) about European elections, it becomes clear that MEPs are clearly not as constrained by upcoming elections as MPs in other European legislatures are. Thus, the nature of elections does not provide an incentive to act within a cohesive party group that is as strong as in genuine national elections. Furthermore, a brand name does not exist either at the European level, since candidates do not use a common European label, but the label of their national party.
This nature of European elections obviously has certain implications for the effects of electoral systems. These vary considerably, as they are determined by the specific member states (notwithstanding efforts to implement a uniform electoral system). Apart from Irelands STV, party lists are generally used. Within these lists, some countries allow preferential voting (i.e. votes for individual candidates within a list), while others do not. In addition, the district magnitude varies considerably (see Bryder 1998, Forschungsgruppe Wahlen 1999). The question is whether that can possibly affect MEPs behaviour in the European Parliament. Although Bowler and Farrell (1993) have shown that it does affect the MEPs' responsiveness to voter requests (in terms of time devoted to constituency work), it still seems questionable. Since voters are said to be hardly interested in the workings of the EP (and hence are unlikely to monitor their MEPs behaviour), MEPs can very well devote time to their constituency without adjusting their voting behaviour. They might instead pursue other goals. After all, this is an empirical question that will be tested below.
As for candidate selection methods, the European Parliament is again characterised by a great variety of practices. The reason is that candidates are selected in each member state by each national party separately (without any European influence). Some parties use a very centralised method (with party officials as the key players), while in other parties, the party members have the final say (Norris/Franklin 1997). A final element has to be controlled for, namely the attention (of national parties) devoted to monitoring their MEPs. Tapio Raunio argues that the linkage between MEPs and national parties is often regarded as one of distance and low status (Raunio 2000a: 212), but he finds an increasing institutionalization of contacts (Raunio 2000a: 220) in many parties.
In any case, this implies that the control of the candidate selection is not in the hands of the party groups and consequently cannot serve as a disciplining instrument within the EP! Instead, some MEPs with strongly monitoring and highly centralised home parties can be expected to display a specific home style behaviour to please their national party whenever this becomes necessary. This should be especially true, since MEPs often pursue a career outside the EP nicely put by Kjaer: the only way up is home (Kjaer 2001: 2). Notwithstanding Scarrows (1997) finding that the number of MEPs that pursue a truly European career is rising, the very high turnover of MEPs one of the more remarkable features of the European Parliament (Corbett et al. 1995: 50) seems to suggest that this is a quite normal pattern (Bryder 1998).
What are the implications of those arguments with respect to party group cohesion in the European Parliament? If national parties care about the voting behaviour of their MEPs and have means to threaten MEPs re-election, MEPs face strong incentives to comply with expectations of their home parties. This is further supported by Raunios recent survey of national parties that showed that about 40% of parties actually issue voting instructions to their MEPs, at least when questions of fundamental importance are at stake (Raunio 2000a: 217). Consequently, we can expect a breakdown of party group cohesion on those questions. However, it means at the same time that MEPs probably have a fair amount of freedom and discretion on questions that are not of fundamental importance, which allows them to devote time and attention to their second goal, office-seeking, which we turn to now.
As was already said above, most MEPs are members of party groups, and party groups possess certain privileges with respect to the internal proceedings of the EP. First of all, party groups control the committee assignments (which are distributed proportionally to the size of the party groups in the plenary) and they seem to have even more discretion in their decisions than their counterparts in the US, as there is no seniority rule in practice (Bowler/Farrell 1995). Secondly (and this is even more important), they also control the distribution of key positions in the EP, i.e. positions in the Bureau of the EP, committee chairs and vice-chairs and rapporteurships. Again, seniority does not matter much, and again, these positions are distributed proportionally to their strength among the party groups (hence, from this point of view the incentive to form a majority party is less strong). Within the party groups, the leadership is in a very strong position, notwithstanding tendencies to distribute the positions proportionally among the national party delegations. After all, the leadership is still left with a choice. The importance of the role played by party groups can be seen from the fact that in cases of controversy, they were always able to get their favourite candidate elected so far (Corbett et al. 1995: 114).
Thus, we can conclude that the party group leadership exerts tremendous influence on the career paths of MEPs within the EP (by distributing influential committee positions and rapporteurships among their members). They . . . decide who gets into positions of power within in the EP (Bowler/Farrell 1995: 241).
What does that mean in terms of party group cohesion? It was argued above that if national parties become involved in the process of voting in the EP (and have the appropriate means to influence the proceedings), national delegations are likely to defect in case of conflict. In these cases, the party group leadership cannot do anything but accept that fact. However, as long as national parties are not involved in the process, the party group leadership has the potential to influence its MEPs and enforce cohesiveness. In terms of frequency of occurrence, this is probably the default case: The party group leadership is not as distant as the national partys leadership. In addition, it probably cares about almost every single decision in the EP, since the EP is its one and only operating domain. Therefore, one can expect cohesive party groups as the default scenario, but also a breakdown of cohesiveness if national parties kick in. One of the aims of the empirical part will consequently be to detect the circumstances under which national parties actually do have a potential to kick in and actually do so.
Turning to the last element, i.e. policy-seeking, there is one element that is especially important in the case of the EP: Its institutional environment is very strong, i.e. when the EP as an institution plays with the Commission and Council (Bowler/Farrell 1995: 221), Commission and Council are very powerful. In order to be (policy) effective under the co-operation and co-decision procedure, an absolute majority of MEPs has to vote in favour of a proposal. This used to very often lead to the so-called grand coalition, since only such a coalition was able to surpass the high threshold (given the low average attendance in the EP). It might also lead to higher cohesiveness, since much more is at stake and party groups exert stronger pressure. At the same time, one could argue that national parties intervene more often, when important policy decisions are at stake, which could result in lower cohesiveness. In any case, different majority requirements and different legislative procedures might result in different behaviour of MEPs.
One last (policy) aspect should be mentioned: The European Parliament is usually regarded as a pro-European institution. Nonetheless, some Euro-sceptic parties are also represented in the EP. These parties can be expected to display their anti-European standing quite frequently. In other words, their home style might require that they permanently meet the (Euro-sceptic) expectations of their home party and thus defect from (pro-European) party group lines.
We have seen that different incentives might pull MEPs in different directions. However, as the incentives are ordered lexicographically (with re-election on top), party group cohesion should not be expected to be generally high. Instead, it can be expected to vary according to the subject matter.
The EP is, of course, aware of the apparent possibility of conflict within party groups. Hence, the standard proceedings of the European Parliament, which operates on a four-week routine, try to cope with it: The first two weeks are dominated by committee work, which is followed by an entire week, in which party groups assemble to agree on internal compromises for the fourth and last week the plenary week. Hence, a great amount of time is devoted to finding compromises and allowing disagreeing members still to vote for the party group line (as it is indicated by the whip). Even then, however, because of the reasons given above, it is still possible for MEPs (and especially national delegations) to vote against a party group line. Party groups tolerate such behaviour. As one senior member of the EPP-group put it: This happens. This is possible; they can do it. But normally we always prefer that they do not do it (quoted in Brzinski 1995: 149).
What kind of results have previous studies revealed? The (empirical) pioneer study about internal politics in the European Parliament was conducted by Attiná (1990). Since then, quite a few (general) studies have followed (for example Brzinski (1995), Kreppel/Tsebelis (1999), Raunio (1996, 1999, 2000b), Kreppel (2000), Hix (2001a, 2001b)). In addition, some individual votes in the European Parliament have been studied very carefully, especially votes on the confirmation of the Commission President (Hix/Lord 1996, Gabel/Hix 1997, Hix 1997) or on EU enlargement (Johannson 1997).
Summarising the results of these studies is, of course, impossible given the limited space of this study. Nonetheless, some common features can be distilled from them. In terms of party cohesion, the studies have revealed a surprisingly high degree of party group cohesion.(11) Large as well as leftist party groups were usually found to be more cohesive than smaller (or rightist) ones. Brzinski found that party groups with a high degree of multi-nationality (members from many different member states and many different national parties) are less cohesive (Brzinski 1995: 148/149), which seems reasonable. It has also been reported that specific national delegations are more likely to defect from the party group line than others. Sometimes, these findings are simply based on nationality (Hix (2001a, 2001b) as well as Kreppel/Tsebelis (1999) report that the British parties are among the usual suspects), but sometimes they are based on structural features. It has been argued that national party delegations whose home parties are in the national government might be more likely to defect from party group lines than others. The underlying argument says that bargaining in the Council is often very difficult. Once a compromise has been found there, the national governments have little interest in an EP that threatens to overturn it. Hence, in case of conflict, they put pressure on their MEPs to vote for the compromise, even if that implies voting against the party group line. Considerable support for this was found in case studies analysing votes in the commission president investiture procedure(12) (Hix/Lord 1996, Hix 1997) and the vote on the EU enlargement in 1995 (Johannson 1997).
In short, the group cohesion picture that has emerged so far is characterised a very high degree of cohesion, but it seems as if this is a rather fragile equilibrium that can easily be distracted as soon as national parties interfere. However, a general, systematic test is still to be conducted to examine this.
The aim of this study is twofold: First of all, it is designed to be the extension of previous studies in reporting the development of patterns of group cohesion in the European Parliament. Secondly (and more important), it is designed to incorporate and operationalise the differing pressures on MEPs in a more thorough and rigid way than previous studies have done by drawing on a variety of data sources. Hence, it is first of all necessary to establish some facts about party cohesion in the present European Parliament and some descriptive findings will be presented first in terms of party group cohesion, national cohesion and the cohesion of national party delegations.(13)
This, however, is obviously not sufficient to explain, why individual MEPs defect from a party group line. In the second step, we will thus explain the defection of national party delegations from the party group line. One can include majority requirements, legislative procedures, and countries as explanatory variables. In addition, one can test whether policy matters at all by including the ideological position of national party delegations relative to their party group. We will also test, whether the methods of selecting candidates and electoral systems matter, whether national parties in government put more pressure on their MEPs than others. In short, we will address and empirically test the questions concerning group cohesion that have been raised in the discussion above.
The data basis of the study consists of roll call votes in the fifth European Parliament. Roll call analysis is a method that has been applied to the US Congress for a long time (see for example Poole/Rosenthal 1985, 1991), but has been underutilized (Corbett et al. 1995: 160) in the case of the EP (although the above mentioned studies were based on analyses of roll call votes). Keeping Bowler and Farrells statement that committees are the sine qua non of legislative power (Bowler/Farrell 1995: 221, see also Neuhold 2001) in mind, all roll call votes taken between July 1999 and December 2000 and based on reports that were adopted in a committee and then tabled and examined in plenary were included.(14) A total of 506 reports was tabled during this period, 232 of them included at least one roll call vote. Four of them had to be removed because of inconsistencies in the online-documentation of the European Parliament, leaving a total of 228 reports under examination. These reports were classified according to the applicable procedure (for example consultation or codecision, including the reading and thus the majority requirement) and the issue area (according to committee responsible).
Voting on tabled reports is basically a two-stage process (Corbett et al. 1995: 161). The voting is organised in a way that amendments to the document under investigation are put to vote first. In the very last step, the final resolution is put to vote, which is the official position of the EP in the inter-institutional game. Thus, one report usually includes quite a number of votes, possibly also quite a number of roll call votes. The 228 reports included a total of 1370 roll call votes. These individual votes were classified according to whether they were on an amendment or a final resolution.
A word of caution with regard to our data basis seem necessary. According to the EPs rules of procedure (rule 133 (1)), the normal voting procedure in the European Parliament is by show of hands. However, according to rule 134 (1), each political group or at least 32 MEPs can request a roll call. In this case it is recorded how each MEP voted. Official statistics suggest that about 15% of votes are taken by roll call. The problem is that those 15% are not necessarily a representative sample of all votes. In fact, we have good reason to believe that roll calls are called for strategic reasons, thus leaving serious doubts about their representativeness. Carrubba and Gabel have set up a model of why roll calls are taken. They argue that roll calls are taken for exactly the reason to produce a voting behaviour that would have been different otherwise. In terms of party cohesion, they actually conclude that cohesion in roll calls votes might be inflated. Their conclusion is that findings based on roll call analysis are conditional on a RCV being requested (Carrubba/Gabel 1999: 5). This is probably true. However, since there is no other way to determine how MEPs voted so far, we have to accept the shortcomings and keep them in mind when we are going to interpret the results.
Based on these roll call votes, the primary dependent variable group cohesion is calculated for each one using Attinàs (1990) index of agreement (IoA):
where M is the mode of the Yes-Votes, No-Votes and abstentions of a group and N is the total number of votes cast by that group. The index theoretical range is from 33 to 100, the former indicating that the group is perfectly split between the three options, the latter indicating unanimity. Positive scores indicate that more MEPs voted for one option than for the other two options together.
Again, a critical remark about the use of party cohesion measures is necessary, as they have not gone without criticism. Especially Krehbiel (2000) has argued that they per se cannot discriminate between cohesion based on similar preferences and cohesion based on organisational power of party groups. Inferring from high cohesion scores that party groups are strong might be a spurious result, since it might be based on genuine preferences as well. Party strength must somehow alter the outcome. There are several ways of how to cope with this problem: Krehbiel (1995) uses the individual preferences of US legislators as estimated by Americans for Democratic Action and the National Taxpayers Union as control mechanisms. Snyder and Groseclose (2000) estimate the preferences of legislators based on heavily lopsided votes. They assume that party group leaders do not try to influence their legislators because of the lopsidedness of the vote. Legislators will consequently reveal their true preferences that can be used as control mechanisms in contested votes.(15) Hix (2001b) finally uses a survey of MEPs for the same reasons. Here, we will also include surrogate measures of preferences in the analysis to cope with Krehbiels point the policy positions of national parties, as estimated by experts (Huber/Inglehart 1995; Ray 1999).
The dependent variable for the second step of our analysis is a dichotomous variable coded 0, if the majority of a national delegation voted the same way as the majority of its party group did, otherwise 1 (meaning defection). As for the explanatory variables, they consist of features of the party groups, of the national party delegations (and their home parties and their political systems), and of the roll call at stake (as already described above). As for the party groups, a measure of their position and diversity on the left-right-dimension and the pro-/anti-Europe dimension is included, as these are the dimensions that make up the shape of the EU political space (Hix 1999: 17).(16) In addition, there are two size measures: the number of MEPs and the number of nationalities for each party group.(17) As for the national delegations and their home systems, we have a measure for the centralisation of the candidate selection process(18), for the party-centeredness of the electoral system(19), for the degree of MEP-monitoring of the national party(20) and one for the national party delegations policy distance to the party group mean(21) as well as a dummy variable capturing whether the home party is in government.(22) In addition, a variable was included that measures the support for European integration among supporters of national parties.(23) This idea is related to the idea of monitoring and home style politics. Since the EP is usually assumed to be very much pro-integration, Euro-sceptic parties and MEPs might be forced to openly display their critical view more often than others. Finally, dummy variables for countries as well as party groups were included when appropriate to include possible elements of a national or a group culture. This yields a total number of cases of 24,660 (one for each of the 18 national delegations(24) and each of the 1,370 roll calls). Logistic Regression is used to explain the parties behaviour.
The picture that emerges concerning the cohesion of party groups looks quite familiar: Party group cohesion is relatively high with index scores exceeding 80 for the four largest party groups in the European Parliament with the EPP having a slightly lower degree of cohesion than the other three as one could expect after the EPPs strategy of opening its borders that caused considerable noise in the party group. The UEN- and the GUE/NGL-group take a middle position. The anti-European EDD-group as well as the non-attached members and the TGI display the lowest degree of cohesion (see Figure 1). If we conceive of MEPs from one member state as a common group, these groups display a much lower level of group cohesion with an IoA ranging from 36 (France) to 64 (Germany). Generally speaking, the countries with a low level of national cohesion tend to be relatively Euro-sceptic (Denmark, France, UK), which probably forces some of their parties to display home style, anti-European voting behaviour (see Figure 2).
If we classify the roll calls according to issues, some interesting features can be observed: In general, questions of agriculture and fishery reveal relatively high overall levels of cohesion (IoA=66, 63). On the national level, questions of culture lead to high national cohesion, which is true for all member states (except UK, Sweden and Denmark). The French MEPs generally with the lowest cohesion display a considerable cohesion on questions of agriculture, as do the Spanish. The Spanish also vote together quite often as do the Portuguese, Greeks and Italians on questions of fishery. Apparently, the structure of national industry influences the voting behaviour of national MEPs to some extent a first hint that some home style exists in the EP. However, this does not apply as one might have expected to questions of regional policy. With regard to differences depending on whether a final resolution is at stake or not, we can see that the overall cohesion rises significantly when a final resolution instead of an amendment is voted on (IoA of 68 instead of 47). Party group cohesion, however, is not affected by that too much, whereas all national groups are much more cohesive on final votes.
Finally, an analysis of the cohesion of national party delegations reveals that these groups are almost perfectly cohesive with IoA-scores of more than 90 throughout, especially if abstentions are left out (Figure 3). Within these groups, conflict hardly exists. Unfortunately, we cannot answer Krehbiels question of whether this pattern emerges because of similar preferences or because of the structural power of national parties with the available data. Be that as it may, it seems justified to use these delegations as our units of analysis in the second stage of the empirical analysis. It is not necessary to disaggregate the data to the level of individual MEPs.
So far, by looking at some preliminary findings concerning party group cohesion, we have found that party groups are (still) more important than national groups, but that there are instances, where the importance might be reversed. However, we need to take a closer look at these incidents to give a more profound account on party group cohesion and defection.
To get a first impression on defections of national party delegations from a party group line, Figure 4 provides an overview of the defection rates for some national delegations with the British Conservatives clearly leading the table. In general, there is a tendency for Euro-sceptic parties to defect quite frequently, but we can test this more thoroughly in our model. Before we actually get to our model, just a quick reminder of what to expect: Perfect party group cohesion (in other words: no defections) would be expected, if office-seeking were to matter exclusively (since party group leaders are in control of office-promotion and have an interest in their party group being cohesive). However, re-election seeking can be expected to interfere and consequently lower the cohesion!
Concerning our model, we can again start off with the overall performance of the model, which is more than satisfactory. Nagelkerkes Pseudo-R2 takes a value of 36.9%.
Starting with the characteristics of the party groups and the national delegations in the European Parliament, we see that larger national delegations are less likely to defect from a party group line.(25) This seems reasonable: If a large national delegation threatens to defect from a party group line, more efforts might be invested to find a compromise, whereas the defection of a small delegation is more easily bearable for a party group. We can also see that the overall cohesion of the party group influences the likelihood of a national delegation to defect. The less cohesive the overall party group in a particular vote is, the more likely the national delegations are to defect. Of course, this is a circular reasoning to some extent(26), but nonetheless: The mere size of the coefficient suggests that there is a tendency speaking very broadly to vote together or to defect together, that is many national delegations defecting and thus lowering the overall cohesion either the whip works well or not all. Again, this seems reasonable: One could argue like Cox/McCubbins (1993): The party group leadership only cares about some of the votes taken (and tries to influence its MEPs), but leaves the MEPs decision open on others, thus yielding very low cohesion scores on those. This finding can also serve as an argument against Krehbiels point: On some votes, party groups definitely are not cohesive, hence they cannot be simply a group of perfectly like-minded people. This is further supported by the fact that the ideological distance between the national delegation and the party group as a whole affects the likelihood of defections as well: The larger the distance, the higher the likelihood. Party groups in the European Parliament are very amorphous creatures: If the policy distances are too large, national delegations do defect more easily. Nonetheless, the high overall cohesion scores show that party groups nonetheless manage to usually maintain their cohesiveness.
Interestingly enough, the circumstances of individual roll calls hardly matter (apart from the first-reading in the co-decision procedure). As for nationalities, Belgians probably among the most pro-European in Europe are significantly less likely to defect, quite contrary to the British MEPs, who clearly lead the table. This holds true even after controlling for support for European integration.
As for the characteristics of the home parties and the features of the home political system, the straightforwardness of the results is astonishing: all expectations are confirmed:
Only the electoral system turns out to be insignificant as could be expected because of the peculiarities of European elections. Finally, MEPs from governing parties are also more likely to defect from a party group line. The crystal-clear pattern of these findings suggests that MEPs are very attentive to their environment and their specific situation. Some MEPs are clearly faced with stronger home pressures than others and react accordingly. Apparently, they care very much about their re-election, just as one would expect from a rational political actor.
Two aims were outlined at the beginning of this study: First, the extension of previous studies in reporting the development of patterns of group cohesion in the European Parliament. Secondly (and more important), the incorporation and operationalisation of the differing pressures that legislators have to cope with in order to explain their behaviour and the resulting patterns of party group cohesion (or the lack thereof). Those differing pressures stem from the fact that legislators want to secure re-election, while, at the same time, they want to achieve internal promotion and policy goals. This becomes especially difficult, if they depend on different external actors to do so, as it is the case in the EP, where national parties are usually in charge of candidate selection (and thus re-election), while party groups are in charge of the distribution of offices in the EP.
As for the first one of these aims, it has been shown that the party groups in the EP, especially the larger ones, still display a very high level of party group cohesion. Hence, the default type of voting in the EP still seems to be along party group lines (and not along national lines). >From a theoretical point of view, this is actually what we have expected: It was argued that national parties (who could induce a breakdown of party group cohesion by putting pressure on their MEPs) will only do so on specific questions that are of special importance to them, but not on a general basis. Hence, we see a high, albeit no perfect level of party group cohesion.
As for the second one of these aims, we have looked at instances when party groups are actually not cohesive. When and why do national party delegations (who turned out to be the appropriate unit of analysis in the EP, since they act almost perfectly cohesive) defect from party group lines? Indirect evidence was provided here to show that national parties are actually the ones that can induce a breakdown of party group cohesion. This evidence included the finding that MEPs from national parties that have a centralised method of candidate selection, that invest more resources in the monitoring of their MEPs and/or that are in the government of their country are more likely to defect from party group lines. Those parties have a higher interest as well as a more effective tool box to influence their MEPs according to their preferences, even if that means voting against a party group line for their MEPs. Consequently, the defection rates of those parties are significantly higher than others. One can thus conclude that MEPs are obviously well aware of their specific situation, they know exactly who deserves primary attention. Although the exact mechanisms of this process are still to be discovered in further studies, these facts have clearly been established in the course of this study. For the future of the EP, this means that as the importance of the EP increases, so will the influence and pressure that national parties will exert on their MEPs. It thus seems questionable whether this is the path to take in order to achieve to a truly European, supranational entity.
Future research will have to confirm the results presented here. In addition, one could include a temporal dimension: re-election seeking behaviour (and thus defections from party group lines) should become more likely as elections approach. To test the influence of party groups on MEPs, one could also pay special attention to those MEPs that switch party groups during a legislative period. If it were true that party groups have considerable influence, as was argued here, one should see a clearly distinctive voting behaviour of those MEPs after they have switched. McCarthy et al. (2001) have actually confirmed this point for the US Congress, but again it still needs to be done for the European Parliament.
J.H. Aldrich (1995): Why Parties? The Origin and Transformation of Party Politics in America, Chicago.
S. Ansolabehere (forthcoming): The Effects of Party and Preferences on Congressional Roll Call Voting, in: Legislative Studies Quarterly, forthcoming.
K.J. Arrow (1951): Social Choice and Individual Values, London.
F. Attiná (1990): The Voting Behavior of the European Parliament Members, and the Problem of Europarties, in: European Journal of Political Research (18), pp. 557-579.
R. Axelrod (1981): The Emergence of Cooperation among Egoists, in: American Political Science Review (75), pp. 306-318.
L. Bardi (1994): Transnational Party Federations, European Parliamentary Groups, and the Building of Europarties, in: R.S. Katz, P. Mair (eds.): How parties organize: change and adaptation in party organizations in western democracies, London.
D.S. Bell, C. Lord (eds.) (1998): Transnational Parties in the European Union, Aldershot.
L. Bille (2001): Democratizing a Democratic Procedure: Myth or Reality? Candidate Selection in Western European Parties, 19670-1990, in: Party Politics (7), pp. 363-380.
S.A. Binder et al. (1999): Uncovering the Hidden Effect of Party, in: Journal of Politics (61), pp. 815-831.
S. Bowler, D.M. Farrell (1993): Legislator Shirking and Voter Monitoring: Impacts of European Parliament Electoral Systems upon Legislator-Voter Relationships, in: Journal of Common Market Studies (31), pp. 45-69.
S. Bowler, D.M. Farrell (1995): The Organizing of the European Parliament: Committees, Specialization and Co-ordination, in: British Journal of Political Studies (25), pp. 219-243.
S. Bowler et al. (1999): Party Cohesion, Party Discipline, and Parliaments, in: S. Bowler et al. (eds.): Party Discipline and Parliamentary Government, Columbus.
T. Bryder (1998): Party Groups in the European Parliament and the Changing Recruitment Patterns of MEPs, in: D.S. Bell, C. Lord (eds.): Transnational Parties in the European Union, Aldershot.
J.B. Brzinski (1995): Political Group Cohesion in the European Parliament, 1989-1994, in: C. Rhodes, S. Mazey (eds.): The State of the European Union, Vol. 3: Building a European Polity, Boulder.
J.M. Carey, M.S. Shugart (1995): Incentives to Cultivate a Personal Vote: a Rank Ordering of Electoral Formulas, in: Electoral Studies (14), pp. 417-439.
C. Carrubba, M. Gabel (1999): Roll-Call Votes and Party Discipline in the European Parliament: Reconsidering MEP Voting Behaviour, London (EPRG Working Paper No. 2-99).
R. Corbett et al. (1995): The European Parliament, London, 3rd edition.
G.W. Cox, M.D. McCubbins (1993): Legislative Leviathan. Party Government in the House, Berkeley.
G.W. Cox, M.D. McCubbins (1994): Bonding, Structure, and the Stability of Political Parties: Party Government in the House, in: Legislative Studies Quarterly (19), pp. 215-231.
A. Denzau et al. (1985): Farquharson and Fenno: Sophisticated Voting and Home Style, in: American Political Science Review (79), pp. 1117-1134.
L. Epstein (1979): Political Parties in Western Democracies, New Brunswick.
R.F. Fenno (1978): Home Style, Boston.
Forschungsgruppe Wahlen (1999): Europawahl 1999, Mannheim.
M. Gabel, S. Hix (1997): The European Parliament and Executive Politics in the EU: Voting Behaviour and the Commission Investiture Procedure, unpublished mimeo.
M. Gallagher (1988a): Introduction, in: M. Gallagher, M. Marsh (eds.): Candidate Selection in Comparative Perspective: The Secret Garden of Politics, London.
M. Gallagher (1988b): Conclusion, in: M. Gallagher, M. Marsh (eds.): Candidate Selection in Comparative Perspective: The Secret Garden of Politics, London.
T. Groseclose (1996): An Examination of the Market for Favors and Votes in Congress, in: Economic Inquiry (34), pp. 320-340.
T.R. Henschel (1994): Die europäischen Parteienzusammenschlüsse, in: W. Wessels, W. Weidenfeld (eds.): Jahrbuch der Europäischen Integration 1993/94, Bonn.
H.J. Hinich, M.C. Munger (1997): Analytical Politics, Cambridge.
S. Hix (1997): Executive Selection in the European Union: Does the Commission President Investiture Procedure Reduce the Democratic Deficit?, European Integration Online Papers (EIOP), Vol. 1, No. 21.
S. Hix (1999): Dimensions and alignments in European Union Politics: Cognitive constraints and partisan responses, in: European Journal of Political Research (35), pp. 69-106.
S. Hix (2001a): Legislative Behaviour and Party Competition in the EU: An Application of NOMINATE to the Post-1999 European Parliament, Journal of Common Market Studies (39), forthcoming.
S. Hix (2001b): Legislators Ideology or Party Organisational Power? What Drives Voting Behaviour in the European Parliament, Paper presented at the Seventh Biennial International Conference of the European Community Studies Association, May 31 June 2, 2001, Madison, WI.
S. Hix, C. Lord (1996): The Making of the President: The European Parliament and the Confirmation of Jacques Santer as the President of the Commission, in: Government and Opposition (31), pp. 62-76.
S. Hix et al. (1999): An Institutional Theory of Behaviour in the European Parliament. EPRG Working Paper 1-1999, London.
J. Huber, R. Inglehart (1995): Expert Interpretations of Party Space and Party Locations in 42 Societies, in: Party Politics (1), pp. 73-111.
J.A. Jenkins (1999): Examining the Bonding Effects of Party: A Comparative Analysis of Roll-Call Voting in the U.S. and Confederate House, in: American Journal of Political Science (43), pp. 1144-1165.
K.M. Johannson (1997): Veränderte Bedeutung der politischen Fraktionen, in: E. Kuper, U. Jun (eds.): Nationales Interesse und integrative Politik in transnationalen parlamentarischen Versammlungen, Opladen 1997.
K.M. Johannson (1998): The Transnationalization of Party Politics, in: D.S. Bell, C. Lord (eds.): Transnational Parties in the European Union, Aldershot.
R.S. Katz (2001): The Problem of Candidate Selection and Models of Party Democracy, in: Party Politics (7), pp. 277-296.
V.O. Key (1964): Politics, Parties, and Pressure Groups, New York, 5th edition.
U. Kjaer (2001): Pathways to the European Parliament National Office as a Stepping Stone?, Paper presented at the 29th ECPR Joint Sessions, April 6 April 11, 2001, Grenoble, France.
K. Krehbiel (1993): Wheres the Party?, in: British Journal of Political Science (23), pp. 235-266.
K. Krehbiel (1995): Cosponsors and Wafflers from A to Z, in: American Journal of Political Science (39), pp. 906-923.
K. Krehbiel (1999a): The Party Effect from A to Z and Beyond, in: Journal of Politics (61), pp. 832-840.
K. Krehbiel (1999b): Paradoxes of Parties in Congress, in: Legislative Studies Quarterly (24), pp. 31-64.
K. Krehbiel (2000): Party Discipline and Measures of Partisanship, in: American Journal of Political Science (44), pp. 206-221.
A. Kreppel, G. Tsebelis (1999): Coalition Formation in the European Parliament, in: Comparative Political Studies (32), pp. 933-966.
A. Kreppel (2000): Rules, Ideology and Coalition Formation in the European Parliament, in: European Union Politics (1), pp. 340-362.
M. Laver (1997): Private Desires, Political Action, London.
M. Laver, K.A. Shepsle (1990): Coalitions and Cabinet Government, in: American Political Science Review (84), pp. 873-890.
M. Laver, K.A. Shepsle (1996): The Making and Breaking of Governments, Cambridge.
M. Laver, K.A. Shepsle (1999): How Political Parties Emerged from the Primeval Slime: Party Cohesion, Party Discipline, and the Formation of Governments, in: S. Bowler et al. (eds.): Party Discipline and Parliamentary Government, Columbus.
S.D. Levitt (1996): How Do Senators Vote? Disentangling the Role of Voter Preferences, Party Affiliation, and Senator Ideology, in: American Economic Review (86), pp. 425-441.
S.M. Lipset, S. Rokkan (1967): Cleavage Structures, Party Systems, and Voter Alignments: An Introduction, in: S.M. Lipset, S. Rokkan (eds.): Party Systems and Voter Alignments: Cross-National Perspectives, New York.
M. Mair (1994): Party Organizations: From Civil Society to the State, in: S. Katz, P. Mair (eds.): How Parties Organize, London.
N. McCarthy (2001): The Hunt for Party Discipline in Congress, in: American Political Science Review (95), pp. 673-688.
R.D. McKelvey (1976): Intransitivities in multidimensional voting models and some implications for agenda control, in: Journal of Economic Theory (2), pp. 472-482.
M. Mezey (1979): Comparative Legislatures, Durham.
P. Mitchell (2000): Voters and their representatives: Electoral institutions and delegation in parliamentary democracies, in: European Journal of Political Research (37), pp. 335-351.
O. Niedermayer (2001): Die europäischen Parteienbünde, in: O.W. Gabriel et al. (eds.): Parteiendemokratie in Deutschland, Bonn, 2nd edition.
V. Nessler (1997): Europäische Willensbildung: die Fraktionen im Europaparlament zwischen nationalen Interessen, Parteipolitik und europäischer Integration, Schwalbach.
P. Norris, M. Franklin (1997): Social representation, in: European Journal of Political Research (32), pp. 185-210.
C. Neuhold (2001): The Legislative Backbone Keeping the Institution upright? The Role of European Parliament Committees in the EU Policy-Making Process, European Integration Online Papers (EIOP), Vol. 5, No. 10.
P. Norton (1990): Parliaments: A Framework for Analysis, in: West European Politics (13), pp. 1-9.
S.C. Patterson, G.A. Caldeira (1988): Party Voting in the United States Congress, in: British Journal of Political Science (18), pp. 111-131.
K.T. Poole, H. Rosenthal (1985): A Spatial Model for Legislative Roll Call Analysis, in: American Journal of Political Science (29), pp. 357-384.
K.T. Poole, H. Rosenthal (1991): Patterns of Congressional Voting, in: American Journal of Political Science (35), pp. 228-278.
G. Rahat, R.Y. Hazan (2001): Candidate Selection Methods: An Analytical Framework, in: Party Politics (7), pp. 297-322.
T. Raunio (1996): Party Group Behaviour in the European Parliament, Tampere.
T. Raunio (1999): The Challenge of Diversity: Party Cohesion in the European Parliament, in: S. Bowler et al. (eds.): Party Discipline and Parliamentary Government, Columbus.
T. Raunio (2000a): Losing Independence or Finally Gaining Recognition, in: Party Politics (6), pp. 211-223.
T. Raunio (2000b): Second-rate Parties? Towards a better understanding of the European Parliaments party groups, in: K. Heidar, R. Koole (eds.): Parliamentary Party Groups in European Democracies: Political Parties Behind Closed Doors, London.
L. Ray (1999): Measuring Party Orientations Towards European Integration: Results from an Expert Survey, in: European Journal of Political Research (36), pp. 283-306.
K. Reif, H. Schmitt (1980): Nine Second-Order National Elections A Conceptual Framework for the Analysis of European Election Results, in: European Journal of Political Research (8), pp. 3-44.
D.W. Rohde (1991): Parties and Leaders in the Postreform House, Chicago.
S. Scarrow (1997): Political Career Paths and the European Parliament, in: Legislative Studies Quarterly (22), pp. 253-263.
E.E. Schattschneider (1942): Party Government, New York.
J.M. Snyder, Jr. (1991): On Buying Legislatures, in: Economics and Politics (3), pp. 93-109.
J.M. Snyder, Jr., T. Groseclose (2000): Estimating Party Influence in Congressional Roll-Call Voting, in: American Journal of Political Science (44), pp. 187-205.
J.M. Snyder, Jr., T. Groseclose (2001): Estimating Party Influence on Roll Call Voting: Regression Coefficients versus Classification Success, in: American Political Science Review (95), pp. 689-698.
K. Strom (1990): A Behavioural Theory of Competitive Political Parties, in: American Journal of Political Science (34), pp. 642-669.
A.L. Teasdale (1999): The Politics of the 1999 European Elections, in: Government and Opposition (34), pp. 435-455.
B.R. Weingast (1979): A Rational Choice Perspective on Congressional Norms, in: American Journal of Political Science (23), pp. 245-262.
M. Westlake (1998): The European Parliaments Emerging Powers of Appointment, in: Journal of Common Market Studies (36), pp. 431-444.
(*) An earlier version of this paper was submitted as a Masters Thesis at the London School of Economics and Political Science (LSE), where the author was enrolled in the Masters Programme European Politics and Policy. I am grateful to Fabio Franchino, Cornelia Hentschel, Zoltan Juhasz, Andreas Wüst and the two anonymous referees, who provided helpful comments on earlier versions of this paper. In addition, I would like to thank Pippa Norris and Tapio Raunio who provided valuable data sources for this study.
(1) I accept this as an empirical fact. I am not going to discuss this from a normative point of view.
(2) The quote is actually a slight modification of Schattschneider (1942: 1), who used unthinkable instead of unworkable.
(3) This is, of course, adopted from Aldrich (1995).
(4) Clearly, this is a quite one-sided view. Parties as Lipset and Rokkan have pointed out in their seminal work also serve macro functions, as they serve as agents of conflicts and instruments of integration (1967: 3).
(5) It is also rational for party leaders to take on that task, as valuable assets are linked to it: money, fame, in short: office, which is generally assumed to be desirable.
(6) However, as Gallagher (1988b: 258) has pointed out, the two do not determine each other entirely!
(7) A strong party can successfully veto any coalition which it is not part of. Hence, it is still in a position to possibly form a minority government even if it lacks a majority in parliament,.
(8) The reader might be inclined to ask why a more favored policy output is discussed under the heading of office-seeking. The reason is that the Laver/Shepsle-model assumes exactly the hierarchy among the three goals: in the end, policy matters, but to achieve policy goals, parties have to gain control over ministerial departments, in other words: office.
(9) In his 1995 article (which is a case study of the US Congress), he provides evidence that non-partisan theories are better predictors. For a comment see Binder et al. (1999) and also Krehbiels reply (1999a). The debate still continues, see for example Levitt (1996), Jenkins (1999), McCarthy (2001), Snyder/Groseclose (2001), Ansolabehere (forthcoming).
(10) The fact that independent MEPs used to regularly form a technical group in order to benefit from being in a parliamentary group proves the effectiveness of these rules.
(11) An exception are those party groups that exist for mere technical reasons, like the Technical Group of Independent Members in the 1999 European Parliament.
(12) After a candidate for the presidency was carefully chosen in the European Council, the (national) governing parties (who agreed on that candidate in the Council meeting) apparently put pressure on their MEPs to vote for that candidate, even if that required defecting from the party group line.
(13) The latter is supposed to show that we can actually treat national party delegations as the unit of analysis for the further steps, since their cohesion is almost perfect.
(14) Speaking in technical terms, only those documents with an A5-number were included.
(15) The underlying idea is very similar to the one outlined by Cox/McCubbins (1991). They argue that one should only focus on those votes that are on the party agenda or that are party leadership votes, because the party group leadership only cares (and invests resources) on those. However, this per se does not solve the problem of differentiating between preference and partisan voting.
(16) Both measures are adopted from Hix (2001a) application of NOMINATE to 126 roll call votes of the 1999 European Parliament taken in 1999.
(17) We could also have included the number of national party delegations. However, all of these size measures are highly correlated. Hence, only two have been included.
(18) This is an index score of two measures: The first is based on the 1994 European Candidate Survey. Candidates running for a seat in the EP were asked to evaluate the influence of different actors in the nomination process. The differences between the estimated influence of national party leaders (1 to 7, the latter meaning strong influence) vs. local party leaders (also 1 to 7) are used here. The second is based on a recent survey by Tapio Raunio (2000a) who surveyed party officials about their candidate selection process. The answers were coded as follows: 1 for those parties that give members or local organizations the final say, 2 for those parties that give party conferences the final say, and 3 for parties, that give central party officials the final say. The final index is the mean of the standardized scores of the two indicators.
(19) This is also an index of two dummy variables: whether regional or local constituencies are used and whether a preferential vote is used, thus ranging from 0 to 2.
(20) This is based on two questions in the recent party survey by Tapio Raunio (2000a). The first is whether parties issue voting instructions (1 never, 2 sometimes/only on questions of fundamental importance, 3 always), the second whether there are other control mechanisms (0 no, 1 yes). Both were added to create the index scores.
(21) The basis for this measure are the expert surveys conducted by Huber/Inglehart (1995) and Ray (1999. The scores are the Euclidean distance to the party groups mean position. The mean position was calculated as a weighted average of the position of the parties it is made up of.
(22) Based on information at Elections around the World (http://www.agora.stm.it/elections/).
(23)Average scores of parties supporters as recorded in the Flash Eurobarometer 60, conducted in November 1997.
(24) The data, which was gathered from very different sources, was completely available for only 18 national delegations, those are: from Belgium CVP, Ecolo, PS, VLD; from Denmark: Venstre; from Germany: CDU, CSU, Grüne; from Italy: AN; from the Netherlands: CDA, GroenLinks, PvdA, VVD; from Spain: IU, PP, PSOE; from UK: Conservatives, Labour.
(25) Notwithstanding that general assessment, the British Conservatives lead the table of defecting national delegations.
(26) It is methodologically inflated: The defection of delegations affects the overall party group cohesion, too.