European Journal of Political Research 37: 309–333, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 309 Political parties in parliamentary democracies: Making delegation and accountability work WOLFGANG C. MÜLLER University of Vienna, Austria Abstract. In modern democracies political parties exist because (1) they reduce transaction costs in the electoral, parliamentary and governmental arenas and (2) help overcome the dilemma of collective action. In Western Europe political parties are the central mechanism to make the constitutional chain of political delegation and accountability work in practice. Party representatives in public office are ultimately the agents of the extra-parliamentary party organization. In order to contain agency loss parties rely on party-internal mechanisms and the institutionalisation of party rights in public rules and, in contrast to US parties, they apply the full range of ex ante and ex post mechanisms. Generally, the role of party is weaker the further down the chain of delegation. Introduction European democracies are not only parliamentary democracies but also party democracies. This fact has been seen both as a problem and a solution to the question of how democracy can be made to work in practice. Criticism was particularly strong in the pre-World War II period. Then political parties were victims of fundamental criticism by academic authors as well as politicians. Some critics contrasted party democracy with an idealised version of democracy based on individual representation. Others had a mythical “whole” – the nation – as their normative reference point. They criticised parties for dividing the nation and promoted non-democratic alternatives. Some of these arguments have been revived every now and then in the post-1945 period and new criticism has been added from a “participatory democracy” perspective (e.g., Barber 1980). Nevertheless, the idea that political parties are essential for practicing democracy in the modern state has become dominant. There is considerable disagreement between the various strands of democratic theory, however, on the precise role political parties should perform in order to make democracy work (see Ware (1987) for an excellent overview). In this article, I take an empirical perspective. I focus on the role of political parties in the context of the delegation and accountability perspective which is fully outlined in Strøm’s (2000) contribution to this special issue (see also 310 WOLFGANG C. MÜLLER Mueller 1996: 17). Of course, an article can neither do justice to the enormous organizational variation of European political parties nor to the considerable institutional variation of parliamentary systems. In this article I first place parties in the chain of delegation in parliamentary systems. Next, employing transaction cost and collective action theory, I discuss why rational actors turn to political party. In the fourth section I discuss the role of extra-parliamentary party organizations. Then I briefly map out the agency problems political parties face when placing their agents in public office. The penultimate section is an inventory of party-internal mechanisms and institutional features which can help to contain agency losses. The conclusion sums up the main arguments of this article. Political parties and the chain of delegation in parliamentary democracies Most constitutions in parliamentary democracies do not mention political parties at all. Those which do mention political parties are deliberately vague on their role in the political delegation process; while the existing “party clauses” in constitutions hint at their presence in the political delegation process, those constitutional clauses which describe the process in some detail tend to ignore political parties. Nevertheless in practical politics two kinds of delegation take place – one constitutional (to the governmental institutions) and one political (to political parties). This is sketched in Figure 1. The left and right sides of Figure 1 show the delegation process as outlined in many constitutions. The middle introduces party or more precisely parties and hence inter-party democracy. Party is central in all steps of the delegation process in the plural. Each step of the delegation process is structured by the interaction of political parties. In all steps of the delegation process we find party competition and cooperation. Figure 1 sketches a two-party coalition. (Note that the relations between government and opposition parties could be visualised in a similar way. It would only require leaving out the governmental institutions in the last three steps for one of the parties.) Figure 1 can help us to discuss normative and empirical questions of the parties’ place in the political delegation process. I begin by briefly summarising the prevailing normative reflections. These are more relevant for the empirical side of the question than deliberations by democratic theorists. To begin with, the legitimacy of parties in Western constitutional systems varies from step to step. Party intervention is most appreciated in the first step of delegation, the one from voters to members of Parliament (MPs). Here the mainstream of modern democratic theory argues that party intervention is necessary to make elections meaningful. POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 311 Party interference is generally considered less legitimate in the second step, from MPs to Government. According to the constitutions, MPs enjoy a free mandate. The interference of political parties hence is often seen as exercising illegitimate power over MPs. Nevertheless the majority of democratic theorists would argue for the need of parties in this step of the delegation process. Without party programmes (if only in the loosest sense) and MPs committed to these programmes, elections would not allow the voters to exercise prospective influence over policy. Moreover, it is generally accepted that parties are essential for making the democratic accountability of MPs meaningful. This critically depends on the voters’ ability to observe the behaviour of their representatives once they are elected. Cohesive political parties make this a manageable job and thus reduce the moral hazard problem (i.e., enhance the principal’s ability to observe the agent’s behaviour). In principle, the same should apply to the next step of the delegation process, the one from government to individual ministers. When it comes to government, however, party intervention is often considered less legitimate. The concept of party government (Rose 1976; Katz 1986) is largely an academic one which does not rally much public support behind it in its fullyfledged version. After all, it should be the government of the nation, not of a party. Government members are exposed to the public, they are personalities in their own right, while parties often appear as anonymous machines. Finally, party intervention into the delegation from ministers to civil servants is simply considered illegitimate according to the dominant Weberian view of bureaucracy. Civil servants should merely implement general rules and they should do so impartially. From a normative point of view, there is no room for party. Figure 1 also provides an assessment of empirical regularities of party intervention in the chain of delegation and accountability. As indicated by the strength of the arrows, party is central in the first step of the political delegation process. Elections are structured by political parties. Voters often do not know their MPs. Even where they do, in Europe the personal vote is weak compared to the party component. In the second step of the delegation chain, governments are created, maintained and brought down by the decisions of party leaders (Laver & Schofield 1990). While MPs are crucial for the implementation of these decisions, they hardly make them in their capacity as deputies. Single-party governments have a clear party agenda, often summarised in electoral manifestos, and within coalition governments the main lines of division are those between their party components. Despite the overwhelming normative conviction in west European democracies that party should not structure the final step in the delegation process, parties 312 WOLFGANG C. MÜLLER Figure 1. Political parties in the delegation chain of parliamentary democracies: a realistic assessment. frequently impinge on this step by making partisan appointments which in some cases (e.g., Belgium, Italy) add up to a fully-fledged patronage system. The rationale for political parties As mentioned above, normative theory is nowadays overwhelmingly positive about political parties’ having a stake in democratic governance (Robertson 1976; Ware 1987). But what is the rationale for political parties intervening in the process of popular representation from a positive point of view? Why do rational individuals – voters, activists, candidates, MPs, cabinet members – play the party game? Building on recent work in the rational choice tradition,1 the argument can be developed in analogy to the theory of the firm. The general answer is that political parties reduce transaction costs and solve collective action problems. POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 313 Reduction of transaction costs Political parties reduce transaction costs in politics, that is, the costs of using the market mechanism, specifically in the pursuit of votes (in elections, parliamentary votes). These are “search and information costs, bargaining and decision costs, policing and enforcement costs” (Dahlman 1979: 148). Internal organization replaces the market by hierarchy. This allows the owner to “invoke fiat to resolve differences”. Moreover it allows “easier and more complete access to the relevant information when dispute settling is needed” (Williamson 1981: 559). According to transaction costs logic, political parties have emerged where individual politicians or short-term alliances would not produce satisfactory solutions. In Europe parties have emerged to enhance work in the electoral, parliamentary and governmental arenas. Indeed, the initial establishment and/or shaping of the current form of these arenas has been a product of party activities since the 19th century. In the electoral arena party candidates take advantage of their party’s “brand name” for reasons of information economisation and strategic voting. The candidates’ party label allows voters to make more informed judgements about how they will behave once elected (Kiewiet & McCubbins 1991: 39; Wittman 1995: 10, 21; Lupia & McCubbins 1998; Jones & Hudson 1998: 184). Once national parties have become the vehicles of political competition there is hardly a market left for individual candidates. Then voters vote for parties (rather than individual candidates) because they do not want to waste their votes. This argument hinges on both the superior electoral viability of party and the dominance of party in the parliamentary arena. In parliament members of the respective parties act cohesively to protect their “brand name” and to maximise their chance of winning legislative battles (Cox & McCubbins 1993). The same logic applies to the governmental arena. This presentation leaves open some important questions. To begin with, where are the boundaries of party? The theory of the firm sees the boundaries of its subject “where the costs of organizing an extra transaction within the firm are equal to the costs involved in carrying this tansaction in the open market” (Coase 1988: 43). In a similar vein Aldrich (1995: 24) has stated: “politicians turn to their political party – that is, use its powers, resources, and institutional forms – when they believe doing so increases their prospects for winning desired outcomes, and they turn from it if it does not”. If individual politicians abandon their respective parties, interest groups, issue networks, personal alliances, etc. could help them to achieve their electoral and legislative goals. At a minimum, it would mean simply that politicians exercise the formal powers which come with their public offices without caring about their party’s preferences. 314 WOLFGANG C. MÜLLER All this does not automatically lead to the political party being superior to “spot market solutions”, i.e. the formation of ad-hoc alliances. In business “the costs of organizing certain transactions within the firm may be greater than the costs of carrying out the exchange transactions in the open market” (Coase [1937] 1988: 45; Williamson 1975: 117–131; 1981: 556–562). Charles Perrow (1986: 24–31) claims that markets are generally superior to hierarchy, because within firms contracts (between employers and employees) still need to be written, opportunism still exists, internal production involves a loss of flexibility, internal coordination has its costs, and market-like mechanisms such as internal pricing and cost systems within organizations are unreliable. Whoever is right with respect to business transactions, the case seems more straightforward in politics. To begin with, voters are interested in policy or patronage. For them it only makes sense to vote for a party if it is likely to make a difference in policy or patronage pay-offs. Political parties therefore have to make credible commitments that they will deliver once elected. Part and parcel of this commitment is their claim that they will indeed control the remainder of the delegation chain after the initial step from voters to MPs. As concerns work within parliament, forming winning alliances is costly. This means that already formed ones will be stabilised (Cox & McCubbins 1994: 227). According to Aldrich (1995: 36) “it is likely that transaction costs for parties are far less (. . . ) than those for forming new majorities for each piece of legislation”. However, in at least one respect, the analogy between political parties and firms does not work well. As mentioned above, in firms the market is replaced by hierarchy which in principle can be enforced easily. Politics is based on hierarchy as far as concerns the constitutional chain of delegation. However, political parties have remained largely outside this formal framework. While they may adopt their internal rules and hierarchy, these are not directly backed up by the formal authority of the state. The crucial question is how parties can enforce their decisions on the holders of public office. I will address this question in the following sections. Political parties and the collective action problem Since the path-breaking work of Olson (1965) it is widely recognized that having a common interest under most circumstances is not a sufficient condition for collective action. Rational individuals will not participate in collective action or will contribute less than they ought to, although successful collective action would benefit them (free-rider problem). Hence, why did rational individuals succeed in forming political parties in the first place? While it is easy to see that individual politicians are better off if they are backed by a party, POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 315 it remains to be explained how parties come into existence. Although this is not the central question of this article, which is concerned with a situation in which parties do exist and hence already shape the strategies of individual actors, this question should not be bypassed. Aldrich (1995: 28), in his magisterial study of American parties, convincingly argues that the formation of parties allow politicians “to win more of what they seek to win, more often, and over a longer period”.2 His historic account of party formation in the legislative and electoral arenas puts party formation (with the chance of winning in the future) against losing (now and also in the future). The few dozen members who made up the first Congressional parties may already have constituted small (or privileged) groups in Olson’s sense, which is to say that their personal interests sufficed to make their individual contributions significant and made their contributions individually rational. Nevertheless, Aldrich (1995: 78–82) places special emphasis on the role of political entrepreneurs who take over the costs of organising the party. These political entrepreneurs constitute the “critical mass” (Marwell & Oliver 1993) required to get parties started. Political entrepreneurs most critically contribute to solving the dilemma of collective action by monitoring other party members in order to ensure that they indeed contribute to the collective goal. Political entrepreneurs are compensated by a claim on some of the expected payoffs from collective action, in particular attractive leadership positions in public institutions (e.g., government office).3 Indeed, party histories as a rule identify party founders (i.e., political entrepreneurs), and they have almost always ended up in the most attractive public positions the respective parties had to fill. Before going into the details of the role political parties play in contemporary politics, I will briefly consider two alternatives: ad hoc alliances of individual politicians within the framework of formal political institutions and parliamentary committees. In the parliamentary and governmental arenas ad hoc alliances would mean vote trading. One of the problems with this strategy is that in politics there is no underlying medium of exchange such as money is in business. As a consequence “trading votes requires future reliance and hence the opportunities for reneging” (Weingast & Marshall 1988: 158). As game theory has proven, repeat play alone often is insufficient to maintain these deals (Weingast & Marshall 1988: 142; Cox & McCubbins 1993: 99– 102). Rather institutional solutions are required which complement reputation building and trigger strategies and “prevent the breakdown of cooperation at precisely the circumstances under which these other strategies fail” (Weingast & Marshall 1988: 142). Weingast & Marshall’s (1988) classic treatment of the US Congress has identified the committee system as the institutional device to protect le- 316 WOLFGANG C. MÜLLER gislative bargains against opportunistic defection. In this perspective the committees are assemblies of “preference outliers”. Congressional rules give committees agenda power. This means that they have a monopoly on bringing new legislation to the floor and hence can choose between new legislation and the status quo. Proceeding from US history and the British present Weingast & Marshall (1988) identify strong political parties as a potential alternative for the institutional underpinning of legislative exchange. Strong parties are those that control entry into competition for individual seats and the positions of power within the legislature and which wield considerable influence over legislative benefits (Weingast & Marshall 1988: 158–159). New research on the US has systematically elaborated and lent weight to this perspective (Cox & McCubbins 1993; Aldrich 1995).4 In the European context, no-one would seriously consider any alternative to political parties as the most important political coordination mechanism. They have set the stage since the First World War or even earlier. “Externally created” political parties emerged as vehicles of social classes excluded from the parliamentary arena by restrictive electoral law. “Internally created” political parties emerged first – in a loose form – to reduce the transaction costs of forming winning legislative alliances. Later they became tighter and also organized outside parliament in order to meet the challenge of the “externally created” political parties (Duverger 1959). In the contemporary context political entrepreneurs enter a stage which is already set by the existence of political parties as the main vehicles of political careers and coordination, and, as Aldrich (1995: 24–25) has remarked, “their existence creates incentives for their use”. Although occasionally political entrepreneurs try to establish new political parties, the usual pattern is to join existing ones. As mentioned above, political parties solve the collective action problem by establishing an institutional arrangement – party organization – which allows the monitoring of the other party members in order to ensure that they indeed contribute to the collective goal. The most elaborate version of this perspective is Cox & McCubbins’ theory of legislative parties. In order to overcome the dilemma of collective action political parties have leaders who (1) internalize the collective interest of the party and (2) monitor their fellow partisans. Party leaders control selective incentives. Cooperating party members get rewarded, while those whom the leaders deem to be “defecting” are punished (Cox & McCubbins 1993: 91). Party leadership positions are both attractive and elective. This means that there is internal competition for these positions and that incumbents can be held accountable if they fail to act in the collective interest. Although there may be disagreement about the best team of party leaders, leadership will be stable (rather than challenged at POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 317 each decision) “because some agreed-upon leadership team is better than no agreed-upon leadership team” (Cox & McCubbins 1993: 135). Cox & McCubbins have developed their theory for and applied it to the US Congress. In this article I use it as a starting point to discuss the role of parties in European parliamentary systems. Their characteristic chain of delegation (Strøm 2000) and strong extra-parliamentary party organizations constitute obvious differences, the consequences of which for political parties will be highlighted in the subsequent sections. Extra-parliamentary party organizations and the chain of delegation in parliamentary democracies Cox & McCubbins (1993, 1994) concentrate on the legislative party. It is only through the Congress members’ considerations of constituency interests that extraparliamentary factors come into play. Thus, their model of party organizations comes close to Downs’ (1957: 25–26) teams of leaders.5 In contrast, the typical European political party has a permanent extra-parliamentary organization which typically parallels the organization of the state. Party activists, who may be recruited from a more passive mass membership, elect party officials. These staff the various party bodies. At the same time public offices at all levels are filled with party representatives. These offices include those of MPs and, if the party is in government, cabinet members. Thus a twotrack system is established (Figure 2) with a party track and a constitutional track. Three types of delegation take place: within the party organization, within the constitutional chain, and between the party organization to the party representatives in public office. This article concentrates on this last type of delegation. So far the relationship between the party organization and the party in public office has remained underspecified. Indeed, the arrows indicating influence between the party organization and the party in public office in Figure 2 point in both directions. The question is: Where are the party leaders, i.e., those who internalize the collective interest of the party and monitor the party’s other office holders? In principle, these may be either the party officials in party office or those in public office. Answering this question the one way or the other would turn principals into agents and vice versa. For the sake of simplicity my point of departure is a set of well-established party models developed in the empirical literature. While the traditional cadre party (or party of individual representation) came close to the US parties, the mass integration party constitutes the most interesting contrast (Neumann 1956; Duverger 1959). This type of party was externally created (in the late 19th century) and its extra-parliamentary layer has remained dominant. 318 WOLFGANG C. MÜLLER Figure 2. Two party models and the chain of delegation. Both the parliamentary party and the party team in government are only instruments of the extra-parliamentary party, which is best exemplified by parties with a class or denominational origin (Duverger 1959). Mass integration parties transformed themselves into catch-all parties around the 1960s (Kirchheimer 1966). More recent changes have been captured in the cartel party and modern cadre party models (Katz & Mair 1995, 1996; Koole 1996). What these three models have in common is that the extra-parliamentary party organization is less important vis-à-vis party representatives in public office than in the mass integration party. As I have argued above, the main prize for the party leaders is the most attractive public offices the party has to fill. Recent empirical work on party-government relations indeed suggests that it is the leaders of the extra-parliamentary party who occupy the most attractive government offices which are available to their party (Blondel & Cotta 1996: 249–254; De Winter 1991). Exceptions to this rule occur mostly when the governments are considered unstable and potentially short-lived, hence when government office is not all that attractive. The leaders of opposition parties also occupy attractive public offices, mostly in parliament. POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 319 To be sure, there are two avenues to the leadership of the extraparliamentary party organization – the party channel and the public office channel. As a rule these channels are tied to each other, with individuals working up both ladders simultaneously. However, occasionally politicians do not advance equally on the party career track as in the public office track or have a head start by being appointed to government office on the basis of an extra-political career or closeness to the party leader. These appointments are nevertheless party-dependent, as the incumbents remain accountable to the party leaders. Even if public office holders of this type take over the leadership of the extra-parliamentary party it is up to the party organization to make this decision. European party organizations typically have a formal party-internal selection mechanism which directly or indirectly also involves the rank and file. A party will select those individuals as leaders who are considered most likely to achieve the party’s collective goals. These can either be policy, office, or votes (Strøm 1990; Müller & Strøm 1999). A vote-seeking party, for instance, may be well advised to select a leader who does not have the typical “party smell” and hence appeals to independent voters. Nevertheless, leadership selection remains a party choice. To put this discussion in a nutshell, despite the transformation of European parties in the post-war period, party representatives in public office ultimately remain the agents of the extra-parliamentary party organization. How can the party’s will be enforced on its representatives running for or in public office – candidates, MPs, cabinet members? Cox & McCubbins (1994) have argued that US legislative parties can withhold electoral benefits (the use of their brand name) and intra-legislative benefits (i.e., things that increase the value of the seat) from those Members of Congress who are not loyal to the party. In the European context the situation is more complex, given the predominance of extra-parliamentary party organizations and the parliamentary chain of delegation, including the executive. Parties can exercise their influence over the holders of public office via party-internal mechanisms and via the institutions and mechanisms of the parliamentary chain of delegation. The latter form of party control can in principle take two forms: (1) the institutionalization of party rights in public rules (such as the constitution, the electoral law, and the parliamentary rules of procedure) and (2) the party’s giving directions to public office holders who in turn use their institutional prerogatives to make other public office holders behave according to the party line. In this article I will concentrate on party-internal mechanisms and on the institutionalization of party rights.6 Before doing so in the penultimate section, I briefly address potential agency problems. 320 WOLFGANG C. MÜLLER Figure 3. Shirking and sabotage. Potential agency problems Party membership of MPs and cabinet ministers is often assumed to rule out conflicting interests between principals and agents, thus solving the agency problem. Common party membership should indeed contain agency problems compared to situations where agents are randomly chosen. Nevertheless, there is reason to believe that party membership of the agents is not sufficient to assume identity of interest. This is reflected, for instance, in the incompatability rules regarding party and public office of the early workers’ movement, which have been revived by the Greens in more recent decades. The fact that these rules have now largely been abolished probably tells more about the impracticability and unintended negative effects of these rules than about the irrelevance of agency problems in political parties. Agency problems result from conflicting interests and assymetric information. According to Brehm & Gates (1997: 50) agents may either work in the interest of their principal (no agency problem) or may engage in leisureshirking, dissent-shirking, or sabotage. These problems will be discussed in the remainder of this article with the help of Figure 3. It shows the range of policy positions which can be found among the members and officials of a political party in one policy dimension. Shirking and sabotage by public office holders – in particular ministers – may affect the party’s ability to move the policy status quo (sq) to its ideal position (p). Leisure-shirking Agents, here: party representatives in public office, may simply not do their best for the principal. Rather than doing their paper work, MPs may prefer to enjoy the pleasures of the capital. Likewise cabinet members can work towards maximising the limelight and having “state dinners in interesting and exotic places” (Laver & Schofield 1990: 40) rather than bringing home policy gains. Cabinet members may also “go native”, that is, they may find it more convenient to subscribe to the conventional wisdoms of their depart- POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 321 ments rather than to work day and night to push through party goals against a civil service displaying reluctant or even negative attitudes towards policy change. In coalition governments or parliamentary negotiations, party agents may settle with less than they could have extracted from their partners (m1), giving their best as negotiators. In Figure 3, leisure-shirking agents would do nothing or not much to move from sq to p, not because they like sq, but because of the costs of pursuing policy change. In the worst case scenario, inactive agents would not prevent other actors from moving policy in the direction of m2. Dissent-shirking Political parties are not monolithic. On most policy dimensions and issues they include a range of positions. To avoid agency loss, the party’s agents (e.g., committee members, ministers) should represent its ideal position on each policy dimension. Dissent-shirking means that party agents do not do their best to achieve the party’s most preferred position because they themselves have different preferences (Brehm & Gates 1997). A cabinet minister, for instance, may prefer sq or m1 to p. This means either that the status quo is preserved or that the party experiences an improvement over the status quo, for instance by moving to m1, which, however, is not enough to satisfy its ambitions. This conflict can also be one over time horizons rather than substance. While the agent may move in the direction the principal wants him or her to move, it may not be fast enough. Political sabotage While shirking leads to insufficient or no policies, political sabotage means the production of negative outputs. Party representatives actively work against the interest of their party. In Figure 3, such a case is illustrated by policy position m2, which a minister may prefer and work towards. While still within the range of ideal points represented within the party, for the party leadership it is inferior not only to ideal position p but also to the status quo sq. Party agents in government office may find this the best opportunity to “correct” the party line. In their capacity as head of government departments they enjoy enormous informational advantages vis-à-vis their party comrades. This is their most important resource for doing sabotage to the goals of their party (Brehm & Gates 1997: 19). Hidden information is particularly relevant to the extent that party goals are not fixed once and forever but are a function of parameters over which the executive members enjoy informational advantages.7 322 WOLFGANG C. MÜLLER Political sabotage is not confined to the governmental arena. In the electoral arena the candidates’ ambition to win single-member districts or preference votes may lead them to present their own sets of policies which are popular among their specific target electorates but which are at odds with the party’s general strategy. While the individual candidate may benefit from this behaviour, it will make the party look unreliable and harm its electoral prospects (Robertson 1976: 50–51; Kiewiet & McCubbins 1991: 43). In the parliamentary arena MPs may sabotage their party through indiscipline. The Third French Republic – the republic de députés – is the prototype of frequent behaviour of this kind, which was also prevalent in the Fourth French and "First" Italian Republics. By strategically abstaining or voting against the party line, an MPs may achieve individual policy and career goals, since if a government fell it would be replaced by a government of the same parties (Hine 1993: 170–171). “Therefore killing one’s own government is not (. . . ) giving the government away, but giving oneself a chance of entering the cabinet”, as Sartori (1997: 111) puts it. Obviously the frequent use of secret votes in parliament encourages this kind of political sabotage (Hine 1993: 190–193). Containing agency losses: party mechanisms and institutional features Argueably, the agency problems outlined in the previous section are the exception rather than the rule in European party democracies. This may indeed be true. However, it does not mean that political parties are immune against agency loss. Rather it may indicate that European parties have developed effective mechanisms to contain agency loss. Indeed, some institutional reforms (such as the severe restrictions on the use of the secret vote in Italy in 1988) were deliberate attempts to solve agency problems. This section maps out the most important party mechanisms and institutional features of parliamentary democracies that contain agency losses. In so doing, I look at four strategies principals (here, political parties) employ for this purpose (Kiewiet & McCubbins 1991: 27–37). Two are ex ante (contract design; screening and selection mechanisms) and two ex post (reporting requirements and monitoring; institutional checks). In their study of US Congressonal parties, Kiewiet & McCubbins (1991: 48) found contract design and reporting requirements and monitoring not to be relevant. The situation is quite different in Europe. In the remainder of this section I discuss the four strategies one by one. In each case I follow the chain of delegation in parliamentary democracies from voters via MPs, Prime Ministers or cabinets, and individual ministers to civil servants. And in each case I look for party-internal mechanisms and the institutionalization of party rights in public rules. POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 323 Contract design Parties intervene forcefully in the delegation from voters to MPs. Jones & Hudson (1998) argue that political parties offer an “implicit contract” between voters and politicians and thereby reduce the scope for opportunism by politicians. While the interests of politicians are short lived, political parties are enduring organizations (Aldrich 1995: 23) interested in maintaining their reputation. In the words of Wittman (1995: 21) “other members of the party have strong incentives to maintain the reputation of the party since the brand name is valuable in attracting votes”. But how can the party keep its deal with the voters? According to the implicit contract (which may be formalised by party rules), MPs are obliged to party discipline. “Party discipline helps curb the ‘excesses’ of politicians” (Jones & Hudson 1998: 177–178). But how to enforce party discipline? First, in most situations party discipline is a necessary condition for satisfying the politicians’ career ambitions. These, in turn, may be progressive, static, or discrete (Schlesinger 1991; Strøm 1997: 159). In the first two cases (i.e., if MPs have the ambition to be returned to parliament for another term, or if they aim at promotion within parliament, to the European Parliament, to executive office, etc.), the party’s ability to withhold these offices is a powerful means to make MPs follow the party line. There is no doubt that European political parties are the gate-keepers to these desired offices. What is less clear, however, is who controls party nomination for parliamentary elections and hence what “party discipline” means in the final consequence. In most parties the vast majority of candidates are nominated by the constituency party organizations (Gallagher & Marsh 1988; Katz & Mair 1992). If MPs are exposed to cross-pressures between local and national party interests, they will presumably do what suits their career best. Probably they will support local demands if they see that their renomination otherwise would be jeopardised. If renomination seems safe, however, ambitious MPs are likely to follow the national party’s line in order to enhance their career chances in and beyond parliament. In contrast to renomination as a parliamentary candidate, which appears on the agenda only infrequently, party and parliamentary offices in many countries can be withdrawn at any time (Damgaard 1995). Empirical studies show that sanctions are not imposed mechanically (Piper 1991) or in a minority of cases only (Saalfeld 1995: 286– 308). Yet the very existence of these sanctions and the MPs’ knowledge that they are indeed imposed at least occasionally help to contain agency loss. The moral hazard problem tends to be not severe in the case of MPs in parties with strong parliamentary representation. As to the former, their behaviour in the electoral arena takes place before a large audience. This also applies to much of their parliamentary activity. It is easy for the party leaders 324 WOLFGANG C. MÜLLER to check whether MPs observe party discipline in the plenary meeting, as well as in committees (Damgaard 1995). While probably more critical, the problem of hidden information also remains within narrowly drawn bounds. In large parties the informational advantage of individual MPs over other party officials (other MPs, cabinet members, etc.) tends to be small, though it can be considerable in smaller ones. Political entrepreneurs have one main incentive to remain loyal to their party: asset specifity. Once they have joined a party they are earmarked, that is in order to maintain credibility they are constrained in only being able to present the package of policies of the respective party (Jones & Hudson 1998: 184). To be sure, there are prominent individual exceptions (e.g., Winston Churchill) as well as new parties which establish themselves as break-away from other parties. Nevertheless, asset specificity tends to be greater in politics than in economics – politicians tend to be “locked into” the transaction (Williams 1981: 555). Politicians’ party loyalty is most difficult to achieve if their ambition is discrete, that is, they desire one particular public office – here that of MP – for one term only. This situation has been discussed as the “last-period problem” (e.g., Wittman 1995). It can be summarized as follows: Politicians cannot transfer their reputation in the way a firm’s goodwill can be capitalized and sold. Consequently, those who will not run again in the next election no longer have the electoral incentive to live up to their promises. Voters anticipate this and do not reelect politicians to their last term in office. Thus what was planned to be the individual politician’s second-to-the-last term becomes the last term, and so on. Thus the last-period problem makes the reputational solution unravel all the way. As a consequence, individual reputation is an insufficient means of linking MPs and voters. Does the same logic apply if political parties replace voters as principals? According to Wittman (1995: 20–21), the answer is ’no’: “the most important role in overcoming the last-period problem is played by the long-lived party”. First, the party has ideological inducements to make the incumbents’ behave in their last terms. To the extent that politicians are policy-seekers, they want “to uphold the party’s reputation so that the party’s candidates will continue to win in the future” and implement their values. However, the policy-seeking argument is not fully convincing. To be valid it would require the individual politician’s policy preferences to be identical with those of the party (p in Figure 3). Second, Wittman (1995: 21) introduces the idea of political heirs, to whom politicians want to transfer their reputation and hence contine to uphold the party reputation. While this may be true in many cases, frequently the relations between incumbents and their potential successors (e.g., of the young-man-in-a-hurry type) are strained and hence not likely to produce the POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 325 required incentive structure. A further agument may be added: there may be uncertainty on what will be the politician’s last term. This requires openended careers, which are more likely in the USA (as exemplified by, e.g., Strom Thurmond and Jesse Helms) than in Europe, where political parties tend to have formally or – in most cases – informally fixed retirement ages. Finally, to the extent that politicians are office-seekers, they may be compensated by other means for not shirking in their last period, e.g., by appointment to non-elective offices after their last term (Wittman 1995: 21). Evidence from Costa Rica, a country with term limits, supports this idea (Carey 1996). In particular, countries with a large public sector should have enough attractive offices available. Although this condition is generally met by European democracies, they differ substantially in the actual use of these resources for compensating (former) politicians, with, for instance, Belgium being much more “generous” than Scandinavian countries. In any case not all parties have access to these resources. A special case to maintain politicians’ loyalty during and beyond their last term is the mass integration party, covering, with its auxillary organizations, the whole life of its members, from cradle to grave (Neumann 1956; Duverger 1959). Turning one’s back on this kind of party puts at risk one’s whole “social existence”. Social relations may collapse with party membership for two reasons: the deliberate exclusion of “traitors” from the party community, or merely because the party had been the place to meet. While the all-encompassing party is now history, elements of it have survived in catch-all parties and beyond. The more parties lose these properties, the more important the last-period problem becomes. Thus far I have concentrated on the career incentives of MPs. To the extent that MPs are policy-seekers, they will consider the fact that without party they have virtually no influence in parliament unless their vote is pivotal (Cox & McCubbins 1994; Wittman 1995: 21). In contrast, maintaining party discipline allows them to win “more often” (Aldrich 1995: 28). This is because of the norm of reciprocity, which means that the other MPs of their party will also follow the party line and help them win in matters where the party’s policy is (almost) identical to their own preferences. The Prime Minister (or, in coalition governments, the party’s most important cabinet member) is likely to be the party leader, i.e., the one who has the prime responsibility for monitoring the other party officials. Thus the constitutional agency relationship is turned around. While the MPs play a role in holding the party leader accountable (i.e., making sure that he or she indeed acts in the interest of the party), it is the party leader who delegates specific tasks to the MPs and monitors how these are implemented. 326 WOLFGANG C. MÜLLER When it comes to delegating to individual cabinet members, the party and constitutional chain of delegation are united again. With individual cabinet members there is a greater potential of agency loss than with individual MPs, because the problems of hidden information and particularly hidden action are more severe. It is much more difficult to observe whether they fulfill their part of the contract. On the other hand, cabinet members are easier to punish than MPs. In the worst case, the Prime Minister can simply dismiss them. As pointed out above, the assumption made in the constitutions is that the governments’ party composition does not lead to civil service agency problems. If parties do not want to rely on the contract the civil servants have with the state, they can offer special partisan “contracts”. These oblige civil servants not just to be loyal to the law and the government of the day, but to a specific political party. In other words, the civil servants’ oblige themselves to work towards p in Figure 3. The party compensates these civil servants by providing their job in the first place and/or helping them advance the career ladder. However, the hidden information and hidden action problems remain. Again a party of the mass integration type is the best safeguard against these dangers. If the patronage strategy works, parties may nevertheless find themselves in office with civil cervants who have a “contract” with a different party. Loyalty to party may produce disloyal civil servants. A patronage strategy is likely to advantage those parties which are in office frequently and therefore can staff the permanent bureaucracy with their trustees. The role of political parties below cabinet can also be institutionalized by the establishment of genuine political appointees. The German beamtete Staatssekretäre, who temporarily retire when the political leadership changes hands, constitute an example. The strategy of providing the ministers with large personal staffs is found more frequently in Western Europe than politically accountable civil servants. However, country differences are vast with respect to building party strongholds in the administration (Müller 2000). I now turn to the direct institutionalization of party rights. Without giving details, Duverger (1959: 198) mentions “constitutional provisions which, in some countries, compel a parliamentary representative expelled from his party to present himself again to the electorate”. Such provisions clearly would be helpful for maintaining party discipline. In most countries, however, parliamentary seats are tied exclusively to individuals. Democratic theorists have occasionally suggested making continued party membership a condition for maintaining a parliamentary seat. Accordingly, those who resign their party membership or who get expelled would automatically lose their seat (Kelsen 1929). These suggestions were echoed in German discussions in the 1960s and 1970s (Müller 1974) without, however, being consequential. The only country where party membership is given preference over the free POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 327 mandate is India. Individual defectors automatically lose their seat. Only in the case of party splits, that is when a party is abandoned by one-third of its MPs, are these allowed to maintain their seat. These and similar proposals are still in the arsenal of institutional engineers (Sartori 1997: 192). The fact that automatic resignation is not available in most countries has not stopped political parties searching for means which would allow them to kick out MPs during their term. Duverger (1959: 198) mentions the device of the blank resignation: “candidates are compelled to sign, before their election, an undated letter of resignation; the party will fill in the blanks and forward the resignation should the representative once elected ever prove insubordinate”. This method has also been a long-standing practice in the major Austrian parties (the so-called Verzichtserklärungen). However, Duverger (1959: 198– 199) is right that this is not (or no longer) an effective means. More important is the social and frequently also economic punishment which often is at the party’s disposal for “traitors” (Maderthaner & Müller 1996). MPs who are party or trade union employees may lose their jobs. Those who work in the public sector may find their further career jeopardized. And even those in the private sector may find it more difficult to do business or to get access to public services. The role of party can also be institutionalized in public rules indirectly, by framing the conditions of political competition in ways which systematically favour political parties over individual candidates or loose electoral alliances and which strengthen the party vis-à-vis its own members. Thus the electoral law may contain a nationwide threshold in percent of valid votes, thus de facto ruling out individual candidates. Likewise, closed-list PR systems preclude candidates who have not been prioritised by their party from getting elected. Political finance rules can stack the deck in favour of permanent parties and concentrate resources in the hands of party leaders. Likewise, parliamentary resources (finance and personnel) can be given to the parliamentary parties rather than to individual MPs and the parliamentary rules of procedure can systematically favour parliamentary parties over individual MPs. Screening and selection mechanisms Principals cannot observe their agents’ actual performance until after having installed them. However, they can look for reliable signals of the potential agents’ qualities and interests. Potential parliamentary candidates often have to go through long periods of apprenticeship before they can win their party’s nomination or move up to “eligible” slots on the party list. In particular they serve in party and subnational public offices before they are nominated as parliamentary candidates. During this process parties check their moral qualifications and make sure 328 WOLFGANG C. MÜLLER that they acquire a certain amount of political professionalism, for instance learn to deliver a speech, argue in a discussion, understand their institutional environment, etc. Parties also allow for the specialisation of candidates and MPs. Both voters and parties are better off if MPs represent a wider spectrum of expertise.8 The screening of cabinet ministers within parties is also a long-term process. The service of potential ministers in the parliamentary party and/or sub-national and party office provides the selectors with ample cues. Even political “outsiders” recruited to cabinet office often have a record as party experts and advisors. The requirement of party screening is institutionalised in the UK and Ireland by the rule that a seat in parliament is a precondition for appointment to the cabinet. As concerns delegation within the cabinet screening and selection can only relate to designating specific cross-departmental tasks, for instance, working out an employment programme, to specific cabinet members. In most cases discretion will be limited, because typically formal competences go together with specific portfolios. When it comes to making civil service appointments, it is mostly ministers who act in behalf of their party. But there are exceptions. Back in the 1950s, the major Austrian parties had special departments in their headquarters which supervised and co-ordinated civil service recruitments. In Belgium, the personal cabinets of the ministers are still staffed by the parties (rather than the minister) (De Winter, Timmermans & Dumont 1997: 402). Reporting requirements and monitoring Parties may in order to avoid hidden action and hidden information, establish procedures which require their agents to report relevant information and actions they have taken. This strategy can have considerable costs since it deflects time and attention away from tasks that agents would otherwise be performing. It may also not be effective since “the agent has incentives to shade things, to make reports that reflect favourably upon himself, or to reveal information in some other strategic manner” (Kiewiet & McCubbins 1991: 32). Therefore principals typically also monitor their agents (“police patrol” oversight) and are interested in receiving information from affected third parties (“fire alarm” oversights) (McCubbins & Schwartz 1984). “Fire alarms” are not costly to the principal, but they can be strategic and hence misleading (Lupia & McCubbins 1994). Reporting is a permanent feature of many parties. MPs regularly report to their constituency organization. Ministers report to cabinet or, in coalition governments, to the party team in government. They also report to the parliamentary party and, less frequently, to the party executive committee and POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 329 the party congress. Some of these reports may be little more than propaganda and all will reflect favourably upon the respective agent. Nevertheless, some of the audiences are well positioned to challenge the agents. They can do so on the basis of their own monitoring (e.g., committee members, parliamentary spokespersons) and what they learn from “fire alarms”. The available empirical information suggests that parties, although generally falling short of text book party government, indeed exercise a considerable amount of permanent control over their cabinet members (De Winter 1993; Laver & Shepsle 1994; Blondel & Cotta 1996). The party organization “on the ground” often functions as a “fire alarm” concerning policy implementation and consequences. These alarms have the advantage of being relatively reliable, since “people with similar interests have little incentive to mislead each other” (Lupia & McCubbins 1994: 108). Institutional checks Institutional checks are defined as follows: when authority has been delegated to an agent, there is at least one other agent with the authority to veto or to block the actions of that agent (Kiewiet & McCubbins 1991: 34). In the context of this article it means that authority is divided or fragmented in the party at large, for instance, in a triangular way between the parliamentary party, the party team in government, and the party executive committee. Notwithstanding the primacy of the extra-parliamentary party organization (see above), this indeed seems to be the case in many parties. Crucial decisions often originate from the party in government, but the parliamentary party and the extra-parliamentary party leadership exercise veto power, at least in the sense that no action is taken until the issue has been reconsidered and an attempt has been made to hammer out a compromise. At lower levels individual ministers, party committee members and occasionally also other party representatives (e.g., sub-national public office holders) often function as institutional checks. Conclusion In this article I have discussed how political parties intervene in the chain of delegation in parliamentary democracies. Rational choice theories suggest that political parties exist because (1) they reduce transaction costs in the electoral, parliamentary and governmental arenas and (2) help overcome the dilemma of collective action. The second purpose is met by the parties having leaders who (1) internalise the party’s collective interest and (2) monitor the other party members. Party leaders have selective incentives available, as Cox 330 WOLFGANG C. MÜLLER & McCubbins (1993) have argued. They reward those party members who cooperate and punished those who don’t. On the basis of this article’s survey some general points can be made about parties in the political delegation process in Western European parliamentary democracies. Here, political parties are the central mechanism to make the constitutional chain of political delegation and accountability work in practice. European parties are clearly more relevant as political coordination mechanisms than their counterparts in the USA, for which the theories discussed in this article have been developed. In Europe party representatives in public office in the last consequence have remained the agents of the extraparliamentary party organization. In order to control the delegation chain political parties rely on party-internal mechanisms and the institutionalisation of party rights in public rules. They apply the full range of ex ante and ex post mechanisms to contain agency loss. In contrast to the findings for the USA (Kiewiet & McCubbins 1991), the ex ante mechanisms of contract design and screening and selection are particularly important. However, there are important differences between the various steps in the delegation chain, between countries, and over time. Generally, the role of party is weaker the further down the chain of delegation. Notwithstanding the dramatic differences between electoral systems, political parties have been very effective in building institutions that protect their control of elections. Party control over MPs is also very high: government formation and maintenance, legislation, and other parliamentary activities are overwhelmingly party-structured. Parties also control the daily work of the executive, though agency problems are certainly greater than with respect to MPs. Parties are least able to control the final stage in the delegation chain – the behaviour of civil servants. The fact that the role of political parties deteriorates the more the delegation chain develops, reflects increasing informational asymmetries and the relevance of normative constraints (i.e., what party influence is considered legitimate or at least tolerable). There are also considerable differences between parliamentary systems with respect to the parties’ control of the delegation chain. There are systems which are, or at some time were, fully-fledged partitocracies (Belgium, Italy, and Austria), systems in which political parties play a central but not all-embracing role (e.g., Germany, Norway), and systems in which parties are less enduring organizations than vehicles tailor-made for the personal ambition of individual politicians, as has been the case in the Fifth French Republic. Finally, there are reasons to belief that there are also differences over time. At the risk of oversimplification, parties have a less central role in the delegation process than 20 years ago both in their legitimacy and effectiveness. However, by and large the party-based mechanisms still work. While far from POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 331 perfect, political parties can still be considered the most important means to make delegation and accountability work in Western European parliamentary systems. Acknowledgements I thank the journal’s reviewers, Torbjörn Bergman, Delia Meth-Cohn, and particularly Kaare Strøm for helpful comments. Notes 1. See Cox & McCubbins (1993), Aldrich (1995), Wittman (1995), Jones & Hudson (1998). 2. Laver (1997: 87) is more specific. According to his theoretical account, political entrepreneurs have at least the following three incentives to form parties: (1) to control sufficient skills and resources to submit realistic tenders for political service contracts in the modern state, (2) to restrict competition (what in turn allows them to get the maximum payoffs from these contracts), and (3) to gain strategic bargaining advantages which stem from combining into larger groups. 3. See Laver (1997: 68–68) for a good general account of political entrepreneurs. 4. It has been questioned by Krehbiel (1993), who distinguishes party from preferences and speaks of party influence on legislative behaviour of MPs only when they act in line with their party but against their preferences. 5. Aldrich (1995: 163–193) has an interesting chapter introducing party activists which echos Robertson’s (1976) theoretical considerations. 6. The means the parliamentary chain provides are discussed in the other articles of this special issue. 7. An example would be a party which wants to maintain a certain level of military capacity, but to concentrate government funds on other activities once this level is achieved. 8. It is doubtful that pre-parliamentary specialisation covering a broad range of subjects could be achieved without national party organizations. Probably, homogeneous electoral districts would select typical representatives (e.g., rural districts would send farmers) while heterogeneous electoral districts would send generalists rather than specialist. References Aldrich, J.A. (1995). Why parties? Chicago: University of Chicago Press. Barber, B.R. (1980). The undemocratic party system: Citizenship in an elite/mass society, pp. 34–49 in R.A. Goldwin (ed.), Political parties in the eighties. Washington, D.C.: American Enterprise Institute. Blondel, J. & Cotta, M. (eds.) (1996). Party and government. London: Macmillan. Brehm, J. & Gates, S. (1997). Working, shirking, and sabotage. Ann Arbor: University of Michigan Press. Carey, J. (1996). Term limits and legislative representation. Cambridge: Cambridge University Press. 332 WOLFGANG C. MÜLLER Coase, R.H. (1988). The firm, the market, and the law. Chicago: University of Chicago Press. Cox, G.W. & McCubbins, M.D. (1993). Legislative Leviathan: Party government in the House. Berkeley: University of California Press. Cox, G.W. & McCubbins, M.D. (1994). Bonding, structure, and the stability of political parties: Party government in the House, Legislative Studies Quarterly 19: 215–231. Dahlman, C.J. (1979). The problem of externality, Journal of Law and Economics 22: 141– 162. Damgaard, E. (1995). How parties control committee members, pp. 308–325 in H. Döring (ed.), Parliaments and majority rule in Western Europe. New York: St. Martin’s Press. De Winter, L. (1991). Parliamentary and party pathways to the cabinet, pp. 44–69 in J. Blondel & J.-L. Thiébault (eds.), The profession of government ministers in Western Europe. London: Macmillan. De Winter, L. (1993). The links between cabinet and parties and cabinet decision-making, pp. 153–176 in J. Blondel & F. Müller-Rommel (eds.), Governing together. Houndmills: Macmillan. De Winter, L., Timmermans, A. & Dumont, P. (1997). Belgien: Über Regierungsabkommen, Evangelisten, Gläubige und Häretiker, pp. 371–442 in W.C. Müller & K. Strøm (eds.), Koalitionsregierungen in Westeuropa. Vienna: Signum. Downs, A. (1957). An economic theory of democracy. New Work: Harper & Row. Duverger, M. (1959). Political parties. London: Methuen. Gallagher, M. & Marsh, M. (eds.) (1988). Candidates selection in comparative perspective. London: Sage. Jones, P. & Hudson, J. (1998). The role of political parties: An analysis based on transaction costs, Public Choice 94: 175–189. Katz, R.S. (1986). Party government: a rationalistic conception, pp. 31–71 in F. Castles & R. Wildenmann (eds.), Visions and realities of party government. Berlin: de Gruyter. Katz, R.S. & Mair, P. (eds.) (1992). Party organizations. London: Sage. Katz, R.S. & Mair, P. (1995). Changing models of party organization and party democracy, Party Politics 1: 5–28. Katz, R.S. & Mair, P. (1996). Cadre, catch-all or cartel? A rejoinder, Party Politics 2: 525–534. Kelsen, H. (1929). Vom Wesen und Wert der Demokratie. Tübingen: Mohr. Kiewiet. R.D. & McCubbins, M.D. (1991). The logic of delegation. Chicago: University of Chicago Press. Kirchheimer, O. (1966). The transformation of the Western European party systems, pp. 177– 200 in J. LaPalombara & M. Weiner (eds.), Political parties and political development. Princeton: Princeton University Press. Koole, R. (1996). Cadre, catch-all or cartel? Party Politics 2: 507–523. Krehbiel, K. (1993). Where’s the party? British Journal of Political Science 23: 256–266. Laver, M. (1997). Private desires, political action. London: Sage. Laver, M. & Schofield, N. (1990). Multiparty government. Oxford: Oxford University Press. Laver, M. & Shepsle, K.A. (eds.) (1994). Cabinet ministers and parliamentary government. Cambridge: Cambridge University Press. Lupia, A. & McCubbins, M.D. (1994). Learning from oversight: Fire alarms and police patrols reconstructed, Journal of Law, Economics, & Organization 10: 96–125. Lupia, A. & McCubbins, M.D. (1998). The democratic dilemma: Can citizens learn what they need to know? Cambridge: Cambridge University Press. Maderthaner, W. & Müller, W.C. (eds.) (1996). Die Organisation der österreichischen Sozialdemokratie 1889–1995. Vienna: Löcker. POLITICAL PARTIES IN PARLIAMENTARY DEMOCRACIES 333 Marwell, G. & Oliver, P. (1993). The critical mass in collective action. Cambridge: Cambridge University Press. McCubbins, M.D. & Schwartz, T. (1984). Congressional oversight overlooked: Police patrols vs. fire alarms, American Journal of Political Science 28: 165–179. Mueller, D.C. (1996). Constitutional democracy. Oxford: Oxford University Press. Müller, M. (1974). Fraktionswechsel im Parteienstaat. Opladen: Westdeutscher Verlag. Müller, W.C. (2000). Patronage by national governments, in J. Blondel & M. Cotta (eds.), The nature of party government. Houndmills: Macmillan (forthcoming). Müller, W.C. & Strøm, K. (eds.) (1999). Policy, office, or votes? Cambridge: Cambridge University Press. Neumann, S. (1956). Toward a comparative study of political parties, pp. 395–421 in S. Neumann (ed.), Modern political parties. Chicago: University of Chicago Press. Olson, M. (1965). The logic of collective action. Cambridge, MA: Harvard University Press. Perrow, C. (1986). Economic theories of organization, Theory and Society 15: 11–45. Piper, J.R. (1991). British backbench rebellion and government appointments, 1945–1987, Legislative Studies Quarterly 16: 219–238. Robertson, D. (1976). A theory of party competition. London: John Wiley. Rose, R. (1976). The problem of party government. Harmondsworth: Penguin. Saalfeld, T. (1995). Parteisoldaten und Rebellen. Opladen: Leske + Budrich. Sartori, G. (1997). Comparative constitutional engineering. Houndmills: Macmillan. Schlesinger, J.A. (1991). Political parties and the winning of office. Ann Arbor: Michigan University Press. Strøm, K. (1990). A behavioral theory of competitive political parties, American Journal of Political Science 34: 565–598. Strøm, K. (1997). Rules, reasons and routines: Legislative roles in parliamentary democracies, Journal of Legislative Studies 3: 155–174. Strøm, K. (2000). Delegation and accountability in parliamentary democracies, European Journal of Political Research 37: 261–289. Ware, A. (1987). Citizens, parties and the state. Cambridge: Policy Press. Weingast, B.R. & Marshall, W.J. (1988). The industrial organization of Congress; or, why legislatures, like firms, are not organized as markets, Journal of Political Economy 96: 132–163. Williamson, O.E. (1981). The economics of organization: the transaction cost approach, American Journal of Sociology 87: 548–577. Williamson, O.E. (1985). Markets and hierarchies. New York: Free Press. Wittman, D.A. (1995). The myth of democratic failure. Chicago: University of Chicago Press. Address for correspondence: Professor Wolfgang C. Müller, Department of Government, University of Vienna, Hohenstaufengasse 9, A-1010 Vienna, Austria Phone: +43 1 4277 38306; Fax: +43 1 4277 9383; E-mail: wolfgang.mueller@univie.ac.at