The objective of this article is to highlight and discuss new approaches regarding the relationship between party politics and the welfare state. These approaches can be qualified as new because they differ from traditional ‘partisan politics theory’. The latter is a well-established theoretical strand in comparative welfare state research and, more broadly, in comparative political economy. In a nutshell, it sees parties as representatives of social constituencies, mostly defined in terms of industrial classes, and as bearers of clear ideological stances for social-democratic or conservative welfare policies. Social policy output, in this view, depends on the partisan composition of government.
New ways of conceptualizing the role and impact of political parties on social and economic policies have been thriving in recent years, but the different contributions have neither been reviewed nor connected. This is what we do in this article. We identify three ways in which recent studies depart from traditional theory. First, it must be taken into account that electoral constituencies have changed and do not correspond to those of the industrial age anymore. Secondly, various contributions have shown that the institutional context, party systems and party competition matter for the kind of welfare policies individual political parties advocate. Thirdly, diverse links between parties and electorates (particularistic versus programmatic) shape the policy strategies adopted by parties.
Traditional partisan politics theory assumes a linear and direct relationship between the type of party in power (e.g. social-democratic or conservative) and policy output. Each of the three strands of research that we are going to address in this article questions this linear relationship in its own way. The empirical analysis of changing socio-structural electoral constituencies alters our expectations, because it means that parties represent different interests from the ones that marked their ideological profiles in the early decades of welfare state growth.Footnote 1 If, in turn, we take context seriously, both in terms of institutions and party competition, the expected policy preference of a party is not ‘given’ by its ideological party family, but is conditional on the interaction with institutions or rival parties.Footnote 2 Similarly, acknowledging the importance of voter–party linkages reminds us that parties often use policies in a particularistic way rather than in the programmatic way implied by the traditional theory of partisan politics. Hence, parties with the same ideological label may make different choices if they act under particularistic competition.Footnote 3 Each of these three strands of research has its own distinct implications. But all of them imply that we cannot adequately grasp the complex relationship between political parties and the welfare state unless we update and expand our model of party politics. Yet, in many respects the old partisan theory remains, of course, valuable. Consequently, in this article, we propose scope conditions in which traditional partisan politics theory may apply, and we also discuss research strategies for those cases where the traditional approach cannot be applied in a straightforward way.
In our discussion of recent research on partisan politics, we do not aim at an exhaustive review of research on political parties and political economy, an endeavour that would be doomed to failure given the breadth of the topic. Rather, we concentrate on recent contributions that satisfy two criteria: first, we focus on the literature in comparative welfare state research. That is to say, we do not discuss similar contributions that belong to the wider field of comparative political economy, such as industrial relations or social pacts.Footnote 4 Secondly, we focus on those welfare state studies that clearly refer to political parties. This excludes studies that focus on government strategies independently of party political considerations.Footnote 5
In the following, we distinguish between an ‘old’ and a ‘new’ school of studies on the role of political parties in welfare state politics. The terms ‘old’ and ‘new’ are, however, not primarily used in a temporal sense, since the traditional partisan politics school is clearly alive and retains a considerable and robust explanatory role for some research questions. In that sense, the ‘new’ approaches we discuss are not aimed at entirely replacing or invalidating the traditional approach. However, they all deviate from the ‘old school’ approach by theorizing and conceptualizing factors that limit or question the key assumption underlying the traditional approach. This key assumption holds that we can deduce the policy preferences of a political party directly from its party family or even from its label. Therefore, one of the main contributions of this review article is to specify the scope conditions under which traditional partisan politics assumptions are justified, and to discuss the empirical and theoretical implications if one or several of these conditions are not present.
Where does the recent questioning of traditional party politics theory come from? We believe that an important impulse can be found in the debate on the ‘New Politics of the Welfare State’, which stirred up comparative welfare state research in the late 1990s and early 2000sFootnote 6 – even if some significant re-considerations of the role of parties are to be found earlier.Footnote 7 The New Politics literature focused primarily on the path-dependent effects of established policies. It argued that the very expansion of the welfare state itself changed the rules of the political game by changing the preferences and expectations of voters and interest organizations. Hence, the effect of partisanship on policy output and outcome would have been weakened, if not erased, by policy legacies. This argument, of course, provoked a wave of reactions, not only in defence of the traditional approach, but also – more constructively – in an effort to re-conceptualize the ways in which parties might still matter.Footnote 8 For instance, it was argued that parties may behave differently depending on the context of policy making (credit-claiming during welfare expansion v. blame-avoidance in times of austerity),Footnote 9 or that social-democratic parties might even be in a better position to reform the welfare state than conservative parties.Footnote 10 Since then, more and more researchers have taken up the task of rethinking party politics and the welfare state, but their contributions have so far been neither systematized nor connected, and this is what we aim at doing in this review article.
The article is structured as follows. In the next section we revisit the traditional partisan politics theories of the welfare state. Then we discuss three strands of new research on party politics and the welfare state, followed by a section on the implications of these new approaches for the use of the traditional theory of partisan politics and for new paths of analysis. A final section concludes by identifying the most promising routes for further research.
‘Old School’: Partisan Politics and the Welfare State
The partisan politics approach to welfare state analysis has emerged as a rival explanation to structural Marxist, functionalist and pluralist theories from the 1970s onwards.Footnote 11 The first thorough comparative analyses regarding the effect of political parties on distributive policies have been presented by Hibbs, Tufte and Hewitt.Footnote 12 These authors demonstrated that left-wing parties took different policy decisions and were associated with more redistribution, tracing these partisan effects back to the distinct class constituency of left-wing and right-wing parties. In a similar, basically symmetrical vein, Borg and Castles, and Castles separately,Footnote 13 have demonstrated effects of the political right on policy output.
Parallel to this ‘parties-matter’ literature, a related but separate theory on the impact of social-democratic parties on the welfare state has been developed by John Stephens and by Walter Korpi.Footnote 14 This power resources theory shares the argument that party politics shape social policy and political economy, but it lays more emphasis on the mobilization of socio-structural classes as the basis of party politics, and on the impact of other class actors such as trade unions. Korpi argued that social policy should be understood as a result of the democratic class struggle, since democracy provides the working-class parties with the opportunity to move ‘the struggle for distribution … into the political arena, where their numerical strength can be used more effectively’.Footnote 15 The social rights won in the parliamentary arena then feed back into the economic sphere as new power resources, counterbalancing the power of capital. In line with power resources theory, Esping-Andersen argued that social-democratic parties not only develop these social policies as resources for wage-earners and the worse-off, but also as distinct instruments to influence class formation and thereby social democracy's mobilization power.Footnote 16 In this manner, power resources theory made the link between class mobilization, party affiliation and partisan government more explicit than the original partisan-politics literature. It added nuanced arguments on the role of collective action through parties and unions, emphasized the feedback effects of policies on collective mobilization and supported the constituency-party link with detailed studies on the class structure of politics.Footnote 17
Both the ‘parties-matter’ studies and power resource theory set out the theoretical framework for a large number of studies using the ‘partisan politics matter’ hypothesis in their work.Footnote 18 Using mostly quantitative methods, these authors test whether stronger left-wing parties – measured by parliamentary or cabinet seats – lead to an expansion of the welfare state. All these studies ‘postulate’Footnote 19 that left parties mobilize the lower wage earners and the ‘working class, as the flagship of redistributive collectivism’.Footnote 20 Thereby, politics is often seen in an overly simplified way, ‘as a simple transmission belt conveying the preferences and demands of various interest groups to the leaders, who implement them’ (as Korpi himself puts it in a criticism of simplified partisan politics scholarship).Footnote 21 In its most reduced version – labelled ‘partisan difference theory’ – the partisan politics approach simply tests whether the complexion of government on a left–right scale matters for policy outputs.Footnote 22 As conditions changed from the era of expansion to times of austerity, a mirror-reasoning was applied to welfare state retrenchment: it was argued that left parties stand for less cuts in the realm of social policy, because they represent the beneficiaries of the welfare state and sustain a pro-welfare ideology.Footnote 23 All these studies share the basic assumptions regarding left parties, their voters and policy preferences: the working class is the main constituency of the left, it holds strong preferences for redistribution and social insurance, and – consequently – the left fights for welfare state expansion in the democratic class struggle.
‘New School’: Post-Industrial Electorates, Context and Party–Voter Linkages
Over the past ten years, an increasing number of studies have shown that party politics often shapes social policy in ways that differ from the hypotheses and mechanisms implied by traditional partisan politics theory. In the following, we group these studies according to the argument that sets them apart from traditional theory: firstly, the electoral constituencies of parties have changed; secondly, electoral institutions, party systems and interaction between parties matter for explaining social policy reforms; thirdly, different kinds of linkages between parties and their voters shape policy making. With these three strands of research, we discuss a range of alternative approaches to the study of partisan politics and the welfare state, which do not refer to each other explicitly, even though they share a common objective in rethinking the role of parties in welfare policy development. We argue that they can and should be brought together in order to gain a better understanding of party politics and social policy. We discuss each of them in turn.
Changing Electoral Constituencies: What Parties Do Depends on Who They Represent
The traditional partisan politics literature on the welfare state sees political parties as the key actors in the ‘democratic class struggle’, a concept that stresses the representative function of parties.Footnote 24 Studies in the tradition of power resource theory have shown that left-wing parties were a driving force of welfare state expansion and assumed that the reason for this was to be found in the working-class interests the left represented. Most of the more recent literature on social policy making since the 1980s still shares that same assumption. However, none of the major studies in the partisan politics tradition actually tests the electoral foundations of the political parties.Footnote 25 Rather, they postulate that the left still represents the same ‘working-class interests’ of redistributive collectivism as it did in the era of welfare state growth. In the light of this assumption on electoral constituencies, they interpret their empirical results: studies that find a positive impact of left-wing power on redistribution and welfare generosity take this as evidence that the model with all its underlying assumptions still holds: welfare politics is still the same democratic class struggle it was fifty years ago.Footnote 26 Many studies indeed do still find such an effect, but the explanatory power of the ‘left party variable’ has become weaker.Footnote 27 This weakening is seen as evidence that parties matter less, and that their programmatic differences are increasingly constrained and narrowed by exogenous forces, such as globalization or fiscal constraints. Finally, studies that point to the absence of party differences or ‘unexpected’ party behaviour (such as left-wing parties cutting back on welfare or different left-wing and right-wing governments pursuing very similar social policy agendas) interpret this as evidence for a loosening link between parties and their electorates,Footnote 28 or even a ‘hollowing’ of the representative function of parties.Footnote 29 In short, if left-wing parties do not defend the ‘old’ welfare state and push for more redistribution, and if right-wing parties do not push for less state intervention and redistribution, this is interpreted as evidence that parties’ policy strategies are not driven by the interests of their core electorates anymore. This may be a reasonable argument, but it rests entirely on the – untested – assumption that the electoral patterns have remained stable.
Such an assumption is obviously problematic because there is ample reason to believe that electorates and voter interests have changed profoundly over recent decades. There is a wide literature on party systems, which shows just how much social structure and electoral behaviour have evolved since the 1970s, entailing a range of electoral dealignments and realignments.Footnote 30 This literature, however, is almost completely ignored by welfare state scholars. The study of electoral change implies that parties may very well still be ‘programmatic organizations with well-developed ties to particular social groups’,Footnote 31 but that we need to re-conceptualize who these particular social groups are. Once we get to a more adequate conceptualization of electoral constituencies, partisan politics variables may explain just as much of social policy development as they did in the heyday of power resource theory.
There are two distinct sides to the argument that recent electoral change needs to be taken into account, one relating to shifts in the electorate (i.e. the same parties today mobilize different social groups than two or three decades ago) and a second relating to shifts in preferences (i.e. even if parties still mobilize their voters in the same socio-structural groups, these ‘traditional voters’ may have changed preferences).
The argument on electoral shifts mostly refers to the fact that left-wing parties increasingly attract highly skilled middle-class voters.Footnote 32 Different left-wing parties (social-democratic parties, green parties, radical left parties) have mobilized a new electorate of younger, highly educated and strongly libertarian voters, many of them women. The shift of these middle-class voters to the left is driven by cultural, rather than economic factors.Footnote 33 However, some authors argue that it nonetheless has profound implications for distributive partisan politics.Footnote 34 Highly skilled women working in the service sector, as an example of typical ‘new left’ voters, obviously incur different social risks and have different social preferences from the average male production worker. They demand social policies linked to education, social investment, universalism and gender equality. By contrast, they have less interest in traditional redistributive schemes (because they are highly skilled and thus have high earnings-power in the market). This implies that to the extent that left-wing parties rely on this ‘new’ electorate, we should expect a more market-liberal social investment agenda, rather than a traditional redistributive policy agenda.
The second development concerns ‘traditional constituencies’ changing their policy preferences. This argument deals with working-class interests. Workers and low-income voters increasingly adhere to traditionalist and anti-immigration stances. Several studies show that these issues have gained saliency among the working-class voters, even leading some of them to abandon the left for the populist right.Footnote 35 This growing threat of electoral loss obviously has implications for the social policy agenda of the left-wing parties, driving it potentially in a more welfare chauvinist direction.
Changing working-class interests, however, are not confined to new cultural orientations. With regard to economic and welfare preferences, a growing literature shows that the ‘working class’ today is divided into at least two groups: insiders and outsiders.Footnote 36 The insider–outsider distinction refers to a divide ‘between a shrinking, largely male core of secure and privileged employees, and a mass of more or less chronically unemployed and marginalized populations’,Footnote 37 which differ with regard to their social needs and their preferences. Different authors have developed distinct hypotheses as to how this divide is mobilized in the party political realm. Rueda and King are so far the clearest voices in this debate.Footnote 38 They argue (but do not test empirically) that labour-market insiders continue to vote for social-democratic parties, while outsiders either abstain from voting or vote for radical right parties. If that was true, we would expect left-wing parties to defend the status quo of insider-protection (mainly through social insurance policies), rather than pushing for more outsider-oriented policies (such as needs-based minimum protection, social investment and universal benefits). In contrast, Häusermann and Walter have analysed the electoral reconfiguration and constituency preferences empirically for Switzerland, showing that it is actually the old insider working class whose party preferences shift to the right, in defence of the status quo and welfare chauvinism.Footnote 39 Outsiders, by contrast, either abstain from voting or vote for left parties that advocate more universal and needs-based social policies. This implies that we should expect the radical right to defend the welfare status quo, while those parties of the left (be they green, social-democratic or radical left parties) that attract the outsider-votes should advocate a universalist and redistributive agenda.
In sum, studies on underlying electoral shifts show that we need to update our assumptions regarding the interests parties actually represent. Only if we formulate expected party policies based on their current and empirically established electoral basis can we test whether parties still fulfil their representative functions or not. Just because parties ‘do different things’ from what they did thirty years ago does not mean they do not represent their voters. Parties may defend different policies simply because they represent different social groups.
Context: What Parties Do Depends on Institutions and Party Competition
In the traditional partisan politics approach the goals and policy choices of a party depend on the ideological ‘family’ it belongs to, which is ultimately a function of its social class constituency. Such a close link between preferences of party voters and policy preferences of parties is also unquestioned in the aforementioned contributions, which study the transformation of these constituencies. Both approaches are largely blind to the way in which the preferences and actions of a party may be conditional on the institutional and partisan environment. However, a range of recent works highlights such contextual determinants, thus potentially overturning the assumptions that guide the traditional partisan politics approach.
With regard to institutions, a growing literature analyses the links between electoral rules, partisan preferences and partisan effects. Iversen and Soskice have argued that left-party governments and, therefore, high levels of redistribution are less likely in majoritarian electoral systems than under proportional representation (PR), because the median voter in a two-party system has a rational incentive to vote for the right, whereas the multiparty system that is typical of PR allows for pro-redistribution coalitions of middle-class and lower-class parties.Footnote 40 In a similar vein, Jusko and also Rodden explain how the electoral rules and the geographical distribution of low-income voters condition electoral mobilization strategies and – ultimately – the power of left parties.Footnote 41 This is an important complement to the partisan politics approach because it shows that two left parties operating under different electoral rules may differ in their effect on welfare policies. However, some of these studies do not question the basic assumption regarding policy preferences of left- and right-wing parties, i.e. they still assume that left-wing parties represent the low-income classes and, consequently, promote more redistribution. In that sense, institutions do not alter partisan preferences, they rather condition power relations. Recent work by Ansell, however, takes the institutionalist argument one decisive step further.Footnote 42 Ansell contends that the actual preferences of parties (not only their relative power) depend on the institutional context, in this case the institutions of the welfare state itself. He shows that whether left parties favour or contest the allocation of resources to higher education depends on the structure of the higher education system (mass v. elite system) in a country. In a mass university system, left parties favour increased spending on tertiary education, because it has a redistributive effect, whereas they contest it in an elite-system that entails a regressive effect. Taking this result seriously means that we cannot simply compare spending preferences across countries because these preferences take a different meaning depending on the institutional context. The implication is that policy preferences of left-wing and right-wing parties cannot be assumed easily, but need to be contextualized.
The second main contextual factor influencing partisan politics is the party system itself. In this respect, we can distinguish between studies that focus – more statically – on the consequences of party system structures (in particular, in terms of cleavages), and those that focus on the dynamics of party competition that are generated by different spatial configurations of party systems.
The ‘old school’ approach presupposes a party system that offers a simple choice between left and right on social and economic issues. The party system most compatible with this perspective is, of course, a two-party system (such as the American or the [former] British party system).Footnote 43 In case of multi-party systems, the analogy would imply that all parties can be allocated to only two camps (forming a bipolar party system), each constituting alternative government options and the main dimension of competition being socio-economic policies. In the terminology of Lipset and Rokkan, these are party systems with one cleavage only: the capital–labour conflict.Footnote 44
However, some authors have stressed that secondary cleavages in addition to capital–labour have shaped welfare state development. In Continental Europe, in addition to social-democratic parties, Christian democratic parties – arising from the state–church cleavage – are strong and have had a large influence in welfare state expansion. This distinct role of Christian democratic parties was stressed early on by van Kersbergen.Footnote 45 In Nordic countries, the urban–rural cleavage has found expression in agrarian parties, which supported and shaped the construction of generous welfare states that was otherwise advanced by social-democratic parties.Footnote 46 In a more general vein, Ferrera has shown that the presence of ethnic, linguistic or religious cleavages, as well as ideological polarization, impedes the introduction of universalistic policies and favours fragmented and occupationally defined programmes.Footnote 47 This view is corroborated by Watson who highlights how a division within the left leads to a struggle for dominance between left-wing parties, in which social policy is used in strategic and targeted ways, impeding encompassing policy solutions.Footnote 48
Other scholars have looked at the different competition dynamics created by various party system configurations. Here, we can distinguish one view that regards party competition rather as an intervening variable, potentially constraining the feasible courses of action of government parties, and another view, which takes party competition as a more fundamental driving factor of parties’ policy preferences. The first perspective has been applied to welfare retrenchment by Kitschelt.Footnote 49 He argues that left-wing competition can hinder – and right-wing competition can facilitate – government decisions to cut back welfare entitlements. In addition, the ability of parties to adopt retrenchment depends on the characteristics of the competitors, such as programmatic credibility or organizational flexibility, and on the salience of socio-economic issues. Kitschelt does not really discuss why parties pursue retrenchment in the first place. But in pursuing this goal they are constrained to varying degrees by party competition.
The second, more fundamental, perspective on party competition holds that social policy making is driven essentially by strategic considerations, which depend on the positioning of parties in political space. According to Picot, policies are often used to appeal to important groups of voters, rather than pursuing long-standing ideological policy programmes.Footnote 50 Who the decisive groups of voters are depends on the spatial configuration of the party system. Green-Pedersen applies a similar reasoning to the parliamentary arena and to the parties’ goal of entering government coalitions.Footnote 51 Parties may adjust their position on social policy issues to the position of a potential coalition partner if that partner is pivotal for forming a coalition. This strategic understanding of party competition departs more clearly from the traditional perspective on partisan politics. In the traditional – bottom-up – approach, parties are the agents of a social class. The strategic view of party competition, by contrast, sees parties as resourceful organizations that are partially autonomous from social structures and – in a top-down manner – deploy social policy to mobilize electoral support. This view is rooted in and supported by the literature on party organizations, which argues that parties used to have a closer connection to their electorate in the past, but are more autonomous today.Footnote 52 Traditional partisan politics theory is based on the model of the socialist mass party that grew out of the labour movement. This type of party did indeed have a clearly defined constituency and was connected to it by close organizational ties. However, it is argued that, since the Second World War, the organizational model of the mass party has been largely superseded by the catch-all party, which has deliberately tried to broaden its electoral basis. Some go even further, suggesting that since the 1970s the catch-all party is being replaced by the cartel party, which has even looser electoral ties and is more strongly incorporated into the state.Footnote 53 In any case, the changes in party organization indicate that political parties have become less dependent on their historical core voters and thus strategic electoral considerations may have become more relevant for policy decisions.Footnote 54
Overall, the studies discussed in this section show that in order to hypothesize a party's policy preferences, it is not enough to know whether it is left or right. Rather, we need to study the context in which the party operates. Electoral institutions condition the way parties come to power, policy institutions influence their policy preferences, cleavages structure the political terrain and coalition-formation, and the spatial configuration of party competition constrains or shapes policy choices in the light of electoral trade-offs.
Party–Voter Linkages: What Parties Do Depends on How They Relate to Voters
The relation between parties and their voters, i.e. party–electorate linkages, can be based on programmatic considerations, on material-particularistic motives, or on socio-cultural ties. In other words, voters may choose a party because they are genuinely convinced of the policies it proposes, because they expect to get a direct benefit in return for their vote, or because they identify with this party due to their social and cultural background.Footnote 55 The traditional partisan politics approach has not conceptualized party–electorate linkages explicitly. Most common is the perspective that sees parties and their voters attached to a broad ideology.Footnote 56 Parties are supposed to promote a certain policy programme (such as generous and universal social benefits in the case of social-democratic parties), which represents the preferences of their voters. Hence, politics is seen in programmatic rather than particularistic or cultural terms.
In contrast to this programmatic perspective, some researchers have stressed that social policy is sometimes shaped by the particularistic motives of parties. This has been indicated early on in national welfare state literatures, such as those on Italy, Greece and Japan.Footnote 57 In a comparative perspective, the study by Orloff and Skocpol on early social policy development in the United States and the United Kingdom was influential.Footnote 58 Drawing on Shefter's work on state building and party patronage,Footnote 59 they show how a lack of administrative autonomy in the United States facilitated clientelist abuse of social programmes in the late nineteenth century. The existence of these clientelist relations in turn kept progressive politicians from demanding social policy expansion, as they feared that new social measures would be hijacked for clientelist purposes as well. Therefore, if particularistic party–electorate linkages prevail, this can have effects even on the policy stance of programmatically minded politicians.
In spite of these early studies, the effects of particularistic party–electorate linkages on social policy have not been more widely recognized on the comparative level until recently. Lynch's widely cited book published in 2006 has shown how particularistic linkage strategies play a role at important junctures of welfare state development. In particular, they can keep governments in occupation-based welfare states from adopting citizenship-based policies. Lynch explains: ‘Occupationally based social insurance programmes plainly lend themselves far more than do universal programmes to the kind of fine-grained targeting of incentives on which particularistic political competition thrives’.Footnote 60 Moreover, particularistic political actors will not promote the development of neutral state capacities as these could put at risk their particularistic strategies. However, universal policies cannot be implemented without adequate state capacity. Consequently, even programmatically oriented politicians refrain from promoting universalist schemes in the context of a weak state.
Estévez-Abe's book on the Japanese welfare state takes the linkage argument a step further by exploring the institutional incentives for different linkage strategies.Footnote 61 According to her, the degree to which parties adopt targeted or universalistic policies depends on the electoral system. Electoral rules determine electoral strategies and, consequently, the policy choices of parties and politicians. The district magnitude and the importance of the personal vote are two crucial parameters. In multi-member districts, electoral competition revolves mainly around organized groups of voters and, therefore, narrower interests. The personal vote, in turn, provides individual candidates with incentives to make particularistic policy choices. Hence, in Estévez-Abe's model the particularistic or programmatic relation between parties and voters is not itself the independent variable but is part of the causal mechanism that links electoral rules and targeted versus universalistic policies.
The crucial insight that we can gain from the literature on party–electorate linkages is that – in contrast to the programmatic bias in partisan theory – the electoral strategies of parties may as well be particularistic. Parties sometimes adopt narrowly defined policies in order to attract specific groups of voters. If particularism prevails, it can keep even parties, which are commonly held to be programmatic, from promoting universal policy solutions. Conditions favouring particularism are, among others, low autonomy of state bureaucracy and specific electoral rules. Under these conditions it is unwarranted to work with hypotheses that are based on broad ideological families, assuming programmatic coherence. Rather, we would have to expect fragmented policies that are targeted towards whichever electoral group is central for gaining or retaining power.
Discussion: Between Updating and Revising
The main general lesson, which can be drawn from the three strands of research we have discussed above is that the traditional partisan politics approach applies to a more limited set of empirical cases than previously thought. For all cases that do not meet the basic assumptions of the traditional approach, the theory of partisan politics needs to be adapted, either by updating some of the underlying assumptions, or by re-conceptualizing the approach more fundamentally. In this last part of our discussion, therefore, we start by proposing scope conditions for traditional partisan theory. Scope conditions specify the empirical conditions under which we expect a theory to hold. Then we go on to discuss the implications of the three strands of research, presented above, for further research. Hence, we address the question: if one or several of the scope conditions do not apply, how should we analyse the relation between party politics and the welfare state?
Each of the three strands of research, outlined in the previous section, can be read in terms of scope conditions for traditional partisan politics theory. The claim that we need to ‘update’ the electoral landscape which drives party politics (the first new approach discussed in this article) is a more moderate critique of the traditional approach because it does not question the representative link between the interests of the electorate and the policy preferences of political parties. Still, it has rather clear implications for the applicability of the traditional approach: the more the social structure and electoral landscape have shifted from an industrial to a post-industrial pattern, the less the traditional assumptions on the nature of represented interests hold. The second strand of research, highlighting contextual influences, contains a much more fundamental critique of the traditional approach: it holds that it is always necessary to take the structure of the party system and the institutional context into account, not just in specific historical circumstances. However, even here the issue can be rephrased in terms of scope conditions. Bipolar party systems that are mainly structured by the socio-economic cleavage come closest to the uni-dimensional structure of party competition the traditional partisan politics approach has in mind. Finally, a theorization of party–voter linkages also implies an important scope condition: where parties use policies as concrete objects in exchange for particular votes, approaches relying on ideological party families fall short.
In sum, the combined scope conditions are the following: an industrial social and electoral structure, a bipolar party system, and programmatic party–voter linkages. As an illustrative example, one could mention Britain during the first three post-war decades as meeting the scope conditions of traditional partisan theory. With Labour and the Conservatives its party system offered clear choices on the left and the right that were primarily distinguished by their positions on social and economic issues. The way they competed was prevailingly programmatic. And although post-industrialization set in early in Britain, the social structure was still largely defined in terms of industrial classes. Otherwise, examples of ‘pure’ two-party, programmatic, industrially structured partisan politics are hard to find. The United States, for instance, meets the two-party system condition. But even when the United States had an industrial social structure, it did not translate into class politics to the extent that this was the case in Europe and particularistic party–electorate linkages have always been strong. On the other side of the spectrum, an example that clearly does not meet the scope conditions of traditional partisan theory would be the First Republic in Italy (1948–93). It had a tripolar party system (centre, far left and far right) with a divided left and a strong religious cleavage, and particularistic competition was pervasive.
The most general and far-reaching obstacle to the adequacy of a simple version of the partisan politics theory today is arguably post-industrialization. In all advanced capitalist countries, tertiary employment today clearly outnumbers industrial employment, which violates the scope condition of an industrial social structure. To the extent that new parties (such as Green and right-wing populist parties) arise in order to meet new interests, post-industrialization also transforms the party system and party competition in ways that do not conform to traditional theory, by shifting politics from a uni-dimensional to a multi-dimensional conflict space.Footnote 62 Moreover, parties have adapted their organizations and have become more flexible in their electoral strategies.Footnote 63 In the light of these pervasive challenges, we would like to use the remainder of this article to discuss concretely how we might adapt our theory and research if one or several of the scope conditions for the traditional partisan politics approach are not met. Every theory of partisan welfare politics implies hypotheses on the preferred policies of particular parties. All three strands of the new school of research we discussed, identify factors that drive parties’ policy preferences in a way or in directions that differ from what we would expect following the traditional approach. Hence, taking them seriously means that we have to adapt our ideas and hypotheses of what parties want.
The first strand of research discussed in this article argues that electoral constituencies of political parties have changed in the wake of post-industrialization. Since it does not question the bottom-up chain of interest representation by political parties, this claim is to a large extent compatible with the underlying assumptions of the traditional approach, but it requires that we update its very foundation: the traditional partisan politics approach assumes that parties represent a class structure that dates back to the age of industrialization and post-war growth. Post-industrialization has brought changes in the class structure and thereby changes of voter preferences with regard to parties and policies.Footnote 64 Hence, while parties’ policy preferences might still be driven by their voter constituencies, we need to adapt our expectations of what parties want to what their current constituencies want. Empirically, this ‘updating’ can take place at the micro or the macro level.
At the micro level, studies integrating individual-level analysis of policy and party preferences in the study of party effects can be seen as a step in the right direction, since they set out actually to observe voters’ policy preferences empirically, rather than assuming them theoretically. Contributions such as those by Cusack, Iversen and Rehm, Rehm, or Rueda all study the preferences of post-industrial risk groups for particular policies, before linking them to what parties and governments actually do.Footnote 65 However, taking this approach seriously implies that we need to observe both policy and party preferences of different post-industrial electoral potentials empirically, since both may have altered.Footnote 66 We can then hypothesize what parties with a particular profile of voters and voter preferences are likely to advocate in terms of welfare policy. Including this in our studies can also yield new insights on how parties prioritize the different demands of their heterogeneous (class-) constituencies,Footnote 67 or how they use this heterogeneity of constituencies when engaging in political exchange.Footnote 68
It is true that large-N macro comparisons may encounter difficulties controlling for such changing electorates, since the extent and structure of electoral reconfigurations varies across countries and cannot easily be generalized and integrated in the standard variables. However, the electoral shifts this literature is concerned with are not random: based on (secondary) studies of electoral dynamics, it may be possible to replace the old variables (such as ‘left-wing party power’ and similar variables) directly with more meaningful variables. For instance, we know by now that we need to distinguish between ‘old left’ and ‘new left’ parties (i.e. between those with a more ‘workerist’ versus those with a more middle-class profile). Hence a simple ‘left power’ variable does not make much sense anymore. We also need to distinguish between radical and moderate right-wing parties’ impact because they mobilize very different segments of the society. Basically, the shifting electoral foundations of post-industrial party systems require us to re-classify parties into new party families, such as workerist left, middle-class left, green parties, etc. Parties may still have the same names as thirty or forty years ago, but that may not tell us much about their voter profile anymore.Footnote 69
Theoretical approaches that emphasize the importance of party competition require a more radical rethinking of the functioning of party politics: while the traditional approach sees party politics as driven by social constituencies and their interests, party competition approaches focus on the relative autonomy of parties as organizations and the independent logic of the interaction between parties. A strong version of the competition argument assumes parties to be mainly focused on ‘contested’ constituencies. Freed from socially structured voter–party links, parties strive to take the ideal policy position that fits the mood of the public, while taking account of the positions of other parties. The objective is to maximize votes and/or office and the way to get there depends always on the actions of the other parties. This differs starkly from the socially grounded logic of the traditional approach. Instead, parties’ policy preferences become strategic, opportunistic and do not remain static as in old school theory. Hence, in this radical form, the highly volatile policy preferences of parties depend on electoral strategies and the specific configurations of party systems. Therefore, this argument is hard to integrate into large-N research designs.
A somewhat softer version of the party competition approach builds on the traditional assumption regarding the socially structured voter–party link, but then adds a competitive element. Parties may have distinct social constituencies, but they opt for only one of them when the others are electorally less important.Footnote 70 The party's strategy changes, but only by narrowing its policy focus down to a subset of its entire constituency. In such a ‘softer’ version of the competition argument, competition may also enter the analysis as a constraining factor: the orientation of a party can be determined by its social constituency (in line with the traditional or updated assumption), but the extent to which this goal is actively pursued, i.e. the policy choice, is limited by party competition.Footnote 71 Thus, if social democrats have to compete with a more radical party to their left, we expect them to advocate more left-wing policies than if they compete only with a centrist or right-wing party. Empirically, such conditional effects can be integrated in regression analysis by means of interaction effects in order to test whether, for example, the impact of social democracy is different when it competes with a more radical left-wing rival party. In that sense, one would be able to capture both the direct and the conditional effect, combining the party competition perspective with the electoral constituency perspective.
Traditional partisan theory assumes that class is the most relevant cleavage in society, if not the only one. Cleavage-based arguments, however, stress that additional cleavages shape parties’ identities and policy priorities. Consequently, two features of party systems should be considered: first, and similar to the ‘competition-as-constraint’ argument, we need to take into account the ideological polarization that goes along with a divided left (such as in France, Italy and Spain) or a divided right (for example, Austria, France and Norway). Secondly, we need to take into account the importance of cultural cleavages that cut across the capital–labour cleavage (as in Germany, the Netherlands and Switzerland) and introduce non-economic, cultural motivations into the politico-economic sphere. Both versions of cleavage-based considerations change the expected orientations and strategies of parties: through polarization and fractionalization of the party system, parties arguably become representatives of more specific groups, so that we would expect their policy preferences to be narrower. This can be integrated in empirical research by using common indicators of party system polarization and fragmentation. However, taking the relevance of non-economic, cultural cleavages in the party system seriously has more consequential implications, since it means that parties’ positions on social and economic policies are not only shaped by their voters’ material interests, but also by cultural values regarding distributive issues. Christian democratic parties, for instance, might advocate poor relief or social insurance for religious reasons, rather than in response to the economic interests of their voters. Similarly, parties of the radical right may defend the welfare state for reasons of national identity. Such factors challenge the traditional partisan politics approach by requiring a careful contextualization of the national policy debates and political processes.
Finally, the literature on programmatic v. particularistic party–voter linkages claims that the traditional partisan politics theory only applies when parties mobilize their voters with programmatic appeals. Politics based fully on particularistic voter–party links cannot be analysed with the traditional approach, because the orientations and strategies of political parties change to such an extent that party labels such as ‘left’ and ‘right’ make no sense. However, as in the case of the party competition argument, the two approaches can be made compatible. One way is to redefine the dependent variable. Lynch, for example, expects particularistic party competition to lead to fragmented social policies and to sustain occupational principles of welfare coverage.Footnote 72 Therefore, the dependent variable does not consist of the usual social spending or benefit generosity measures.Footnote 73 Similarly, we can control for the degree of particularism by using a gradual measure, for example with expert survey data that provides information on clientelistic practices of different parties in particular countries.Footnote 74
A more indirect way of adapting the traditional approach to the insights of studies on programmatic and clientelistic competition is to trace the origins of different kinds of linkages back to institutional causes. Estévez-Abe's argument that multi-member electoral districts and the personal vote in Japan led parties to focus on narrow interest groups is a case in point.Footnote 75 It implies that vote fragmentation leads to more particularistic policies, and thus uses institutional variables to control for the type of linkages that prevail in a system. Similarly, we may account for the degree of state autonomy, which determines to what extent parties and politicians can influence the assignment of welfare benefits for particularistic electoral motives.Footnote 76 Overall, taking different linkages into account (either directly or by focusing on their causes) should make us aware that these are different strategies from those which the partisan approach assumes. And these strategies imply different (more fragmented) orientations and policy choices.
Table 1 provides a brief overview of the arguments and implications developed in this article. Our discussion of possible strategies to deal empirically with the theoretical challenges raised in much of the recent literature is not and cannot be exhaustive, since research in this area is just starting to grow. However, it is intended to provide a systematization of the problems we are dealing with and of possible strategies to solve them.
Table 1 Different Theoretical Approaches to Partisan Welfare Politics and Their Implications
The new theoretical approaches, in particular the second (party competition) and third (party–electorate linkages) are hardly compatible with traditional partisan politics theory if we focus on their more radical implications. However, all of them can be adapted to or even combined with the traditional approach and doing so leads to important insights into the complex interplay of party politics and welfare state development, as shown by the numerous works cited in this review. Even more importantly, when these approaches are combined with the traditional partisan politics approach, intriguing research questions emerge: do parties act as representatives of social constituencies or is party politics best described as a semi-autonomous and strategic elite behaviour? How much do cultural values (religion or post-materialism) leave their footprint on social policy making? When do politicians opt for different linkage strategies? And how do parties integrate or exclude preferences of their heterogeneous electorate?
Conclusion: A ‘New School’ of Partisan Politics Theory with ‘Old School’ Roots
Partisan politics has been a powerful approach in comparative welfare state research and has contributed significantly to our understanding of the expansion of welfare states and the emergence of different welfare regimes. However, the rising tide of studies indicating different ways in which party politics influences social policy development shows that traditional partisan theory needs to be reconsidered. In this article, we have discussed three strands of theorizing and research, which point to different ‘deviations’ from the assumptions of traditional partisan theory. A first strand of research argues that post-industrial socio-structural change has altered the electoral constituencies of parties, so expecting them still to be advocating the same policies as in the industrial age is unwarranted. A second strand maintains that what parties want depends on the context of party competition and (electoral) institutions with which they are confronted. Hence, we cannot theorize about the impact of isolated parties unless we account for the environment in which they operate. Finally, a third strand of research argues that the underlying assumption of programmatic voter–party linkages in partisan politics theory does not always hold. Even in developed democracies, some parties mobilize voters with clientelistic or particularistic practices, which alter our expectations of the policies they advocate.
We presented two kinds of consequences that follow from these three ‘new schools’ of partisan politics research. First, they suggest that the traditional approach should be applied to a more limited set of empirical cases only, i.e. to countries with class-related electoral constituencies that correspond to the social structure of the industrial era, bipolar party systems, and a programmatic mode of party competition. Secondly, we discussed how the theory of party politics and the welfare state outside of these scope conditions should be adapted. As we have pointed out, it may suffice to update partisan theory to take into account the changes in the composition of electoral constituencies, or to acknowledge constraints on party policies stemming from particular patterns of party competition. However, it may be necessary to re-conceptualize our model of party politics more fundamentally, in order to account for strategic motivations of policy decisions that are conditioned by the configuration of party competition and that are sometimes guided by particularistic electoral strategies.
One of the most important questions for further research is to investigate to what extent parties’ policy decisions are still based on the representation of identifiable constituencies and social interests, and to what extent policy choices are motivated by the need to compete with rival parties over specific groups of voters. The answer to this question requires that we integrate the study of voters, parties and policies. Welfare state research can benefit greatly from adopting insights of electoral studies and party research. In the debates that we have outlined, we already see that part of the welfare studies literature has moved closer to these other fields of comparative politics and we believe this to be both necessary and beneficial for our understanding of welfare state development.
