To link to this article:

Transcription

1 This article was downloaded by: [University Of Pittsburgh] On: 30 May 2013, At: 11:25 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK West European Politics Publication details, including instructions for authors and subscription information: In Search of Successful Reform: The Politics of Opposition and Consensus in OECD Parliamentary Democracies Despina Alexiadou Published online: 22 May To cite this article: Despina Alexiadou (2013): In Search of Successful Reform: The Politics of Opposition and Consensus in OECD Parliamentary Democracies, West European Politics, DOI: / To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sublicensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

2 West European Politics, In Search of Successful Reform: The Politics of Opposition and Consensus in OECD Parliamentary Democracies DESPINA ALEXIADOU Why are some governments able to undertake controversial policy reforms and others are not? Conventional wisdom argues that single-party majority governments are best able to implement reforms because there are fewer veto actors within the government that can block the reforms. However, these accounts fail to consider the veto power of societal actors and particularly of trade unions, which can stall reform even in the presence of a unified executive. This paper argues that controversial reforms require broad societal and, consequently, political consensus, which are easier to achieve under minority governments or governments of broad coalitions. Evidence from 22 OECD parliamentary democracies over 35 years shows that minority and large coalition governments have been more successful in reducing social security contributions and pensions than narrower majority governments. This is especially true in countries where trade unions are militant and often resort to industrial action. On 6 May 2012 a large majority of Greek voters elected ideologically radical political parties, which opposed the IMF (International Monetary Fund)/EUbacked austerity measures causing tremor in the international markets. On the same day, French voters elected the Social Democratic candidate Francois Hollande, who ran on a pro-growth, anti-austerity manifesto, as their president. The Greek and French elections are illustrations of a much broader phenomenon of public discontent against cuts in pensions and social security reforms across Europe, from Romania and Poland to Italy and the UK. European governments have been trying for decades to promote pro-market policies to boost employment and increase economic growth but have found strong political resistance. Pension and social security contribution reforms have been particularly controversial. Governments have been targeting non-wage labour costs as a way to make labour more competitive internationally and to boost employment and economic growth. Cuts in pension replacement rates and increases in Correspondence Address: dalexiad@pitt.edu Ó 2013 Taylor & Francis

3 2 D. Alexiadou retirement age have also been a central policy objective to reduce social security contributions and public debt. However, these reforms are not easy. Reducing social security contributions is an unpopular reform, especially for trade unions. Lower social security contributions involve cuts in social welfare provisions or unpopular new taxes on income or on goods and services. Reducing pensions is even more unpopular as it affects almost everyone, and especially trade unions in continental European countries, the self-appointed defenders of the pension status-quo (Natali and Rhodes 2008: 25 27), as the recent pension reform in Poland illustrates. 1 According to some, the current fiscal crisis and the accompanying austerity measures are the outcome of the failure of consecutive governments in many European countries, such as Greece, Italy and France, to implement the necessary reforms in pensions and social security systems. At the same time governments in Denmark or the Netherlands have been able to resolve major tax and pension dilemmas successfully. Why are some governments more successful in implementing reforms than others? Although governments all over the world struggle with policy reform, existing theories give us some indication of when reform will be possible. According to conventional wisdom and veto actor theories, more ideologically diverse and polarised coalition governments should have a harder time implementing reforms than single-party majority governments, because policy conflict is more likely to occur in polarised coalitions than in single-party governments (Alesina and Drazen 1991; Basinger and Hallerberg 2004; Ganghof 2006; Tsebelis 2002). Others argue the contrary: that consensus governments, understood as governments that seek cooperation with opposition parties, can implement reforms more easily because they can build broad political and societal coalitions, which are necessary for controversial reforms (Katzenstein 1985; Lijphart 1984; Luebbert 1984). More recent theoretical approaches integrate veto actor and consensus approaches to stress the complex interaction of governments, opposition parties and social partners in the policy-making process. Explaining pension reform in continental Europe, these accounts emphasise the role of trade unions as important veto players, with policy- and office-seeking goals, whose consent is necessary, but not a given, for policy reform (Bonoli 2000; Natali and Rhodes 2008; Palier 2000; Schludi 2005). When trade unions and governments fail to reach an agreement, policy concessions (Bonoli 2000) and cross-party cooperation between the government and the opposition become necessary to build political support for reform (Schludi 2005). More broadly, a comprehensive study of 16 European countries shows that political competition is as critical in explaining pension reform as the number of veto players in a country s political system (Immergut et al. 2007). Building on this literature, this paper argues that the type of government, whether it is a near-majority or consensus government, 2 is critical in predicting policy change in the presence of militant trade unions. The success of a policy reform rests on a government s ability to co-opt opposition parties and mil-

4 In Search of Successful Reform 3 itant interest groups which oppose the reform proposal. As already illustrated by Schludi (2005: 53), when trade unions oppose a proposed reform, they have more legitimacy and bargaining power over the government if they are supported by opposition parties. When the opposition supports these interest groups, then the issue is politicised and the political stakes are higher for the government as it fears high future electoral costs. Under single-party majority governments and cohesive minimal winning coalitions, cross-party cooperation between the government and the opposition is less likely. Under these types of government, the opposition seeks to benefit electorally from the government s policy dispute with the unions and might find it profitable to align with the unions and against the government. This explains why, in many instances, single-party governments with solid majorities in the parliament have failed to move forward with unpopular reforms. In contrast, when opposition parties cooperate with the government to pass the bill typically under minority governments or when the government is a itself a broad coalition of parties, such as a grand coalition, then trade unions are politically isolated and, consequently, have less bargaining power over the government. Therefore, contested policy reforms should be easier to pass in more consensual political systems than in adversarial political systems, in the presence of societal resistance to the reforms. This argument is in line with Katzenstein (1985) and Luebbert (1984, 1986) who have argued that consensus governments, understood as governments that seek cooperation with opposition parties, can implement reforms more easily because they can build broad political and societal coalitions. However, the present theoretical account differentiates from and goes beyond existing consensus accounts, which fail to explain policy change when governments and social partners are unable to reach agreement on controversial reforms. Existing consensus accounts (Katzenstein 1985; Lijphart 1999) assume that societal consensus and political consensus require strong corporatist 3 norms and institutions. It is argued here that consensus governments can implement reforms not only because they are supported by corporatist arrangements but also because they form a political front against militant trade unions. Thus, the proposed theory makes a novel contribution by explicitly differentiating between corporatism and the type of government, which varies over time and within countries. In doing so, it explains first why there has been both policy failure and success in countries that lack corporatist institutions but have had consensus governments, as well as the occurrence of policy failure in corporatist countries under near-majority coalition governments. Furthermore, theproposedtheoryoffersaclearexplanationforwhyandwhensingle-party majority governments fail to implement reforms despite government unity. Havingasastartingpointthatgovernmentsfeartheelectoralconsequencesofunpopular reforms when trade unions find allies amongst opposition parties (Schludi 2005), thepaperarguesthatthetypeofgovernmentiscriticalinovercomingsocietalandpoliticalresistanceagainstunpopularreforms. Whentrade unionsare strongandfindallies amongst opposition parties, as has been the case in Greece, single-party majority governments are politically weakened and often have to postpone or retire their reform

5 4 D. Alexiadou programme.itisinthesepoliticalsystemswhichtendtoproducesingle-partymajority governmentsthatcross-partycooperationistheleastpossibleandinwhichpoliticsare best described as adversarial, even when the political parties are ideologically close to each other. On the other hand, single-party majority governments can implement unpopular reforms successfully and with minimal societal resistance when trade unionsareweakandlackthecapacitytomobilisethepublicagainstthereforms. Thisis bestillustratedbythesocialwelfarereformsimplementedbythatcher sgovernments inthe1980sintheuk(bonoli 2000). Inother words,theinteraction between thetype of government and the type of the trade union movement determines the success or failureofunpopularreforms. Pension and social security contribution cuts are suitable policy reforms to test the theoretical argument. Reducing social insurance payments and pension generosity are among the most politically contested policy changes any government could undertake (Natali and Rhodes 2008). One of the most unpopular tax reforms, particularly in contributory systems, is the reduction in social security contributions, which finance social insurance programmes such as unemployment, sickness insurance and pensions. A reform to decrease social security contributions is politically contested and unpopular because trade unions and voters see such reforms as cuts to popular welfare state entitlements (Pieson 2001; Schludi 2005). Cuts in state pensions are even more unpopular as they are more visible and directly affect voters and trade unions (Natali and Rhodes 2008; Schludi 2005). The next section presents the theoretical argument in more detail and further explains why and when the type of government matters for cross-party cooperation and the implementation of unpopular reforms. A hypothesis is derived from the theoretical argument, which is tested on time-series data that span in 22 parliamentary democracies. Type of Government and Consensus Building The importance of the party system for policy reforms is well established in the literature. Yet the ways in which the party system affects reforms is contested. Veto actor theories (Tsebelis 2002) stress the inability of systems with multiple and ideologically distant veto players to undertake reforms. Recent contributions on changes in capital and income taxation confirm that countries with multiple veto players experienced smaller changes in their tax system than countries with fewer veto players (Basinger and Hallerberg 2004; Ganghof 2006). Specifically, Ganghof (2006) shows, through detailed case studies, that the unified executives in Australia and New Zealand were able to change income tax faster than Scandinavian minority governments or majority coalitions in Germany. However, these accounts cannot explain the inability of the unified executives found in countries like Greece or France to implement tax reforms. 4 Consensus approaches, on the other hand, reject the argument that strong and ideologically compact majorities are sufficient for reform. Instead they stress the advantages of cooperation between government and opposition, which is

6 In Search of Successful Reform 5 easier to achieve under minority governments and corporatist practices (Katzenstein 1985; Luebbert 1984). Katzenstein (1985: 103) argues that small open market countries benefited from minority party systems: A system of minority governments is well suited to the party system of the small European states because, far from penalizing opposition parties, it offers these parties significant influence over policy. Luebbert (1984: 231) takes a similar position, although he stresses the critical role of norms of interparty cooperation, which derives from corporatization of interests and an un-polarized party system. Both authors stress the importance of corporatist methods (consultation with trade unions) and of cross-party coalitions in economic restructuring. In fact, both authors regard corporatism as a necessary condition for successful reforms, as corporatism contributes to the existence of a consensual political system by structuring interests in a manner that encourages compromise and by lessening the representational burden on the parties and party system (Luebbert 1984: 236). Consensus theories take into account the critical role of trade unions in policy reform more seriously than veto actor approaches, which tend to focus on the role and preferences of political actors. Yet consensus approaches go to the other extreme when they argue that corporatism, in the form of coordinated wage bargaining between governments, unions and employers, is a necessary condition for consensus building. Recent accounts of the politics of pension reform in Europe show that not only has corporatism weakened but trade unions have continued to act as major veto players against economic reform (Bonoli 2000; Natali and Rhodes 2008; Palier 2000; Schludi 2005). Indeed, in many instances governments have failed to move forward with policy changes due to resistance from trade unions in corporatist societies such as in Sweden or Denmark. In fact, cooperation with opposition parties has been critical in convincing the unions of the necessity of reforms and thus minimising their resistance (Bonoli 2000: 4; Schludi 2005: 53). However, consensus approaches are right to stress the importance of political institutions in reaching cross-party cooperation. Most of the literature on pension reform concentrates on the changing preferences and strategic considerations of governments and unions to explain policy change (Hausermann 2010; Natali and Rhodes 2008). The theoretical argument that is advanced in this paper brings together the literature on consensus governments, which stresses the importance of formal political institutions, and the political economy literature on pension reform, which has primarily focused on the changing interaction between unions and governments. This paper argues that consensus governments, defined as minority or large majority coalition governments, are better able to undertake reforms in the presence of militant trade unions, with or without corporatist arrangements, because they can work together with the opposition. Consensus building between governments and the opposition can be an important condition for exerting pressure on particular interests (e.g. trade unions) who oppose reforms that are beneficial for the society as a whole (Green-Pedersen 2002;

7 6 D. Alexiadou Schludi 2005). In numerous instances governments have failed to pass unpopular reforms due to the resistance of trade unions which can mobilise public opinion against the reforms. In adversarial party systems, opposition parties are more likely to side with the unions and distance themselves from the government s proposals for their own political gain. Even if their vote in parliament does not matter for passing the proposed bill, their stance towards the government s proposals has an impact on the public s attitude towards the proposed legislation. If the opposition supports, or even just condones the unions resistance for its own political gain, then the government is more likely to water down the bill and even to stall the process. If, on the other hand, the opposition supports the proposed reforms, then the unions are forced to cooperate, since they will be unable to find political allies against the government and thus will have a harder time to convince the public about their cause. The Greek political experience in the early 2000s illustrates the point. In Greece, pension, labour market and social security reforms have been incremental and socially contentious. From 1993 to 2004, Greek Socialist governments undertook only marginal labour market and tax reforms due to the high political cost of the proposed changes. The Giannitsis law which, among other things, sought to reduce employers contributions and to increase the retirement age was fiercely opposed by public and private sector trade unions and eventually by ministers of the governing party, who were close to the union movement (Soumeli 2000; Papadimitriou 2003; Triantafillou 2007; Nɛρζής 2002). In the meantime, the centre-right (and main party in opposition) celebrated its higher ratings in the polls as its appeal to the lower social classes increased (Nαμπακόπoυλoς 2002). Ultimately, the prime minister reshuffled his cabinet, retracted the original reforms and replaced them with milder ones, in preparation for the upcoming local elections (Παπαδιoχoς 2002). In other words, the adversarial character of the political party system in Greece and the militant trade unions created the perfect mix for reform blockage, despite consecutive strong single-party majority governments. If Greece is the prototype example of lack of reform and political and social conflict, Denmark is the prototype of political and social consensus. The Danish minority governments achieved unpopular tax and pension reforms through their cross-party cooperation and broad political consensus during the 1980s and 1990s. The Danes had their difficult times during the 1970s when the Social Democrats were unable to implement economic reform, partly due to tensions in their relationship with the unions. However, cross-party cooperation in the 1980s and 1990s allowed them to pass an array of tax and pension reforms. When the Liberals came to power in 1982, the Social Democrats offered their legislative support twice to reforms they favoured; on private pensions in 1982 and on the tax reform in 1985 (Green-Pedersen 2002: 15). 5 Similarly, when the minority Social Democratic government came to power in the 1990s, major tax reforms and a restructuring of the welfare state were

8 In Search of Successful Reform 7 achieved with the help of the right-wing opposition, but also with the support of the trade unions (Ganghof 2006: 19). One could ask why the opposition would want to be involved in controversial legislation since such a reform could lead to loss of office for the incumbent government. While this is true for party systems that produce single-party majority governments, in party systems that tend to produce minority governments, opposition parties are often incentivised to cooperate with the government for two reasons. Firstly, the opposition has the opportunity to co-determine policy and, secondly, it receives the credit and shares the cost of the reform. In systems with minority governments, opposition parties will be faced with the same problems once in government and will have to seek the new opposition s support. Thus, they have an interest in helping the government pass controversial legislation as long as that legislation does not violate policy interests that are vital to their identity. 6 In fact, cooperating with the government and having policy input is one of the reasons political parties choose to stay out of office and support minority governments (Strom 1984). Cooperation between governments and oppositions on unpopular reforms is especially important as it minimises the political costs of the new legislation. The government and the opposition split the political cost of the unpopular reform, reducing the political cost to each group (Natali and Rhodes 2008). This way, the opposition party not only avoids taking the full blame for tax reform, but can also claim credit for acting responsibly and helping the government revive the economy (ibid.). Minority governments are not the only types of consensus government. Governments of broad and diverse party coalitions can also pass controversial legislation by reducing the chances of interest groups finding political allies in parliament. The Dutch case illustrates the point well. Despite the relative ideological cohesion between the Christian Democrats and the Liberals in the Lubbers I government ( ), the coalition avoided reforms in taxation and pensions. The largest pension reform was undertaken under the grand coalition of the Christian Democrats and the Social Democrats in the early 1990s, and the first major reform to social security contributions was implemented by the Purple coalition between the Social Democrats, the Liberals and D66 (Democrats66) (Timmermans 2003: 91 98, ). To sum up, when trade unions are militant, policy change especially when it involves unpopular reforms opposed by the unions is harder and slower to achieve in adversarial types of government, defined here as narrow majority governments. In contrast, militant trade unions have less veto power when they operate under consensus governments, which I define as minority and large coalition governments. Social Security Contributions and Pensions: The Battleground for Governments and Trade Unions Which reforms are particularly divisive? There is a long and substantial literature on the politics of pension and tax reform (Basinger and Hallerberg 2004;

9 8 D. Alexiadou Ganghof 2006; Hausermann 2010; Kato 2003; Natali and Rhodes 2008; Palier 2000; Schludi 2005). Trade unions have particularly strong views and positions regarding reform of pensions and social security contributions, especially in systems where social insurance is linked to earnings and contributions over time (Natali and Rhodes 2008; Palier 2000; Schludi 2005). Reforming social insurance contributions is particularly controversial as trade unions perceive cuts in employees and employers contributions as cuts in social benefits (Schludi 2005). Reforming pensions is even more controversial since cuts in pensions replacement rates are perceived as an assault on workers insurance and entitlements for which they have paid through their salaries (Palier 2000: 121). Thus, trade unions have generally resisted cuts in both social security contributions and pensions. In the majority of European countries, their position has been one that favours the status quo regarding pensions, or that favours increased social security contributions when the issue of fiscal sustainability of the social security funds arises (Hausermann 2010; Natali and Rhodes 2008; Palier 2000; Papadimitriou 2003; Schludi 2005, 2008; Visser and Hemerijck 1997). But we do find governments, of both the left and the right, which have tried to reform social security contributions and pensions during the last two decades, regardless of the attendant difficulties. Social Democrats and Christian Democrats both broadly supported higher pensions until the 1980s, when questions about the financial sustainability of social security systems and their ability to address the social needs of new groups of citizens appeared on governments policy agendas (Esping-Andersen 1990; Hausermann 2010; Van Kersbergen 1995). Consequently, in the 1990s and 2000s even Social Democrats chose to break ranks with the unions and push forward their modernising reform agenda on pensions, which often led to severe clashes between governments and unions (Hausermann 2010; Palier 2000; Visser and Hemerijck 1997). Over time, partisan differences have decreased with respect to social security contributions as well. Historically, employers organisations and Liberal parties have been in favour of lower social security and, particularly, employers contributions, while trade unions and Social Democratic parties have favoured higher contributions as the best way to provide sustainable funding for the welfare system. In the late 1970s the conservative coalition in Sweden eliminated the payroll tax (Mares 2006: 107). While both trade unions and employers agreed that the tax structure had to be reformed, they disagreed about the magnitude and the direction of the changes. Employers called for an across-the-board reduction in taxes while the unions demanded tax cuts on low-income earners. Moreover, the unions also called for the reintroduction of the payroll tax (ibid.). Similarly, the proposal for reducing employers costs in Denmark in 1998 came from the employers organisations and was brought to the parliament and the government by the opposition Liberal Democratic and the Conservative parties (EIRO, 28 December 1998). However, the political conflict over reforming social security contributions cannot be understood along partisan lines. Although traditionally parties of the right have favoured lower contributions, more recently Social Democrats have

10 In Search of Successful Reform 9 also tried to reduce contributions in order to reduce labour costs and boost employment. During the first purple coalition in the Netherlands, between the Liberals, Social Democrats and D66, the most urgent policy goal was to increase employment. The Liberals demanded a smaller welfare state and lower taxes to boost job creation. On the other hand, the Social Democrats wanted to keep the welfare state intact while boosting low-paid jobs. The Social Democrats proposed to cut the costs of labour by reducing employers contributions (Timmermans 2003: ). Diverse experiences among European countries and reports by the OECD became a source for learning and emulation (Batturo and Gray 2007) with respect to reforming social insurance contributions. Increased globalisation and technological change pose challenges to the high tax-revenue OECD economies, which have to compete with the low wage, low tax emerging economies. Most of the European countries that fund social insurance through employers and employees contributions (accounting for 74 per cent of social expenditures in Germany and 82 per cent in France in the mid-1990s) failed to create low-skilled/low-pay jobs, which partly explains the high unemployment rates in these countries (Blanchard 2006: 45; Crepon and Desplatz 2001; Kemmerling 2005; OECD 1998; Scharpf 2001). Generally, high social insurance contributions increase non-wage labour costs and can have a negative impact on employment creation (Blanchard 2006: 45; Scharpf 2001). Why do trade unions not want the reduction of non-wage labour costs to boost employment? After all, employment creation is one of the major goals of trade unions for their members. While employment creation is an important policy objective for trade unions, safeguarding the system of social insurance is also critical, if not more important. Most unionised members are employed and in some countries many are pensioners. 7 Cutting social security contributions translates into lower social insurance benefits, and especially lower pensions. Alternatively, shifting the burden of taxation from social security contributions to general income taxation breaks the link between social security contributions and social insurance benefits, by right. In other words, a shift from contributory to general taxation weakens employees rights to receive benefits and could thus endanger future funding of the social security system. This is why in most European countries with Bismarckian welfare states, trade unions support the preservation of the link between social security contributions and contributory benefits while they have accepted the funding of non-contributory social security benefits through general taxation (Natali and Rhodes 2008; Schludi 2008). Moreover, trade unions should particularly favour social security contributions to fund social insurance benefits over general taxation in more adversarial party systems. Tax revenues are broken down into income, capital and property tax, VAT and social contributions. While social insurance can be funded by many different taxes, trade unions are more likely to resist forcefully a reform that shifts pensions funding from social contributions to general taxation the less they trust future governments.

11 10 D. Alexiadou To conclude, trade unions have a strong preference for preserving high social security contributions and pensions. In contrast, governments prefer cuts in both contributions and pensions for fiscal stabilisation and employment growth. Agreement between unions and governments is more likely under consensus governments, while it is almost impossible in adversarial political systems. This is because under more adversarial types of government (understood as single-party near-majority governments or narrow minimal winning coalitions) opposition parties prefer to exploit for their own political gain the political costs to governments of pushing forward with unpopular reforms. When trade unions are militant they have higher chances of stalling a reform if opposition parties support them. Thus, H1: When trade unions are militant, single-party majority or narrow majority coalition governments are less able to cut social insurance contributions and pensions than are consensus governments. Data, Independent and Dependent Variables According to Hypothesis 1, when trade unions are militant they will resist cuts in social security contributions and pensions, especially under adversarial types of government. In contrast, they will be more willing to accept cuts in contributions and pensions in more consensual types of government. The operationalisation of adversarial and consensual governments is critical for testing the argument. Consensus government is a binary variable that is coded as 1 when the government is consensual and 0 when it is not. Consensus is defined as single- or multiparty governments with less than 49.5 per cent of parliamentary seats or multiparty governments with more than 60 per cent of the seats in the parliament. This variable captures the cooperation of different parties in the policy-making process. This operationalisation is preferred over using a categorical variable that codes types of governments for two reasons. The first reason is that not all minority governments are similar. In order to code minority governments as consensual one has to make sure that the minority governments require genuine cooperation with the opposition, as is often the case, for example, in Scandinavian minority governments. By doing this, we ensure that the variable consensus governments does not include near-majority minority governments that require only the support of a couple of independent legislators, as is often the case in Ireland or in Spain. Minority governments that control less than 49.5 per cent of the parliamentary seats cannot behave like majority governments but have to reach consensus with at least one party of the opposition. Consensus governments are also defined as multiparty governments that have more than 60 per cent of the parliamentary seats. Thus, consensus government does not include single-party majority governments with large parliamentary majorities (accounting for only 1 per cent of all single-party majority governments in the sample) or narrow majority coalition governments. Table 1 shows the frequency of consensus government for the different types of government. Ninety per cent of single-party minority governments are consensus governments. Under half of the minimal winning coalitions are consensus governments,

12 In Search of Successful Reform 11 TABLE 1 THE BREAKDOWN OF CONSENSUS GOVERNMENTS BY TYPES OF GOVERNMENT Government type Consensus government Single-party majority 0.01 Minimal winning coalition 0.44 Oversized coalition 0.71 Single-party minority 0.90 Coalition minority 0.68 as are 70 per cent of the oversized coalitions. If consensus government was coded as a non-single-party majority government, we would be losing a lot of variation in the breadth of the different types of majority governments that have formed over time. To recapitulate, the objective here is to capture how many political actors are forced to be included in the decision-making process. Thus, the larger the coalition or the smaller the minority government, the more political parties should be included in decision-making. Strikes are the other important explanatory variable that captures trade union militancy. By using strikes instead of union density we capture the noise trade unions make and their mobilisation instead of their numerical power. This is further supported by the fact that the correlation coefficient between union density and strikes is zero. Strikes are measured as working days lost due to strike action per 1,000 workers (Armingeon et al. 2011). The dependent variables are social security contributions as a percentage of total tax revenue (OECD revenue statistics) and pension generosity (Scruggs and Allan 2006). 8 Dividing contributions by total tax revenue is the most direct way of testing how much government revenue relies on social security contributions, and consequently captures the trade-off between social insurance contributions and other types of taxes. This measure also allows us to compare non-wage labour costs across countries while at the same time being insensitive to changes in GDP. The second dependent variable is pension generosity as constructed by Scruggs and Allan (2006). The index contains the following six indicators: the minimum net replacement rate for singles, the minimum net replacement rate for couples, the standard net replacement rate for singles, the standard net replacement rate for couples, years of contribution required to receive a standard pension (scored inversely), and the individual s share of pension financing. 9 Each of these six characteristics is then given a score of 0 4, so the variable theoretically varies from 0 to 24. Both dependent variables are regressed on change instead of levels for both theoretical and empirical reasons. Theoretically, the argument is about changes instead of levels. Empirically both indicators change slowly over time (a unit-root test fails to reject the null of non-stationarity for social security contributions). By regressing the dependent variable as change safeguards us against spurious results.

13 12 D. Alexiadou Finally, the control variables that are included in the model are the change and lag of GDP growth, unemployment rate, deficit, social transfers (only for the social contributions model) and the lag of left-wing cabinet. Both pension generosity and social security contributions are expected to be influenced by the economic conditions in the country. An increase in unemployment could lead to higher contributions if the state needs to raise money to fund social welfare; alternatively, higher unemployment means that fewer people work and thus there are fewer social security contributions. Moreover, high unemployment might put pressure on governments to lower social security contributions in an effort to bring down unemployment. Higher unemployment is expected to have a negative impact on pension generosity as it negatively affects the government s tax revenues. Similarly, a large deficit might force governments to raise social insurance contributions to increase the government s revenues in the short run; but it might also have a negative impact in the long run, as higher deficits might force governments to cut down on social benefits. Deficits should have a negative long-run effect on pensions as well. GPD growth is expected to be positively correlated with both contributions and pension generosity, since growth raises revenue and allows more spending. Finally, countries with higher social transfers as a percentage of GDP usually have higher social security contributions. 10 Left-wing cabinet is the percentage of left-wing cabinet posts over the total cabinet posts weighed by the number of days. The variable ranges from zero to 100 (Armingeon et al. 2011). The role of left-wing governments in social security contributions and in pensions is not clear, as already discussed. On the one hand, left-wing governments could be associated with lower social insurance contributions as they favour progressive/redistributive forms of taxation (Ganghof 2006: 48). But on the other hand, according to partisan theories (Korpi and Palme 2003), left-wing governments wish to safeguard the welfare state and, given pressures to reduce the marginal tax rate on labour and capital, they could be associated with higher contributions. With respect to pensions, there is no particular reason to believe that left-wing governments are more willing to protect pension generosity over other social benefits, especially when they distance themselves from trade unions, as shown by Hauserman (2010). The statistical model employed to test the argument is an interactive model as described below. Strikes, which capture union militancy, are interacted with consensus government. According to the theoretical argument, strikes should be positively correlated with contributions or pensions under near-majority governments, since these governments are not able to reduce contributions or pension. On the other hand, strikes should be negatively correlated with contributions or pensions under consensus governments. In other words, according to the theoretical argument, b 1 > 0 and b 3 <0.

14 In Search of Successful Reform 13 Change in social security contributions = lag of s.s. contributions + b 1 Strikes + b 2 Consensus government + b 3 Strikes consensus + Controls + e The model is estimated using a random effects estimator with clustered errors by country and a country fixed effects estimator to address concerns that TABLE 2 THE EFFECT OF UNION MILITANCY ON SOCIAL SECURITY CONTRIBUTIONS AND PENSION GENEROSITY UNDER NARROW-MAJORITY AND CONSENSUS GOVERNMENTS (1) (2) (3) (4) D. SS D. Pension contributions generosity D. SS contributions D. Pension generosity Lag ss contributions (0.006) (0.021) Lag pension generosity (0.012) (0.021) D. GDP growth (0.043) (0.033) (0.022) (0.020) Lag GDP growth (0.046) (0.043) (0.027) (0.027) D. unemployment (0.083) (0.076) (0.056) (0.047) Lag unemployment (0.009) (0.029) (0.010) (0.015) D. deficit (0.055) (0.034) (0.015) (0.020) Lag deficit (0.043) (0.034) (0.020) (0.020) Lag 2 deficit (0.044) (0.035) (0.015) (0.020) D. Social transfers (0.111) (0.071) Lag social transfers (0.014) (0.033) Lag left-wing cabinet (0.001) (0.002) (0.001) (0.001) Lag strikes (0.020) (0.027) (0.010) (0.018) Lag consensus (0.142) (0.145) (0.058) (0.085) Consensus strikes (0.034) (0.033) (0.026) (0.021) Constant (0.275) (0.581) (0.181) (0.281) Observations Number of countries Country effects No Yes No Yes R-squared Significant at 10%; significant at 5%; significant at 1%. Standard (FE model) and Robust standard errors (RE model) in parentheses.

15 14 D. Alexiadou unobserved country-specific characteristics cause omitted variable bias (Hsiao 2003). All the regressors are lagged by one year while two lags are included for deficits. Finally, the proposed theory is compared to consensus theories, which argue that unpopular reforms are politically more feasible under minority cabinets and corporatist practices. Thus, the model is compared to alternative empirical specifications that capture the consensus models more directly by interacting union militancy (strikes) with minority governments, coded as single-party and multiparty minority governments, and with wage bargaining coordination 11 (Visser 2009). Discussion Table 2 and Figures 1 and 2 provide strong evidence in support of the hypothesis derived by the theoretical argument that consensus governments are better able to implement unpopular cuts when faced with militant unions. Specifically, Figures 1 and 2 show that under consensus governments, changes in social security contributions and pensions are negative. In contrast, under near-majority governments changes in social security contributions and pension generosity are positive. However, the results are not identical for social security FIGURE 1 MARGINAL EFFECT OF UNION MILITANCY ON SOCIAL SECURITY CONTRIBUTIONS IN NARROW MAJORITY AND CONSENSUS GOVERNMENTS

16 In Search of Successful Reform 15 FIGURE 2 MARGINAL EFFECT OF UNION MILITANCY ON PENSION GENEROSITY IN NARROW MAJORITY AND CONSENSUS GOVERNMENTS contributions and for pensions. With respect to social security contributions, the conditional effect of consensus governments on contributions is negative and statistically significant at the 95 per cent level, while the conditional effect of narrow-majority governments is positive, as expected, but not statistically significant. The opposite is the case for pension generosity. The conditional effect of narrow-majority governments on pension generosity is positive and statistically significant at the 95 per cent level, but consensus government has no statistically significant impact on pension generosity change. Nonetheless, overall, the graphs clearly show that the type of government has an important conditional effect on changes in social security contributions and pension generosity when trade union militancy is high. The long-term conditional effects of consensus governments on contributions are substantial. Strikes are associated with a drop in social security contributions by more than 4 per cent of total tax revenue in the long run under consensus governments. This is estimated by dividing the marginal conditional effect of social security contributions by the coefficient of the lag of social security contributions. The effect on pension generosity is slightly different. Specifically, strikes have a large and positive effect of eight points on pension generosity under narrow majority governments. To summarise, a change in government from narrow majority to consensus decreases social security contributions by 4 per

17 16 D. Alexiadou cent of total tax revenue; and a change in government from consensus to narrow majority increases pension generosity by eight points in the 24 point index, which is equivalent to a one-quarter increase in pension generosity. The control variables are statistically significant and in the expected direction. Higher unemployment and deficits have a negative impact on contribu- TABLE 3 THE EFFECT OF UNION MILITANCY ON SOCIAL SECURITY CONTRIBUTIONS AND PENSION GENEROSITY UNDER MINORITY GOVERNMENTS AND CORPORATISM (1) (2) (3) (4) D. SS D. Pension contributions generosity D. SS contributions D. Pension generosity Lag ss contributions (0.005) (0.007) Lag pension generosity (0.013) (0.015) D. GDP growth (0.039) (0.054) (0.022) (0.024) Lag GDP growth (0.049) (0.062) (0.026) (0.030) D. unemployment (0.092) (0.084) (0.057) (0.059) Lag unemployment (0.015) (0.014) (0.013) (0.013) D. deficit (0.052) (0.054) (0.017) (0.017) Lag deficit (0.044) (0.049) (0.020) (0.023) Lag 2 deficit (0.044) (0.048) (0.014) (0.016) D. Social transfers (0.135) (0.136) Lag social transfers (0.014) (0.018) Lag left-wing cabinet (0.001) (0.001) (0.001) (0.001) Lag strikes (0.017) (0.059) (0.014) (0.065) Lag minority government (0.127) (0.091) Strikes minority Lag wage bargaining coordination (0.051) (0.017) (0.390) (0.476) Strikes coordination (0.145) (0.148) Constant (0.310) (0.406) (0.206) (0.269) Observations Number of countries R-squared Significant at 10%; significant at 5%; significant at 1%. Robust standard errors in parentheses; random effects estimator.

18 In Search of Successful Reform 17 tions and pension generosity. Also, higher social transfers are positively correlated with social security contributions, as expected although this is only significant in the fixed effects model. Finally, GDP growth has a negative effect on both contributions and pensions but only in fixed effects models. This finding goes against our expectations and it indicates that when economic growth is higher governments might have more tax revenues and thus are more able to reduce social security contributions. The same is true for pension generosity. These findings give support to the argument that unpopular reforms are politically more feasible during good times. Finally, left-wing governments have no effect on either contributions or pension generosity, as expected. Overall, it becomes clear that the explanatory interactive effect of the type of government and trade union militancy is as large as the effect of changes in public deficit, GDP growth and unemployment. These effects are substantively very large, especially given that the variable left-wing governments finds no statistical or substantive significance. How does this model fare compared to alternative explanations? Table 3 provides the empirical results of the alternative conditional hypothesis that unpopular changes are easier under minority governments and corporatism. As Table 3 and Figure 3 show, this hypothesis finds strong support with respect to FIGURE 3 MARGINAL EFFECT OF UNION MILITANCY ON PENSION GENEROSITY IN MAJORITY AND MINORITY GOVERNMENTS

19 18 D. Alexiadou pension generosity but not with respect to social security contributions. Indeed, the decrease in pension generosity is large and statistically significant for minority governments, while the effect for majority governments is identical to the effect found under narrow-majority governments in Figure 2. Furthermore, wage bargaining coordination also has the expected conditional effect on changes in pension generosity, but it fails to reach statistical significance at the 95 per cent level. 12 Moreover, neither minority governments nor wage bargaining coordination have an effect, either positive or negative, on social security contributions. Finally, the alternative models explain half of the variance in changes in social security contributions than the proposed model in Table 2 (14 per cent and 25 per cent respectively). Conclusion Politicians have, for decades, called voters to support strong and unified majorities in government. Similarly, researchers have argued that single-party majority governments are more able to push forward with reforms and are more accountable to voters, since they can identify those responsible for their happiness or suffering. Yet both history and recent events in Europe fail to provide ample support to this argument. On the contrary, we have seen the European Commission and the International Monetary Fund (IMF) call for broad consensus governments in Greece to implement the tough austerity measures outlined in the 2010 and 2012 Troika loans. In light of the tough fiscal and debt crisis in Europe, European governments from Greece to Italy and the Netherlands have been trying to build political and societal consensus through minority governments or broad, encompassing coalitions. This paper sheds light on the politics of policy reform and the mechanics of consensus building and thus explains why political institutions are critical in passing unpopular reforms. Unpopular reforms require the support of important societal actors, namely trade unions. This support is easier to secure by consensus governments, defined as minority or large majority coalition governments. The type of government can play a critical role in policy reform as it allows alliances that can either block or enhance cooperation. Under narrow majority governments the opposition secures greater electoral benefit from supporting militant trade unions that oppose reform. In contrast, under minority or large majority coalition governments, the opposition is more likely to be included in the decision-making process and thus will tend to align with the government to support its reform plans. The theoretical argument put forward in this paper echoes that of consensus-building approaches and is in agreement with their theoretical claim that minority governments and large coalition governments are, at times, the most efficient types of governments in implementing policy changes. Yet the argument advanced here differs from consensus theories in that it adds an important caveat to the sufficiency of tripartite cooperation in achieving reforms. Cooperation between governments and oppositions can be helpful in undertaking