The Electoral Consequences of Welfare State Reforms in Denmark and Sweden

Explaining the realignment of social democratic core voters in Denmark

Department of Political Science Aarhus University

Paper to be presented at the joint NordWel and Re-Assess International Summer School, August 15-20, 2010, Odense. Word Count: 7,839 words. 1. Introduction

Scandinavian social democracy has traditionally advocated a decommodifying and universalistic welfare state with generous social security benefits. So did the Danish as well as the Swedish social democratic parties in their welfare policy the post-war era. In the 1990s, both parties engaged in welfare state reforms to tackle high unemployment and an economic crisis.1 However, only the Danish social democrats (SD) experienced a serious electoral setback and a dealignment of parts of its core constituency in this period, whereas Swedish social democracy (SAP) was by and large able to maintain its voter base. What explains this difference and what role do welfare state reforms in the 1990s play for the diverging electoral performance of the two parties in the period 1993-2002? This paper thus examines the effects of the social policy reforms implemented by the Danish and Swedish social democratic parties on their partisan alignments. The classical theoretical framework of Esping-Andersen (1985, 1990) has argued that social democratic parties tie their core constituency by its welfare policy. Since manual workers and lower white-collar employees have an interest in alleviating their dependency on market forces and social democracy has traditionally advocated a welfare state which decommodifies the wage-earner, social democratic parties have aligned a particular social stratum as core constituency. As long as social democrats perform a welfare policy congruent with the decommodification principle, they are able to maintain the alignment of its core constituents. However, if social democrats not longer perform decommodification in welfare policy, the alignment is questioned and other parties now have a chance to lure those voters away, which may come along with a realignment of social democratic core constituents (Esping-Andersen 1985: 148f). I argue in this paper that the Danish social democrats experienced a lasting electoral setback as consequence of their welfare state reforms in the 1990s since these broke with decommodification principles. On the contrary, the policy of the Swedish social democrats did not fundamentally collide with the party‟s traditional principles in welfare policy allowing the SAP to largely maintain the alignment of its core constituency. Furthermore, I claim that the SD not only experienced a dealignment due to the reforms, but that another party, the right-wing Danish People‟s Party (DF), was able to foster a realignment of parts of the SD‟s core voter base. This is at odds with established arguments insisting left-wing competitors to be the likely benefiters of unpopular welfare state reforms of social democratic governments (e.g. Kitschelt 2001), but can be explained by the party competition when the SD held

1 The Danish SD ruled from 1993 to 2001 and the Swedish SAP from 1994 to 2006. For Sweden I limit the analysis to the years 1994-2002 in order to have a similar period of investigation for both cases.

1 office from 1993 to 2001. The reason why voters shifted to a right-wing, but not to a left-wing competitor, is that the DF has a social protectionist stance in social policy as well as electoral appeals on non-economic issues such as immigration reflecting the preferences and political attitudes of social democratic core constituents. Hence, I hypothesize the recommodifying welfare state reforms of the Danish social democrats to have lead to a realignment of social democratic core voters to a right-wing contender party, whereas the Swedish SAP kept its voter base as the party did not break with its traditional aspirations in social policy in the same time. The remainder of this paper is structured as follows. The second section outlines the theoretical framework on the class-based alignment of to social democratic parties advocating a decommodification of the wage-earner. Afterwards, I describe and compare the social policy and welfare state reforms of the social democratic governments in the two countries in 1993-2002. The following fourth section illustrates the party competition in Denmark and Sweden and develops an argument for the expected realignment of social democratic core constituents towards the Danish People‟s Party as a consequence of the welfare policy of the Danish social democrats. In the fifth section, I test this hypothesis empirically using statistical analysis. The final section sums up and discusses the findings in a comparative perspective.

2. Theoretical Background 2.1.The welfare state as base for social democratic alignment

Esping-Andersen (1985, 1990) regards the welfare state as an agency to tie and mobilize the social democratic core constituency. As social democrats advocate an encompassing and generous welfare state, they attract a particular social group, the working class. This is because the welfare state has a risk-hedging function and thus decommodifies individuals from pure market dependency. In turn, manual classes claim for decommodification due to their limited resource endowment resulting in precarious positions in the labour market and low living standards if exposed to market forces. Furthermore, decommodification works as one element of egalitarianism – a key theme in social democracy – improving the living conditions of the working class and other less well-off segments in society (Esping-Andersen 1985; Huo 2009: chap. 2). Thus, the social democrats formed an alignment with the manual classes by offering politics of decommodification and the welfare state as the respective agency. This was also expanded to lower white-collar employees sharing similar positions in the labour market and thus demand for risk-hedging policies (Esping-Andersen 1985).

The basic claim here is that social democrats have successfully mobilized these classes by politics of decommodification. These voters have a particular individual interest in decommodification of labour given their position in the labour market. A pure exposure to market forces would create particular hardships for the individual, as in this case, labour is only a discrete commodity depending on labour demand and the accompanied price for labour (Esping-Andersen 1985: 31). Therefore, labour as commodity is the most conflictual issue in social policy (Esping- Andersen 1990: 35). Moreover, the risk of being unemployed is contingent on the economic context and the individual‟s endowment with human capital. A related feature is the effect of commodification on individual living standards as living standards are contingent on the wage-earners‟ skills and the demand thereof. As manual classes are endowed with limited skills being nevertheless abundant on the labour market, the logic of pure competition on manual labour is likely to decrease their wages and thus the living standards. Moreover, a low skill level is accompanied by limited chances for individual promotion. Hence, as manual workers and lower white-collar employees are expected to have below-average incomes, they also have an interest in income redistribution, wage compression and generous welfare entitlements increasing their living standards by decommodification of labour. Decommodification in favour of the wage-earner is thus understood as establishment of generous and unconditional income maintenance in case of unemployment, old-age or sickness. If this is fulfilled, wage-earners are not longer contingent on pure market forces, but have a potential to improve their living conditions independently of market forces (Esping-Andersen 1990: 37). In this respect, decommodification is institutionalized by a universal welfare state and legislation guaranteeing generous and unconditional benefits including high income replacement rates (Esping- Andersen 1985: 33). Therefore, by offering politics that mitigate market forces, social democrats represent the natural political ally of classes claiming for risk-hedging. This match between the working class‟ interested-based claim for decommodification and the social democratic policy not only accounts for simple electoral mobilization but also for the formation of the party‟s core constituency and its long-term mobilization as long as the social democrats represent politics against markets and pursue the respective social and economic policy (Esping-Andersen 1985). The next section demonstrates that these claims apply to a considerable degree to the social democratic parties in Denmark and Sweden, their welfare policy, and consequently their core constituency.

2.2. Social democratic policy and welfare state development in Denmark and Sweden

In the post-war era until 2002, the Danish and Swedish social democratic parties brought forward four respectively three party constitutions (SAP 1960, 1975, 1990, 2001; SD 1961, 1977, 1992). All party constitutions contain traditional social democratic commitments to social policy and the welfare state in line with the theoretical framework on social democratic parties sketched in the previous section. The effects of pure market allocation of goods and labour are criticized and the welfare state is seen as an agency to alleviate those effects. More concretely, income maintenance in case of accidents, sickness, old-age and unemployment ought to be secured through universal social security schemes and are often formulated as social rights. Furthermore, the manifestos are committed to solidarity and egalitarianism to be realized through taxation and wage policy to improve the living standards of the less well-off. Two diverging things have to be mentioned, though. First, what makes the SAP different from its Danish counterpart is the reference to active labour market policy (ALMP) even in the earlier 1960 and 1975 programs, but also in the later ones. ALMP is seen as in instrument to fight unemployment by upgrading and supporting the labour force in adapting to changing working conditions. Second, the 1992 constitution of the SD is still committed to universal social rights, egalitarianism and solidarity, but it departs from the earlier ones as it accentuates individual responsibility and individual duties in the provision of welfare as the party‟s principles and thus indicates the shift of the SD to a third way agenda (Petersen 2001: 45). In sum, the Danish and Swedish social democracy‟s programmatic commitments represent Esping-Andersen‟s claim for decommodifying welfare policies and Huo‟s conception of traditional social democratic core values, but do contain two elements which will be further discussed in section 3. In line with its ideological principles, the social policy record of the SAP and SD reveals that both parties were responsible for the development of advanced social security schemes and the institutionalization of social rights leading to a highly decommodifying welfare state.2 In the immediate post-war era, both parties institutionalized state-financed universal flat-rate pension schemes (Sweden 1948, Denmark 1957) and abolished the hitherto existing eligibility criteria for old-age pensions (Esping-Andersen/Korpi 1987: 49). Afterwards, the private sickness and accident insurances were replaced by universal and compulsory schemes under social

2 The SAP had been in much more dominant position in Sweden, whereas the SD often had to cooperate with other parties (especially the Social Liberals) in welfare legislation.

4 democratic reign.3 The SAP also managed to implement a compulsory and universal superannuation scheme for pensions (ATP), whereas the Danish scheme is of a more limited nature (ibid). Both parties further engaged in expanding and supplementing the existing schemes throughout the years making them more generous and encompassing. Most importantly, the replacement rates of the social security schemes had been increased up to 90 per cent of previous income levels. This brief historical brief overview demonstrates that the Danish and Swedish social democracy established welfare states with universal social security schemes following the income maintenance principle. Thus, the SD and SAP institutionalized their programmatic commitment to decommodification in line with Esping-Andersen‟s (1985) theoretical framework and should have aligned a particular social stratum as core constituency in the second half of the 20th century. This is demonstrated in Table 1.

Table 1: Vote shares of Danish and Swedish social democratic parties among core constituency

Denmark: SD 1973 1975 1977 1979 1981 1984 1987 1988 1990 1994 1998 2001 2005 RNM n.a. 32 39 38 25 25 24 29 36 34 34 27 33 Skilled workers 41 51 59 46 37 45 34 40 55 44 37 34 27 Unskilled workers 57 45 57 56 59 47 44 44 57 45 44 33 34 Sweden: SAP 1960 1968 1976 1979 1982 1985 - 1988 1991 1994 1998 2002 2006 RNM 55 54 43 42 43 43 - 44 36 50 39 47 37 Industrial workers 83 79 68 71 75 71 - 73 61 73 58 56 59 Other workers 71 69 58 62 62 56 - 60 49 57 49 54 48 Sources: Denmark: Own calculations based on Danish election studies 1973-2005; Sweden: for 1960 and 1968: Oskarson 1994; for 1976-2006: Oscarsson/Holmberg 2008. Notes: RNM: Routine non-manual; Swedish studies distinguish between industrial workers and other workers; n.a.: not available.

Table 1 shows that both parties indeed mobilized the support of the manual workers over a long period of time in the 20th century. In contrast to its Danish sister party, which did not align this class to such a strong degree, the Swedish party also mobilized the lower white-collar employees as part of the party‟s core voter base. This has been particularly attributed to the introduction of the earnings-related pension scheme (ATP) in 1959 by the SAP creating a broader wage-earner alliance in Sweden (Esping-Andersen 1985: 108ff; Arter 1999: 152; 2003: 97). The 1990s, however, indicate a tendency towards a dealignment of the manual workers in both countries, but especially

3 The exception being the health care reform in Denmark implemented under a centre-right government, but largely supported by the SD.

5 in the Danish case. After 1998, the SD is not longer able to maintain previous levels of at least 40 per cent of electoral support among the blue-collar electorate. On the contrary, the party share drops below one third of the votes among workers. The Swedish party is also harmed by a decline of votes from the party‟s core voter base, but this not as distinct as in case of SD since the SAP still attracted at least every second worker in its worst performance in 1998 and there are signs of a recovery in the elections afterwards. The respective periods of a suspected dealignment of social democratic core voters coincide with welfare state reforms and austerity policies under social democratic governments in the 1990s in both countries. This raises the question whether these reforms do account for the weakening of the core constituency‟s alignment to the two social democratic parties and the variation between the SD and the SAP, respectively. In this regard, Esping-Andersen (1985) not only theorized the role of welfare policies for the alignment of particular social classes to social democracy, but also for a possible dealignment of those voters as a response to social democratic policy. This will be discussed in the following section.

2.3. Welfare State Reforms as catalyst for social democratic dealignment

For Esping-Andersen (1985: 35) social policy does not only account for partisan alignments, but can also foster the opposite, partisan dealignments and realignments. Policies that reflect a recommodification of labour are simply not in line with the interests and preferences of social democracy‟s core constituency. Consequently, if the social democrats engage in policies regarded as a departure from the politics of decommodification, they face a serious electoral threat (Esping- Andersen 1985: 148f). This implies not only short-term effects due to dissatisfaction, but a dealignment of the party‟s core electorate, or in Esping-Andersen‟s words a “social democratic decomposition” (ibid: 35) as the mechanism tying voter and party is suspended. As pointed out, a decommodifying welfare state and the accompanied social policy is characterized by generous and unconditional benefits as well as high replacement rates in order to reduce the individual‟s dependency on the market as main source of income and to protect one‟s skills and assets. Consequently, reforms that tighten eligibility criteria, increase conditionality and decrease replacement rates clearly mark a departure from politics of decommodification. In sum, social democrats engaging in welfare state reforms that imply a recommodification of labour do not longer fulfil the conditions for the long-term alignments of their core constituency since it stands against the social policy demands of these traditional supporters. In their research on

6 social democratic parties, Merkel et al. (2008: 14) claimed that “the decommodification objective has been partially substituted by a principle of conditionality”. To see which extent the Danish and Swedish party followed this proposition in their welfare policy, the next section examines the welfare state reforms of social democratic government in Denmark and Sweden in the 1990s in terms of the theoretical arguments on welfare policy and decommodification.

3. Welfare state reforms under social democratic reign in the 1990s

Both parties returned to power in the mid1990s replacing centre-right governments. In Sweden, this took place during a severe economic crisis which forced the SAP to perform a policy of fiscal prudence, which also went in line with welfare state reforms and cut-backs. In case of the SD, their policy revealed the transformation towards a third way agenda as already indicated with the 1992 party constitution as mentioned above.

3.1. Denmark

During their incumbency 1993-2001, the SD was responsible for three rounds of labour market reforms reflecting a changed stance on welfare policy and a paradigm shift in Danish social policy (Goul Andersen 2002, 2003; Green-Pedersen 2002). The eligibility period for unemployment benefits had successively been curtailed from 8 to 4 yours and similarly activation became mandatory after one year on benefits. Moreover, activation criteria to qualify for benefits have been tightened by the SD-lead governments after 1993. For instance, the period needed to qualify for unemployment benefits was doubled from 26 to 52 weeks of ordinary employment within three years. This was accompanied by tighter suitability criteria and increased conditionality as participation in active labour market programs became mandatory for the persons concerned otherwise their benefits are withdrawn. Further reforms concerned younger, older, and long-term unemployed as well as social assistance recipients since they were subject to mandatory activation programs. The SD was also responsible for a highly unpopular reform of the early retirement scheme as eligibility criteria have been tightened and the scheme‟s generosity was reduced (Green- Pedersen 2002: 81f). The replacement rates were not directly targeted, but due to changed indexation rules and the occasional non-adjustment of the maximum amounts, the scheme‟s generosity is likely to decrease (Christiansen et al. 2010).

The reforms in the 1990s have been characterized as clear break with existing social policy traditions of Danish social democracy as they content clear signs of recommodification (Goul Andersen 2002: 76). Furthermore, the reforms were evaluated as curtailing rights by tighter conditionality and a changed balance between rights and duties as the latter had been clearly emphasized by the reforms (Goul Andersen 2002: 69ff; Clasen/Clegg 2007). More generally, the policy change of the SD after 1993 has been seen as the shift towards a third way agenda which partly abandons existing programmatic commitments in economic and social policy (Green- Pedersen/van Kersbergen 2002). In sum, the SD implemented labour market reform implying a recommodification of labour and also realized a paradigm shift on the programmatic level. This can be expected to threaten the alignment of the party‟s core constituency as outlined above.

3.2. Sweden

After the SAP regained power in 1994, the party withdrew some even minor reforms of the previous centre-right government (Bergh/Erlingsson 2009: 77). However, the SAP stuck to the reductions of the replacement rates of the major social security schemes and the modified indexation rules in accordance with its primary aim to consolidate the budget. This is not in line with the SAP‟s decommodification aspiration, but was accompanied by other measures in the 1994 austerity package stressing two other social democratic core principles, egalitarianism and solidarity.4 The SAP (supported by the Left Party) implemented higher taxes on capital gains, wealth, property and a supplementary 5 per cent tax on high incomes labelled as “solidarity tax” (Merkel et al. 2008: 161f). In 1995, the SAP implemented a further reduction of the replacement rate of all welfare benefits from 80 to 75 per cent to come into effect in 1996, but fierce resistance from the trade unions implied that these measures were said to have only a temporary duration. Accordingly, the replacement rates returned to its previous level of 80 per cent in September 1997 as promised (Palme/Wennemo 1998: 27ff; Timonen 2003: 95). This was further supplemented by increasing the benefits for the first 100 days of unemployment (Huo 2009: 209). Other proposals were to tighten the eligibility criteria for the unemployment insurance, which were again met with fierce resistance of the unions and the unemployed. The proposals were

4 In addition, the effects of the changed indexation rules are more difficult to detect and will unfold their effects in the long-run. The symbolic raise of maximum amounts also obfuscates this type of austerity measures (Christiansen et al. 2010; cf. also Timonen 2003: 98).

8 designed to increase qualification periods, shorten the eligibility period and to manipulate the calculation rules for the benefits implying lower benefits (Timonen 2003: 96f). Given their unpopular character and the public resistance of the unions, the SAP finally abstained from implementing the proposals, which would have implied a considerable degree of recommodification. The SAP only accomplished a limitation of the eligibility period to three years, a fourth year being contingent on participation in active labour market or qualification programs (Merkel et al. 2008: 168). But even this reform was postponed to 2001 when unemployment and thus the number of persons concerned has decreased sharply. Mainly in the period 1995 to 1999, the SAP government also maintained and expanded the existing Swedish tradition of ALMPs. Some of the new measures aimed at upgrading and qualifying the labour force in times of high unemployment and to provide long-term unemployed with public service employment or private sector employment by using wage-subsidies. Others concerned younger unemployed (under 25 years) as these faced tighter eligibility criteria and only receive flat-rate activation benefit. Furthermore, the mandatory participation did not (re-)qualify the persons concerned for unemployment benefits as previous legislation did. Both aspects are at odds with existing traditions in Swedish ALMP (see Hort 2001: 259; Timonen 2003: 96, 102f). On the other hands, these measures concerned the margins of the labour market, but not the core labour force. Thus, most commentators argued that the reforms were neither a dismantling or retrenchment of the Swedish welfare state, but rather a response the deteriorating economic situation or a liberalization limited to economic policy (Hort 2001; Lindbom 2001; Timonen 2003; Merkel et al. 2008: chap. 8; Bergh/Erlingsson 2009). Other aspects of the reforms were their temporary nature or their postponement to better context conditions. The reforms had also been supplemented by symbolic policies stressing other fundamental social democratic principles such as higher taxes for the well-off. There had been no substantial reforms which implied that the core work force will lose its benefits rights through tighter eligibility criteria or increasing conditionality as those reforms concerned the labour market fringes.5 Active labour market programs have always existed and were used continuously by the SAP under the crisis and did not imply a break with existing policy patterns as its introduction under other social democratic parties (Huo 2009: 112).

5 The work of Lindbom (2001) also shows that even though cuts have been made, Swedish replacement rates after the reforms remained roughly at the same level as in 1980.

3.3. Balance

On balance, the SAP implemented some reforms and measures pointing to limited recommodification in times of an economic crisis, whereas its Danish sister party was responsible for a considerable degree of recommodification. Moreover, while the SD broke with existing social policy commitments, the SAP did not carry out any fundamental breaks with its principles in social policy or with the main pillars of the Swedish welfare state. That is, in contrast to the SD which turned towards a third way agenda, the policy of the SAP is to a considerable degree consistent with traditional social democratic principles and has been characterized as coherent with the party‟s programmatic stance (Merkel et al. 2008: 180ff). Some reforms in Sweden reflect incremental changes that have to be seen in light of existing traditions of active labour market policies and the workfare element that always played a role in Swedish welfare policies and the SAP‟s commitment to social policy (e.g. Hort 2001; Timonen 2003: 7). The Danish turn to ALMPs is more seminal since previous ALMPs had a largely voluntary character (Goul Andersen 2002). Therefore, we can expect the Danish party to have performed policies against the interests of its core constituency as the reforms point to a considerable recommodification of the wage-earner. This should lead to a dealignment of core voters as argued in the theoretical framework above. By contrast, the policy of the Swedish SAP in the 1990s had been more in line with traditional social democratic welfare policy commitment and the defections from its core constituency are expected to be more limited. This will be further elaborated in the next section on party competition in both countries.

4. Party Competition in Denmark and Sweden

The party systems in Denmark and Sweden have resembled each other to a considerable degree during the period under consideration, but also differ in crucial aspects structuring the electoral consequences of the social democratic reforms in both countries. Therefore, I apply a most similar systems design for the empirical investigation. Political competition in both countries is organized around two blocs. The type of competition has been characterized as a divided centre-right bloc against united social democrats (Kitschelt 2001, cf. also Arter 1999). The political left consists of a dominant social democratic party (particularly in Sweden) and smaller left socialist parties, the Socialist People‟s Party (SF) in Denmark and the Left Party (VP) in Sweden. This camp had been supplemented by the radical left Unity List in Denmark and a small green party in Sweden in the last two decades. The bourgeois parties are medium-sized conservative and liberal parties whose

10 electoral performance had been highly volatile over the last decades. Often crucial for obtaining legislative majorities is the support of centre parties for both camps. In Denmark, the Social Liberal Party and the Centre-Democrats represented the political centre, in Sweden it has been the Centre Party (CP) even though it became integrated into the bourgeois bloc after 1998/2002 (Aylott/Bolin 2007). The party systems were supplemented by small Christian Democratic parties even though the Danish party virtually disappeared during the period under examination. The crucial difference is the emergence of the right-wing Danish People‟s Party (DF) in 1995. The establishment of a far right party in Sweden failed in the early 1990s when the Ny Demokrati gained parliamentary representation in 1991, but disappeared again after the 1994 election due to internal quarrels and its political marginalization. In Denmark, the DF managed to establish itself in Danish politics as a splinter from the far-right Progress Party (FrP). In contrast to the market-liberal FrP, the DF developed centrist views on socio-economic issues. In particular, the party can be characterized as social protectionist in terms of Kitschelt‟s work (2003, 2004) allowing the party to woo those voters interested in maintaining core welfare programs by a the respective programmatic and populist appeals (Nielsen 1999; Kitschelt 2004; Rydgren 2004). The party has also presented itself as social corrective for the liberal-conservative government after 2001 for which the DF is the supporting party (Spier/Wirries 2007: 101f). Thus, the DF represents a contender for the SD due to the party‟s social protectionist stance making it possible to attract social democratic core constituents. Moreover, the DF represents authoritarian positions on socio-cultural matters, especially a sceptical view towards immigration. Previous research has shown that right-wing parties have electoral appeals on socio-cultural issues reflecting the attitudes of social democratic core supporters such as blue-collar workers (e.g. Scheuregger/Spier 2007; Houtman et al. 2008). Thus, for the Danish case, the presence of the right-wing DF implies a potential for a realignment of the social democratic core voters. This does not apply to Sweden as a right-wing competitor failed to establish after 1991 and the Sweden Democrats only gained importance after the period under examination. As mentioned, parties to the left of social democracy are normally seen as the most dangerous competitors for social democratic parties engaging in welfare retrenchment. Hence, the two major left contenders SF and VP deserve special attention here.6 Contrary to standard

6 I do not consider the Danish Unity List here as all arguments apply to this party to the same or even stronger degree. Moreover, due to the party’s far left stance, the party is not electable for large parts of the Danish electorate.

11 theoretical assumptions, the SF is not expected to foster a realignment of social democratic core voters. There are two aspects making it difficult for the SF to capitalize on social democratic welfare state reforms and pulling social democratic core voters further to the left. The first concerns the behaviour of SF under conditions of bloc competition during the 1990s. As the party supported the SD-lead government after 1994, the SF had limited room to manoeuvre to attack the social democrats by engaging in a more pronounced opposition strategy. The latter would imply a threat of toppling the government and a bourgeois majority if the centre parties shift side again which is not in the interest of the SF at last (cf. Nielsen 1999: 68ff). This distinguishes the SF from the VP which had not been a supporter party in the retrenchment packages. Accordingly, the SF leadership had been criticized for being too compliant in the cooperation with the social democratic government (cf. Spier/Wirries 2007: 104). The second factor concerns the position of SF on socio-cultural issues. As the party has had libertarian views on matters such as immigration, the Socialist People‟s Party‟s prospects of attracting social democratic core voters should be rather limited (cf. Spier/Wirries 2007: 115). The Swedish VP is also not expected to have fostered a lasting realignment of core voters of the SAP which owing to both the limited nature of the reforms in Sweden and the party‟s own strategy in the period under consideration. After the SAP shifted to the CP as main partner in welfare state reforms in the years 1995-1998, the VP found itself in a favourable opportunity structure as it was possible to mobilize against the reforms and to appeal to disaffected SAP voters with a strong welfare profile (Arter 2002: 11, 2003: 93; Blomqvist/Green-Pedersen 2004: 606). This can be expected to have attracted a share of SAP core supporters in the 1998 election as the VP doubled its vote share. However, after the 1998 election and the SAP‟s bad performance, the social democrats were able to remain in office thanks to a legislative coalition with the Left Party, and the Greens. This meant that the VP rescued the government which policies it has fiercely criticized before and which voters it absorbed. The legislative coalition also implied no direct policies in return for the cooperation of the left (Spier/Wirries 2007: 84, 114). Moreover, the SAP changed back to an expansion of the welfare state in the electoral term 1998-2002 after the crisis was overcome allowing the party to claim credit not to share with potential coalition partners. In this respect, the strategy of the Left Party to back an SAP minority government without getting some policies in return has been characterized as useless and the party was seen a Babylonian captive of the SAP (Spier/Wirries 2007: 84; Merkel et al. 2008: 160). Thus we can expect the Left Party to make

12 temporary gains in the 1998 elections as consequence of the austerity policies under the SAP government, but not having been able to foster a realignment. In sum, I expect that Danish social democracy suffered a loss of core constituents having realigned with the Danish People‟s Party, a right-wing, but social-protectionist party. Against theoretical odds, I expect neither left-wing competitor to make inroads into the social democratic core voter base. This is due to the parties‟ strategic behaviour in both countries and in Sweden due to the albeit limited character of the reforms of the SAP which did not imply a considerable degree of recommodification as did the reforms of its Danish sister party. This next section examines these propositions of the electoral consequences of the welfare state reforms under social democratic governments in Denmark and Sweden in the 1990s.

5. Results

The aim of this section is to prove the hypothesis that the Danish social democrats suffered a realignment of parts the party‟s core constituency to the right-wing Danish People‟s Party. In contrast, the Swedish SAP is not expected to suffer from a lasting realignment as the gains for the Left Party among the party‟s core constituency should be of a temporary nature. This will be tested by using descriptive statistics on class voting as well as by presenting conditional probabilities obtained from multinomial logistic regression models for the two elections where a de- /realignments could have taken place (Sweden 1998, Denmark 2001). Afterwards, I use descriptive statistics on voter migrations for the following elections in both countries to control whether the party shifts had been lasting and thus indicate a realignment. I used the Danish National Election Studies (Valgundersøgelsen 2001, 2005) and the exit polls for the Swedish general election studies (VALU 1998, 2002) for the respective elections as data source. The dependent variable is party choice, where I focus on the results for the two social democratic parties and the suspected competitors on welfare grounds discussed in the previous section. The independent variable is a dummy variable on belonging to the social democratic core constituency operationalized via social class and a previous voting decision for the social democratic party in the two countries. In the logistic models I further used variables measuring the attitude towards the policy of the social democrats to model dissatisfaction with the SD and SAP, respectively. To assess whether the shifts to the competitor parties in the next election have been

13 lasting, I also added a category for a voting decision for the DF and VP in the previous election to check whether those voters stuck to the parties. First, Table 2 shows the party choice among those voter groups identified as social democratic core voters during and after the welfare reforms in both countries took place.

Table 2: Party shares for selected parties by selected social classes, Denmark and Sweden 1990-2006

Denmark Sweden 1990 1994 1998 2001 2005 1991 1994 1998 2002 2006 SD RNM 36 34 34 27 33 SAP RNM 36 50 39 47 37 Skilled 55 44 37 34 27 Industrial 61 73 58 56 59 Unskilled 57 45 44 33 34 Other 49 57 49 54 48 SF RNM 13 11 13 7 5 VP RNM 3 8 11 8 8 Skilled 2 5 2 5 4 Industrial 3 7 17 12 4 Unskilled 9 8 10 3 3 Other 4 9 14 11 7 DF RNM - - 5 10 14 NyD/ RNM 7 1 - - 2 Skilled - - 13 24 21 SvDem Industrial 10 1 - - 5 Unskilled - - 6 19 21 Other 6 1 - - 4 Source: Denmark: Own calculations based on Danish election studies 1973-2005 Sweden: Oscarsson/Holmberg 2008. Notes: RNM: Routine non-manual. Swedish studies distinguish between industrial workers and other workers. SD and SAP: Social Democrats, SF: Socialist People‟s Party, VP: Left Party, DF: Danish People‟s Party, NyD: New Democracy, SvDem: Sweden Democrats.

The figures demonstrate a sharp decline for the SD among manual workers from the beginning of the 1990s, whereas the SAP‟s setback among its core electorate is clearly more limited. As expected, the Socialist People‟s Party is not able to make inroads into the social democratic core voter base in this period. The Swedish Left Party is able to secure a considerable share of social democratic core constituents after the SAP engaged in welfare state retrenchment in the mid1990s, but the party is not able to keep this share and falls back to the level of support from the early 1990s. Concerning the right-wing competitors, we can observe a considerable support for the Danish People‟s Party among the social democratic core after 1998 which does also not decline afterwards. As mentioned, right-wing parties failed to establish in Sweden as the figures indicate. In next step, I present conditional probabilities obtained from logistic models to investigate whether dissatisfaction with the social democrats explains the defection of core voters to other parties.7 The Danish data set for 2001 contains a more precise item, though, as the respondent is

7 Due to space restrictions, I do not present the results from the logistic models, but focus on the conditional probabilities obtained from these. The coefficients are available on request from the author.

14 asked about the social policy record of the SD in the last term. In Sweden, respondents were asked to assess the policy of the SAP in general. Given the cutbacks and welfare issues being the dominant political issues during this period, I regard this as an acceptable proxy variable.

Table 3: Predicted probabilities of vote choice for attitudes towards welfare policy among aligned core constituents1

Denmark Assessment welfare policy: Well/Neither Assessment welfare policy: badly SD 72.7 % 20.9 % * Venstre 8.5 % 31.4 % * SF 4.3 % 3.1 % DF 4.3 % 23.3 % * All other parties 10.2 % 14.3 % Non-voting 3.6 % 7.0 % Sweden Assessment policy: Well/Neither Assessment policy: badly SAP 81.3 % 15.1 % * Moderates 1.5 % 19.7 % * VP 11.7 % 26.7 % * Christian Democrats 1.6 % 14.8 % * All other parties 3.9 % 23.7 % Source: Predicted probabilities obtained from multinomial logistic regression models using Clarify (Tomz et al. 2001). Note: 1Aligned core constituency is defined as being a worker or a lower white-collar employee and having voted for the social democrats in the previous general election. The number of simulations to obtain the probabilities was set to 1,000. Figures do not always sum up 100 per cent due to rounding. The Swedish data does not include reliable information on non-voting and was thus excluded in the models. * indicates that the party share reported in the right- hand side is significantly distinguishable from the share in the left-hand side at the 0.05 level.

The results for the voting behaviour of core constituents which actually voted social democratic in the last national election in the two countries are largely in line with expectations.8 As long as the core constituents were not disaffected by their parties‟ policy, they stuck to the SD and SAP. Among dissatisfied core constituents, the two parties got a setback as they did not mobilize more than one out of five votes. In both countries, this has lead to increased support for the biggest competitor party for the social democrats as Venstre (Denmark) and the Moderates (Sweden) got significantly higher support here. In Sweden, the small Christian Democratic party was also to gain support from discontent SAP voters, surprisingly. More interestingly and in line with the theoretical

8 For the Danish case, I also controlled for the attitudes towards immigration. Even though negative attitudes towards immigration explain voter migration towards the DF, it does not substantially alter the findings here.

15 expectations are the results for SF, DF, and VP. In Denmark, many SD-core voters shifted to the DF whose support among the dissatisfied is almost 20 per cent higher compared to the other group. The Socialists in Denmark have not been able to benefit among dissatisfied social democratic core voters, whereas the Swedish Left Party absorbed about one fourth of unsatisfied SAP core supporters. In sum, these results have shown that a considerable degree of discontent social democratic core constituents have changed its alignment as they went to the rightist DF in Denmark and the leftist VP in Sweden in line with the expectations. To see whether this has been lasting, I control for the voter migration of the relevant parties among the classes conceptualized as social democratic core constituency in the following elections.

Table 4: Voter migration in Danish 2005 and Swedish 2002 general election by selected voter groups

Denmark 2005 election SD Venstre SF DF All other N=1,892 parties1 Non-core 18.7 % 30.7 % 7.2 % 9.3 % 34.0 % 1,014 constituency 100 % Non-aligned core 11.2 % 39.2 % 4.8 % 13.5 % 47.6 % 254 constituency2 100 % Core constituency 73.5 % 6.1 % 3.0 % 1.8 % 8.4 % 269 aligned to SD 100 % Core constituency 6.6 % 0.3 % 0.1 % 86.7 % 6.3 % 124 aligned to DF 100 % Sweden 2002 election SAP Moderates VP Christian All other N=8,062 Democrats parties Non-core 26.9 % 21.8 % 8.0 % 9.3 % 34.0 % 2,462 constituency 100 % Non-aligned core 14.6 % 18.6 % 5.7 % 13.5 % 47.6 % 2,675 constituency2 100 % Core constituency 83.5 % 1.5 % 4.8 % 1.8 % 8.4 % 2,193 aligned to SAP 100 % Core constituency 25.3 % 0.3 % 60.1 % 2.1 % 12.2 % 732 aligned to VP 100 % Source: Own calculations based on Danish election study 2005 and exit polls for Swedish General election 2002 (VALU 2002). Notes: 1includes non-voters in the Danish case. 2indicates voters of all other parties.

Table 4 reveals the respective party shares for the voter groups of analytical relevance. The figures reveal that the DF was able to keep more than 8 out of 10 voters from this voter milieu and that the SD was only able to (re-)gain less than 7 per cent in this group. In Sweden, the VP did loss 4 out 10 supporters from the social democratic core voter base in the 2002. This is largely to the benefit of the SAP indicating that the party had been able to recapture a considerable share of core constituents which defected to its left competitor in 1998. This confirms the patterns in Table 3 showing a stable and even somewhat increasing share for the DF among the social democratic core constituency after 1998, whereas the SAP only “lent” their dissatisfied voters to the leftist VP in 1998. Summing up, the results reveal that the Danish social democrats suffered a lasting realignment after they engaged in recommodifying welfare state reforms in the 1990s. The party lost core supporters opposing the welfare policy under the SD-lead government to the rightist DF. Their Swedish sister party did not implemented such encompassing reforms and thus were able to regain ground among its core constituency after disappointed core supporters shifted to the Left Party in 1998, which failed to realign these voters afterwards. Hence, the results are largely in line with the expectations formulated above. However, two further results which were not embedded in the theoretical framework used appear. These concern the gains for the Danish Liberals and the Swedish Christian Democrats which seem to have a longer lasting nature

6. Conclusion

This paper investigated the electoral consequences of welfare state reforms under social democratic government in the 1990s in Denmark and Sweden. The paper‟s general theoretical claim was that social democrats reforming the welfare state face an electoral threat as they act against the interests and preferences of their core constituency. Drawing on Esping-Andersen‟s work, it was argued that this is because reforms reflecting recommodification of labour stand against the core voters‟ political preferences. The actual extent of the electoral backlash is contingent on the extent of the reforms implemented and the presence and behaviour of particular competitor parties. Normally, left-wing parties constitute the most dangerous contender for social democracy as they can attract social democratic voters by drawing on the decommodification mechanism. However, I expected that neither the Danish nor the Swedish social democratic parties suffered from a core constituency realignment to their left-wing competitors. This is due to the left parties‟ electoral appeal and their

17 strategic behaviour. Instead, in case of Denmark, I expected a realignment towards a right-wing party – the Danish People‟s Party – as this party has accommodated the preferences of social democratic core voters after the SD engaged in recommodifying reforms. As the reforms under the SAP where more limited and no other party represented as serious competitor on welfare grounds, I proposed the Swedish party to maintain a great deal of support among its core constituents in the period of examination. The findings demonstrated that the Danish SD experienced a realignment of core voters to its right-wing contender whereas the Swedish SAP did not suffer from a dealignment. The Danish People‟s Party had been able to woo a considerable degree of core supporters of the SD being disaffected with the reforms and the alignment of manual workers to Danish social democracy clearly weakened after 1998. Moreover, the change proved to be lasting as the DF kept its support among the social democratic voter base. On the contrary, the SAP experienced a setback in 1998 as dissatisfied core supporters turned towards the Left Party, but regained a considerable share of these votes later. The Left Party failed in establishing a lasting realignment of social democratic core constituents after the exceptional 1998 election due its own strategy and the SAP‟s return to a expansionary welfare policy after the crisis was overcome. In this respect, the Danish 2001 election may be regarded as a critical election as a lasting realignment inside the electorate took place. The Swedish 1998 seemed to have been a deviating election as the electoral change did not consolidate and the voting patterns of the social democratic core constituency turned back to the status quo ante afterwards. However, there are things left to be examined in future work The first concerns the gains for the Swedish Christian democrats and the Danish Venstre in the period of investigation. For the latter, first investigations have found that the Liberals accommodated towards the preferences of social democratic core voters in Denmark and were able to make inroads in this voter group (Arndt 2010). For the Swedish Christian Democrats, there are no verifiable statements why this party was able to gain ground after the SAP engaged in welfare state retrenchment and reforms in the 1990s. One may speculate whether this party followed other continental Christian democratic parties and acquired a welfare profile over the years making it possible to attract some support among ordinary wage-earners. Second and more generally, further research is needed to examine the electoral consequences of welfare state reforms under social democratic reign. The same is true for the ideological transformation of social democratic parties towards third way agendas opening the

18 space for contenders to mobilize traditional social democratic voters (Huo 2009: 11ff). The role of competitor parties and party systems can be seen as one important aspect as these determine the opportunities and constraints in welfare state reforms and their electoral effects (cf. Kitschelt 2001; Green-Pedersen 2002). For instance, the Dutch and German social democrats faced left-wing and social populist contenders after having engaged in social policy reforms and consequently lost considerable vote shares to these parties. In contrast, New Labour in the UK faced no serious competitor on welfare grounds and thus no systematic defection towards other parties even though the party abandoned its traditional stance in welfare policy. Recently, the Swedish party faced a backlash as the Moderates softened their stance in economic and welfare policy and were able to gain some social democratic votes. This mirrors the shift of the Danish Liberals towards the centre in the 1990s, but these phenomena have not been seriously examined yet by linking welfare policy to partisan alignments. Thus the interplay of social democratic welfare state reforms and the party system in determining the electoral consequences of those reforms need more attention in future research as we lack empirical investigations in this respect so far.

References

Arndt, Christoph (2010): Chapter Denmark. In: The Electoral Consequences of Welfare State Reforms, PhD project, Department of Political Science, Aarhus University. Arter, David (1999): Sweden: A Mild Case of „Electoral Instability Syndrom‟? In: Broughton, David/Donovan, Mark (eds.): Changing Party Systems in Western Europe. London/New York, Pinter, pp. 143-162. Arter, David (2002): „Communists we are no longer, Social Democrats we can never be‟: The Evolution of the Leftist Parties in Finland and Sweden, in: Journal of Communist Studies and Transition Politics 18, 3, pp. 1-28. Arter, David (2003): Scandinavia: What‟s Left is the Social Democratic Welfare Consensus, in: Parliamentary Affairs 56, pp. 75-98. Aylott, Nicholas/Bolin, Niklas (2007): Towards a Two-Party System? The Swedish Parliamentary Election of September 2006, in: West European Politics 30, 3, pp. 621-633. Bergh, Andreas/Erlingsson, Gissur Ó. (2009): Liberalization without Retrenchment: Understanding the Consensus on Swedish Welfare State Reforms, in: Scandinavian Political Studies 32, 1, pp. 71-92.

Blomqvist, Paula/Green-Pedersen, Christoffer (2004): Defeat at Home? Issue-Ownership and Social Democratic Support in Scandinavia, in: Government and Opposition 39, 4, pp. 587-613. Clasen, Jochen/Clegg, Daniel (2007): Levels and Levers of Conditionality: Measuring Change within Welfare States. In: Clasen, Jochen/Siegel, Nico A. (eds.): Investigating Welfare State Change. The „Dependent Variable Problem‟ in Comparative Analysis. Cheltenham, Edward Elgar, pp. 166-197. Christiansen; Flemming Juul/Green-Pedersen, Christoffer/Jensen, Carsten (2010): Dismantling Strategies in a Comparative Perspective. Department of Political Science, Aarhus University. Esping-Andersen, Gøsta (1985): Politics against Markets. The Social Democratic Road to Power. Princeton, Princeton University Press. Esping-Andersen, Gøsta (1990): The Three Worlds of Welfare Capitalism. Cambridge, Polity Press. Esping-Andersen, Gøsta/Korpi, Walter (1987): From Poor Relief to Institutional Welfare States: The Development of Scandinavian Social Policy. In: Erikson, Robert/Hansen, Erik Jørgen/Ringen, Stein/Uusitalo, Hannu (eds.): The Scandinavian Model. Welfare States and Research. New York, M.E. Sharpe, Inc, pp. 39-74. Goul Andersen, Jørgen (2002): Work and Citizenship: Unemployment and Unemployment Policies in Denmark 1980-2000. In: Goul Andersen, Jørgen/Jensen, Per H. (eds.): Changing Labour Markets, Welfare Policies and Citizenship. Bristol, Polity, pp. 59-84. Goul Andersen, Jørgen (2003): Operationen Lykkedes, Men Lægen Døde. Socialdemokratiets Kriser fra Krag til Lykketoft, in: Politica 35, 4, pp. 413-432. Green-Pedersen, Christoffer (2002): The Politics of Justification. Party Competition and Welfare- State Retrenchment in Denmark and the Netherlands from 1982 to 1998. Amsterdam, Amsterdam University Press. Green-Pedersen, Christoffer/van Kersbergen, Kees (2002): The Politics of the „Third Way‟. The Transformation of Social Democracy in Denmark and the Netherlands, in: Party Politics 8, 5, pp. 507-524. Hort, Sven E.O. (2001): Sweden – Still a Civilized Version of Workfare? In: Gilbert, Neil/Van Voorhis, Rebecca A. (eds.): Activating the Unemployed: A Comparative Appraisal of Work-Oriented Policies. New Brunswick/London, Transaction Publishers, pp. 243-266. Houtman, Dick/Achterberg, Peter/Derks, Anton (2008): Farewell to the Leftist Working Class. New Brunswick, Transaction Publishers.

Huo, Jingjing (2009): Third Way Reforms. Social Democracy after the Golden Age. Cambridge/New York, Cambridge University Press. Kitschelt, Herbert (2001): Partisan Competition and Welfare State Retrenchment, When Do Politicians Choose Unpopular Policies? In: Pierson, Paul (ed.): The New Politics of the Welfare State. Oxford, Oxford University Press, pp. 265-302. Kitschelt, Herbert (2003): Political-Economic Context and Partisan Strategies in the German Federal Elections, 1990-2002, in: West European Politics 26, Special Issue on Germany, pp. 125-152. Kitschelt, Herbert (2004): Diversification and Reconfiguration of Party Systems in Postindustrial Democracies. Bonn, Friedrich Ebert-Stiftung. Lindbom, Anders (2001): Dismantling the Social Democratic Welfare Model? Has the Swedish Welfare Lost its Defining Characteristics?, in: Scandinavian Political Studies 24, 3, pp. 171- 193. Merkel, Wolfgang/Petring, Alexander/Henkes, Christian/Egle, Christoph (2008): Social Democracy in Power. The Capacity to Reform. London/New York, Routledge. Nielsen, Hans Jørgen (1999): The Danish Election 1998, in: Scandinavian Political Studies 22, 1, pp. 67-81. Oscarsson, Henrik/Holmberg, Sören (2008): Regeringsskifte. Väljarna och Valet 2006. Stockholm, Norstedts Juridik. Oskarson, Maria (1994): Klassröstning i Sverige. Stockholm, Nerenius & Santérud Förlag. Palme, Joakim/Wennemo, Irene (1998): Swedish Social Security in the 1990s: Reform and Retrenchment. Stockholm, Välfärdsprojektet. Petersen, Klaus (2001): Programmeret til Velfærd? Om Ideologi og Velfærdspolitik i Socialdemokratiske Partiprogrammer efter 1945, in: Arbejderhistorie 2/2001, pp. 23-51. Rydgren, Jens (2004): Explaining the Emergence of Radical Right-wing Populist Parties: The Case of Denmark, in: West European Politics 27, 3, pp 474-502. SAP (1960): Program för Sveriges Socialdemokratiska Arbetareparti. SAP (1975): Socialdemokraterna, Program. SAP (1990): Socialdemokraternas Partiprogram. SAP (2001): Partiprogram för Socialdemokraterna 2001.

Särlvik, Bo (2002): Party and Electoral System in Sweden. In: Grofman, Bernard/Lijphart, Arend (eds.): The Evolution of Electoral and Party Systems in the Nordic Countries. New York, Agathon Press, pp. 225-269. Scheuregger, Daniel/Spier, Tim (2007): Working-class Authoritarianism und die Wahl Rechtspopulistischer Parteien, in: Kölner Zeitschrift für Soziologie und Sozialpsychologie 59, 1, pp. 59-80. SD (1961): Vejen Frem. Socialdemokratiets Principprogram 1961. SD (1977): Solidaritet, Lighed og Trivsel. Socialdemokratiets 3. Principprogram. SD (1992): Det Ny Århundrede. Socialdemokratiets Principprogram. Spier, Tim/Wirries, Clemens (2007): Ausnahmeerscheinung oder Normalität? Linksparteien in Westeuropa. In: Spier, Tim/Butzlaff, Felix/Micus, Matthias/Walter, Franz (eds.): Die Linkspartei. Zeitgemäße Idee oder Bündnis ohne Zukunft? Wiesbaden, Verlag für Sozialwissenschaften, pp. 71-116. Timonen, Virpi (2003): Restructuring the Welfare State. Globalization and Social Policy Reform in Finland and Sweden. Cheltenham, Edward Elgar. Tomz, Michael/Wittenberg, Jason/King, Gary (2001): CLARIFY: Software for Interpreting and Presenting Statistical Results. Version 2.0 Cambridge, MA: Harvard University. Valgundersøgelsen (2001): Valgundersøgelsen 2001. Institut for Økonomi, Politik og Forvaltning, Aalborg Universitet. Valgundersøgelsen (2005): Valgundersøgelsen 2005. Institut for Økonomi, Politik og Forvaltning, Aalborg Universitet. VALU (1998): VALU 1998 - SVT exit poll survey parliamentary election 1998. Swedish National Data Service. VALU (2002): VALU 2002 - SVT exit poll survey parliamentary election 2002. Swedish National Data Service.