There is power in a union? Union members' preferences and the conditional effect of labour unions on left parties in different welfare state programmes

Engler, Fabian; Voigt, Linda Article — Published Version There is power in a union? Union members' preferences and the conditional effect of labour unions on left parties in different welfare state programmes British Journal of Industrial Relations Provided in Cooperation with: John Wiley & Sons Suggested Citation: Engler, Fabian; Voigt, Linda (2022) : There is power in a union? Union members' preferences and the conditional effect of labour unions on left parties in different welfare state programmes, British Journal of Industrial Relations, ISSN 1467-8543, Wiley, Hoboken, NJ, Vol. 61, Iss. 1, pp. 89-109, https://doi.org/10.1111/bjir.12665 This Version is available at: https://hdl.handle.net/10419/287927 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/4.0/ Accepted: 13 January 2022 DOI: 10.1111/bjir.12665 ORIGINAL ARTICLE There is power in a union? Union members’ preferences and the conditional effect of labour unions on left parties in different welfare state programmes Fabian Engler Linda Voigt Heidelberg University, Institute of Political Science, Heidelberg, Germany Correspondence LindaVoigt, Heidelberg University, InstituteofPolitical Science,BergheimerStr. 58, 69115 Heidelberg, Germany. Email:[email protected] Bothauthorscontributedequallytothe paper. Abstract This article studies the effect of labour unions on policymaking in six different parts of the welfare state (passive and active labour market policy, employment protection, old-age pensions, health care and education) in OECD countries after 1980 with a two-level strategy: At the micro-level, we investigate union members’ preferences. Ordered logit regression analyses indicate that union members favour generous social policies more strongly than non-members. Moreover, this effect is stronger for programmes closely related to the labour market than for programmes without a strong labour market link. At the macro-level, we investigate the conditional effect of unions on left parties expecting the former to push the left towards more generous labour marketrelated (but not towards less-labour market-related) programmes. Regression analyses essentially provide evidence for such a relationship. Overall, unions have been powerful in promoting theirmembers’ social policy preferences via left parties in government but their power is recently vanishing. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2022 The Authors. British Journal of Industrial Relations published by John Wiley & Sons Ltd. Br J Ind Relat. 2023;61:89–109. wileyonlinelibrary.com/journal/bjir 89 90 BRITISH JOURNAL OF INDUSTRIAL RELATIONS 1 INTRODUCTION For a long time, labour unions and left parties have been found to be natural allies when it comes to the promotion of working-class interests (see, e.g. Korpi, 1983; Stephens, 1979). However, they do not act at the same level of representation. Unions as intermediary organizations are often understood as the major linkage between the welfare state preferences of (organized) workers at the micro-level and the social policy-making of left parties in government at the macro-level. This article aims at an empirical evaluation of this mechanism in times of tremendously declining union membership. In investigating this relationship, we answer the following research questions: What exactly are the welfare state preferences of union members and how effective are labour unions in pushing left parties1to be responsive to union members’ interests in different parts of the welfare state? What can we learn from existing literature? At the level of individual welfare state preferences, studies often include union membership but mostly only as a control variable. Quite consistently, union members favour more redistribution or higher aggregate social spending (e.g. Cusack et al., 2006; Finseraas, 2009; Häusermann et al., 2016; Iversen & Soskice, 2001; Kitschelt & Rehm, 2014; Mosimann & Pontusson, 2017; Rehm, 2009). Furthermore, they prefer more government intervention and higher spending on individual social policies (e.g. Busemeyer & Neimanns, 2017 for unemployment support and childcare; Engler & Zohlnhöfer, 2019 for industrial aid and market regulation; Gingrich & Ansell, 2012 for health care and unemployment support; Häusermann et al., 2016 for job creation). However, what this group of studies lacks is a systematical comparison of union members’ preferences across various disaggregated welfare state programmes. When it comes to the macro-level and unions’ direct impact on policy-making, labour power mostly has a positive effect on unemployment benefits (e.g. Jensen, 2012a), active labour market policies and employment protection (e.g. Rueda, 2006 for both indicators). However, it has no impact on (or even negatively affects) old-age provision and health care spending (e.g. Jensen, 2012a). Regarding the conditional effect of labour unions on left parties, the literature is fairly thin: For aggregated data, some studies provide evidence for a positive conditional relationship (Hicks et al., 1989) or show that unions push left parties to globalization-induced welfare state expansion (Garrett, 1998; Kwon & Pontusson, 2010). For disaggregated data, existing empirical evidence is even thinner: Only recently, Swank (2020) shows that strong unions reinforce the left’s positive effect on minimum income protection and active labour market policies, and Engler (2021) reports that labour power conditions left parties’ reactions to globalization regarding unemployment benefits. Besides, Zohlnhöfer et al. (2013) provide evidence that in the 1990s, left parties have rather reduced pension generosity with increasing union density. Yet, the dependent variables being studied in the existing literature are not representative for the welfare state in general. Therefore, what the existing literature lacks is a theoretical elaboration and a simultaneous empirical test of the intermediary role of labour unions and their impact on left parties in government across several welfare state programmes. This article aims at filling these gaps and contributes to the existing literature in a number of ways: Firstly, we study every step of the causal mechanism that runs from labour union members to social policy-making. This is, to the best of our knowledge, the first article that explicitly connects the macro-level (unions’ effect on left parties’ social policy-making) to its micro-foundations (union members’ welfare state preferences). Secondly, we systematically compare various programmes. This allows drawing a broader picture than studying individual welfare state policies separately. To this end, leaning on Jensen’s (2012a) categorization, we deduce and test theoretical expectations for both programmes that are close and programmes that are less close to the THERE IS POWER IN A UNION? 91 labour market as well as both policies of income protection and policies of activation. Overall, our findings have more general implications regarding labour unions’ role as intermediary organizations between workers’ (and other employees’) interests and left parties in government by learning whether unions are able to effectively push the left to satisfy their members’ preferences. We proceed as follows: In the next section, we discuss theoretically how the social policy preferences of union members should look like, why labour unions should care about their members’ preferences and why left parties should be responsive to labour unions. Subsequently, we present a two-step empirical analysis. Firstly, we analyse the social policy preferences of union members. Secondly, we investigate the conditional effect of labour power on left parties in different welfare state programmes. The final section concludes. 2THEORY Before outlining our theoretical arguments on union members’ preferences and unions’ impact on left parties’ social policy-making, we discuss our choice of welfare state programmes. In order to distinguish between the wide range of programmes, we primarily lean on Jensen (2012a, b), who separates labour market-related from life course-related programmes without a strong link to the labour market. In addition, we further subdivide these categories and make use of the differentiation between income protection and activation measures (see, e.g. Häusermann, 2012). As we elaborate in the following, these dimensions allow deducing arguments on the social policy preferences of labour unions and their members. In order to paint a broad picture of welfare state policies, our study includes six programmes. With regard to labour market-related programmes, we investigate employment protection as well as passive and active labour market policies. Concerning less-labour market-related policies,2we look at old-age pensions, health care and education. This choice of programmes allows investigating a wide spectrum of welfare state areas and gives an idea of what kind of programmes union members exactly prefer and what unions should stand up for in their interaction with left parties. While studying the relationship between labour unions, left parties and the welfare state at the macro-level is popular among scholars, connecting such cross-country analyses to their micro-foundations is largely unexplored. However, questions such as ‘Which social policies do union members actually prefer?’ and ‘Why should left parties in government be responsive to the demands of labour unions at all?’ are relevant questions that need to be addressed before investigating the interplay between labour power and left parties. To this end, we assume that unions have a certain self-interest: They are member-seeking and -maintaining and, thus, have to act in accordance with their (potential) members’ preferences in order to claim credit for their implementation (see also Jensen, 2012a). Even if unions prefer more generous welfare state policies in general, they face a first trade-off where to employ their own scarce human resources as well as a second trade-off concerning the expansion of which programmes to demand from their left allies in government, who, in turn, have to deal with fiscal constraints. Hence, we expect unions to focus on those social policy fields that are of prior interest to their members, namely programmes that are related to their labour market status. To be clear, unions are not necessarily unitary actors and union members may very well have different social policy preferences. On the one hand, for example, Arndt (2018) argues that the preferences of white-collar workers should differ from those of blue-collar workers (against vs. in favour of more redistribution and intervention) and, thus, there are different union federations representing these varying preferences resulting from differences in the composition of their 92 BRITISH JOURNAL OF INDUSTRIAL RELATIONS membership. Similarly, Nijhuis (2009, 2011) reports that the union structure affects its position towards social policy. On the other hand, we know that at least parts of the white-collar group (socio-cultural (semi-)professionals) tend to support redistribution (Kitschelt & Rehm, 2014)or share the same policy positions regarding cuts in public spending as blue-collar workers (Engler & Zohlnhöfer, 2019), and we know that union members – even when controlling for income or public employment – support more redistribution (e.g. Häusermann et al., 2016; Kitschelt & Rehm, 2014) and the generosity of programmes that rather target low-income groups, such as unemployment benefits or subsidies (Busemeyer & Neimanns, 2017; Gingrich & Ansell, 2012). Moreover, we argue that differences between the social policy preferences of blueand white-collar union members may only exist for labour market-related programmes rather than for programmes that are not primarily labour marketbut rather life course-related. While both groups should benefit from an expansion of the latter, the former particularly target the needs of the more vulnerable blue-collar workers (see also Svallfors, 2004). Furthermore, it is also possible that union membership comes along with the internalization of egalitarian norms – even among their white-collar, high-income members (Mosimann & Pontusson, 2017). Turning to the question on union members’ preferences, we fall back on the dimensions discussed before. This allows painting a more fine-grained picture than previous literature’s finding that union members generally demand for more generous social policies (e.g. Busemeyer & Neimanns, 2017; Gingrich & Ansell, 2012): We argue that union members (compared to nonmembers) prefer those social policy programmes that are the closest to their organizational affiliation.3As unions particularly operate in labour market-related policy fields, we expect their members to have strong preferences regarding passive as well as active labour market policy and regarding employment protection legislation. Conversely, the impact of union membership should be rather small when it comes to not primarily labour marketbut rather life course-related policies, expecting such issues to be equally relevant for both members and non-members. However, when it comes to policies without a strong link to the labour market, we expect different logics for the three programmes under investigation, as it should be additionally relevant whether a measure aims at income protection or activation. The argument being that union members are interested in securing their economic status quo. Accordingly, we expect that those policies of this category that provide social security to individuals in need most directly and that ensure future benefits in the case of need are characterized by a stronger support of union members than activation programmes. This should be particularly the case for old-age pensions but also for health care. While the former provides future benefits in general and income protection in particular, the latter provides benefits in the case of need, too, but with no element of income protection.4In contrast, activation policies, such as education, provide no or rather few future benefits for union members, as most of them have completed their training periods in the past. Thus, their selfinterest in encompassing education policies should be rather low. In sum, we expect union members to favour labour market-related programmes over less-labour market-related programmes with future benefits over less-labour market-related programmes without future benefits. Correspondingly, our micro-level hypothesis reads as follows: The effect of union membership is the strongest when it comes to passive and active labour market policy as well as employment protection and it is stronger for old-age pensions and (to a smaller extent) health care than for education. Why and by whom should these preferences be translated into policies? Our argument builds on the strong link between the microand macro-level, namely individuals and their organization in labour unions on the one hand and left parties on the other hand. There is a broad literature on the disintegrating link between the working-class and the left, identifying two trends: First, the number of (both organized and non-organized) workers rapidly declines due to changes in THERE IS POWER IN A UNION? 93 the economy and the occupational structure (e.g. Oesch, 2006). Second, the social structure of left parties’ electorate has changed with the share of workers decreasing over the last decades (e.g. Kitschelt, 1999; Oesch & Rennwald, 2018). As a consequence, and assuming that left parties are primarily office-seeking because they can influence policies more effectively when being in office, they have to mobilize more than only their (former) core constituency. Accordingly, a priori, left parties might have an electoral incentive to implement those welfare state policies that not only address the needs of workers and other employees but that appeal to broad voter segments (see also Jensen, 2012a). Still, we expect the effect of left parties on social policies to be conditioned by workers’ organizational representation for several reasons: Firstly, these two organizations have been found as natural allies, with the left representing the interests of workers and other employees for decades (e.g. Garrett, 1998; Korpi, 1983; Stephens, 1979). Secondly, union members have just until recently been more supportive (voters) of left parties (e.g. Arndt & Rennwald, 2016; see also Ramiro (2016) for the support of radical left parties). Hence, from a voteand office-seeking perspective, it is electorally rewarding for the left to be responsive to unions’ social policy claims even in times of socio-structural changes within their electorate. Thirdly, the literature also shows that strong unions push left parties to adopt more leftist policy positions (Pontusson & Rueda, 2010), and it is plausible to assume that, ultimately, these more leftist positions result in more leftist policies when the left is holding office. Therefore, in accordance with our micro-level expectations and based on the assumption that labour unions transfer their members’ preferences into the political arena,5we argue that the conditional effect of unions on left parties differs between labour market-related and not primarily labour market-related policies. For the former group of policies, we hypothesize that the effect of left parties increases with labour union power. This should hold true for passive and active labour market policy as well as for employment protection legislation. While some scholars argue that labour unions mostly represent labour market insiders (Rueda, 2005, 2006) and, hence, should particularly push for employment protection of regular contracts, others expect the representation of outsider interests to increase with union density (Becher & Pontusson, 2011; Martin & Swank, 2012;Swank,2014), which should, in turn, rather result in a conditional effect of labour power on left parties when it comes to the employment protection of temporary contracts. Yet, we expect no substantial differences between these two categories of employment protection, arguing that, first and foremost, employment protection legislation is a regulatory policy.6Thus, the trade-off regarding invested resources is smaller than when it comes to (re-)distributive policies. Accordingly, in theory, employment protection can be improved for both regular and temporary contracts simultaneously. Conversely, we hypothesize that the effect of left parties is not conditioned by labour union power when it comes to less-labour market-related policies, as we expect such programmes to be of minor concern to union members. In addition, unions facing a trade-off where to employ their own resources should rather focus on policies that are most directly related to their organizational raison d’être than on policies that are less targeted on their members. 3MICRO-LEVEL ANALYSIS: UNION MEMBERS’ SOCIAL POLICY PREFERENCES 3.1 Method and data We test our micro-level hypothesis for 13 OECD countries7using all waves of the International Social Survey Programme’s (ISSP) Role of Government data set. We use five waves, namely 1985, 1990, 1996, 2006 and 2016 (ISSP, 1986,1992,1999, 2008,2018). 94 BRITISH JOURNAL OF INDUSTRIAL RELATIONS TABLE 1 Results of the pooled ordered logit regressions Passive labour market policy Active labour market policy Health care Pension Education Union member 0.361*** 0.247*** 0.248*** 0.181*** 0.097** (0.067) (0.026) (0.046) (0.051) (0.042) Sex –0.162*** –0.267*** –0.285*** –0.198*** –0.086** (0.047) (0.042) (0.054) (0.046) (0.039) Age 0.004*** –0.000 0.002** 0.008*** –0.001 (0.001) (0.001) (0.001) (0.003) (0.001) Religion –0.204*** –0.084 –0.196*** 0.147*** –0.212*** (0.064) (0.063) (0.033) (0.043) (0.055) Income –0.240*** –0.271*** –0.158*** –0.219*** –0.012 (0.031) (0.027) (0.033) (0.029) (0.023) Education –0.092*** –0.176*** –0.081*** –0.197*** 0.097*** (0.032) (0.025) (0.028) (0.030) (0.021) Observations 34,743 34,508 35,418 35,003 35,217 Pseudo R20.061 0.061 0.049 0.047 0.036 Note:*=p<0.1; ** =p<0.05; *** =p<0.01; standard errors clustered by country in parentheses; cut-off points, country dummies and wave dummies not reported. As our investigation units are individuals, our dependent variables are their social policy preferences. More precisely, we use ISSP questions with five response categories asking whether the government should spend more or less for health care, pension, education and unemployment benefits. For easier interpretation, we re-code the categories originally ranging from 1 (spend much more) to 5 (spend much less) into 1 (spend much less) to 5 (spend much more). The most accurate indicator to operationalize active labour market policy is whether it is the government’s responsibility to provide a job for everyone.8We also re-code the four categories of this indicator in a more intuitive way from 1 (definitely should not be) to 4 (definitely should be). Unfortunately, there are no data available regarding employment protection legislation preferences. Our independent variable is an individual’s present labour union membership operationalized as a dummy variable (1 =member; 0 =non-member). We include several control variables. We control for age in years. Sex is added as a dummyvariable (1 =male; 0 =female). Education is measured by an individual’s degree in several categories. We test for the effect of religiosity by adding a dummy variable with the value 0 if individuals do not attend any religious services and the value 1 if individuals do so. Additionally, we included a country-specific standardized (z-transformed) income variable.9 As our dependent variables are ordinal, we run pooled ordered logit regressions.10 All models include country as well as wave dummies and robust standard errors clustered by country. 3.2 Empirical results What do the social policy preferences of union members actually look like? Table 1shows the results of the pooled ordered logit regressions. Firstly, we have a closer look at the control variables, which either show the expected sign or do not reach statistical significance: Women more THERE IS POWER IN A UNION? 95 strongly support generous social policies, as they have a higher labour market vulnerability and a higher risk to become dependent on social security programmes (see also Häusermann et al., 2016). High income negatively affects preferences for more intervention as those individuals are more independent from the welfare state. The same holds true for a higher educational degree. Reversely, a higher age comes along with stronger welfare state preferences. In sum, these effects corroborate recent literature on preference formation, in which these findings are explained by self-interest as those people with lower educational degrees, lower income, higher age and women have a higher risk to become dependent on social programmes. With regard to our main variable of interest, union membership has a statistically significant positive effect on individual preferences in all policy fields. Accordingly, union members favour generous social policies more strongly than non-members. These findings are perfectly in line with previous studies reporting that union membership affects individuals’ preferences on unemployment support, job creation or health care (e.g. Busemeyer & Neimanns, 2017; Gingrich & Ansell, 2012; Häusermann et al., 2016).11 In order to investigate our micro-level hypothesis more explicitly, we check whether our expected order of preferences can be corroborated, that is the impact of union membership on programmes closely connected to the labour market is stronger than on less labour market-related programmes and that the effect on less-labour market-related programmes is stronger for those that provide future benefits than for those that do not. To this end, we estimate average marginal effects of union membership for binary logistic regression models. We dichotomize our dependent variablescoding 1 for the categories ‘spend more’and ‘spend much more’and coding 0 for the categories ‘spend less’ and ‘spend much less’ for the indicators on passive labour market policy, health care, pensions and education. For the active labour market policy indicator, the response options ‘probably should be’ and ‘definitely should be’ are coded 1 and the options ‘probably should not be’ and ‘definitely should not be’ are coded 0. Accordingly, the number of observations drops substantially for most indicators, as we exclude the middle category (‘spend the same’). We decided to exclude this category when focusing on differences between the various programmes for several reasons: Firstly, while all but one of the original dependent variables include a middle category, the original active labour market policy indicator does not. Thus, our coding decision leads to a better comparability between the programmes. Secondly, we argue theoretically that the middle category has no substantial claim for our analysis and should not shift partisan effects and unions’ actions. Thirdly, we follow existing literature due to comparability (see, e.g. Häusermann et al., 2016), and fourthly, presenting average marginal effects from logistic regressions allows for an easier and more intuitive presentation. Figure 1shows the average marginal effects of union membership and their confidence intervals in different programmes. All coefficients are statistically significant, indicating that union membership enhances the chance to be in favour of more generous state intervention for all social programmes (for the regression results, see online appendix Table A2). However, the size of the average marginal effect differs substantially across programmes. Regarding passive labour market policy, union membership increases the chance to be in favour of more spending by about 12 percentage points, while for active labour market policy, it is increased by 6 percentage points. Conversely, the average marginal effect of union membership is much smaller for health care, pensions and education, ranging between 1 and 3 percentage points. Interestingly, the confidence intervals of active and passive labour market policy do not overlap, with the coefficient for passive labour market policy being significantly higher. This result indicates that union members focus on income replacement more than on activation, even in the field of labour market-related programmes. Furthermore, the confidence intervals of these two programmes do not overlap 96 BRITISH JOURNAL OF INDUSTRIAL RELATIONS FIGURE 1 Marginal effect of union membership on individuals’ social policy preferences [Colour figure can be viewed at wileyonlinelibrary.com] Note: The average marginal effects are taken from the regression results in Table A2 in the online appendix. The lines show the 95 per cent confidence intervals. Abbreviations: ALMP =active labour market policy; PLMP =passive labour market policy with the programmes being not closely related to the labour market. This corroborates our micro-level hypothesis. Indeed, union membership has a stronger effect on active and passive labour market policies than on health care, pension and education. Moreover, results substantiate the hypothesized differences between the latter three programmes: Non-overlapping confidence intervals show that union membership has a significantly stronger effect on preferences for old-age pensions (future benefits and income protection) than on preferences for education (no future benefits and activation). In contrast, the confidence interval for health care preferences overlaps with the interval of pension and education, which perfectly reflects our line of reasoning that the income-protective element of this programme is much smaller than that of old-age pensions. In sum, the order of preferences in Figure 1fits perfectly with the existing literature: The median voter, indeed, has higher preferences for health care as the risk to become in need of that measure is not predictable. At the same time, health care and other life course-related programmes, such as education, for middleand high-income earners are (almost) as popular as for low-income earners. Thus, the non-existing or less clear differentiation for life course-related programmes as well as the gap between these programmes and labour-market-related programmes reflects the theoretical arguments and the macro-level results in other studies, most notably Jensen (2012b). As robustness checks, we specify several re-estimations (see online appendix Tables A3–A12). Firstly, we use a different measurement of religiosity (belonging to a religious group). We see that the effects remain the same and that religious respondents tend to ask for less generosity. Secondly, we additionally include partisan left-right orientation (scaled from 1 =far left to 5 =far right). Results remain almost the same but the positive effect for education misses conventional significance levels, which at least corroborates our theoretical expectations regarding the lower THERE IS POWER IN A UNION? 103 against life course-related risks (Jensen, 2012a, b), sick pay follows a different pattern. Here, the conditional relationship between unions and left parties is positive with the left unfolding a positive effect on this programme at high levels of union density. This is not surprising given that sick pay protects wage earners’ income and, in doing so, it is much closer to the needs of organized (and non-organized) workers. Accordingly, sick pay rather reflects the pattern of labour market-related policies. Overall, the reported findings essentially corroborate our macro-level hypotheses: Labour unions, indeed, condition left parties’ effect in particular parts of the welfare state, that is when it comes to labour market-related programmes but not when it comes to less-labour market-related programmes. To assess the sensitivity of these results, we run several robustness checks (see online appendix Tables A14–A24): Firstly, we change our labour power indicator in order to study whether unions condition left parties via mechanisms other than their organizational degree. To this end, we take Ghent systems, strike activity and the share of union members among left party voters into account. Concerning Ghent systems, it is possible to argue that unions should only be able to push left parties to provide generous labour market-related programmes when they are legally guaranteed administrative and decision-making powers in these areas. Concerning strike activity, one could argue that unions’ willingness to promote their members’ interests by striking affects left parties’ policy-making, too. Concerning the share of union members among left party voters, we fell back on data from all ISSP waves and Eurobarometer. This indicator should relate our micro-level discussion more closely to the macro-level, expecting left parties to be the more responsive to the demands of unions and their members’ preferences, the higher the share the latter make up among the left’s electorate. Empirically, we can only substantiate expectations for Ghent systems and the micro-level indicator: Labour unions’ access to (and responsibility for) the administration of welfare state programmes as well as the share of union members among left party voters are alternative mechanisms that allow unions and their members pushing left parties towards more generous labour market-related social policies. Secondly, we test whether our findings are robust when using a broader definition of left parties, additionally including the share of green cabinet members. Here, we expect that the conditional effect of labour unions on left parties is weaker because green parties should be less close (and, therefore, less responsive) to workers’ organizations than parties of the ‘old left’. Thirdly, we extend our main models by including additional controls: We take deindustrialization processes, the role of female labour force participation and Ghent systems into account. Fourthly, we control for country-specific effects by including country dummies, which is not implausible when analyzing several cabinets within each of the 21 countries. Fifthly, we re-estimate the main models for the smaller micro-level sample, which could be argued to be more appropriate when comparing the two steps of our empirical analysis. Sixthly, we re-run our main models for the more commonly used country-years as unit of observation. To this end, we specify time-series cross-country regression models with panel-corrected standard errors and both with and without country and year fixed effects. In sum, these sensitivity analyses support our main findings for both labour market-related and less-labour market-related programmes. Finally, we split our sample in two sub-periods. Our findings show that the conditional effect of unions concerning labour market-related welfare state programmes has vanished over time: While labour unions were successful in pushing left parties to promote their members’ main social policy preferences up until the mid-1990s, they seem to have lost this capacity ever since. Yet, this finding may not necessarily solely be related to the decrease in the number of union members over the past decades but could also be related to several other developments: Firstly, the tremendous changes in left parties’ electorates with decreasing shares of working-class voters 104 BRITISH JOURNAL OF INDUSTRIAL RELATIONS (e.g. Gingrich & Häusermann, 2015; Oesch & Rennwald, 2018) might have negatively affected the left’s responsiveness to workers’ demands and shifted the lefts’ attention towards other groups. Secondly, social democratic parties themselves have changed ideologically from supporting traditional social security instruments to promoting cross-class policy measures or even engaging in retrenchment reforms (e.g. Merkel et al., 2011). Lastly, recent socio-economic challenges, such as permanent austerity or globalization, limit the political actors’ fiscal room for manoeuvere, which might have made it difficult for left parties to be responsive to unions’ demands as well. What is more, concerning programmes that are not primarily related to the labour market, we find positive conditional effects in the post-1995 sub-period for health care and education spending. This evidence, in turn, might hint to changes in the social structure of unions’ membership or to efforts of union leaderships seeking new members, especially middleor high-income individuals that should demand these programmes rather than class-based issues, such as unemployment benefits. 5 CONCLUSION This study assessed the role of labour unions in different parts of the welfare state in advanced democracies, answering the two questions which social policies union members actually prefer and whether unions (still) effectively influence left parties in government. We did so by connecting the micro-level of individuals’ preferences to the macro-level of governments’ policy-making in several labour market-related programmes (passive and active labour market policy, and employment protection legislation) and less-labour market-related programmes (old-age pensions, health care and education). Overall, our empirical analyses provide evidence for our theoretical arguments: At the microlevel, we show that union members have stronger preferences for more state interventions across all social policies under investigation than non-members. More specifically, we find that the effect of union membership is stronger for labour market-related programmes than for those that are not primarily labour market-related. Moreover, our findings provide evidence that union members prefer the increase of less-labour market-related programmes with future benefits (oldage pensions) over the increase of programmes without future benefits (education). For the macro-level, we can report that labour unions, indeed, condition left parties’ social policy-making when it comes to labour market-related policies. At least regarding passive and active labour market measures, we find that strong unions push the left towards more generous policies. Yet, our analysis of sub-periods indicates that this conditional relationship has vanished in recent years. Conversely, labour power has largely no such effect when it comes to programmes that are less relevant for union members, especially health care and education. These findings have several implications for our understanding of the political role of labour unions: Firstly, this article identifies important differences between welfare state programmes. Regarding individual-level preferences, it goes beyond previous research, which has reported in separate contributions that union membership affects some individual programmes. In contrast, our study not only corroborates that this holds true across major parts of the welfare state but that the degree of union members’ social policy preferences actually varies between issues. Regarding the macro-level of governments’ policy-making, we show that other studies’ findings on the interplay between labour unions and left parties for aggregate social spending (e.g. Garrett, 1998; Hicks et al., 1989; Kwon & Pontusson, 2010) are essentially driven by labour market-related policy instruments and are not representative for all parts of the welfare state. Secondly, and more THERE IS POWER IN A UNION? 105 generally, our results imply that labour unions have been powerful intermediary organizations and effective promoters of their members’ interests by influencing their left allies in government in the past. However, our results for sub-periods indicate that their power is vanishing – even though our micro-level results show that individual preferences of union members are perfectly stable over time. Finally, we put forward some subjects that still need to be addressed in more depth. Starting at the micro-level, further research should take a closer look at the possible reversed causality regarding social policy preferences and union membership. Analyses focusing on the temporal dimension and even in-depth interviews could shed light on the causal relationship. Second, the interplay between labour unions and left parties surely is not confined to the core of welfare state policies. At least with regard to economic policy instruments, we know that union members’ support for industrial aid, market regulation or state ownership of enterprises is stronger than among non-members (Engler & Zohlnhöfer, 2019; Gingrich & Ansell, 2012). Hence, future research should go beyond the social policy-centred view that we took here by studying other policy fields. Third, recent decades were not only characterized by a substantial decline in union membership across almost all advanced democracies but also by tremendous changes in both left parties’ electorates and the national and international socio-economic conditions. In addition, two changes are observed by Häusermann (2006): On the one hand, ‘old’ social risks are increasingly accompanied by ‘new’ social risks that change the policy outputs. On the other hand, this leads to changing alliances between unions and left parties regarding the expansion or retrenchment of the different policies. While it is well established that these developments affect left parties’ social policy-making in general, we still know too little about their consequences for labour unions’ capacity to push left parties effectively towards more generous policies. ACKNOWLEDGEMENTS We thank Reimut Zohlnhöfer for his valuable comments on our paper. In addition, we thank Lisa Galvagno, Johanna Hornung and Philipp Mai for their impulses. Open Access funding enabled and organized by Projekt DEAL. DATA AVAILABILITY STATEMENT Data that support the findings of this study are available upon request. FUNDING Fabian Engler gratefully acknowledges funding by the German Research Foundation (DFG, grant number ZO 126/5-1). ORCID FabianEngler https://orcid.org/0000-0002-4253-2481 LindaVoigt https://orcid.org/0000-0003-4621-3979 ENDNOTES 1We conceptualize left parties as those parties that historically have strong ties to the labour movement, that is social democrats and socialists. Häusermann (2010) and others call this group of parties the ‘old left’ in contrast to the ‘new left’ that is much more libertarian and that is much less characterized by working-class voters, such as green parties. 2Terms like ‘less’ or also ‘not primarily’ are rather simplistic and are used for convenience only. Of course, pension generosity usually depends on an individual’s previous position in the labour market. Not least, programmes may 106 BRITISH JOURNAL OF INDUSTRIAL RELATIONS be financed by taxes or social security contributions with the latter being relatively closer to both individuals’ labour market position and labour market developments in general. Nonetheless, we argue that the social risks that pensions, health care and education address are rather life course-related (see also Jensen 2012a, b) and less closely related to the labour market than passive and active labour market policies or employment protection. 3One could object that individuals should have certain social policy preferences before becoming union members and that these preferences are the reason for their membership in the first place (e.g. Checchi et al. 2010; Cusack et al. 2006). Yet, as others have shown (see, e.g. Ahlquist & Levi 2013;Kim&Margalit2017; Pontusson & Rueda 2010), union membership itself transforms or at least sharpens individuals’ preferences through socialization processes or the provision of information. We assess the possibility of reversed causality in more detail in the micro-level analysis. 4Of course, sick pay is an important but fiscally rather minor part of health care policies aiming at income protection of individuals being currently employed. Hence, this part of health care should rather follow the pattern of labour market-related programmes. We come back to this argument in our macro-level analysis. 5Of course, unions do not only function as ‘transmission belts’ (Jensen 2012a: 221) but also act in their own self-interests (see also Pontusson 2013: 799). However, our assumption is based on Jensen’s argument (2012a: 221): Unions, and especially union leaders, are interested in gaining and maintaining members, as spoils are provided by a large membership, which is similar to the benefits of political parties when gaining office. To achieve this goal, unions fight effectively for their current and future members’ preferences in the visible political arena. Additionally, unions failing to represent workers’ interests may face decreasing membership and, in turn, lower organizational power resources. Empirically, previous studies have shown that union members not only favour redistribution (e.g. Häusermann et al. 2016; Kitschelt & Rehm 2014; Mosimann & Pontusson 2017)but that unions successfully represent these preferences in the political arena, especially up until the 1990s (Iversen & Soskice 2006; Pontusson 2013). 6Of course, employment protection includes (re)distributive elements, such as severance payments, too. However, relying on the OECD’s (2019) Employment Protection Legislation indicators’ dimensions, there are more regulatory elements included, such as notification procedures in the case of individual dismissal of a worker with a regular contract, delay involved before notice can start and a definition of unfair dismissal. 7Belgium, Denmark, Finland, France, Germany, Japan, New Zealand, Norway, Spain, Sweden, Switzerland, the United Kingdom and the United States. The exact sample of countries varies between the different Role of Government waves (see online appendix Tables A8, A9). 8Although this indicator is suboptimal, it is the most appropriate one in the Role of Government questionnaire for our research interest. We argue that the provision of a job includes offering training programmes and other active labour market policies in order to provide a job for everyone. 9Descriptive statistics for the independent variables are summarized in online appendix Table A25. 10 We refrain from applying multi-level analyses because our small sample of countries could bias confidence intervals (see Stegmueller 2013 for details). 11 We assess the possibility of reversed causality by running logistic regressions with union membership as dependent variable and social policy preferences as independent variable. Our findings suggest that we cannot rule out that causality is reversed (see online appendix Table A1), as preferences for more generous state interventions are statistically significant determinants of union membership. By far, however, this result is no unequivocal proof of reversed causality because there is no temporal dimension in our data that allows concluding that the social policy preferences antecede union membership. 12 Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States. Aiming at the inclusion of the most comprehensive sample of established and advanced OECD democracies, this sample is larger than the micro-level sample. For reasons of comparability, we discuss the macro-level results for the sample of the micro-level in the robustness checks. 13 The period of observation is limited due to data availability. The main models cover the longest possible periods for each indicator. 14 Online appendix Table A13 provides more detailed variable descriptions and the data sources and Table A26 reports descriptive statistics. THERE IS POWER IN A UNION? 107 REFERENCES Ahlquist, J. & Levi, M. (2013) In the interests of others: leaders, governance, and political activism in membership organizations. Princeton, NJ: Princeton University Press. Arndt, C. (2018) White-collar unions and attitudes towards income inequality, redistribution, and state-market relations. European Sociological Review, 34(6), 675–693. Arndt, C. & Rennwald, L. (2016) Union members at the polls in diverse trade union landscapes. European Journal of Political Research,55(4),702–722. Becher, M. & Pontusson, J. (2011) Whose interests do unions represent? Unionization by income in Western Europe. In Brady, D. (Ed.). Comparing European workers. Part B: Policies and institutions. Bingley: Emerald Group Publishing Limited, pp. 181–211. Boix, C. (1997) Privatizing the public business sector in the eighties: economic performance, partisan responses and divided governments. British Journal of Political Science, 27(4), 473–496. Busemeyer, M. R. & Neimanns, E. (2017) Conflictive preferences towards social investments and transfers in mature welfare states: the cases of unemployment benefits and childcare provision. Journal of European Social Policy, 27(3), 229–246. Checchi, D., Visser, J. & van de Werfhorst, H.G. (2010) Inequality and union membership: the influence of relative earnings and inequality attitudes. British Journal of Industrial Relations, 48(1), 84–108. Cusack, T., Iversen, T. & Rehm, P. (2006) Risks at work. The demand and supply sides of government redistribution. Oxford Review of Economic Policy,22(3),365–389. Engler, F. (2021) Political parties, globalization, and labor strength: assessing differences across welfare state programs. European Journal of Political Research, 60(3), 670–693. Engler, F. & Zohlnhöfer, R. (2019) Left parties, voter preferences, and economic policy-making in Europe. Journal of European Public Policy, 26(11), 1620–1638. Esping-Andersen, G. (1990) Three worlds if welfare capitalism. Princeton, NJ: Princeton University Press. Finseraas, H. (2009) Income inequality and demand for redistribution: a multilevel analysis of European public opinion. Scandinavian Political Studies,32(1),94–119. Garrett, G. (1998) Partisan politics in the global economy. Cambridge: Cambridge University Press. Garritzmann, J.L. & Seng, K. (2016) Party politics and education spending: challenging some common wisdom. Journal of European Public Policy,23(4),510–530. Gingrich, J. & Ansell, B. (2012) Preferences in context: micro preferences, macro contexts, and the demand for social policy. Comparative Political Studies,45(12),1624–1654. Gingrich, J. & Häusermann, S. (2015) The decline of the working-class vote, the reconfiguration of the welfare support coalition and consequences for the welfare state. Journal of European Social Policy,25(1),50–75. Häusermann, S. (2006) Changing coalitions in social policy reforms: the politics of new social needs and demands. Journal of European Social Policy, 16(1), 5–21. Häusermann, S. (2010) Solidarity with whom? Why organised labour is losing ground in continental pension politics. European Journal of Political Research, 49(2), 223–256. Häusermann, S. (2012) The politics of new and old social risks. In Bonoli, G. & Natali, D. (Eds.). The politics of the new welfare state. Oxford: Oxford University Press, pp. 111–132. Häusermann, S., Kurer, T. & Schwander, H. (2016) Sharing the risk? Households, labour market vulnerability, and social policy preferences in Western Europe. Journal of Politics, 78(4), 1045–1060. Hicks, A.M., Swank, D. H. & Ambuhl, M. (1989) Welfare expansion revisited: policy routines and their mediation by party, class and crisis. European Journal of Political Research, 17(4), 401–430. ISSP. (1986) International Social Survey Programme: Role of Government I - ISSP 1985. Version 1.0.0. GESIS Data Archive, Cologne. ISSP. (1992) International Social Survey Programme: Role of Government II - ISSP 1990. Version 1.0.0. GESIS Data Archive, Cologne. ISSP. (1999) International Social Survey Programme: Role of Government III – ISSP 1996. Version 1.0.0. GESIS Data Archive, Cologne. ISSP. (2008) International Social Survey Programme: Role of Government IV – ISSP 2006. Version 1.0.0. GESIS Data Archive, Cologne. 108 BRITISH JOURNAL OF INDUSTRIAL RELATIONS ISSP. (2018) International Social Survey Programme: Role of Government V – ISSP 2016. Version 2.0.0. GESIS Data Archive, Cologne. Iversen, T. & Soskice, D. (2001) An asset theory of social policy preferences. American Political Science Review, 95(4), 875–893. Iversen, T. & Soskice, D. (2006) Electoral institutions and the politics of coalitions: why some democracies redistribute more than others. American Political Science Review, 100(2), 165–181. Jensen, C. (2012a) Two sides of the same coin? Left-wing governments and labour unions as determinants of public spending. Socio-Economic Review, 10(2), 217–240. Jensen, C. (2012b) Labour marketversus life course-related social policies: understanding cross-programme differences. Journal of European Public Policy, 19(2), 275–291. Kim, S.E. & Margalit, Y. (2017) Informed preferences? The impact of unions on workers’ policy views. American Journal of Political Science,61(3),728–743. Kitschelt, H. (1999) The transformation of European social democracy. Cambridge: Cambridge University Press. Kitschelt, H. & Rehm, P. (2014) Occupations as a site of political preference formation. Comparative Political Studies, 47(12), 1670–1706. Korpi, W. (1983) The democratic class struggle. Boston, MA: Routledge & Kegan Paul. Kwon, H.Y. & Pontusson, J. (2010) Globalization, labour power and partisan politics revisited. Socio-Economic Review, 8(2), 251–281. Martin, C.J. & Swank, D.H. (2012) The political construction of business interests. Coordination, growth, and equality. Cambridge: Cambridge University Press. Merkel, W., Petring, A., Henkes, C. & Egle, C. (2011) Social democracy in power. The capacity to reform.London: Routledge. Mosimann, N. & Pontusson, J. (2017) Solidaristic unionism and support for redistribution in contemporary Europe. World Politics, 69(3), 448–492. Nijhuis, D.O. (2009) Revisiting the role of labor. Worker solidarity, employer opposition, and the development of old-age pensions in the Netherlands and the United Kingdom. World Politics, 61(2), 296–329. Nijhuis, D.O. (2011) Explaining British voluntarism. Labor History,52(4),373–398. Obinger, H., Schmitt, C. & Zohlnhöfer, R. (2014) Partisan politics and privatization in OECD countries. Comparative Political Studies, 47(9), 1294–1323. OECD. (2019) Detailed Description of Employment Protection Legislation, 2019 OECD countries. Available at: https://www.oecd.org/els/emp/All%202019.pdf [Accessed 17th March 2021). Oesch, D. (2006) Coming to grips with a changing class structure. An analysis of employment stratification in Britain, Germany, Sweden and Switzerland. International Sociology,21(2),263–288. Oesch, D. & Rennwald, L. (2018) Electoral competition in Europe’s new tripolar political space: class voting for the left, centre-right and radical right. European Journal of Political Research, 57(4), 783–807. Pontusson, J. (2013) Unionization, inequality and redistribution. British Journal of Industrial Relations,51(4),797– 825. Pontusson, J. & Rueda, D. (2010) The politics of inequality: voter mobilization and left parties in advanced industrial states. Comparative Political Studies, 43(6), 675–705. Ramiro, L. (2016) Support for radical left parties in Western Europe: social background, ideology and political orientations. European Political Science Review, 8(1), 1–23. Rehm, P. (2009) Risks and redistribution. An individual-level analysis. Comparative Political Studies, 42(7), 855–881. Ross, F. (2000) “Beyond left and right”: the new partisan politics of welfare. Governance,13(2),155–183. Rueda, D. (2005) Insider-outsider politics in industrialized democracies. American Political Science Review,99(1), 61–74. Rueda, D. (2006) Social democracy and active labour-market policies: insiders, outsiders and the politics of employment promotion. British Journal of Political Science, 36(3), 385–406. Scharpf, F.W. (1999) Governing in Europe: effective and democratic? Oxford: Oxford University Press. Schmitt, C. (2016) Panel data analysis and partisan variables: how periodization does influence partisan effects. Journal of European Public Policy, 23(10), 1442–1459. Schmitt, C. & Zohlnhöfer, R. (2019) Partisan differences and the interventionist state in advanced democracies. Socio-Economic Review, 17(4), 969–992. THERE IS POWER IN A UNION? 109 Simoni, M. & Vlandas, T. (2021) Labour market liberalization and the rise of dualism in Europe as the interplay between governments, trade unions and the economy. Social Policy & Administration, 55(4), 637–658. Stegmueller, D. (2013) How many countries for multilevel modeling? A comparison of Frequentist and Bayesian approaches. American Journal of Political Science,57(3),748–761. Stephens, J.D. (1979) The transition from capitalism to socialism. Urbana and Chicago, IL: University of Illinois Press. Svallfors, S. (2004) Class, attitudes and the welfare state: Sweden in comparative perspective. Social Policy & Administration,38(2),119–138. Swank, D.H. (2014) The political sources of labor market dualism in postindustrial democracies, 1980–2011. Review of Keynesian Economics, 2(2), 234–257. Swank, D.H. (2020) The partisan politics of new social risks in advanced postindustrial democracies: social protection for labour market outsiders. In: Careja, R., Emmenegger, P. & Giger, N. (Eds.). The European social model in times of crisis. Wiesbaden: Springer VS, pp. 139–157. Zohlnhöfer, R., Wolf, F. & Wenzelburger, G. (2013) Political parties and pension generosity in times of permanent austerity. World Political Science Review, 9(1), 291–318. SUPPORTING INFORMATION Additional supporting information may be found in the online version of the article at the publisher’s website. How to cite this article: Engler, F. & Voigt, L. (2023) There is power in a union? Union members’ preferences and the conditional effect of labour unions on left parties in different welfare state programmes. British Journal of Industrial Relations, 61, 89–109. https://doi.org/10.1111/bjir.12665