Trade Unions and Life Satisfaction in Germany

Becker, Björn; Goerke, Laszlo; Huang, Yue Article — Published Version Trade Unions and Life Satisfaction in Germany British Journal of Industrial Relations Provided in Cooperation with: John Wiley & Sons Suggested Citation: Becker, Björn; Goerke, Laszlo; Huang, Yue (2024) : Trade Unions and Life Satisfaction in Germany, British Journal of Industrial Relations, ISSN 1467-8543, Wiley, Hoboken, NJ, Vol. 63, Iss. 3, pp. 462-477, https://doi.org/10.1111/bjir.12861 This Version is available at: https://hdl.handle.net/10419/330208 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. https://creativecommons.org/licenses/by/4.0/ British Journal of Industrial Relations ORIGINAL ARTICLE Trade Unions and Life Satisfaction in Germany Björn Becker1Laszlo Goerke2Yue Huang3 1Trier University - IAAEU (Institute for Labour Law and Industrial Relations in the European Union), Trier, Germany 2Trier University - IAAEU (Institute for Labour Law and Industrial Relations in the European Union), IZA, Bonn, CESifo, München and GLO 3Trier University - IAAEU (Institute for Labour Law and Industrial Relations in the European Union), and GLO Correspondence: Laszlo Goerke ([email protected]) Received: 17 February 2024 Revised: 23 August 2024 Accepted: 22 November 2024 Keywords: German socio-economic panel | industrial relations in Germany | life satisfaction | trade union membership ABSTRACT The effects of trade union membership on wages and job satisfaction have been studied extensively. Arguably, life satisfaction serves as a more comprehensive measure of the benefits of union membership and warrants closer examination. Using all relevant waves from the German Socio-Economic Panel between 1985 and 2019, we find a negative correlation between trade union membership and life satisfaction in OLS and FE specifications. The association may arise because union members are more concerned about their job and the economic situation and less satisfied with their work. Social capital and wages also perform as channels between membership and life satisfaction. The negative correlation is more pronounced in settings in which trade unions are relatively weak. JEL Classification: I 31, J 28, J 51 1 Introduction Trade unions, like many other interest groups, aim to improve the well-being of their constituency. Since one of central objectives is to raise wages, the wage mark-up represents an important indicator of union success. While this wage gain for union members varies across countries and over time, it is mostly positive (Fang and Hartley 2022) and often substantial enough to exceed the costs due to membership fees. The wage premium may, however, be an inadequate indicator of the well-being consequences of union membership. First, unions affect other working conditions. This impact could be positive if, for example, trade unions bargain for better safety standards. The effect could be negative if higher wages induce employers to reduce nonwage benefits, in order to slow down the increase in labour costs. Second, benefits associated with trade union membership, such as greater vacation entitlements or support in case of conflicts with the employer, may have non-monetary elements. Third, the gains from union membership may be distributed unevenly and non-monetary benefits may accrue to other members than to those who obtain the wage increases. Fourth, for the employees’ decision whether to be a union member, the subjective evaluation of the consequences of membership may be more relevant than objective indicators, such as higher wages or greater fringe benefits. Accordingly, analyses of the impact of trade unions on their members’ well-being have frequently focused on job satisfaction (see, e.g., the meta-analysis by Laroche 2016 and the surveys by Artz and Heywood 2021 and Goerke 2021). However, as it is the case for the union wage mark-up, job satisfaction may also be an insufficient indicator of the entire well-being effects of union membership. First, work-related correlates of union membership can have broader satisfaction effects. Second, employees may trade off a lower utility from the job against a higher non-work utility resulting from union membership, or vice versa. Accordingly, investigating the life satisfaction effects of union membership can offer insights beyond those gained from This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Author(s). British Journal of Industrial Relations published by John Wiley & Sons Ltd. 462 British Journal of Industrial Relations,2025;63:462–477 https://doi.org/10.1111/bjir.12861 examining wages or job satisfaction. Consequently, this paper explores the relationship between trade union membership and life satisfaction. We focus on Germany, the largest economy in the European Union (EU), using data from all waves of the German Socio-Economic Panel (SOEP) between 1985 and 2019 that include information about a respondent’s trade union membership. In Germany, union membership is not tied to a job and, therefore, distinct from coverage by a collective bargaining contract. While a collective agreement binds the firms it covers and the members of the contracting trade union, covered employers usually apply agreements to the entire workforce, irrespective of an individual’s union membership status. In consequence, many employees free-ride on the benefits of union-negotiated bargaining agreements. These features make the German case particularly interesting as the benefits of union membership and the well-being consequences are less obvious than in countries where membership and bargaining coverage are more closely aligned as, for example, in the United States. As our first contribution, we document a robust negative correlation between trade union membership and life satisfaction for pooled cross-sectional data and when considering time-invariant unobservable individual characteristics. This finding is consistent with the results concerning job satisfaction for Germany, but is not in line with the scarce evidence on the life satisfaction effects of union membership for other countries. Previous investigations have put forward several explanations for a positive correlation. If, for example, membership results in a wage differential, this can represent a channel by which membership contributes to life satisfaction. Empirically, the explanation can be assessed by comparing estimation results with and without wage information as a covariate. As our second contribution, we analyse various explanations for the observed negative correlation between union membership and life satisfaction. Our analysis suggests that union members are more concerned about job security and their own economic situation than comparable non-members. These complementary explanations can help to clarify why union members are less satisfied with their life. Moreover, these effects are not fully reflected in differential levels of job satisfaction, suggesting that life satisfaction may be a more comprehensive indicator of the well-being consequences of union membership. We, furthermore, find that union members earn higher wages and have greater social capital than non-members. Controlling for these two complementary aspects strengthens the negative correlation between union membership and life satisfaction. When accounting for all five explanations, we continue to observe a negative association, suggesting the need for further work on the causes of a negative correlation. As our third contribution, we analyse the role of the two central pillars of industrial relations in Germany. The first pillar is collective bargaining. The second pillar is co-determination, particularly through works councils in the private sector and personnel councils in the public sector. Our findings suggest that the negative correlation between union membership and life satisfaction is much weaker or does not arise in surroundings where unions have a more significant impact on working conditions, that is in co-determined firms or those covered by collective bargaining. Instead, it is observable primarily in settings where trade unions are relatively weak. The further paper proceeds as follows: Section 2provides some background information on the German labour market and the role of trade unions. Section 3outlines the theoretical considerations, which constitute the basis for our analysis. Section 4 summarizes the previous literature. In Section 5, we introduce the data and describe the empirical model. Section 6presents our findings, while Section 7concludes the paper. 2Institutional Background In Germany, bargaining coverage in the private sector has declined to 50% in 2021, while it still hovers around 90% in the public domain. Roughly, four out of five employees subject to a collective agreement are covered by one negotiated at the sectoral level, while the remaining 20% benefit from a contract signed by a single firm and a trade union (Ellguth and Kohaut 2022). Moreover, there are substantial differences in coverage rates according to firm size. While about 80% of firms with more than 500 employees have signed a collective agreement or belong to an employer association that has done so, this is true for only 2 out of 3 firms with 200–500 employees and less than 50% with 50–199 staff members (Ellguth and Kohaut 2022). Firms not subject to collective agreements can negotiate wages and working conditions individually with their staff. In about 50% of the relevant contracts, at least pay is aligned to a collective agreement (Hohendanner and Kohaut 2023). Net union density equalled 16.3% in 2019. It has declined substantially over the last decades, from a peak of about 36% just after re-unification in 1990 (OECD and AIAS 2021). Union density among males is about 50% higher than among females. Union members generally pay a (tax-deductible) membership fee of 1% of their gross wage. In exchange, trade unions, for example, provide them with strike pay and advice in case of employment-related conflicts. Furthermore, union members benefit from financial advantages, such as reduced insurance contribution rates or price reductions for some leisure activities. Additionally, there is evidence that union members are better protected against dismissals and able to utilize their vacation entitlements to a greater extent than non-members (Goerke and Pannenberg 2011; Goerke, Jeworrek, and Pannenberg 2015). One of the main benefits of union membership can be a wage premium. There is substantial evidence for the United States and other countries of such a premium of 10% or more (surveyed, e.g., by Fang and Hartley 2022). In contrast, most studies for Germany suggest the absence of wage differences between union members and non-members (see, e.g., Schmidt and Zimmermann 1991, Fitzenberger, Ernst, and Haggeney 1999, and Goerke and Pannenberg 2004). Somewhat in contrast, Wagner (1991)reportsa positive wage effect of union membership for blue-collar workers of about 4.5% and a negative one for white-collar employees of a comparable magnitude, using data for West Germany in 1985. More recently, Bonaccolto-Töpfer and Schnabel (2023)document a union membership wage premium of 2.5% for 2015 and 2019. 463 One reason for the absence of or a relatively small union membership wage effect is that membership is not tied to a job and, therefore, distinct from coverage by a collective bargaining contract. This can already be observed when comparing the bargaining coverage to union density. From a legal perspective, collective bargaining agreements apply to all firms covered by the contract and union members who work in the covered firm. Individuals in covered firms cannot be forced to join a trade union, or to support unions financially. According to Hirsch, Lentge, and Schnabel (2022), more than 90% of employees working in firms covered by a collective bargaining agreement are actually paid according to it. Due to data limitations, the study cannot relate the non-payment of collectively bargained wages to the union status of employees. Nonetheless, the substantial difference between bargaining coverage and union density implies that many employees free-ride on the benefits of union-negotiated bargaining agreements in Germany.1 While (sectoral) collective bargaining constitutes one of the main pillars of the German system of industrial relations, codetermination represents the second such cornerstone (Jäger, Noy, and Schoefer 2022). Works councils can be established in private-sector firms with at least five employees. Works councils have information, consultation and co-determination rights. The latter are more extensive concerning personnel policy and social affairs and less pronounced regarding financial and economic aspects. Co-determination is no substitute but rather a complement to collective bargaining. Works councils are, for example, not allowed to negotiate wages if a collective bargaining determines them and they are forbidden to initiate a strike. Since works councils are not compulsory, they exist in less than 10% of all eligible private-sector establishments. Personnel councils, the public-sector equivalent to works councils, have broadly the same rights and are much more widespread. As works councils are more likely to exist in large than in small establishments, about 40% of all private-sector employees are employed in firms with a works council.2In consequence, the two pillars of the industrial relations system co-exist especially in large establishments, and more than 80% of the employees in plants with a workforce of more than 500 individuals are covered by a collective bargaining agreement and simultaneously represented by a works council (Ellguth and Kohaut 2022). Trade unions and works and personnel councils often closely cooperate and more than half of works council members actually belong to a trade union (Goerke and Pannenberg 2024). However, councils and unions are distinct entities and, therefore, deserve separate consideration. 3Theoretical Considerations 3.1 Individual Union Membership and Life Satisfaction If the union membership decision is based on a rational choice, an employee will be a member if the resulting (expected) utility exceeds the (expected) utility from non-membership (Booth 1985; Booth and Chatterji 1993; Barth et al. 2020;Oswald1982). While utility is difficult to measure, there is extensive evidence on people’s life satisfaction. If such satisfaction measure constitutes a suitable proxy for utility (see, inter alia, Blanchflower and Oswald 2004 and Rayo and Becker 2007), we can expect life satisfaction to increase with trade union membership (see, e.g., Flavin, Pacek, and Radcliff 2010; Flavin and Shufeldt 2016; Keane, Pacek, and Radcliff 2012; Radcliff 2005). Because union members and non-members can differ, for example, in demographic and job-related aspects, which, in turn, impact on life satisfaction, the expected correlation may not always be discernible in cross-sectional analyses. An empirical specification that accounts for time-invariant individual characteristics enables us to relate empirical results and theoretical predictions more closely. Therefore, it may be more likely that individual fixed effects (FE) specifications reveal a positive correlation between an individual’s union membership and life satisfaction. Empirical studies, while often implicitly or explicitly based on the above line of argument, have mostly looked at job satisfaction.3 However, focussing on this linkage can neglect important determinants of well-being, which are not directly related to the job. If, for example, union membership provides individuals with valuable information about aspects not related to work, offers them the opportunity to purchase goods at reduced prices or enables people to lead a socially more active life, this can result in an increase in life satisfaction without having an impact on the evaluation of the job. Moreover, even job-related aspects, such as job security, may affect job and life satisfaction differently if, for example, job-related uncertainty has an impact on the family. If, therefore, union membership involves benefits–or costs–not associated with or extending beyond the job, life satisfaction will be a strictly better proxy for the utility derived from being a union member than wages or job satisfaction. In addition, life satisfaction can reflect an individual’s evaluation of the consequences of union membership, which is likely to be a more relevant determinant of behaviour than actual effects, such as the union wage mark-up. 3.2 Channels In earlier contributions, various channels have been proposed through which union membership could positively affect life satisfaction. First, union membership and life satisfaction may be correlated because of a union membership wage premium, which has been documented for many countries (see Bryson 2014 and Fang and Hartley 2022). If wages have a positive impact on life satisfaction, union members will exhibit a higher level of satisfaction than non-members (Lee 2022). For Germany, most studies find no evidence of a union membership wage premium, once observable characteristics are accounted for (Schmidt and Zimmermann 1991; Fitzenberger, Ernst, and Haggeney 1999; Goerke and Pannenberg 2004;Bryson2014). Somewhat in contrast, Wagner (1991) reports a positive correlation between wages and union membership for blue-collar workers in West Germany, and Bonaccolto-Töpfer and Schnabel (2023)documentawage premium in recent years. In consequence, the wage channel either plays no role or could give rise to a positive, though not a negative, correlation between trade union membership and life satisfaction in Germany. 464 British Journal of Industrial Relations,2025 Second, some contributions argue that union members have more intensive social interactions, greater social capital and are more integrated into life (see, e.g., Radcliff 2005). These expectations are often based on the assumption that unions organize meetings for their members, foster collaboration among colleagues and help employees to cope with work-related stress.4Moreover, life satisfaction is positively associated with social capital (Helliwell and Putnam 2004; Winkelmann 2009). The above reasoning suggests that social capital contributes to a positive relationship between union membership and life satisfaction (Flavin, Pacek, and Radcliff 2010; Flavin and Shufeldt 2016; Keane, Pacek, and Radcliff 2012; Radcliff 2005).5 Third, it is often argued that union members are better protected against job losses (Berglund and Furåker 2016; Freeman 1980; Goerke and Pannenberg 2011; Ivlevs and Veliziotis 2017; Pierse and McHale 2015) and, therefore, experience a higher level of job security than non-members. If subjectively perceived job security is a good proxy for its actual level and the perception has a positive impact on life satisfaction, as Clark, Knabe, and Rätzel (2010) document, union members can be expected to exhibit higher life satisfaction than non-members through this channel (Flavin, Pacek, and Radcliff 2010; Flavin and Shufeldt 2016; Keane, Pacek, and Radcliff 2012; Radcliff 2005). The evidence for Canada and the United States is in conflict with the assumption that perceived job security of union members is higher than that of comparable non-members (Brochu and Zhou 2009;BrochuandMorin2012). In the United Kingdom, the relationship varies across groups of workers and depends on tenure (Bender and Sloane 1999). Moreover, union members’ satisfaction with job security, which is probably highly correlated with perceived job security, may be lower than that of non-members (Bender and Sloane 1998;Bryson and White 2016; Green and Heywood 2015; Powdthavee 2011). For Germany, there is evidence of a positive impact of perceived job insecurity on union membership (Chadi and Goerke 2023). In consequence, the direction of the relationship between trade union membership and life satisfaction due to job security considerations is difficult to predict. Fourth, union membership may affect the employees’ view of their own economic and financial situation. Such an effect can arise as unions provide members with detailed information about economic developments both for the entire economy and individually. Moreover, trade unions can furnish job-related information, which allows members to better evaluate the prospects of their professional future. To garner support for their activities, trade unions are more likely to emphasize adverse aspects of the economic and financial situation (see, e.g., Hammer and Avgar 2005; Borjas 1979; Heywood, Siebert, and Wei 2002; Freeman and Medoff 1984, pp. 139f). If a more sceptical evaluation of one’s own economic position has negative consequences for life satisfaction, union members can be hypothesized to exhibit a lower level of life satisfaction than non-members. Finally, a common starting point of many analyses considering the linkage between union membership and life satisfaction is the expectation that job satisfaction of union members exceeds the respective level of non-members. If job and life satisfaction are correlated positively, life satisfaction of union members is likely to exceed life satisfaction of comparable non-members (Flavin, Pacek, and Radcliff 2010; Flavin and Shufeldt 2016; Keane, Pacek, and Radcliff 2012).6In the only comprehensive study of job satisfaction and trade union membership for Germany, Goerke and Huang (2022) document negative correlations in most OLS specifications, which often become insignificant in FE models.7 Therefore, the job satisfaction channel suggests a non-positive correlation between union membership and life satisfaction. We conclude the exposition of the channels, which may give rise to an impact of union membership on life satisfaction, with four observations: First, the channels outlined above are not mutually exclusive but rather complementary as, for example, higher pay may affect life satisfaction directly and also indirectly via job satisfaction. Nonetheless, for analytical clarity, we discuss them separately. Second, the enumeration of potential channels is unlikely to be exhaustive. Consequently, we expect a correlation between trade union membership and life satisfaction to persist even when we empirically account for all of the above channels. Third, the channels we look at may be based on a causal impact of union membership or indicate selection effects. A higher wage may, for example, result from joining a trade union. Alternatively, those employees who are highly motivated and better paid may also be more inclined to become a union member. Our analysis of the channels that influence the relationship between trade union membership and life satisfaction yields insights irrespective of the nature of the linkage. Fourth, a comparison of the findings for the various channels can provide valuable insights: The first and the second channels, higher wages and greater social capital, relate to actual outcomes. In contrast, the third and fourth channels refer to expectations, namely about job security and the economic situation. Systematic differences between these two groups of channels can be informative about whether trade union membership actually yields benefits or primarily affects the perception of outcomes. The fourth and fifth channels, concerns about the economic situation and job satisfaction, represent encompassing measures of the potential effects of union membership. Their analysis can provide information about whether the dissatisfaction of members is based on broader job-related and economic aspects of union membership, beyond the effects on wages and job security. Finally, channels 2 and 4 relate to social capital and the economic situation and, therefore, to potential consequences of union membership not directly associated with the job. The findings for these outcomes can help to ascertain whether the analysis of life satisfaction provides insights, which extend beyond those obtained from considering job satisfaction. 4Previous Analyses Radcliff (2005) provides an early analysis of the linkage between an employee’s trade union membership and life satisfaction. He estimates a linear model, using data for 17 countries from the 1990 wave of the World Value Survey (WVS) and finds a positive correlation. A number of further studies also use the WVS, focusing on different waves and countries. Flavin, Pacek, and Radcliff (2010) document a positive linkage for 14 465 industrialized countries in OLS specifications, particularly for low-income individuals. Charman and Owen (2014)usealarger sample covering 49 countries. In ordered probit specifications, they find a positive correlation between union membership and life satisfaction, which they no longer obtain for a subsample of high-income countries. Flavin and Shufeldt (2016) focus on the United States and observe a positive correlation in OLS and Ordered Probit models. Employing data from the United States Gallup Daily Tracker, Artz, Blanchflower, and Bryson (2022)and Blanchflower, Bryson, and Green (2022)alsoreportapositive correlation for OLS models. Moreover, Blanchflower, Bryson, and Green (2022) document a qualitatively comparable relationship for almost 40 European countries, based on the European Social Survey, without providing results for individual countries. Guzi and de Pedraza García (2015) utilize a web-based survey for Spain and observe a positive correlation in an OLS specification. In the context of a completely different industrial relations system and employing OLS and FE specifications, Booth et al. (2022)show that rural-urban migrant workers in China who belong to real (and not to so-called paper) unions exhibit higher happiness than non-members. Finally, using South Korean household panel data, Lee (2022) finds no correlation between union membership and life satisfaction in firms where a union is present in a correlated random effects panel ordered probit model. In sum, most previous investigations present findings consistent with the theoretical expectation of a positive impact of union membership on life satisfaction. However, the correlations do not always appear to be robust to variations in the sample or estimation method. Moreover, most studies base their insights on inter-personal comparisons.8 For Germany, there is no thorough study of the relation between trade union membership and life satisfaction. Humpert (2013) documents a negative (insignificant) relationship for females (males) when analysing gender differences in the effects of social participation, using the German General Social Survey (ALLBUS). Goerke and Huang (2022) analyse the relationship between trade union membership and job satisfaction. Employing SOEP data, they find a negative association in OLS models that can no longer be observed in FE specifications. The authors conjecture that individuals with time-invariant unobservable characteristics negatively associated with job satisfaction may sort into union membership. To scrutinize this hypothesis, they also look at the relationship between union membership and life satisfaction and report significantly negative correlations for OLS and FE models. The present contribution builds on these initial estimates by Goerke and Huang (2022) and extends their findings to subgroups, evaluates potential mechanisms that bring about the negative correlation and investigates the impact of labour market institutions, such as collective bargaining and co-determination. 5 Data and Methodology We use data from the German Socio-Economic Panel (SOEP, version 36). The SOEP is a representative and rich longitudinal dataset of private households in Germany (see Goebel et al. 2019 for more information). It provides information on individuals’ demographic and work-related characteristics. Our study focuses on employed individuals aged between 16 and 65. The observation period ranges from 1985 to 2019, the first and the last year for which information on union membership status is available. 5.1 Definition of Variables The dependent variable measures a respondent’s life satisfaction. The respective question is asked in each wave and possible answers range from 0 (completely dissatisfied) to 10 (completely satisfied). The main explanatory variable, union membership status, is a dummy variable that takes the value of one if an individual is a member at the time of the interview and zero otherwise. The questionnaire contains the corresponding question at irregular intervals, namely in 1985 and 1989 for West Germany, in 1990 for Eastern Germany, and in 1993, 1998, 2001, 2003, 2007, 2011, 2015 and 2019 for the entire country, which was reunited in October 1990. We separate further independent variables into three broad categories. The first includes demographic and labour market variables. The former include age (level & quadratic terms), being female, being married, years of education (level & quadratic terms) and the number of children living in the household. We differentiate labour market variables into those which are relatively exogenous from an individual’s perspective and a second group, for which this is more difficult to argue. The first group includes dummy variables taking the value of one if a respondent works in the public sector, is a civil servant, a blue-collar worker, and 16 dummy variables indicating the industry affiliation (at the 2-digit NACE level). The second set of labour market-related variables incorporates information on actual weekly working hours, tenure (level & quadratic terms), and a dummy variable indicating that the respondent has joined the firm within the last year. The second category of variables is utilized to analyse the channels outlined in Section 3.2. We use the natural logarithm of the gross labour income in the previous month to compute the wage mark-up. It includes overtime remuneration, but does not incorporate one-time payments, such as vacation pay or bonuses. To derive a measure of social connectedness, we follow Winkelmann (2009) and Bauernschuster, Falck, and Woessmann (2014), and focus on the networking aspect of social capital. In particular, we utilize information on leisure-related activities and the participation in organized events. To do so, we employ a range of questions about how individuals spend their leisure time.9The answers to each question range from 1 ‘at least once per week’ to4‘never’. We recode them such that the more one performs an activity, the higher the score is. Moreover, we average the answers over all responses in each year, such that our indicator of social capital can range from 1 to 4.10 We derive indicators of an individual’s perceptions of job security and the economic situation from the following questions: ‘How concerned are you about the following issues?’. The list of topics includes ‘Your job security’and‘Your own economic situation’. The possible answers are 1 ‘very concerned’, 2 ‘somewhat concerned’and3‘not concerned at all’, and higher values indicate fewer concerns. Finally, job satisfaction is measured on an 11-point scale ranging from 0 (completely dissatisfied) to 10 (completely satisfied). 466 British Journal of Industrial Relations,2025 In a subset of SOEP waves, there is information about collective bargaining coverage and the existence of a works or personnel council in the respondent’s establishment. From these, we construct our third category of independent variables. First, we use the question ‘Are you paid according to a collectively agreed wage agreement?’. The SOEP contains this enquiry and information about union membership simultaneously in 2015 and 2019. Second, the information concerning co-determination at the establishment level is derived from: ‘Does a works or a personnel council exist at your place of work?’. This question and the one on union membership are asked contemporaneously in 2001, 2011 and 2019. In all specifications, we control for federal state and survey year fixed effects. We exclude all observations with missing values. Our main estimating sample consists of 88,739 observations from 39,218 individuals. Of these, 59,537 observations stem from 29,162 individuals who never belonged to a trade union, while 8430 observations relate to 4093 individuals who were a member of a trade union in all survey waves in which we observe them. Moreover, 5963 individuals (20,772 observations) reported changes in the membership status. In particular, 1352 individuals joined, 2990 left, and 1621 reported both entry and exit. Since the information needed to scrutinize the various channels is not available in all years and is not provided by all respondents, the respective estimations are based on 73,151 observations (from 35,920 individuals). The sample incorporating information about collective bargaining consists of 22,678 observations (from 18,117 individuals), while the co-determination sample includes 27,415 observations (from 22,024 individuals). 5.2 Model We initially report OLS estimates. In order to control for timeinvariant observable and unobservable factors that might be correlated with union membership and life satisfaction, we also employ FE regressions. We estimate the following model: 𝐿𝑆𝑖𝑡 =𝛽0+𝛽1𝑇𝑈𝑖𝑡 +𝜷𝟐𝑿𝒊𝒕 +𝜆𝑖+𝜆𝑡+𝜀𝑖𝑡,(1) where 𝐿𝑆𝑖𝑡 is life satisfaction of individual 𝑖interviewed in year 𝑡and 𝑇𝑈𝑖𝑡 indicates the union member status. 𝑿𝒊𝒕 is a vector of control variables, including federal state fixed effects. 𝜆𝑖indicates individual fixed effects (omitted in OLS specifications), while 𝜆𝑡 represents a set of year dummy variables. 𝜀𝑖𝑡 is the error term. We cluster standard errors at the individual level. We focus on linear models since they are more straightforward to interpret. Using SOEP data, Ferrer-i-Carbonell and Frijters (2004) show that OLS models yield results equivalent to those for nonlinear models. To substantiate this equivalence for our analysis, we provide results from an Ordered Probit model as a robustness check for the main analysis. 6Results 6.1 Descriptive Evidence Table 1reports the summary statistics for the main variables. In our main estimating sample, 21.37% of all observations are union members. There is a substantial gender gap in union membership. While 43.54% of the observations are female, among union members, this share amounts to 32.91%. Union members are, on average, older than non-members, have longer tenure, are more likely to work in the public sector, in blue-collar occupations and large companies. Most importantly, from our perspective, union members exhibit a level of life satisfaction that is around 0.15 points lower than the level reported by non-members.11 When we look at the distribution of replies to the life satisfaction question, Figure 1indicates that the most frequent answers are those to the left and right of the mean value. Moreover, union members are less likely to state that they are almost completely satisfied with their life (particularly responses 8 and 9), whereas they are more likely to provide intermediate answers (especially responses 5 and 6). 6.2 Main Findings Table 2depicts estimation results from OLS models (Panel (A)) and FE models (Panel (B)) for the main estimating sample. Column (1) shows a simple correlation between trade union membership and life satisfaction, including year and federal state fixed effects as covariates. The estimated coefficients are negative and statistically significant. Adding demographic controls (in column (2)) and information on labour market characteristics (in column (3)) does not qualitatively change the estimates.12 The specification in column (3) contains both groups of labour market characteristics, the relatively exogenous ones and those which can be argued to be affected by union membership. If we only include the former, the estimated coefficient on the union membership dummy (not depicted) is not statistically different from the one depicted in column (3) of Table 2. Hence, our findings hold even if bad controls existed. The results show that life satisfaction of union members is substantially lower than that of non-members. The estimated effects are quantitatively sizeable and, focusing on the FE model, roughly equivalent to moving from a medium or large firm to one with less than 20 employees (see Table A6 in the Supporting Appendix, which contains the full set of results for the specification in column (3)). The feature that the estimated coefficient in the FE model is somewhat smaller in magnitude than in the OLS specification is compatible with the view that individuals who generally express lower levels of satisfaction sort into union membership. However, the significant correlation in FE specifications clarifies that sorting cannot entirely explain for the observed negative relationship. Table 1indicates that a relatively large majority of union members in Germany is male (see also OECD and AIAS 2021). In consequence, trade unions may better represent the interests of male members, such that the life satisfaction effect of union 467 TABLE 1 Summary statistics of the main estimation sample. (1) (2) (3) (4) (5) (6) (7) All observations Members Non-members Difference in the mean mean s.d. min max mean mean (col. 5 −col. 6) life satisfaction 7.2035 1.6670 0 10 7.0870 7.2351 −0.1481*** TU member 0.2137 0.4099 0 1 1 0 age 41.1792 11.7733 16 65 42.8811 40.7166 2.1645*** female 0.4354 0.4958 0 1 0.3291 0.4643 −0.1352*** married 0.5739 0.4945 0 1 0.6336 0.5577 0.0759*** education 12.1561 2.5946 7 18 11.7898 12.2556 −0.4658*** children in HH 0.5767 0.8874 0 10 0.5660 0.5796 −0.0136 hours worked 38.9923 12.1470 0.5000 80 40.1111 38.6882 1.4229*** public sector 0.2623 0.4399 0 1 0.3568 0.2366 0.1203*** civil servant 0.0709 0.2566 0 1 0.1164 0.0585 0.0578*** blue-collar worker 0.3017 0.4590 0 1 0.4354 0.2653 0.1701*** tenure 10.6454 10.1204 0 50.9170 14.5978 9.5711 5.0266*** new job 0.1226 0.3280 0 1 0.0650 0.1383 −0.0733*** firm size self-employed or w/o employees 0.0215 0.1449 0 1 0.0025 0.0266 −0.0242*** <20 0.2378 0.4258 0 1 0.0706 0.2833 −0.2126*** [20, 200) 0.2683 0.4431 0 1 0.2279 0.2793 −0.0514*** [200, 2000) 0.2213 0.4151 0 1 0.2817 0.2049 0.0769*** ≥2000 0.2511 0.4337 0 1 0.4173 0.2059 0.2114*** observations 88,739 18,077 70,662 individuals 39,218 9353 33,763 Note: This table shows summary statistics of the main estimation sample using individual-level weighting factors. Observation period is 1985–2019. In columns (1)–(4), we show statistics for the whole sample, in column (5) for trade union members, and in column (6) for non-members. Column (7) shows the difference in means between columns (5) and (6). *** 𝑝<0.01. FIGURE 1 Distribution of life satisfaction responses for union members and non-members. [Colour figure can be viewed at wileyonlinelibrary.com] 468 British Journal of Industrial Relations,2025 TABLE 2 Correlation between trade union membership and life satisfaction. (1) (2) (3) Panel (A): OLS estimates TU member −0.0863*** −0.0489*** −0.0966*** (0.0188) (0.0189) (0.0195) Panel (B): FE estimates TU member −0.0579** −0.0578** −0.0636** (0.0252) (0.0251) (0.0252) control variables year/state fixed effects X X X demographic controls X X labour market controls X observations 88,739 individuals 39,218 Note: This table shows the relationship between trade union membership and life satisfaction. The observation period is 1985–2019. Demographic characteristics include age (level & quadratic terms), being female, being married, years of schooling (level & quadratic terms) and the number of children in the household. As labour market factors, we include relatively exogenous factors like working in the public sector, being a civil servant, being a blue-collar worker, and dummy variables for firm size and industrial sectors, as well as less exogenous factors, such as actual work hours per week, years of tenure (level & quadratic terms) and having a new job (tenure shorter than 1 year). Federal state fixed effects and year fixed effects are accounted for in all specifications. Standard errors (clustered at the individual level) in parentheses: *** 𝑝<0.01;**𝑝<0.05. membership could be gender-specific.13 Panel (A) of Table 3 documents a significant and negative correlation of about the same magnitude for males and females in OLS models (see columns (1) and (3)). Accounting for individual fixed effects does not qualitatively alter the findings for males, while the estimated negative coefficient for females becomes statistically insignificant (see columns (2) and (4)). Therefore, our findings do not support the conjecture that unions act primarily on behalf of their male membership. Recent studies show that the association between union membership and job satisfaction varies across time and birth cohorts (Blanchflower and Bryson 2022; Blanchflower, Bryson, and Green 2022; Goerke and Huang 2022). Moreover, the relationship with life satisfaction turns positive in the United States and the EU after the Great Recession (Blanchflower, Bryson, and Green 2022). To analyse whether a similar feature is observable for Germany, we create two sub-samples of about equal size; one consisting of early observations surveyed in and before 2003, and a second one with responses collected in the second half of our observation period. Both OLS and FE estimates are statistically significant and negative for individuals surveyed in or before 2003 (see columns (1) and (2) of Panel (B) in Table 3). When looking at the employees surveyed in later years (see columns (3) and (4)), the OLS estimate is negative and marginally significant, while the magnitude of the estimated coefficient is substantially smaller than in column (1). The FE estimate is statistically insignificant.14 Furthermore, we have split the sample according to birth year and age. To ascertain whether the relationship between union membership and life satisfaction varies across birth cohorts as, for example, Blanchflower, Bryson, and Green (2022)document for the linkage between membership and job satisfaction in the United States, we consider separately individuals born in or before the median birth year 1964 and afterwards. To look at the possibility of age effects, we split the sample into observations aged 41 or younger at the time of the interview and a second group aged 42 or older. Using these samples, we observe that both OLS and FE estimates are statistically significant and negative for the older cohort and insignificant for the younger. All four estimated coefficients for the different age groups are negative, with the one resulting from the FE specification for the older individuals being statistically insignificant (see Table A7 in the Supporting Appendix). The findings that the negative relationships between union membership and life satisfaction become weaker or statistically insignificant after 2003 and for individuals born later than 1964, while there are no clearly identifiable patterns across age groups, suggest the absence of strong changes over time or age cohorts. We may instead interpret the results outlined above in conjunction with the feature of a declining union density over time. This fall could imply, in contrast to the hypothesis usually put forward when investigating the exit-voice perspective, that only those individuals remain in a trade union, or join it, whose benefits from membership are particularly large as the costs of membership have remained basically unchanged. Their dissatisfaction is likely to be smaller than of individuals who benefit to a lesser extent from membership. From this perspective, the negative relationship between union membership and life satisfaction has become weaker over the years because the remaining members perceive the benefits of membership to be larger than in earlier periods. However, in contrast to other countries, we obtain no evidence for Germany that the relationship between union membership and life satisfaction has turned positive since the Great Recession or in recent years (Artz, Blanchflower, and Bryson 2022; Blanchflower, Bryson, and Green 2022). 6.3 Potential Channels In each of the subsequent subsections, we proceed as follows: First, we document the relationship between union membership and the variable under scrutiny, both descriptively and based on OLS and FE models. This step helps to establish whether the potential channel exists. We also re-estimate our baseline specification with life satisfaction as the dependent variable for the smaller sample that we use to analyse the various channels. This clarifies whether the change in the estimating sample has an impact. Second, we include the respective variable as an additional covariate into the baseline specification. By comparing the estimated coefficients for the union membership dummy with and without the additional covariate, we can ascertain whether the relationship between union membership and life satisfaction operates via the channel considered. Finally, we discuss the insights from combining the results for alternative combinations of explanations and we consider all potential channels at once.15 469 Blanchflower, D. G., and A. J. Oswald. 2004. “Well-Being Over Time in Britain and the USA.” Journal of Public Economics 88, no. 7–8: 1359–1386. Bonaccolto-Töpfer, M., and C. Schnabel. 2023. “Is There a Union Wage Premium in Germany and Which Workers Benefit Most?” Economies 11, no. 2: 50. Booth, A. L. 1985. “The Free Rider Problem and a Social Custom Model of Trade Union Membership.” Quarterly Journal of Economics 100, no. 1: 253–261. Booth, A. L., and M. Chatterji. 1993. “Reputation, Membership and Wages in an Open Shop Trade Union.” Oxford Economic Papers 45, no. 1: 23–41. Booth, A., R. Freeman, X. Meng, and J. Zhang. 2022. “Trade Unions and the Welfare of Rural-Urban Migrant Workers in China.” ILR Review 75, no. 4: 974–1000. Borjas, G. J. 1979. “Job Satisfaction, Wages, and Unions.” Journal of Human Resources 14, no. 1: 21–40. Brochu, P., and L.-P. Morin. 2012. “Union Membership and Perceived Job Insecurity: Thirty Years of Evidence from the American General Social Survey.” ILR Review 65, no. 2: 263–285. Brochu, P., and L. Zhou. 2009. “Is Job Insecurity on the Rise? Evidence from Canadian Perception Data.” Canadian Journal of Economics/Revue canadienne d’économique 42, no. 4: 1305–1325. Bryson, A. 2014. “Union Wage Effects.” IZA World of Labor 1, no. 35: 1–10. Bryson, A., and M. White. 2016. “Not So Dissatisfied After All? The Impact of Union Coverage on Job Satisfaction.” Oxford Economic Papers 68, no. 4: 898–919. Chadi, A., and L. Goerke. 2023. “Seeking Shelter in Times of Crisis? Unemployment, Perceived Job Insecurity and Trade Union Membership.” Economica 90, no. 359: 1041–1088. Charman, C., and A. L. Owen. 2014. “Collective Happiness: Labour Union Membership and Life Satisfaction.” Applied Economics Letters 21, no. 13: 924–927. Chen, K., and S. Islam. 2023. “Declining Unionization and the Despair of the Working Class.” Journal of Law and Economics 66, no. 2: 279–307. Clark, A., A. Knabe, and S. Rätzel. 2010. “Boon or Bane? Others’ Unemployment, Well-Being and Job Insecurity.” Labour Economics 17, no. 1: 52–61. Coshow, S. M., and B. Radcliff. 2009. “Labor Organization and the Quality of Life in the American States.” In Happiness, Economics and Politics, edited by A. K. Dutt and B. Radcliff, 285–298. Edward Elgar Publishing. Ellguth, P., and S. Kohaut. 2022. “Tarifbindung und Betriebliche Interessenvertretung: Ergebnisse aus dem IAB-Betriebspanel 2021.” WSIMitteilungen 75, no. 4: 328–336. Erdogan, B., T. N. Bauer, D. M. Truxillo, and L. R. Mansfield. 2012. “Whistle While You Work: A Review of the Life Satisfaction Literature.” Journal of Management 38, no. 4: 1038–1083. Fang, T., and J. Hartley. 2022. “Evolution of Union Wages and Determinants.” In Handbook of Labor, Human Resources and Population Economics, edited by K. F. Zimmermann, 1–40. Berlin and Heidelberg: Springer-Verlag. Ferrer-i-Carbonell, A., and P. Frijters. 2004. “How Important is Methodology for the Estimates of the Determinants of Happiness?” Economic Journal 114, no. 497: 641–659. Fitzenberger, B., M. Ernst, and I. Haggeney. 1999. “Wer ist noch Mitglied in Gewerkschaften? Eine Panelanalyse für Westdeutschland.” Zeitschrift für Wirtschaftsund Sozialwissenschaften 119, no. 2: 223–263. Flavin, P., A. C. Pacek, and B. Radcliff. 2010. “Labor Unions and Life Satisfaction: Evidence from New Data.” Social Indicators Research 98, no. 3: 435–449. Flavin, P., and G. Shufeldt. 2016. “Labor Union Membership and Life Satisfaction in the United States.” Labor Studies Journal 41, no. 2: 171–184. Freeman, R. B. 1980. “The Exit-Voice Tradeoff in the Labor Market: Unionism, Job Tenure, Quits, and Separations.” Quarterly Journal of Economics 94, no. 4: 643–673. Freeman, R. B., and J. L. Medoff. 1984. What Do Unions Do? New York: Basic Books. Goebel, J., M. M. Grabka, S. Liebig, et al. 2019. “The German SocioEconomic Panel Study (SOEP).” Jahrbücher für Nationalökonomie und Statistik / Journal of Economics and Statistics 239, no. 2: 345–360. Goerke, L. 2021. “Unions and Workers’ Well-Being.” In Handbook of Labor, Human Resources and Population Economics, edited by K. F. Zimmermann. Berlin and Heidelberg: Springer-Verlag. Goerke, L., and M. Pannenberg. 2004. “Norm-Based Trade Union Membership: Evidence for Germany.” German Economic Review 5, no. 4: 481–504. Goerke, L., and M. Pannenberg. 2011. “Trade Union Membership and Dismissals.” Labour Economics 18, no. 6: 810–821. Goerke, L., and M. Pannenberg. 2024. “Wage Determination in the Shadow of the Law: The Case of Works Councilors in Germany.” Economic and Industrial Democracy 45, no. 1: 83–115. Goerke, L., and Y. Huang. 2022. “Job Satisfaction and Trade Union Membership in Germany.” Labour Economics 78: 102238. Goerke, L., S. Jeworrek, and M. Pannenberg. 2015. “Trade Union Membership and Paid Vacation in Germany.” IZA Journal of Labor Economics 4, no. 1: 1–26. Green, C. P., and J. S. Heywood. 2015. “Dissatisfied Union Workers: Sorting Revisited.” British Journal of Industrial Relations 53, no. 3: 580–600. Gürtzgen, N. 2009. “Rent-Sharing and Collective Bargaining Coverage: Evidence from Linked Employer–Employee Data.” Scandinavian Journal of Economics 111, no. 2: 323–349. Guzi, M., and P. de PedrazaGarcía. 2015. “A Web Survey Analysis of Subjective Well-Being.” International Journal of Manpower 36, no. 1: 48–67. Hammer, T. H., and A. Avgar. 2005. “The Impact of Unions on Job Satisfaction, Organizational Commitment, and Turnover.” Journal of Labor Research 26, no. 2: 241–266. Helliwell, J. F., and R. D. Putnam. 2004. “The Social Context of Well– Being.” Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 359, no. 1449: 1435–1446. Heywood, J. S., W. S. Siebert, and X. Wei. 2002. “Worker Sorting and Job Satisfaction: The Case of Union and Government Jobs.” Industrial and Labor Relations Review 55, no. 4: 595–609. Hipp, L., and R. K. Givan. 2015. “What Do Unions Do? A CrossNational Reexamination of the Relationship between Unionization and Job Satisfaction.” Social Forces 94, no. 1: 349–377. Hirsch, B., and S. Mueller. 2020. “Firm Wage Premia, Industrial Relations, and Rent Sharing in Germany.” ILR Review 73, no. 5: 1119–1146. Hirsch, B., P. Lentge, and C. Schnabel. 2022. “Uncovered Workers in Plants Covered by Collective Bargaining: Who Are They and How Do They Fare?” British Journal of Industrial Relations 60, no. 4: 929–945. Hohendanner, C., and S. Kohaut. 2023. Tarifbindung und Mitbestimmung: Keine Trendumkehr in Westdeutschland, Stabilisierung in Ostdeutschland. In IAB Forum. July 19th. Humpert, S. 2013. “Gender Differences in Life Satisfaction and Social Participation.” International Journal of Economic Sciences and Applied Research 6, no. 3: 123–142. Ivlevs, A., and M. Veliziotis. 2017. “What Do Unions Do in Times of Economic Crisis? Evidence from Central and Eastern Europe.” European Journal of Industrial Relations 23, no. 1: 81–96. 476 British Journal of Industrial Relations,2025 Jäger, S., S. Noy, and B. Schoefer. 2022. “The German Model of Industrial Relations: Balancing Flexibility and Collective Action.” Journal of Economic Perspectives 36, no. 4: 53–80. Jirjahn, U., and G. Tsertsvadze. 2006. “Betriebsräte und Arbeitszufriedenheit.” Jahrbücher für Nationalökonomie und Statistik/Journal of Economics and Statistics 226, no. 5: 537–561. Keane, L., A. Pacek, and B. Radcliff. 2012. “Organized Labor, Democracy, and Life Satisfaction: A Cross-National Analysis.” Labor Studies Journal 37, no. 3: 253–270. Laroche, P. 2016. “A Meta-Analysis of the Union–Job Satisfaction Relationship.” British Journal of Industrial Relations 54, no. 4: 709–741. Lee, Y.-W. 2022. “Trade Unions and Workers’ Life Satisfaction in South Korea.” Applied Research in Quality of Life 17, no. 1: 277–298. OECD. 2001. The Well-Being of Nations. Paris: OECD Publishing. OECD, and AIAS. 2021. “Institutional Characteristics of Trade Unions, Wage Setting, State Intervention and Social Pacts.” Oswald, A. J. 1982. “The Microeconomic Theory of the Trade Union.” Economic Journal 92, no. 367: 576–595. Pierse, T., and J. McHale.2015. “Unions and Involuntary Job Separations.” Human Resource Management Journal 25, no. 4: 496–515. Powdthavee, N. 2011. “Anticipation, Free-Rider Problems, and Adaptation to Trade Unions: Re-Examining the Curious Case of Dissatisfied Union Members.” Industrial and Labor Relations Review 64, no. 5: 1000–1019. Putnam, R. D. 2020. Bowling Alone: The Collapse and Revival of American Community. Simon and Schuster. Radcliff, B. 2005. “Class Organization and Subjective Well-Being: A CrossNational Analysis.” Social Forces 84, no. 1: 513–530. Rayo, L., and G. S. Becker. 2007. “Habits, Peers, and Happiness: An Evolutionary Perspective.” American Economic Review 97, no. 2: 487–491. Schmidt, C. M., and K. F. Zimmermann. 1991. “Work Characteristics, Firm Size and Wages.” Review of Economics and Statistics 73, no. 4: 705–710. Wagner, J. 1991. “Gewerkschaftsmitgliedschaft und Arbeitseinkommen in der Bundesrepublik Deutschland: Eine ökonometrische Analyse mit Individualdaten.” ifo Studien: Zeitschrift für empirische Wirtschaftsforschung 37, no. 2: 109–140. Winkelmann, R. 2009. “Unemployment, Social Capital, and Subjective Well-Being.” Journal of Happiness Studies 10: 421–430. Supporting Information Additional supporting information can be found online in the Supporting Information section. 477