What types of social policies does business want? Examining German employer associations’ positions toward labor-protective and labor-activating social policies

Abstract

Studies inspired by the varieties of capitalism (VoC) approach suggest that in coordinated market economies, some employer associations support public social policies to encourage the workforce to invest in company and industry-specific skills (VoC thesis). Yet the VoC thesis remains disputed. We present and assess an alternative thesis that builds on employers’ interest in the protection of labor supply (labor supply thesis). We test the labor supply thesis using a systematic content analysis of 370 press releases issued from 2002 to 2017 and find evidence of moderate employer support for more labor-activating social policies and less labor-protective social policies. Moreover, the analysis shows a decline in preference heterogeneity, with the positions of the four German employer associations converging toward the end of the period analyzed. Our findings have theoretical and methodological implications: First, they point to the relevance of labor supply as a source of employers’ social policy preferences. Second, they point to the need for a more systematic measurement of employer policy positions to be able to compare positions accurately.

Introduction

How do employer associations view public social policies? Employer interests are often at odds with generous social protection since the latter increases labor costs and may weaken work incentives. Yet an expanding literature on the social policy preferences of employer associations indicates that those preferences vary considerably and cannot be characterized as uniform opposition.¹

Theoretical debates over the role of employer interests in social policymaking have revolved since the 1980s around a controversy between the power resource approach and what Walter Korpi has called employer-centered explanations of welfare state development.² These two approaches differ in terms of how they characterize the role of capital and labor in the politics of welfare state development. The first emphasizes labor mobilization and class conflict as drivers of social policy expansion, while the latter emphasizes employer support as decisive for the adoption of social reforms. The employer-centered appraoch suggests that studies inspired by the power resource approach overestimate the importance of labor mobilization and class conflict and that cross-class alliances—that is, alliances between segments of capital and segments of labor—are empirically more prevalent than expected by the power resource approach.³

As we elaborate in the next section, the view that some employers have an interest in the social protection of workers through public social policies draws on arguments about the role of social protection as an incentive for skill investments by workers. These theoretical considerations allow the prediction of sectoral differences in employers’ preferences for social protection, depending on sector-level differences in skill needs. We contrast this thesis with an alternative characterization of employer social policy preferences that identifies the maintenance and enlargement of labor supply through activation, rather than the incentivization of skill investments by workers, as the main motive underlying the positions of employer associations towards different types of social policies.

From the perspective of employers, we argue, public social policy has two functions: a labor-protective function and a labor-activating function. While existing employer-centered explanations attribute employer support for public social policy, where it exists, to the benefits of the labor-protective function for skill investments, we argue that it is better explained by the labor-activating function of social policy and its expected positive effects on labor supply. During periods of overstaffing and labor shedding, employers may support labor-protective policies, while during periods of labor shortages—characteristic of the last two decades—they will be more supportive of labor-activating policies. We thus expect German employers today to be more supportive of labor-activating social policies than of labor-protective ones.

We investigate the validity of this thesis, elaborated in detail in the next section, by analyzing the positions of four nationwide employer associations in Germany on four types of social policy, drawing on a systematic content analysis of press releases issued by these associations. We decided to focus on Germany because of its archetypical character as a coordinated market economy in which employers are most likely to have an interest in skill promotion through social protection, thereby making it a critical case for the testing of competing theoretical expectations about employer preferences.

We use the level of unemployment insurance benefits and employment protection legislation to operationalize labor-protective policies, and active labor market policies and sanctions for benefit recipients to operationalize labor-activating policies. The analysis is based on 370 press releases issued by four employer associations from 2002 to 2017. We use data from the Reform Monitor on Political Conflict⁴ since this dataset captures the policy positions of different associations as well as their changes over time.

Our findings by and large support our thesis. We find all employer associations to be more supportive of labor-activating policies than of labor-protective ones, in line with what the labor supply thesis predicts. At the same time, the preferences of the four associations are rather similar overall, and the differences among them are limited. In addition, we find that the preferences become more similar over time.

The article is structured as follows: After an overview over the research design of the study in the next section, we present the method and coding scheme, followed by our findings. The conclusion provides an outlook on further needs for research.

Theoretical expectations: What types of social policies do employers want?

Theoretical expectations of employer support for public social policies are frequently grounded in the varieties of capitalism (VoC) approach developed by Hall and Soskice.⁵ Studies of employers’ social policy preferences inspired by this approach point to the advantages of public social policy to dominant groups of employers in coordinated market economies (CMEs).⁶ According to this approach, the production strategies of dominant types of firms in CMEs rely on workers with skills that are specific to a firm or sector (specific skills), rather than general skills (i.e., skills applicable across a wide range of firms and sectors).⁷ Some types of social programs provide incentives to workers to invest in specific skills, rather than general skills, by minimizing the risks that these investments may be devalued, particularly as a result of job loss and structural changes in the economy.

Estevez-Abe, Iversen, and Soskice, in a seminal contribution to the development of this thesis,⁸ highlight three types of policy as particularly relevant for inducing workers to invest in specific skills: unemployment insurance, employment protection, and wage protection through collective bargaining. Even though the authors point out that they do not intend to make an argument about the historical origins of social policies, others have used employer interests in CMEs in specific skill investments to explain the origins of public welfare programs. We refer to the argument that a need for specific skills motivates employers in CMEs to back public policies that protect workers from social risks as the VoC thesis. An important feature of this thesis is that it locates the sources of employer preferences for social policies in the labor-protective function of these policies, rather than in their labor-activating function. According to this thesis, employers will support social policy because they value the labor-protective effects of social policy, rather than the labor-activating effects.

The VoC thesis identifies firm size as one of the main determinants of the type of skills firms need and infers from this consideration that firm size also determines the social policy preferences of firms. As Estevez-Abe et al. argue, “one of the most salient divisions in employers’ preferences over types of social protection is firm size.”⁹ The VoC thesis thus expects employers’ social policy preferences not to be uniform, but to vary depending on firm size. As we will elaborate, our research design makes use of this implication to formulate two hypotheses with distinct expectations about preference heterogeneity (see Table 1)¹⁰.

Table 1. Expected social policy positions of employers’ associations: Two alternative theses.

Note: Gray shading indicates positions moderately or strongly supportive of that policy.

Recent research on national growth models and growth strategies has revived the argument that some types of firms benefit from labor-protective policies by protecting investments in specific skills. This literature identifies specific growth models and relates them to types of welfare state systems, whereby the latter are seen as a part of government strategies to promote economic growth.¹¹

Like the VoC approach, the growth model approach characterizes employers’ social policy preferences as derived from their needs for different types of skills. Following this logic, Avlijaš et al., for instance, argue that business interests “do not always favor welfare retrenchment and liberalization. For example, in some contexts, social protection can serve as insurance for investment in skills, which business might support.”¹² Where growth is driven by exports of high-quality manufacturing goods and firms rely predominantly on specific skills—as in Germany—“the employees’ specialization is not easily transferable to other sectors, thus employers aim to ensure job security and social security for them.”¹³ As in the VoC approach, Avlijaš et al. thus expect skill needs to drive employer support for social protection.

The thesis that skill needs drive employers to support social protection for workers has been investigated quantitatively and qualitatively, but findings differ. Based on historical research on employer positions toward the adoption of new social programs in France and Germany, Mares found that employer support for social reforms varied depending on skill profile and average firm size in a sector.¹⁴ An event study analysis of stock market responses to policy changes in France by Watson and Arunachalem found greater support for the skill-centered approach than for a class-centered one.¹⁵ A study of employer support for short-time work in Germany and Italy between 2008 and 2010 by Pancaldi found that skill intensity stimulated support.¹⁶

Yet other studies have cast doubts on the validity of the VoC thesis of skill-driven employer support for social policy. Several studies have found either that German employers did not historically support the adoption of policies that the VoC thesis would predict them to support,¹⁷ or that they turned against these policies in recent decades.¹⁸ These studies are rich in historical source evidence, but they are limited by their lack of systematic analysis of variation across sectors that stems from their primary reliance on statements by the national peak federations and by business-funded think tanks.

Moreover, while the VoC model of employer social policy preferences has been criticized frequently, alternative models have remained underdeveloped so far. We present a developed alternative and test it for Germany. This alternative thesis treats the labor-activating function of social policy, rather than its labor-protective function, as fundamental to the way that employers assess the advantages and disadvantages of public social policies. Labor-activating features of social programs are likely to increase labor supply, while labor-protective features are likely to reduce it.

Whether employers prefer an increase or a reduction of labor supply will depend on labor market conditions and firms’ human resource strategies. As several studies have shown, during the 1970s and 1980s, when industry faced overstaffing and wanted to shed labor, German employers favored policies that facilitated labor shedding, such as early retirement schemes, disability pensions, or unemployment insurance.¹⁹ Yet during periods of labor shortages and high demand for skilled labor, we argue, they will support labor-activating policies and oppose labor-protective ones in order to enhance the supply of skilled workers—a thesis we call the labor supply thesis. An absence of oversupply of labor is an important scope condition for the validity of the labor supply thesis.

The labor supply thesis and the VoC thesis both assume that firms have an interest in skilled labor but differ in the causal beliefs they assume underlie employers’ assessments of the impact of public social policies on the supply of skilled labor. According to the VoC thesis, employers will expect a larger supply of skilled labor from labor-protective policies; according to the labor supply thesis, they will expect a larger supply from labor-activating policies. The two theses also differ in the dimensions along which they expect employer preferences on social policy to vary. The VoC thesis expects variation across sectors, with sectors relying on specific skills and larger firms showing more support. The labor supply thesis does not expect variation across sectors but does expect variation across types of social policy—that is, more support for labor-activating policies than for labor-protective ones.

Our distinction between labor-protective and labor-activating types of social policies captures differences in decommodificaton. The former decommodify labor, albeit to varying extents, while the latter do not. Labor-protective policies are, for instance, policies that protect workers form income loss due to unemployment, sickness, or old age. We include in this article two labor-protective policies: unemployment insurance and job security regulations. These two policies benefit the part of the population fit to work and thus are likely to reduce labor supply. In contrast, health insurance, work injury insurance, disability benefits, and old-age pensions are types of labor-protective policies that do not benefit those fit to work, or expected to work, and for this reason, we do not include them.

Labor-activating policies are policies that are intended to act as incentives for the unemployed to seek work or to enhance their employability. We include in this article two types of labor-activating policies: first, active labor market policy (ALMP) instruments that focus on upgrading skills, building human capital for the unemployed, and making direct payments for firms to employ unemployed persons; and second, sanctions or other forms of pressure on workers or the unemployed to invest in new skills or to search for and accept gainful employment.

The extant literature supports the expectation that employers tend to favor active labor market policies, albeit with variation across countries and types of firms. Using pooled time-series analysis, Martin and Swank found that centralized and highly coordinated employer associations were associated with higher spending on active labor market policy, which suggests that organized employers may play a positive role in the expansion of active labor market policy.²⁰ Geyer found that under certain conditions, employers supported youth-oriented active labor market policies to fight skill shortages or support firms that train apprentices.²¹ Based on comparative case studies of seven European countries, Bonoli argues that “active social policy,” a term broader than “active labor market policy” and including childcare, is an area in which the preferences of labor and capital can converge.²² Bender also found consensus by German employer associations and trade unions on ALMP and childcare services and showed that social investment policies tend to be multifaceted, as they address both union and business issues and therefore receive joint support by the social partners.²³

However, other studies have highlighted employer opposition to active labor market policy. In a comparative historical analysis of Sweden and the United States, Gordon found important instances of employer opposition to active labor market policy.²⁴ For Denmark, a micro-level study of the participation by individual firms in active labor market programs by Bredgaard found that a minority of Danish employers were committed to participation.²⁵ Overall, the findings in the literature on employer preferences for active labor market policies and activation policies are heterogeneous, whereby existing studies are limited by the fact that they do not compare employer preferences for labor-activating (active) and labor-protective (passive) social policies, but rather look at active policies in isolation.

In short, the VoC thesis and the labor supply thesis predict different employer preferences for labor-protective and for labor-activating social policies. The VoC thesis expects support for labor-protective policies from sectors that rely on specific skills, and it does not formulate any expectations regarding employer preferences on labor-activating policies. The labor supply thesis expects employers across all sectors to oppose labor-protective policies and to support labor-activating policies (see Table 1). The differences in theoretical expectations are thus most marked regarding labor-protective policies.

Selection of associations

To capture the differences in positions within the business community expected by the VoC thesis, we include two nationwide sector-level associations and two nationwide multisector associations. The two sector-level national associations are the national association of metalworking (Gesamtmetall) and of chemical engineering (BAVC, Bundesarbeitgeberverband Chemie); the two multisector peak associations are the association of German craft firms (ZDH, Zentralverband deusches Handwerk) and the association of German employer associations (BDA; Bundesvereinigung deutscher Arbeitgeberverbände). The ZDH represents craft firms; the BDA aspires to represent all firms but in effect is dominated by larger firms in industry.²⁶

Our decision to focus on nationwide associations, rather than regional or subsector associations, is motivated by the fact that these associations are the ones that are most active and the most influential in the fields of social policy and labor market policy. A downside of the focus on nationwide peak federations is that it does not allow us to disaggregate preferences to more specific types of firms. Nationwide peak federations are internally heterogeneous in terms of the skill needs and the firm size of their members, among other things. However, regional and subsector associations in Germany often do not have the resources to deal with social policy issues and rarely publish press statements in this policy field, thus making them unsuitable for the purpose of our analysis.

Operationalization of the hypotheses

The labor supply thesis thus leads us to predict that all four associations included in our study will oppose labor-protective policies and support labor-activating policies, with positions being uniform across sectors (see Table 1). The VoC thesis leads us to predict variation in preferences for labor-protective policies depending on firm size and skill needs. The four associations included in our study represent firms of different sizes and different skill needs, and thus, following the VoC thesis, they should differ in their policy positions. The two associations representing manufacturing sectors—that is, metalworking (Gesamtmetall) and chemical engineering (BAVC)—should be most supportive of labor-protective policies because their members are predominantly large firms that are dependent on specific skills.

The ZDH, in contrast, should be least supportive of labor-protective policies. It represents craft firms, which are, compared with manufacturing firms, on average smaller, often in service sectors, and more reliant on general skills. Seventy-nine percent of all craft firms have a maximum of nine employees (2020), and a further 18 percent have between ten and forty-nine employees.²⁷ We thus use the ZDH as a proxy for the type of firms that the VoC thesis characterizes as opposing labor-protective policies. Furthermore, while the level of education of craft employees is lower and the focus is on general skills (apprenticeships are specific to occupation, e.g., car mechanic, carpenter, heating or sanitary service), training programs are not company specific. This contrasts with the company-specific skills required in the chemical or automobile industry, where the machines to be operated differ greatly from company to company. Therefore, following the VoC thesis, we should expect the ZDH to be least supportive of labor-protective policies²⁸ (Table 1).

As a nationwide umbrella federation, the BDA represents a wide range of firms, yet it is dominated by large firms in manufacturing.²⁹ Therefore, following the VoC thesis, we would expect it to be moderately supportive of labor-protective policies. Yet it is important to note that its public policy positions are likely to reflect internal compromises even more so than those of the other associations, and its positions should thus not be seen as reflecting the interests of a single sector.³⁰ We include the BDA in the analysis because of its dominant role in social policymaking, as it is the most influential voice of German employers in this policy field, but because of the dominance of large manufacturing firms among its members, we expect it to side more with the manufacturing associations rather than the ZDH.

In short, the two theses generate distinct expectations regarding the dimensions of variation in employer social policy preferences. Following the VoC thesis, we expect variation between types of firms, and following the labor supply thesis, we expect variation between policy types. Regarding the VoC thesis, we would like to note, though, that our research design is not intended for testing the causal relevance of the different firm characteristics—that is, firm size, sector, and skill specificity. Since these dimensions overlap empirically, our analysis is not suitable for disentangling the relevance of these three variables. Rather, our research design is intended to allows us to distinguish the relative importance of considerations relevant to all types of firms and thus predicting homogenous positions (i.e., labor supply) and considerations relevant only to specific types of firms (e.g., skill specificity, which is associated with firm size).³¹

Method, data, and coding scheme

To identify the policy positions of the four associations, we rely on a systematic content analysis of press releases (Reform Monitor on Political Conflict, ReMoPo),³² whereby we hand-coded their degree of support using a procedure that we outline in this section.³³

Systematic content analysis allows a systematic comparison of positions across sectors and across time, relying on transparent criteria for the coding of positions.³⁴ Compared with narrative forms of content analysis, which are used widely in the study of employer social policy preferences, this method enables us to more systematically assess an association's degree of support for, or opposition to, particular policy provisions. As shown by Lehmann et al., systematic content analysis is a common approach in the study of political parties’ manifestos, as well as political statements and speeches from parliamentary debates.³⁵ However, to our knowledge, it has not been used frequently to study the social policy positions of employers’ associations. We complement our use of systematic content analysis with brief narrative accounts of employer positions on selected policy issues in order to illustrate the character of these positions in a manner that is more intuitively understandable.

Press releases as a source of data

As a source of data for research, press releases have both advantages and disadvantages. Their disadvantages are twofold. First, press releases do not convey the internal deliberations and conflicts within an employer association. As Behrens argues, employers’ associations “need to convert the interest of individual entrepreneurs and firms into collective action and representation.”³⁶ Thus, the policy stances articulated in press releases often reflect compromises between different segments of members within an association. Second, like other political actors, employer associations act strategically, and thus they formulate their positions so as to maximize their expected influence.³⁷ Press releases are intended to influence public opinion and the views of policymakers. As a result, their content does to some extent reflect strategic considerations, rather than being an unfiltered expression of views held by members.³⁸ To give an example, an association may articulate cautious, conditional support for a policy that enjoys strong support among voters, even if its members oppose this policy, since taking a cooperative stance may enable the association to shape details of that policy in a way that outright opposition would not. Notwithstanding these limitations, press releases do indicate the direction of change that an organization wants to achieve—for instance, whether it wants to reduce or to increase benefit generosity in a program.

The advantages of press releases are threefold. First and foremost, their public availability for a wide range of policy issues facilitates systematic analysis compared with types of documents that are available only sporadically or for a very limited range of policy issues, such as internal reports or memos, which the associations may or may not disclose to the public. Second, compared with media interviews and media reports, press releases have the advantage of not being filtered by the editorial decisions of the medium. Even when media reports and media interviews with representatives of an association represent the views of that association accurately, they do not necessarily reflect its policy priorities accurately. In contrast, associations control the publication of press releases, which allows the researcher to avoid selection bias in terms of topics and viewpoints, which may occur in news media coverage, as well as distorted representations of viewpoints. Third, press releases are likely to focus on policy issues that an association considers important and that involve public controversy, rather than being the low-conflict issues characteristic of “quiet politics” and discussed in committees behind closed doors.³⁹

We selected 370 press statements from the ReMoPo dataset in the period from 2002 to 2017 that included references to unemployment benefits, job security regulation, ALMP, sanctions, and other disincentives to control job search behavior. We divided the period of analysis into three subperiods (2002–07, 2008–11, 2012–17) to investigate whether the Great Recession that started in 2008 had an impact on the associations’ social policy positions.⁴⁰ Since the level of unemployment is higher during periods of crisis, the risk of loss of specific skills is also higher; therefore, employers might be more willing to support labor-protective policies during periods of crisis.

Table 2 gives an overview of the press releases coded for our analysis. As the table shows, they are not equally distributed. Some organizations published many press releases and others very few. The national umbrella organization (BDA) published the most communications that were relevant to our research question: 105 press releases with respect to labor-protective policies and 51 with regard to labor-activating policies. This is not surprising, since the umbrella organization is the most relevant employer association for social policy issues. Therefore, it was important to include the BDA in our analysis in addition to the sectoral organizations. The second most frequent publisher of press releases was the federation of German crafts firms (ZDH), followed by the national federation for the metalworking sector (Gesamtmetall). The national federation of chemical engineering firms (BAVC) published the lowest number of press releases.

Table 2. Relevant press releases from selected German employer associations.

Table 2 also shows that the overall number of press releases (sum, last row) decreased across time. Looking at individual organizations, the only exception to this decreasing trend is the number of BDA press releases that focused on labor-activating policies, which increased from 8 to 25. For all other organizations, there is a clear downward trend from the first to the second period and again from the second to the third period. Overall, the number of press releases published by the four federations went down from 216 in the first to 91 in the second, and 63 press releases in the third subperiod.

The unusually large number of press releases during the first subperiod is due to the Agenda 2010 reform initiative by the German government, which triggered substantial controversy. Mass protests took place and the German political landscape changed.⁴¹ Therefore, we argue that the number of press releases on social policy mainly reflects the salience of social policy issues during a particular period. Employer associations increase their publication via press releases in order to inform and shape the opinions of other actors, such as members, other political actors, and public opinion.⁴² In the two subsequent periods, no other reform initiative or proposal was debated as controversially as the Agenda 2010 proposals. As a result, the federations had less reason to intervene in public debate.

Coding scheme

We coded statements according to the extent of support for the respective policy expressed in them, using a coding scheme with four categories: pronounced pro, moderate pro, moderate contra, and pronounced contra.⁴³ For example, statements on labor-protective social policies were coded as pronounced contra when employer associations were in favor of a very low level of unemployment benefits or a hire-and-fire system without protection against unfair dismissal. The moderate contra code was used when employer associations favored reducing the duration of unemployment benefits, for example, to only six to twelve months. The moderate contra code was also used when employers favored the reduction of job security regulations without recommending that they be abolished, and when employer associations argued for limited job security regulations for small companies. By contrast, moderate pro coding was used when employer associations spoke favorably of job security policy but some limits were also stated (e.g., job security measures are necessary for companies with more than twenty employees). Notably, the difference between moderate contra and moderate pro reflects the overall attitude voiced toward a policy. The practical limitations on policy may be the same, but the attitudes expressed are very different.

For press releases that were relevant to labor-activating social policies (such as qualification or further training measures, job creation programs or wage subsidies for private firms hiring the unemployed), the contra activation codes were used if the employer associations argued that assistance for the unemployed should be reduced (moderate contra) or eliminated (pronounced contra). The pro activation codes were used if employer associations favored the use of sanctions (e.g., reduction of benefits) or other ways of activating job search behavior.

For the program of unemployment insurance, we decided to code benefit generosity and sanctions for noncompliance with rules separately, treating benefit generosity as labor protective and sanctions as labor activating. Seen as a whole, unemployment insurance thus contains both labor-protective and labor-activating features, which our conceptualization, and associated way of coding, allows us to separate (see also Table A1 in the appendix).

We converted the codes to numerical values (+1, +2 for moderate/pronounced pro; –1, –2 for moderate/pronounced contra) for each of the social policies. These values were then added together to provide an average score (pro/contra) for the three policy areas,⁴⁴ using the three subperiods 2002–07, 2008–11, and 2012–17. A pronounced contra position on labor-protective and labor-activating policies is represented by –2; by contrast, a strong position in favor is shown by a score of +2. The neutral codes were excluded after the pre-test because we did not find any employer associations that published a neutral positions in their press releases.

Results: The position of employer associations on labor-protective and labor-activating policies

This section presents the findings for the four policy fields: unemployment benefits and job security regulations (both labor-protective policies), and active labor market policies and sanctions (both labor-activating policies). As explained earlier, we analyze the extent of variation along two dimensions: (a) variation across sectors, relevant for testing the VoC thesis, and (b) variation across the three policy fields, relevant for testing the labor supply thesis. We use the relative size of variation along these two dimensions to infer the importance of the two assumed underlying determinants of employers’ social policy positions: inducing worker investments in specific skills by offering social protection, and enhancement of labor supply through activation.

The level of unemployment benefits (labor-protective social policy)

Figure 1 displays the positions of the four employer associations. In line with the labor supply thesis, but contrary to the VoC thesis, we find no systematic support for a high level of unemployment benefits by employers’ associations across sectors. Prior to the financial crisis (2002–07), the employer associations’ position is moderate contra, as well as after the crisis until 2017.⁴⁵ The only exception to the overall contra position toward a higher level of unemployment benefits is Gesamtmetall during the financial market shock between 2008 and 2011. As Figure 1 shows, Gesamtmetall took a moderate positive stance on unemployment benefits. This positive stance is attributable to a specific debate about the level of unemployment benefits related to the short-time work program (Kurzarbeit). Workers subject to Kurzarbeit in the metal and electronics industry received the highest level of cash payments, but only if they were covered by collective bargaining.⁴⁶ However, after the financial market shock, Gesamtmetall reverted to its former position, opposing a higher level of unemployment benefits. Employer associations in all sectors argued that the unemployed would be less motivated to find a new job if passive transfers were too high. Furthermore, they argued that a higher level of unemployment benefits would have negative economic effects for their member companies because they would have to finance welfare state expansion through higher taxes.

Figure 1. Positions of employers’ associations on the level of unemployment benefits, 2002–17 (cumulative period).

After 2002, the BAVC, Gesamtmetall, ZDH, and BDA all repeatedly expressed interest in combining the different benefits systems in Germany into one system. All four organizations supported the plans by the Red-Green government under Chancellor Gerhard Schröder of the Social Democratic Party (SPD) to combine the security systems Arbeitslosengeld and Arbeitslosenhilfe into Arbeitslosengeld II (known as Hartz-IV) with reduced benefits and a shorter duration of benefits. To illustrate this position, the BDA published statements supporting Hartz-IV, as well as the merger of unemployment assistance (Arbeitslosenhilfe) and social assistance (Sozialhilfe) to create a single program.⁴⁷ Furthermore, and regardless of the demand for specific skills, all four associations criticized the reduction of the level of unemployment benefits as not going far enough. These positions on providing a low level of unemployment benefits by all sectors were also articulated after the general election in 2005. The employers welcomed the grand coalition of the Christian Democratic Union, Christian Social Union, and SPD and its commitment to maintain the Agenda 2010 policy of the Red-Green government until 2005. In 2006–07, all organizations criticized the SPD's new proposal to extend unemployment benefits for persons fifty-eight years or older to a maximum of twenty-four months as a backward step. Common arguments made by the BAVC, Gesamtmetall, ZDH, and BDA were that higher unemployment benefits would lead to higher nonwage labor costs for companies. The BAVC supported the Hartz-IV reform on the grounds that labor costs should be reduced and the best way to do so would be to lower the employers’ payments for cash benefits for the unemployed.⁴⁸ Furthermore, they argued that the unemployed would be less motivated to find a new job if the level of unemployment benefits was too high.

Therefore, during the pre–financial crisis period from 2002 to 2007, all four associations were in favor of less unemployment benefits, and they did not support a more generous level of passive transfers. This is in line with the labor supply thesis but contrary to the expectation of the VoC thesis that employers’ associations will support a higher level of unemployment benefits if they have a high demand for specific skilled workers. By the end of 2017, the associations praised the positive long-term effects of the Agenda 2010 reforms and expressed their interest in not taking any measures that would lead to unstable economic development. Apart from the positive stance taken by Gesamtmetall between 2008 and 2011, the results provide no evidence that the positions of employers’ associations varied based on the demand for specific skills. In contrast, all four associations were in favor of a low level of unemployment benefits between 2002 and 2017.

Job security regulations (labor-protective policy)

Following the labor supply thesis, we would expect uniform opposition to job security regulations by employers in all sectors. In contrast, following the VoC thesis, we would expect employers’ associations in the chemical and metal industries, and to a lesser extent the BDA, to support job security regulations to provide attractive working conditions for workers with specific skills. We find that all four associations moderately opposed stricter job security regulations, with the chemical engineering association until 2011 being less opposed than the other three associations.

As Figure 2 shows, all four associations tended to oppose stricter job security regulations. All four organizations advocated in their press releases for a moderate relaxing of job security regulations as a way to lower unemployment. To give an example, BDA president Dieter Hundt welcomed a proposal by the Christian Democrats to restrict dismissal protection rules to companies with at least twenty employees.⁴⁹

Figure 2. Positioning of employers’ associations on job security regulations, 2002–17 (cumulative period).

The chemical industry (BAVC) slightly deviated from this overall contra position toward stricter dismissal protection. As the detailed analysis of the press releases shows, during the financial market crisis, the BAVC sometimes expressed moderate support for job security regulation. More specifically, the BAVC argued during that period that dismissal protection could help prevent skilled workers from becoming unemployed. Furthermore, the BAVC argued during that period that dismissal protection could help to protect human capital during economically hard times and would lead to more economic success in the postcrisis period.

However, these arguments by the BAVC were limited to the period of economic recession (2008 and 2011). As Figure 2 shows, afterward, labor market flexibility became BAVC's main concern—like that of Gesamtmetall, ZDH, and BDA throughout the entire period of analysis. We thus find some support for the VoC thesis regarding job security regulations during the financial market crisis (2008–09). However, taking into account that Gesamtmetall, which also represents predominantly large firms that rely on specific skills, moderately opposed job security regulations, support for the VoC thesis is overall weak. Overall, our findings for job security regulations offer more support for the labor supply thesis.

Labor-activating policies (active labor market policy and sanctions)

Employer preferences on labor-activating policies are crucial for testing the labor supply thesis, since this is the only type of policy for which the labor supply thesis predicts support from employers. Figure 3 shows the federations’ stances on labor-activating policies. As a reminder, this indicator combines attitudes toward ALMP and sanctions on benefit recipients for noncompliance. A more detailed analysis of the press releases shows that the ZDH was opposed to ALMP while favoring sanctions and other negative consequences of unsatisfactory job search activity. Contrary to the ZDH, Gesamtmetall and the BAVC were in favor of both ALMP and sanctioning policies. The umbrella organization, BDA, was in favor of sanctions and other disciplinary measures but against ALMP. During the financial crisis (2008–11), the ZDH shifted its overall stance, and the BDA began to support ALMP as well.

Figure 3. Positioning of employers’ associations on labor-activating policies, 2002–17 (cumulative period).

The skilled craft organizations ZDH and BDA saw state-funded job creation programs, part of ALMP, as competition for their member firms. To give an example, the ZDH rejected proposals to extend the Ich-AG scheme for another six months.⁵⁰ The Ich-AG and other ALMP instruments, such as the One-Euro Job, were created in sectors in which they were in competition with other sector companies (e.g., gardening, carpentry, cleaning services). These measures were said to have negative effects on current members’ competitiveness.

During the second period (2008–11), the ZDH and BDA shifted their stances on ALMP, a shift that is rooted in but not explained by the economic and financial market crisis. The shift occurred for two reasons. The first was opposition to state-funded job creation programs, which were cut back by the government in 2009 before being abolished in 2011. After their abolition, the ZDH no longer criticized ALMP measures in its press releases. Second, the government also introduced new ALMP measures in the form of wage subsidies (Lohnkostenzuschüsse) for companies giving work to the long-term unemployed. In contrast with One-Euro Jobs and Ich-AG, this new program was favored by the ZDH and in particular the BDA because wage subsidies gave direct financial benefits to employers. Taken together, these two changes in the character of ALMPs resulted in an overall shift toward a pro stance on ALMP policies.

The second component in the position measure relates to employer stances on the use of sanctions and other disincentives to control job search behavior. All four associations supported this type of social policy measure. The Fördern und Fordern (“encourage and expect”) policy as well as the Job-Active-Act, both introduced by the Red-Green coalition government (2002–05), were welcomed by all employers’ associations. The umbrella organization BDA even postulated in 2006 that the rules on what job offers the unemployed were allowed to reject (Zumutbarkeitsregelungen) should be revised so that the unemployed would have to accept any and every job offer to avoid losing their benefit rights. To give an example, the BDA demanded that “benefits should be reduced, or withdrawn entirely, for those rejecting job offers presented by the job centers.”⁵¹

Because of substantial public debate and legal action between 2011 and 2012 concerning the rights of those affected by these sanctions, they became a more salient topic in employer organization press releases. The positive relationship between sanctions and increasing employment rates was highlighted. Sanctions and other disciplinary measures in the job search process served as push factors whereby the unemployed were given extra motivation to find and accept a new job.

Overall, employer support for sanctions and benefit conditionality reflects a distinct interest in labor activation, rather than simply reflecting support for welfare retrenchment or cost containment in general. This is evident from the fact that the BDA in particular did not demand indiscriminate cuts to all social programs but targeted its demands for cuts specifically to those programs that it suspected of reducing the labor supply. The debate in 2014 about a package of government reform proposals in public pensions (RV-Leistungsverbesserungsgesetz) illustrates this point. This omnibus law addressed a set of issues related to benefit adequacy and the retirement age. One component of the proposed law consisted of an option for workers with long periods of insurance to retire ahead of the statutory retirement age, at the age of 63 (Rente mit 63), without an actuarial reduction in benefits. The BDA firmly opposed that proposal because it would create incentives to retire early. At the same time, the BDA supported higher pensions for persons with reduced earnings capacity, or those incapable of working at least six hours per day (Erwerbsminderungsrente).⁵² While this example is merely illustrative, it indicates that the BDA targeted its opposition at social programs that decommodified those capable of work, rather than indiscriminately opposing any and all type of labor-protective social policies. In short, the BDA cared not just about welfare retrenchment, but about labor supply.

To sum up, after some initial disparity, the overall stance toward labor-activating policies was positive for employer associations from all sectors, both during and after the financial crisis. We see some convergence over time toward moderate support for labor activation. However, the implications of our findings for our two theses are mixed, depending on the period. Overall, though, support for labor-activating policies among German sector employers is stronger than for the two labor-protective policies in our study.

Comparison of policy positions across policy fields

Figure 4 shows the observed positions of the four associations in a two-dimensional space. The horizontal dimension shows positions on two labor-activating policies—ALMP, sanctions, and other push factors, and the vertical axis preferences on the two labor-protective policies—level of unemployment insurance benefits and job security regulations. Despite some variation, all four associations are in the lower-right corner, meaning that they tended to support labor-activating policies and oppose the two labor-protective policies.

Figure 4. Positioning of employers’ associations for the metal and electronics industry (Gesamtmetall), chemical industry (BAVC), craft sector (ZDH), and the nationwide umbrella federation (BDA); two-dimensional policy space, 2002–17 (cumulative period).

According to the VoC thesis, associations representing sectors that depend on specific skills will be more supportive of labor-protective policies. With respect to unemployment benefits and job security regulations, however, we find no evidence of variation in positions across sectors: all associations tended to oppose these policies, and we thus assess support for the VoC thesis as relatively weak.

We assess the evidence for the labor supply thesis to be stronger since employer support for the two activating policies is stronger than that for the two labor-protective policies. However, we did find differences across the subperiods: support for ALMP was not uniform, and during the first subperiod from 2002 to 2007, the ZDH was less supportive of ALMP than the sector associations for metalworking and chemical engineering. Therefore, the ZDH is a bit more toward the lower-left corner in Figure 4 than the other organizations. However, as the government cut back temporary job creation programs, all associations turned in favor of labor-activating policies.

Overall, the heterogeneity in positions on employment protection and on labor-activating policies declined in the third of our three subperiods—the positions of the four associations in these two policy fields tended to converge. Furthermore, on unemployment insurance, we see a return to unity, after the temporary emergence of differences during the global financial crisis (cf. Figures 1, 2, and 3). In the last subperiod, employers were more uniformly in favor of labor-activating policies and more uniformly opposed to labor-protective policies. Overall, the associations’ positions on employment protection and on labor-activating policies became more similar across the three periods (Figures 2 and 3).

Conclusion

Our study aimed to describe and explain social policy preferences by German employers’ associations. Based on an analysis of 370 press releases, we find greater employer support for labor-activating policies (ALMP and sanctions) than for labor-protective policies (level of unemployment benefits and employment protection legislation), meaning that support for the labor supply thesis is stronger. Differences between the four associations were either small overall or tended to decline over time. This may suggest that sectoral differences in policy positions have become less important. The arguments used by the associations suggest that an interest in increasing labor supply drove their preferences for labor-activating over labor-protective social policies.

To conclude, our article contributes to the study of employer social policy preferences in two ways. First, we used a method hitherto not used widely in this research field, systematic content analysis of press releases, which allowed us to compare policy positions across associations and across policy fields more systematically than narrative accounts. Second, the article presented a theoretical contribution in the form of the labor supply thesis. We suggest that this thesis offers an alternative to the VoC account of employer social policy preferences that can incorporate variation in employer preferences, while at the same time being compatible with the power resource approach.

Understanding what types of social policy employers want can be relevant for social policymakers if they aim to identify areas of cross-class consensus. If political actors are interested in forging alliances with business in favor of social policy reforms, then labor-activating polices are the most promising. Yet our study could not determine whether policymakers should prioritize policies that have employer support or instead pursue policies independent of the level of employer support. This is ultimately a normative question.

The generalizability of our findings is limited in three ways: our focus on a single country, our choice of policy issues, and our choice of period of analysis. First, even though Germany is a crucial case for testing the VoC thesis, as we argued earlier, further research should test the validity of the labor supply thesis in other countries. Second, labor supply may matter less to employers with regard to social policies further removed from the labor market, such as health insurance, work injury insurance, or pensions. Third, the importance of enhancing labor supply will also be affected by the conditions in the labor market, which vary across time.⁵³ The relevance of the labor supply thesis may thus be weaker in periods of oversupply of labor.

Further research is thus needed to test the labor supply thesis in a larger range of countries, including liberal market economies and countries with growth regimes different from the German one. One research question could be, How does the enlargement of labor supply through activation contribute to the German export-led growth model? Baccaro and Höpner underline the importance of wage restraint and currency undervaluation for Germany's export-led growth model.⁵⁴ An increase in labor supply through activation could be seen as facilitating wage restraint during periods of labor shortage, reflecting a complementarity of these two goals. Yet, whether German employers’ associations do perceive such a complementarity, and whether they do purposefully promote labor-activating policies as tools to facilitate wage restraint is beyond the scope of this study and would require further research. Moreover, we need further research to find out whether employers’ associations share the understanding of the causal mechanisms between social policies and labor supply that is assumed by the labor supply thesis.