Why does Germany abstain from statutory bargaining extensions? Explaining the exceptional German erosion of collective wage bargaining

Against the European trend, German statutory collective bargaining extensions (SBEs) have decreased in the last two decades, contributing to the exceptional erosion of German wage-bargaining coverage. This article distinguishes between two liberalization dynamics: an intrasectoral dynamic that started with the introduction of employers’ association memberships outside the scope of collective agreements, and an intersectoral dynamic. The latter is the result of an abnormal German institutional feature, the veto power of the employers’ umbrella association in the committees that have to approve SBE applications. Activation of this veto enabled employers to promote collective bargaining erosion in sectors other than their own, in order to contain cost pressures. This intersectoral liberalization dynamic has been part of Germany’s transition into an asymmetrically export-driven growth regime and could be stopped by means of political reforms.


Introduction
The decline of trade union membership is a secular trend. It is happening in almost all industrialized countries, including European countries such as Ireland and the UK, Finland and Sweden, Greece and Portugal, and Austria and Germany. Membership of trade unions (or employers' associations alike), however, must not be confused with collective bargaining coverage. In Austria, Belgium, Denmark, Finland, France, Italy, the Netherlands, and Spain, among other countries, coverage rates have remained almost stable since the early 1990s, with 80% or more of the workforce covered by collective agreements. This is because statutory bargaining extensions (henceforth: SBEs) make sectoral agreements binding for all firms within the sector. In most cases, national or regional labor ministries declare agreements binding after formal applications from sectoral trade unions, employers' associations, or both. As a result, employees remain covered, and business associations maintain their representativeness (Brandl and Lehr, 2019).
No such stabilization has taken place in Germany. The Collective Agreements Act (Tarifvertragsgesetz) that came into force in 1949 enabled SBEs, but they were never used as frequently as in other countries. In the postwar decades, a well-functioning collective bargaining system with high coverage emerged in many sectors, especially manufacturing. In those sectors, extensions were not needed. However, other sectors, such as construction or retail, made use of SBEs (Wonneberger, 1992: 28-32). As recently as the early 1990s, about 5.4% of sectoral agreements were declared generally binding -a comparatively low, but nevertheless significant number (Bispinck, 2012: 500). Since the 1990s, however, and against the international trend, German SBEs have been eroded further. Only 1.6% of sectoral agreements were extended in 2017 (Schulten, 2018: 76), compared to about 47% in the Netherlands, 80% in Finland, and 34% in Switzerland (Visser, 2018: 48). 1 Without a safety net in the form of SBEs, German coverage is now down to 51%. 2 This is an incredibly low rate.
The amount of erosion is puzzling not only empirically, but also theoretically. The 'varieties of capitalism' (VoC) literature portrayed Germany as the paradigmatic case of a coordinated market economy. According to this school, increased international competition translates into liberalization pressure in liberal market economies, and into functional pressure to maintain (or restore) coordination capacity in coordinated market economies (Hall and Soskice, 2001: 56-60). In the German understanding, the right to freedom of association also guarantees 'negative' freedom, which means that the state cannot press firms or employees into associations against their will. But wage coordination could be maintained by means of SBEs or functional equivalents. Why has no such stabilization taken place in Germany? This article offers a novel explanation for the declining use of SBEs in Germany and offers policy recommendations to reformers who are striving to reverse this trend.

Competing explanations
In a recent contribution, Paster et al. (2020) explore the stark differences between the uses of SBEs in the Netherlands and in Germany. The lack of stabilization by means of SBEs in Germany is, they argue, mainly attributable to the behavior of employers. We are in complete agreement with this assessment. If SBE applications fail, the reason for this is almost never that the trade unions oppose them. Therefore, an analysis of the use of SBEs primarily requires, in addition to knowledge of the political-institutional rules within which SBEs come about, analysis of the attitudes and strategies of employers and their associations.
Why do German employers' associations allow SBEs to erode, rather than make use of them in order to maintain their own representativeness? Paster et al. propose an ideacentered explanation. According to this view, today's employers maintain a double ideational legacy, consisting of ' [f]irst, ideas about the desirability of wage competition by outsiders; and, second, ideas about the appropriateness in principle of state compulsion in industrial relations. These ideas are in turn affected by historically evolved experiences with state compulsion and wage competition and have emerged over a period that goes back to the inter-war years' (Paster et al., 2020: 535). 'German associations,' they argue, 'view any form of state intervention as problematic ' (p. 551). In line with this explanation, Paster et al. conclude that a restabilization of German collective agreements would require ideational reorientation on the part of employers' associations. The authors consequently propose a regulatory dialogue between German employers' peak associations and trade unions (Paster et al., 2021: 104).
We believe that Paster et al.'s ideational explanation is adequate given their comparative research puzzle, though we find the first part of their argument, the ideational legacy relating to the desirability of wage competition, more convincing than the second part, the legacy relating to the desirability of state intervention. Consider the following example: during the so-called 'production location debate' that took place between the mid-1990s and the early 2000s, employers' associations and trade unions fought hard over the introduction of opening clauses in sectoral collective agreements (Bispinck, 2003). The German government intended to steer social partners toward decentralization. Chancellor Schröder made clear that he was willing to legalize firm-level derogations if the social partners failed to agree on the flexibilization of collective agreements on their own (Fehmel, 2010: 222-233).
Rather than rejecting interference by the state, and somewhat unexpectedly in light of theories based on the power of the idea of state-free self-regulation, the peak employers' association BDA (Bundesverband Deutscher Arbeitgeberverbände) asked the government to legislate. Specifically, the BDA asked the state to legalize firm-level derogations from sectoral collective agreements if management and work councils agreed on this (BDA, 2003: 21). This episode indicates that employers do ask for state action in collective bargaining matters if to do so is in their time-specific strategic interest, despite the historically and culturally entrenched idea of wage-bargaining autonomy.
Nevertheless, with regard to ideas about the desirability of wage competition, Paster et al. have a valid point: here, employers from different countries perceive their interests in different ways and translate them into different strategies. As we will show below, however, a change of perspective reveals additional, and in part diverging insights. Although our data also stem from a comparative research project, this article will explore the intertemporal and sectoral variation within Germany in detail. We offer a historicalinstitutionalist perspective that places emphasis on timing and sequence, on the triggers of sectoral path changes at exceptional moments, and on subsequent path dependencies up to the point at which path corrections become extraordinarily costly and therefore highly unlikely (see Capoccia and Kelemen, 2007;Mahoney, 2000;Pierson, 2000, among others). This change in perspective will also affect the practical recommendations for reformers striving to restabilize the German collective bargaining system. We make our argument by tracing the liberalization path of the German wagebargaining regime and by investigating the SBE-related decisions made by employers' associations. For the first, we consult the research literature on German wage bargaining as well as original documents and databases. For the second, we draw on interview material with collective bargaining experts from sectoral employers' associations and trade unions, both at the sectoral level and at the level of the peak associations. This material originates from a research project on extension practices in Germany, the Netherlands, and Finland (see Günther, 2021). The interviews with collective bargaining experts were conducted in 2016 and 2017. 3 Respondents were asked about structural characteristics and frequency of SBEs in their sectors, their perceptions of SBEs, and their SBE-related intentions and strategies. The 15 German respondents represent the construction, trades, electrical trades, retail, and security services sectors, and the peak federations BDA and DGB (Deutscher Gewerkschaftsbund), as well as the Federal Ministry of Labor and Social Affairs.
After discussing the sectoral distribution of SBEs in Germany in the following section, we will go on to show that some employers' associations -initially only a few and with some hesitation -responded to the exceptional cumulation of shocks in postunification Germany with the introduction of new weaponry, the so-called OT memberships (ohne Tarif: associational membership without collective agreement application). This phenomenon became more widespread than had initially been intended. The diffusion of OT memberships effectively blocked the SBE option: no association can offer OT memberships and afterwards introduce SBEs in order to make collective agreements generally binding. Concerning future options, the respective employers' associations have thereby tied their own hands. SBEs therefore only have a chance in sectors without significant OT memberships, such as construction.
But these intrasectoral dynamics are only half the story. In addition, an intersectoral dynamic has emerged that has driven wage-bargaining liberalization even further. Here we pick up on Hassel's (2014) insights on the specific German form of industrial relations liberalization, which led to a segmented and dualist political economy that, in the terminology of Baccaro and Pontusson (2016), favors export-driven growth. The export sector's cost-cutting imperative and its liberalizing effects also shaped the transformation of SBEs. Even under conditions of a sectoral consensus favoring extensions, the intersectoral umbrella association BDA (if the federal level is concerned), and other sectoral employers' associations (if the subnational Länder are concerned) can veto German SBEs in the federal or regional collective bargaining committees that decide on them (Tarifausschüsse). This institutionalized veto point is a German singularity that also marks an important institutional difference between Germany and the Netherlands, the two countries compared by Paster et al. Here we object to Paster et al.'s (2020: 534) claim that 'employer attitudes are the only variable that co-varies with the outcome.' In this setting, SBEs require an intersectoral consensus, or at least, sufficient intersectoral acceptance. The BDA, which coordinates the work of these committees, adopted an increasingly critical attitude toward SBEs the more widespread OT became, and the more Germany turned into an export-oriented economy. Within the last two decades, around 200 applications have been vetoed or withdrawn within the process due to a low likelihood of success, despite a sectoral consensus in favor of extension. Institutional reforms could halt this dynamic and reverse it. In the concluding section, we also discuss how the analysis of an exceptional case, Germany, speaks to comparative research.

The sectoral distribution of German SBEs
With regard to the acceptance of extensions, German social partners rank somewhere in the middle between Sweden (broad rejection) and the Netherlands or Finland (broad acceptance). Paster et al. correctly point out that ideational legacies are important here (Nautz, 1991). As indicated above, SBEs never diffused widely in Germany. Nevertheless, the social partners of some sectors used and appreciated the instrument. This held true for sectors that were labor-cost intensive, had low barriers for market entry, were fragmented, and were hence prone to wage competition (see Wonneberger, 1992: 28-36, 41). From the 1950s onward, SBE practices spread in the construction, trades, textile, and security service industries, and -until the OT shift described later in the article -in retail as well. Between 1950 and 1990, SBEs in these sectors made up 75% of all SBEs. In the 2000s, the majority of SBEs (70%) were to be found in construction, electrical trades, textiles, security services, cleaning services, and hairdressing. The interest these sectors shared in extensions, in addition to several sector-specific challenges, 4 was to prevent outsider companies from entering the market by undermining existing wage and working standards (Günther, 2020).
In the first few years after German reunification, SBEs were used to transfer collective bargaining institutions to the new Länder. Unions and employers sought to create a level playing field and to eliminate competition from Eastern German regions by incrementally introducing Western German pay norms through collective agreements (the Stufenplan in manufacturing) or through SBEs. Extensions of agreements in retail, wholesale, hotels, and catering (Kreimer-de Fries, 1995: 222-223), but also agreements in construction and trades like baking, hairdressing, cleaning services, painting and decorating, roofing, scaffolding, stonemasonry, or the electrical trade, now applied to Eastern regions or the whole federal territory (see BMAS, 2000). The number of SBEs in Eastern Germany rose from 7 in 19917 in to 122 in 19957 in , and 179 in 19997 in (Günther, 2021.
Until the 1990s, the guiding principle behind German SBEs could have been described as employer voluntarism: sectoral employers could and did use SBEs if they wished to do so. During these years, no one would have predicted an erosion of the instrument in the near future. Why did this erosion occur nevertheless?

The cumulation of shocks in the 1990s
All times are special; all times are times of change. There are nevertheless exceptional times that mark the emergence of new paths. With regard to the German growth regime and its industrial relations sphere in particular, the years after the reunification boom were such a period (Baccaro and Benassi, 2017). In the terminology of historical institutionalism, these years were a 'critical juncture'; this is defined by Capoccia and Kelemen (2007: 348) as a short period of time during which there is a heightened probability that agents will change course. Several crises cumulated and led to path-breaking behavioral change among employers, such as: the introduction of OT memberships, exceptional macroeconomic conditions, harsh struggles with trade unions, especially in the metals industry, and changing relations between suppliers and purchasers.
The Bundesbank reacted to the boom in demand that came with the introduction of the Deutsche Mark (West) in East Germany with a sharp increase of the Lombard interest rate. This reached an all-time high of 8.75% in July 1992, which made borrowing exceptionally costly for all firms, especially those in the East, which were at the same time confronted with a harsh supply-side shock that many of them did not survive. The interest-rate increase ended the unification boom immediately and drove Germany and its European neighbors into a sharp depression in 1993, followed by years of modest growth that lasted until the first years of the euro. Another shock made the situation for export firms even worse: German firms had been used to having a currency that was almost always undervalued (Höpner, 2019), a circumstance that changed over the course of the 1990s.
With the collapse of the East German economy after its supply-side shock, manufacturing firms began to dismantle the newly established bargaining structures. Neither willing nor able to accept the pay rises resulting from the planned harmonization of Eastern and Western wages in the Stufenplan agreed in 1991, they undercut sectorally agreed wages and working-time standards, pushed for opening clauses, or left the employers' associations altogether (French, 2000;Silvia, 1997).
With regard to wage policy, exceptional frustration emerged in the year 1995. The moderate wage agreements of the year before had disappointed many trade unionists and led to an increased willingness to strike for higher wages that year. The strikes took place in the Bavarian metals industry and brought about a wage increase of 5.1%, a concession for which the employer association Gesamtmetall was harshly criticized by many of its members (Behrens, 2011: 168-169). This led to critical discussions about the lack of flexibility that collective agreements offered to firms under cost pressure. Yet another incident of high symbolic significance added fuel to the fire: many collective agreements asked firms to complete their phased transition to the 35-hour week, a measure that small and medium-sized firms had strongly opposed. 5 The protest against the 1995 metals industry agreement was particularly pronounced among small and medium-sized firms. In the metals sector, smaller firms tend to supply larger ones, rather than the other way around. Given the quasi-monopolistic relationship between purchasers and suppliers in the sector, the purchasers could dictate prices, and did so increasingly ruthlessly (Haipeter, 2017: 308). 'We have to deliver on terms that are brutally forced upon us' (Spiegel, 13 March 1995: 105), 6 was a typical statement in that context. Increased import competition by Chinese and Eastern European firms exposed smaller, less productive companies to further cost pressure (Baumgarten and Lehwald, 2019).
There was not much the small and medium-sized firms could do against this unusual cumulation of uncomfortable conditions, but at least they could ask their employers' associations for more wage flexibility. Many of them not only expressed their dissatisfaction, but also left the associations, while many new firms did not join them from the outset. The associations had to respond in some way.

A new path: OT memberships
As Brandl and Bechter (2019: 472) point out, significant institutional transformations of wage-bargaining regimes are often 'triggered' by harsh economic shocks. The new path chosen by employers' associations was the introduction of OT memberships. The intention was to send signals in two directions: to trade unions and to smaller member firms. With OT, employers' associations could credibly make the threat that if trade unions made wage demands that were too ambitious and resisted the introduction of more opening clauses, this would push coverage rates down -a strategy pursued in particular by Werner Stumpfe, Gesamtmetall president between 1996 and 2000 (Haipeter, 2017: 312). 7 Equally important was the signal this move sent to small members: 'See, we have finally discovered an instrument that will discipline trade unions. 8 You can therefore stay.' Hence the intention was partly that OT would work largely as a threat, and would not need to be regularly put into practice.
But OTs became more widespread than anticipated, and thereby became 'locked in.' Not so long ago, in the year 1992, Gesamtmetall had been warning its member associations against experimenting with such forms of membership. The OT pioneers in the metal sector, located in the districts of Baden-Württemberg, Hesse, and East Germany, introduced the new instrument, in the words of Silvia andSchroeder (2007: 1454), 'with reluctance and even embarrassment' (see also Behrens, 2011: 174). OTs, however, diffused across regions and several sectors, including nonexposed sectors such as retail, in part because small and medium-sized firms were beginning to proactively ask their employers' associations to introduce these memberships too (Behrens and Helfen, 2016: 454-455).
The logic of OTs runs counter to that of SBEs. OTs allow firms to choose whether to apply collective agreements voluntarily while remaining association members, whereas SBEs make the application of collective agreements binding for all companies, even those which remain outside of the associations. Hence, branches that rely on SBEs consciously reject the introduction of associational OT-membership statuses, as confirmed by the following quotes from representatives of the construction industry and security services, the two most prominent SBE sectors: ZDB (Zentralverband des Deutschen Baugewerbes, construction industry): There are other branches in which OT associations are of relevance, the metal industry is the best example. Such structures barely exist in construction. In other words, the bargaining coverage is, in contrast to other branches, not an obstacle to joining employers' associations -because we have universal coverage. . . . [Others] have looked for alternatives that only strengthen the associations but do not strengthen the Tarifautonomie. These are the OT structures. 9 Ver.di (Vereinigte Dienstleistungsgewerkschaft, service sector union): We do not have [OTs]. This concerns other branches. This does not exist in security services, either you are in or you are out. 10 The actual motivations for and the effects of OT nevertheless varied across sectors. For some, like chemical industries and banking, amending association statutes to OT was mainly an emulation of Gesamtmetall's approach toward a 'modern' organizational form, designed to secure members' loyalty and enhance recruiting capabilities. In these sectors, actual OT memberships remained the exception (Behrens and Helfen, 2019;Haipeter, 2016: 88), and bargaining coverage remained rather stable. Other sectors, however, used OT as a new way to cast sectoral agreements off altogether. This became evident in the retail sector and had far-reaching consequences for SBEs.
Traditionally, most framework and wage agreements in the retail sector were extended. During the 1990s, many retail firms expressed critical views on their framework agreements, due to their outdated job specifications and wage groupings, both of which partially stemmed from the 1950s. But all sectoral attempts to reform the respective framework agreements had failed since the 1980s. Additionally, the dissatisfaction with multi-employer bargaining peaked after a number of illegal deviations from collective agreements became public and after retail firms faced high charges for unpaid social security contributions (Behrens, 2011: 174-186). Also, in 1999, the splitting up of what had previously been the joint bargaining associations Handelsverband Deutschland (HDE) and Bundesarbeitsgemeinschaft der Mittel-und Großbetriebe des Einzelhandels (BAG) fragmented the bargaining cartel and fostered competition between them. Under these circumstances, it became the explicit goal of retail firms to get rid of SBEsinstruments that, in their view, imposed harmful and outdated collective agreements on them. OT memberships became a welcome tool with which to dismantle multi-employer bargaining. The traditional retail practice of applying for extensions ended in the year 2000 and gave way to a low-wage, high-flexibility strategy. The retail episode is a good illustration of how the 'reprogramming' (Behrens and Helfen, 2016: 452) of employers' associations by means of OT undermined SBEs. 11

The path not chosen
With their OT experiments, employers in manufacturing and further sectors embarked on a new path that changed the basic principles of German industrial relations. This path was not without an alternative. For example, employers could have threatened trade unions with the OT scenario -uncontrolled decentralization, the worst outcome from the perspective of trade unions -while at the same time offering a scenario such as the following: use of SBEs in the export sector as well, in exchange for, firstly, extensive internal (controlled) flexibilization of collective agreements and, secondly, a mid-term commitment to restrained wage policies. From today's perspective, all those involved would probably have been better off in the end in this scenario, particularly with regard to organizational representativeness. The first, ultimately unsuccessful Bündnis für Arbeit (Alliance for Jobs), suggested by IG Metall chairman Klaus Zwickel in the autumn of 1995, could have provided the framework for such an agreement.
Based on the logic of influence, there would have been good arguments for such a solution. However, the associations on both sides were also subject to a logic of membership (Schmitter and Streeck, 1999). On the part of the unions, the wage restraint that had already taken place since the end of the unification boom had dissatisfied members, leading to the 1995 Bavarian strike. Similarly, employers' organizations had already, as we saw above, faced waves of member dissatisfaction since the early 1990s. To them, the 1995 wage round had signaled the need to quickly find leverage to discipline trade unions. In addition, there had been no historical experience of the SBE instrument in most areas of the export sector. Paster et al.'s argument that ideational legacies matter is very plausible here.
Even if one considers that the chances for SBEs in the export sectors were poor, an acceptance of controlled decentralization on the part of IG Metall, like that which would take place in 2004 with the Pforzheim agreement, might have encouraged those on the employers' side to correct the OT path before wide diffusion. This might at least have slowed down uncontrolled decentralization. From the point of view of the trade unions, however, this would have been a highly unequal exchange with no immediate gain. It would therefore have been extremely difficult to sell this to the members.
In sum, the shocks reported above did not dictate the path taken. There were alternatives, but they were not chosen. The crucial benefit of OTs for firms dissatisfied with their sectors' collective bargaining system -flexibility on demand -became enshrined in large parts of the German system. In 2019, according to its president Rainer Dulger, Gesamtmetall had 3400 firms as 'normal' members (with 1.9 million employees in total) and 3900 firms as OT members (with 570,000 employees). 12 In this context, a path correction is highly unlikely. This lock-in effect is only partly the result of 'increasing returns' (Pierson, 2000: 257-259) in the form of membership stabilization. 13 More important are the increasing costs of potential path correction, given that OT members would most likely rather leave associations entirely than return into the scope of application of collective agreements if OT memberships were to be abolished by the means of SBEs. Implementation of a Dutch-style economy-wide use of SBEs in order to prevent coverage rates from falling further is therefore almost impossible.
If any other European case were concerned, we could end our analysis here and conclude that the remaining German SBEs remain a matter of 'sectoral voluntarism.' But this is only half the story. Even in sectors with a strong consensus in favor of SBEs, their survival is by no means guaranteed: the path change analyzed above has also set in motion an intersectoral liberalization dynamic, to which we now turn.

The BDA exercises its veto power
Sectoral divides over wage-bargaining matters are by no means a uniquely German feature, but also exist in countries with stable SBE practices (more details in the Conclusion). In Finland, for example, export sectors are applying pressure to further liberalize the wage-bargaining system (Bergholm and Bieler, 2013;Günther, 2021: Ch. 7). However, they cannot veto SBEs, because an independent commission checks collective agreements and makes decisions about extensions. In Switzerland, the social partners in construction managed to revitalize SBEs, against resistance from export-oriented employers (Afonso, 2017;Fischer, 2002). Germany, however, is different: SBEs can be vetoed by representatives from other sectors. This is due to a German singularity (Schulten, 2012: 490-491): the formal 14 veto point of the intersectoral umbrella organization BDA and its representatives in the ministries' collective agreement committees (Tarifausschüsse). These committees are located in the federal and Länder employment ministries and need to approve sectoral SBE requests. In every other EU country, SBEs can be approved by the labor minister or impartial commissions alone, which often, as in the Finnish example above, happens quasi-automatically. In the Netherlands, social partners informally solve potential conflicts before an application, especially if an agreement within one sector collides with one in another; in case of conflict, the bipartite Labor Foundation (Stichting van de Arbeid, StvdA) works as a consensual coordination platform, yet open disputes are very rare. 15 Over decades, this veto point had been a 'sleeping resource' without frequent activation. But in the course of the reorientations within the employers' camp that followed the shocks described earlier in the article, the BDA began to make strategic use of that resource. In the year 2000, the BDA adopted 'coordination guidelines' and thereby announced a more critical stance toward SBEs: extensions, according to the guidelines, must remain an 'exceptional instrument.' Competition from outsider firms alone does not justify extensions. Rather, applicants have to prove that a considerable number of outside firms pay substantially less than the sectorally agreed wages. The BDA also denies that there is a public interest in extending entire wage schemes, and higher-wage groups in particular (BDA, 2000, cited in Bispinck et al., 2003. In its amended guidelines for the 2010s, the BDA announced that it would tolerate extensions of wage agreements for the lowest wage groups only, and only if the sectoral minimum wage in question were not above that of comparable sectors (BDA, 2010). 16 This change in strategy had a significant impact. SBE applicants have reported frequent conflicts, especially at the regional level. The number of collective agreements newly declared generally binding has decreased sharply. In 1996, 145 agreements were extended. This number dropped to 50 in 2003 and went down even further to an all-time low of only 10 in 2010, followed by figures fluctuating between 27 and 38 up to 2016 (BMAS, 2017: 7). The share of rejected applications had been 1-6% between 1991 and 1997 only (Bispinck, 2012: 501). Between 2000 and 2018, however, up to 30% of all applications were rejected or withdrawn by the applicants due to limited prospects of success (Deutscher Bundestag, 2019). 17 The prospects of success vary across regional committees, depending on the sitting representatives. In order to succeed, informal concessions are often required before a formal application is made. Employer representatives usually coordinate SBEs with the intersectoral (regional or national) BDA before they are passed on to the Tarifausschuss (Vogel, 2019: 46, 67). Hence, the veto option allows third parties to rule on matters of other sectors' bargaining arrangements, as the following quotes show. BDSW (Bundesverband der Sicherheitswirtschaft, security services): We had identical collective agreements for Hesse and Rhineland-Palatinate. The committee meetings were held on two consecutive days. On Monday, the committee in Rhineland-Palatinate accepted the SBE for a part of the agreement. It became common practice to not extend entire wage schemes but selected wage groups or wages up to 10, 11, or 12 euro only. On Tuesday in Hesse, the employers rejected the entire application, although we are a member association of the Hessian peak employers' federation. 18 BDSW (Bundesverband der Sicherheitswirtschaft, security services): We used to make entire wage schemes universally binding in the past. Today, in some Länder we only get the lowest wage, in others selected wage groups only, in yet others like Hamburg we get the entire wage range. It varies a lot.
Interviewer: And what would you have to do in order to make all wage groups universally binding? What is the obstacle? Would you have to make concessions to the employers' representatives before you file the application? BDSW: Well, such incidents should not happen, but yes, it is exactly like this. In North Rhine-Westphalia, for example, we needed to identify one wage group to leave out, in order to accommodate the employers' wishes in the committee. So we can say: See, we'll take away another part here, is it okay then? 19 DGB (Deutscher Gewerkschaftsbund, German Trade Union Federation): An important issue is the voting process in the committee. This concerns the famous veto option. Formally, the employers' representatives can block an SBE application with a veto. This has not been directly used in the recent past. Yet, we are repeatedly confronted with it, for example with respect to the restriction of the [sectoral] range of application of an SBE. And it is clear, if we do not go along with it, we run the risk that the SBE will be blocked entirely. It is like the sword of Damocles constantly hanging above us. 20 The last quote makes clear that the BDA's veto power has implications even where it is not openly applied. The awareness that the BDA can always exercise the veto is enough to enforce concessions. But why would employers care about extensions in sectors other than their own?

An intersectoral liberalization dynamic
Two closely related factors can encourage employers to promote the erosion of coverage in sectors other than their own. The respective incentive is unequally distributed between exposed sectors on the one hand, and sheltered sectors on the other.
First, wage pressure can diffuse across sectors. If a sector holds nominal wage increases down in a first round while others engage in an inflationary wage policy, the sectoral association can choose between two bad options: sticking to the below-average nominal wage increases by further resisting trade unions' demands, and thereby having to sell real wage losses due to increased overall inflation to the employees within the sector, or accepting higher wages in the next round; in other words, letting the spillover happen. International competition promotes wage restraint in exposed sectors, but this mechanism is absent in sheltered sectors. This is why inflationary wage pressure often originates in nonexposed sectors, including public sectors, as happened in several Eurozone countries during the first 10 years of the euro (Di Carlo, 2018;Hancké, 2013;Höpner and Lutter, 2018;Johnston, 2012). 21 Exposed sectors therefore have a particularly pronounced interest in ruling on nonexposed sectors' wage policy. 22 Second, not only wage pressure, but also cost pressure can diffuse across sectors. Remember that the transmission of cost-cutting pressure by means of squeezing suppliers had become a widely used management tool in Germany's cost-sensitive export sector in the 1990s. From the perspective of this sector, the wage level in the internal sector is among the determinants of its own cost pressure. In light of these considerations, it does not come as a surprise that all filed -and all rejected -applications for extensions within the last two decades stemmed from the sheltered service and trade sectors (BMAS, 2017; Deutscher Bundestag, 2019).
The following example illustrates this problem. In the mid-1990s, the introduction of the German Posted Workers Act (Arbeitnehmerentsendegesetz) brought about conflict between Gesamtmetall and the intersectoral umbrella organization BDA, on the one hand, and employers' associations in construction (mainly the HDB, or Hauptverband der Bauindustrie), on the other. In the course of this intersectoral conflict, the BDA, in consensus with the export-oriented associations Gesamtmetall and Gesamttextil (textile industries), vetoed the application for extending a minimum wage agreement at the national collective bargaining committee in 1996 (Eichhorst, 2000: 250-252).
The remarkable aspect here is the high visibility of the 'smoking gun'; that is, of the actual motives behind the BDA's resistance. Given the export sectors' interest in cheap construction services, the BDA neither wanted an extension of the sectoral minimum wage agreement in question, nor did it want that agreement to be applied to posted employees. In addition to its concerns about the risk of a general wage increase beyond the construction sector, the BDA expressed 'concerns that construction services would become more expensive as a result, with negative effects on willingness to invest and employment.' 23 Two further rounds of negotiations in the Tarifausschuss and an arbitration by Labor Minister Norbert Blüm were necessary before the BDA finally agreed to the SBE based on a 17% lower minimum wage: 17.00 instead of 18.60 Deutsche Mark.
This episode makes the logic behind the intersectoral liberalization dynamic clear. In the German context, cost-cutting on the part of the export sector creates incentives to frustrate SBEs in the domestic sector. The BDA's veto point, a political-institutional feature, enables such frustration. As we have seen above, the historically inherited idea of the desirability of wage competition can help us to understand why Gesamtmetall, faced with economic shocks and discontent on the side of its members, refrained from responding with attempts to increase its representativeness by means of SBEs. The same could be said about the lack of stability of the pro-SBE consensus in the retail sector. 24 The intersectoral dynamic, however, is of a different kind. It expresses a purely material, cost-related interest. In the construction conflict, the BDA applied pressure until a lower construction minimum wage was extended.
To sum up these insights, the path changes among manufacturing employers that occurred in the 1990s not only led to the spread of OT in exposed sectors but also to a willingness to frustrate SBEs in nonexposed sectors. This intersectoral liberalization dynamic, part of the transformation of the German economy into what Baccaro and Pontusson (2016) refer to as an asymmetrically export-driven growth model, should be distinguished from the historical-ideational legacy that led to the nondiffusion of SBEs in German manufacturing after the Second World War (Paster et al., 2020). As we make clear in the conclusion, taking this dynamic into account is important in terms of its implications for reformers seeking to reverse the trend.

Conclusion
We conclude with a brief discussion of the policy-advisory implications of this study and of the broader lessons for research in comparative labor relations.
We have distinguished two dynamics that have eroded the use of SBEs in Germany since around the turn of the twenty-first century: an intrasectoral and an intersectoral dynamic. Firstly, there are sectors in which employers and their associations have had no historical experience with SBEs, as Paster et al. (2020) rightly emphasize, and that have particularly come to appreciate wage competition from outsiders as a means of disciplining trade unions. This applies in particular to the leading export-sector association, Gesamtmetall. Given these circumstances, even a fundamental rethink of the advantages and disadvantages of wage competition on the part of association leaders would not result in a more positive stance toward SBEs, because the associations have tied their own hands through the introduction of OT memberships. The introduction of SBEs in the metals sector, or likewise in the retail sector, would almost certainly lead to mass withdrawals by current OT members. Attempts at persuasion, therefore, have little chance of success.
But the remaining alternative, the implementation of SBEs against the will of the representative employers' associations, does not appear realistic in the short-to-medium term either. Even if governing parties were to pass a corresponding reform, a comprehensive rethinking on the part of jurisprudence and courts would be necessary. Unless that happens, the extensive interpretation of the principle of 'negative freedom of association' would frustrate such a measure. Against this background, regaining a high level of collective bargaining coverage in these sectors can only be achieved through regained union power, and most likely not through SBEs.
Secondly, we have distinguished the above dynamic from the intersectoral liberalization dynamic. In the course of the reorientations of the 1990s, the BDA began to make extensive use of its veto power in the collective bargaining committees. The veto frustrates SBEs in the internal sectors even against the will of the sectoral employers' associations or, as in the case of the conflict over generally binding minimum wages in the construction sector discussed above, at least enforces lower wages. This dynamic is not about inherited ideas about desirable modes of wage determination in one's own sector, but about the cost advantages that derive from the maximization of wage competition in other sectors.
This dynamic could definitely be addressed through political reforms, provided that reformers strive to turn the trend around. A reform of the Collective Bargaining Act could yield to European standards by eliminating the BDA's veto point in cases where the SBE had been applied for jointly by the sectoral associations on both sides. For such cases, the reform could rule that the rejection of applications requires a majority in the collective bargaining committees. In contrast, if the SBE was only requested by one side (a possibility that existed until a reform in 2014 and that could be reintroduced), the requirement of an approving majority could be maintained. 25 Such a reform could at least bring about a return to sectoral voluntarism: the approximately 200 26 SBE applications that have failed because of the BDA's veto since 2000, or that have been withdrawn from the process due to a low likelihood of success, would be given a new chance. 27 Is it possible to generalize these findings on the German erosion of SBEs? Germany is not a typical case with regard to the chosen research question. On the contrary, the extent and the speed of the erosion of German collective bargaining coverage are, in comparison to other European coordinated market economies, exceptional. The analysis of exceptional cases can nevertheless provide lessons for comparative research.
The sectoral interest divide that has resulted in an exceptional liberalization dynamic in Germany is not country-specific as such; it is universal. Export sectors have a particular interest in containing the costs emanating from the sheltered sectors. This conflict line is a potential trigger of change in all European countries with coordinated wage bargaining. Every industrial relations regime can and should therefore be analyzed from the perspective of this conflict as well. In recent industrial relations research, the conflict has primarily been analyzed with reference to the Nordic countries, where it resulted in the controlled decentralization of formerly centralized collective bargaining systems (Baccaro and Howell, 2017: 152-158;Ibsen, 2016;Marginson and Dølvik, 2020;Thelen, 2014: 60-67). The conflict can also be observed in disputes about the transnational posting of workers, for example in Denmark (Arnholtz et al., 2018: 347). In Sweden, the export employers' struggle for low wage standards for posted workers culminated in the Laval 28 conflict (Müller et al., 2018: 367;Seikel, 2015Seikel, : 1176Seikel, -1177.
The universality of this conflict does not imply, however, that the liberalization of wage-setting in general, or attacks against SBEs in particular, are likely outcomes. The containment of cost pressures emanating from sheltered sectors can take different forms. It can happen through the centralization of collective bargaining systems and wage guidelines controlled by the export sector, through controlled decentralization within the existing collective bargaining systems, or, as in Germany, through their tolerated or consciously promoted erosion. Which strategies are pursued, and especially whether they also include the frustration of SBEs against the will of employers' associations in the respective domestic sectors, depends on parameters that are calibrated differently in different countries. In Germany, this calibration is 'extreme' and has led to an exceptional outcome.
The first parameter is wage drift. It decides whether exits from collective agreements promise falling wage pressure, and whether an erosion of coverage in other sectors may lead to lower cost pressure. The Netherlands and Finland, for example, both stable SBE cases, have positive wage drifts: firms pay more than agreed in the collective agreements, on average. Germany shows a different pattern. As data from the DGB indicate, collectively agreed wages rose more sharply in the years 2000-2015 than those in noncovered areas (Deutscher Gewerkschaftsbund, 2021). Lübker and Schulten (2021) assume that covered firms pay 11% higher wages than noncovered ones, under statistical control for firm size, economic sector, qualification structure, and technical status of machinery. However, it should be noted that the economy-wide averages hide sectoral and regional variance, in Germany as well as in countries with a positive wage drift.
The second parameter concerns the ideational sphere emphasized by Paster et al. (2020). Even if wage and cost pressure can be minimized through forced erosion of coverage, this strategy presents employers' associations with a disadvantage: they forfeit their representativeness. The weighting of the pros and cons presupposes assessment of their relative importance, which necessarily has a normative, idea-based dimension. Inherited ideational legacies may be decisive here, but do not have to be -new ideas can also emerge erratically, as overreactions in exceptional sets of circumstances. Interestingly, experiments with OT strategies based on the German experience can currently be observed in Denmark ('differentiated memberships'; see Ibsen and Navrbjerg, 2019: 46). Potential positive payoffs and ideas that legitimize the choice of such a strategy are nevertheless not sufficient. The strategy must also have a chance of success. A third parameter is, therefore, the trade union and political countervailing power. In Finland, for example, export employers' actions were directed against SBEs as well (Müller et al., 2018: 361-362), and the same can be observed in the case of Norway (Alsos and Eldring, 2008;Arnholtz et al., 2018: 348;Dølvik and Marginson, 2018: 420). But in neither of these countries could these struggles lead to the frustration of SBEs, due to the countervailing power of SBE supporters (on Finland: Günther, 2021: Ch. 7).
The fight for the introduction of simplified SBEs in the Swiss construction sector was fiercely contested, too -a fight that the unions and employers in construction were ultimately able to win, due to given power relations (Afonso, 2012: 721-723;Fischer, 2002). Since then, in contrast to Germany, Switzerland has significantly more SBEs and an increasing coverage rate (Afonso, 2017;Oesch, 2007: 344-350). Fourthly, and closely related, political-institutional parameters are part of the opportunity structure. As we have shown in detail, the German employers' associations were able to take advantage of an institutional singularity: the institutionalized veto power of the BDA.
In more abstract terms, the intersectoral SBE dynamic can be understood as a special form of dualization. Hassel (2014), among others, has argued that coordination and liberalization can go hand in hand: large German export firms are typically covered by sectoral collective agreements and maintain coordination with their core workforce by means of codetermination (see also Thelen, 2000;Wood, 2001); at the same time, however, they profit from the deregulation of peripheral labor markets because such deregulation reduces cost pressure. Although coordination and liberalization seem contradictory at first glance, they can be functionally complementary from the point of view of export firms (Hassel, 2014: 70). The SBE puzzle discussed in this contribution shows that export employers not only passively profit from this kind of dualization; they also proactively contribute to the erosion of wage coordination in sectors other than their own. 20. Interview DGB, 26 July 2016. 21. Swenson (1991 has used this model to explain Swedish export firms' preference for centralized wage bargaining between the mid-1960s and the late 1980s. 22. Lesch et al. (2017: 12-15) from the employer-financed Institut der deutschen Wirtschaft (IW) make this point particularly clear: sheltered sectors need sufficient wage competition to be exerted by outsiders in order to compensate for the lack of international product market competition. 23. This is a quote from the protocols of the Tarifausschuss, cited in a retrospective DGB (trade union) document from 14 June 2007 (available upon request). 24. However, as Paster et al. (2020: 545) correctly note, the BDA supported the opponents of SBEs in the retail conflict. Therefore, this conflict had an intersectoral dimension, too. 25. This reform proposal corresponds to an initiative that the Länder Berlin, Bremen, and Thuringia introduced to the German Federal Council (the second chamber within Germany's bicameralism) in May 2021. 26. Data source: WSI Tarifarchiv. 27. As an aside, it should be mentioned that current case law of the Court of Justice of the European Union (CJEU) opens up an alternative option: the reform path could also run through the Posted Workers Act, the legal tool that makes collective agreements generally binding and extends them to posted workers. With the Laval ruling (C-341/05), the CJEU declared in 2008 that this possibility applies to sectoral minimum wages, but not to entire tariff grids. In 2015, however, the CJEU retreated from this strict reading (C-396/13), and the reform of the Posted Workers Directive in 2018 clarified that the transfer of complete tariff grids is now unequivocally possible. 28. The Laval ruling of the CJEU originated from a case that the Swedish employers' associations had brought to the Swedish courts.