State-led bricolage and the extension of collective governance: Hybridity in the Swiss skill formation system

This paper explores the extension of collective governance to sectors without collective governance tradition. We introduce the concept of state-led bricolage to analyze the expansion of the Swiss apprenticeship training system – in which employer associations ful ﬁ ll core collective governance tasks – to economic sectors in which training had previously followed a school-based and state-oriented logic. In deindustrializing societies, these sectors are key for the survival of collectively governed training systems. Through a mixed-methods analysis, we examine the reform process that led to the creation of new intermediary organizations that enable collective governance in these sectors. In addition, we compare the organizational features of these organizations with the respective organizations in the traditional crafts and industry sectors. We ﬁ nd that the new organizations result from state-led bricolage. They are hybrid organizations that re ﬂ ect some of the bricoleur ’ s core policy goals and critically build on the combination of associational and state-oriented institutional logics.


Introduction
Institutional orders emphasizing collective governance have attracted considerable attention as alternatives to more market-based or state-based orders because of their ability to produce desirable socio-economic outcomes (Hall & Soskice 2001;Streeck & Kenworthy 2005;Busemeyer 2015). In terms of institutional logics (Deeg 2005;Carstensen 2017), collective governance relies on a specific logic that has evolved historically (Streeck & Schmitter 1985). This associational logic has an important effect on employers' capacity to provide collective goods and contribute to social goals (Culpepper 2003;Campbell 2007). Yet, due to increasing competitive pressures in a globalizing economy and the decline of manufacturing as stronghold of the associational logic, systems based on employers' provision of collective goods are under pressure to adjust (Martin & Thelen 2007). Prominent examples are dual vocational education and training (VET) systems in coordinated market economies (CMEs), which rely on the associational logic to foster the voluntary provision of training by firms (apprenticeships). In these apprenticeship systems, employer associations are assigned key governance roles as they develop training content that fits labor market needs and foster firms' participation in training (Busemeyer & Apprenticeship training in CMEs is widely acknowledged as a prototypical case of collective governance. However, given that the traditional base of apprenticeship training is in the crafts and manufacturing sectors, which are in relative decline, maintaining both strong collective governance and high participation rates requires deepening its linkage to the expanding service sector (Busemeyer & Trampusch 2012). How can the state extend collective governance to sectors that lack the intermediary organizations needed for collective governance? What does transferring the associational logic of traditional apprenticeships imply for collective organizing in the new sectors? And most importantly in the context of this analysis, which kind of reform process manifests itself in the characteristics of the newly created associations? Answers to these questions will provide a better understanding of new forms of collective organizing outside of the sectors in which collective governance is already well established.
In contrast to the emphasis on institutional stability and coherence in the classical conception of varieties of capitalism (Hall & Soskice 2001), we argue that the state may strategically create and enable hybrid organizational forms to solve the problem of transferring collective governance to new sectors. Through what we call state-led bricolage, the state can transfer the associational logic to sectors without collective governance tradition. In this process, the state as bricoleur develops institutional solutions by combining elements of existing institutional principles and practices in new ways, which results in organizations that are different from but potentially resemble the old ones (Campbell 2004). Crucially, the state retreats to its role as enabling state (Schmidt 2009) once the bricolage is achieved. Furthermore, given the state's high capacity as agent of institutional change, the state leaves footprints on the outcome of the change process, most notably by making the new organizations reflect some of its core policy goals. Together, these two features of the reform process lead to hybrid arrangements, as the associational logic gets blended with the logic characterizing the targeted sectors and the state's policy goals are partially imprinted on these logics.
Empirically, our study examines a crucial reform of the Swiss apprenticeship system, which is proportionally the world's largest one (OECD 2017a). Collectively governed apprenticeship training represents an unlikely case for state-led bricolage given that in these systems, private actors have a strong sense of ownership (Busemeyer & Trampusch 2012) and thus typically resist state intervention. Yet, by extending the associational logic to new economic sectors through bricolage, the Swiss state intervened to stabilize the apprenticeship system. While in most traditional dual VET countries, the rate of apprentices relative to students has dropped in recent years, Switzerland stands out with a high and stable rate of around 60 percent of all learners in their first year of upper secondary education enrolled in apprenticeships (OECD 2017a). To achieve this stabilization, the state needed to adapt the system to socio-economic change. When in the 1990s, firms' provision of apprenticeship positions dropped dramatically, the system was reformed with the aim of maintaining it. A crucial measure was expanding the VET system to economic sectors that were not yet part of the system: agriculture, forestry, arts, health and social care (see Appendix S1, Table A1). Training in these sectors had never followed an associational logic but a stateoriented one and primarily took place in schools rather than firms. Thus, the reform addressed the challenge to transfer the associational logic of traditional apprenticeships to sectors that followed different organizing principles and lacked traditional employer associations in charge of VET.
To transfer the associational logic to the new sectors, the state sponsored the creation of new organizations to fulfill employer associations' tasks. At the formal level, authorities introduced a new umbrella term to include these organizations in the system. Both the new organizations and the traditional employer associations are called "Organizations of the World of Work" (Organisationen der Arbeitswelt, OdA). This deliberately broad term allowed including organizations in VET governance that do not necessarily correspond to the typical organizational form of employer associations in the VET field.
Our case study of one New-Sector-OdA (i.e., OdA in the newly integrated sectors) traces the change process and shows that the combination of elements of different institutional logics and the imprinting of some of the state's core policy goals made the resulting New-Sector-OdA a hybrid organization. We find that this organization (OdASanté) dependedat least initiallyon the capacity of the state to act as bricoleur to encourage private actors to collectively organize in new ways. This allowed to integrate an important economic sector into the VET system, which was essential to the survival of collective governance in a changing environment. To explore whether New-Sector-OdA more generally display hybrid characteristics, we subsequently compare the organizational characteristics of all New-Sector-OdA to the Old-Sector-OdA (i.e., OdA in the traditional sectors for apprenticeship training) based on an original survey of all 146 Swiss OdA responsible for initial VET. The statistical analysis supports our argument that the creation of hybrid organizational forms helps to overcome public policy problems related to collective governance.
In the next section, we develop our theoretical framework. First, we introduce the institutional logics of the state and the association and we theorize hybridity in the context of state-led bricolage. Then, we operationalize hybridity and describe our methods and data. Next, we present our findings on New-Sector-OdA through an historical account of the relevant VET reform, a case study of an exemplary New-Sector-OdA (OdASanté), and a cross-sectional comparison of New-and Old-Sector-OdA. A final section concludes.

Theoretical framework: State-led bricolage and hybridity
This section develops our core argument around state-led bricolage and hybrid organizations. We start by introducing the institutional logics of the state and the association as two central building blocks for this argument.
2.1. Institutional logics: The state logic and the associational logic Each of the major institutional orders of contemporary Western societies has a key institutional logic. Deeg (2005, p. 172) refers to logics as "typical strategies, routine approaches to problems, and shared decision rules that produce predictable patterns of behavior by actors" within an institutional system (see also Streeck & Thelen 2005, pp. 12-13;Carstensen 2017, pp. 144-145). We follow Streeck and Schmitter's (1985) famous distinction between four institutional orders and respective logics. The guiding principle of interaction in the logic of the state is hierarchical control, in the logic of the association it is organizational concertation, in the logic of the market the authors refer to dispersed competition, and in the logic of the community to spontaneous solidarity. In this paper, the state and associational logics are of relevance, which is why we introduce them in greater detail (see Appendix S1, Table A2, for additional information).
As Streeck and Schmitter (1985, p. 6) put it, "in the ideal-typical state bureaucracy, allocational decisions are made through public policies that are enforced, with the ultimate backing of the state's monopoly on legitimate coercion, by civil servants striving to satisfy their dominant interest in career advancement and bureaucratic stability, on subjects which strive to avoid punishment." In this depiction, the focus lies on the state as bureaucracy rather than partisan aspects. State agencies are the main actors that devise authoritative regulations to which other actors abide because otherwise they face sanctions. The state logic hence resonates with the concept of a regulatory state that devises rules, monitors, and sanctions them (Koop & Lodge 2017). In statist skill formation, VET is typically school based and employers' involvement in training provision limited (Busemeyer & Trampusch 2012), which together with significant involvement of public actors reflects a statist logic.
The state logic captures well the pre-reform state-oriented logic of training in economic sectors in Switzerland which were not yet part of the apprenticeship training system. In these sectors, training was carried out under the leadership of regional administrations, while being strongly school-oriented and characterized by weak involvement of employers and their associations (see case study below for details). Consequently, we refer to the pre-reform logic of the more school-based forms of training in the economic sectors that were to be integrated into the collective governance system of apprenticeship training as state logic.
In contrast, apprenticeship training in collectivist systems in countries like Germany and Switzerland is typically characterized by an associational logic (Busemeyer & Trampusch 2012). In this logic, "actors are contingently or strategically interdependent in the sense that actions of organized collectivities can have a predictable and determinant effect (positive or negative) on the satisfaction of other collectivities' interests, and this induces them to search for relatively stable pacts" (Streeck & Schmitter 1985, p. 11). In this logic, coordination and networks among employers are strongly developed and provide an alternative, for instance, to coordination by the market. The associational logic allows firms to overcome cooperation dilemmas in various institutional arenas, such as industrial relations or skill formation (Culpepper 2003).
The associational logic is characterized by the systematic involvement of employer associations and labor unions in policymaking and implementation (Streeck & Kenworthy 2005). Focusing on employers, 1 ideal-typical associations are membership-based with associations representing employer interests in a broad array of policy fields often along sectoral lines (Farago & Kriesi 1986;Behrens & Helfen 2009). For example, they negotiate collective labor agreements with unions. Typically, such associations of firms are members in higher-level umbrella associations. This allows a hierarchically structured and coordinated aggregation and representation of employer interests (Schmitter & Streeck 1999). Umbrella associations represent employer interests in policy forums such as consultative commissions with public authorities. Moreover, the state conveys public tasks to employer associations, which is thought to improve the efficiency of implementation and lower its costs (Streeck & Schmitter 1985). This cooperation with the state is said to promote the organizational development of employer associations (Traxler & Huemer 2007).
However, finding a balance between cooperating with the state to gain influence in policymaking and remaining faithful to members' interests is challenging for such intermediary associations (Schmitter & Streeck 1999). For example, cooperating with the state may imply pursuing more social goals than firms would otherwise consider (Campbell 2007). Although the associational logic increases the organizational capacity of associations, not all associations are equally developed. Many of the traditional associations have existed for a long time (some date back to the mid-19th century), but they can vary, for instance, regarding their degree of professionalism. For example, many associations unite small groups of firms and rely on voluntary member support, while some are highly professionalized and employ paid staff.

The role of the state for fostering the associational logic
In countries and policy fields that follow the associational logic, employer associations capable of finding a balance between cooperation with the state and representing members' interests are crucial actors (Schmitter & Streeck 1999). Collectively governed skill formation, which combines training at the workplace and in schools, traditionally builds on associations that foster cooperation among private actors as well as with public actors for the administration and reform of training (Culpepper 2003;Busemeyer & Trampusch 2012). However, such capable associations do not exist everywhere. Typically, institutions fostering collective organizing of employers have their origins in the crafts and industry sectors (Thelen 2007).
Yet, the importance of the crafts and industry sectors in terms of employment is decreasing. Moreover, economies' skill needs are changing not least due to technological change and demographic developments, which trigger an increased need of, for instance, health and care professionals (OECD 2017b). Consequently, if the involved stakeholders want to maintain the associational logic in skill formation, they need to adapt it to these developments. This adaptation is challenging, however. In countries that rely on the associational logic, the state may be reluctant to or lack the capacity for coercive state intervention. Due to the institutional arrangements between private actors and the state, the state may need to negotiate with the involved actors and gain agreement on reforms, otherwise reforms may be blocked or fail (Schmidt 2009, p. 525). Moreover, as the state has delegated important public tasks to private actors, such as the provision of the collective good of transferable training, it now depends on the private actors' continuous provision of these services (Streeck & Schmitter 1985).
Hence, traditionally, under the associational logic, the state acts as enabler. It facilitates private actors' decentralized cooperation (Culpepper 2003). It delegates public tasks to private actors, to business associations, and labor unions, and these actors are involved in defining rules in areas such as wage bargaining or apprenticeships (Hall & Soskice 2001;Streeck & Kenworthy 2005). Thus, the enabling state fosters nonmarket coordinating institutions (Schmidt 2009, pp. 521-522). In the heyday of this associational logic, the state could retreat to a partly passive, reactive, and indirect role, giving substantial autonomy to private actors to manage their own affairs (Streeck & Kenworthy 2005).
Yet this institutional arrangement was based on the capacity of intermediary associations to enforce collective decisions among members. In a context of deindustrialization and liberalizing pressures, state inactivity may lead to the exhaustion of institutions such as collective skill formation. Hence, the state may be pushed to adopt a more active role to ensure the sufficient provision of vocational training. Following Schmidt (2009, p. 526), the state could, theoretically, take over the functions it delegated to private actors or coerce employers to offer training slots. However, this option would imply covering important costs and losing the close connection between the labor market and training for which dual VET is valued. In addition, such coercive state intervention may be blocked in the context of a system that is marked by cooperation and negotiation between the state and private actors (Hall & Soskice 2001). However, inactivity is not a desirable option either, as private actors left on their own encounter important obstacles to engage in the provision of collective goods (Culpepper 2003).
Consequently, to maintain the existing associational logic in VET, we argue that the state needs to adopt a more active roleat least temporarily. One approach is to expand the system to new economic sectors. 2 Yet, in these sectors the state may lack partners for the delegation of tasks, most importantly intermediary associations of employers that help overcoming cooperation dilemmas among private actors. Moreover, previously involved stakeholders in training in the new sectors might not be easily displaced. In the next section, we argue that through bricolage, the state may attempt combining the associational logic of the apprenticeship system with previous institutional logics in newly integrated sectors. In this way, and in contrast to approaches that emphasize the importance of keeping institutional logics distinct (Hall & Soskice 2001), the state may be able to foster private actors' organizational capacity to fulfill traditional governance functions before resorting to its primarily enabling role and leaving collective good provision to private actors.
2.3. Theorizing state-led bricolage and hybridity 2.3.1. Institutional variety and state-led bricolage While ideal types such as associational and state logics help to understand and explain differences between countries, in reality, a variety of institutional principles and practices can coexist in a single political economy (Campbell 2010). For example, high-quality apprenticeships may exist outside countries characterized by the associational logic in specific industries or regions (Fortwengel et al. 2019). Hence, institutional variety refers to the coexistence of institutional principles and practices that remain distinct. If one sees institutions as resources to accomplish goals (Streeck & Thelen 2005), a next logical step is to understand such institutional variety as enabling institutional change processes as it provides actors with a range of institutional elements to work with (Crouch 2005). In this context, the basic idea behind the concept of bricolage is that actors frequently develop institutional solutions by combining elements of existing institutional principles and practices in new ways, which results in institutions that are different from but potentially resemble the old ones (Campbell 2004). Bricolage thus typically represents change in a rather evolutionary and gradual way (Carstensen 2017). Here, rather than one institutional logic replacing another completely, a new arrangement is created. If such an arrangement blends elements of the two pre-existing institutional logics, these arrangements are defined as hybrids (Haveman & Rao 2006). While bricolage was originally associated with bottom-up, ad hoc, or situationally improvised change (Lévi-Strauss 1962), we follow authors like Campbell (2004) and Carstensen (2017) who suggest that bricolage can also be promoted more strategically by change agents. Carstensen (2017) highlights that bricolage is a key tool of policymakers to respond to policy challenges. Carstensen (2017) shows that in the case of disruptions to a socio-economic systemin his case a financial crisisstates often resort to bricolage to address policy problems. As a result, the pre-reform institutional logic in the relevant policy field may change gradually but significantly. Like Carstensen (2017), we study a case in which bricolage is a response to a policy challengethe intricate task to extend collective governance to new economic sectorswhich led policymakers to seek new policy solutions. In our empirical case, the respective policy processes are linked to policy solutions crystalized in the 2002 VET law and especially the creation of associations that are needed to extend collective governance to sectors previously governed by a state-oriented logic.
Importantly, it matters who the bricoleur is. The capacity of a bricoleur to creatively combine institutional elements depends on its social, organizational, and institutional connections to relevant institutions and actorswhich enlarges the institutional repertoire available for bricolage. Furthermore, this capacity is increased by the bricoleur's tangible resources, like financial means or political clout, which are needed, for instance, to make institutional changes stick (Campbell 2004, pp. 74-77). In our empirical case, the bricoleur is not anyone, but the state. The respective Swiss state agencies are in a privileged position not only in terms of their connections to relevant stakeholders in collective skill formation, but also in terms of the resources available to them to facilitate and maintain institutional changes. Hence, our bricoleur has a relatively high capacity to promote bricolage. However, this state-led bricolage is not a new type of bricolage, but rather a bricolage process for which we identify the state as the key change agent. This high capacity also implies that the state can strongly influence the process of gradual change. Therefore, the result of bricolage is likely to reflect, in one way or another, core policy goals of the state, which are likely to deviate somewhat from the goals of associations typically in charge of training occupations in collectivist systems (Bonoli & Wilson 2019). In this way, the state as a bricoleur may leave a footprint on the outcome of the change process. This, in turn, implies that the state's policy goals get partially imprinted on the respective other logic(s)with specific consequences that we are conceptualizing in the following.
In our empirical case, the relevant institutional variety available to the bricoleur consists of the state-oriented logic of VET in the sectors to be integrated in apprenticeship training and the associational logic in the traditional industry and crafts sectors. Importantly, the integration of previously state-oriented VET into the dominant dual VET system based on the associational logic potentially reduces institutional variety at the national level. However, in the process of bricolage the state created new collective actors in the affected sectors, which differ from traditional intermediary associations. Hence, we focus our attention on the characteristics of these new organizations, which enabled the expansion of collective governance to new sectors, by building on the rich literature on hybrid organizations presented in the next section.

Conceptualizing hybridity and state-led bricolage at the organizational level
While the concept of bricolage speaks to both institutions and their organizational carriers, we focus on measuring the outcomes of bricolage for the latter. In our empirical analysis, the main unit of analysis is the "Organizations of the World of Work" (OdA) as essential intermediary organizations within VET governance in Switzerland. The OdA can be seen as the crystallizing points through which the change process related to the 2002 VET reform is institutionalized at the organizational level. More specifically, if our argument around stateled bricolage holds water, we should observe that the newly created OdA, which result from a creative combination of state and associational logics through state-led bricolage, are hybrid organizational forms.
To test our argument empirically, we zoom in on the organizational level and the concept of hybrid organizations. Broadly speaking, one can understand hybrid organizations as entities that arise from the blending of two or more institutional logics, that is, "[h]ybrid organizations combine the institutional logics that are materialized in two or more organizational forms" (Haveman & Rao 2006, p. 974). Thereby, hybrid organizations allow overcoming strict boundaries between two institutional logics. Hybrid organizations can play a central role in processes of institutional change and maintenance (Graf 2013). Crucially, while an organizational field may be exposed to considerable tensions between logics, hybrid organizations can contribute to stabilizing the relevant (sub-)field over time as they develop organizational features to accommodate heterogeneity and thus handle tensions internally (Reay & Hinings 2009), as they can "develop structures and processes that allow them to involve a variety of stakeholders, pursue multiple and often conflicting goals and engage in divergent or (seemingly) inconsistent activities" (Mair et al. 2015, p. 733). Thus, we adopt the perspective that hybridsinstead of necessarily being temporary organizational arrangementscan be long-lasting and have the potential to contribute to permanent policy solutions.
Institutionalist research has shown that disposing of coercive authority, the state is a potential driver of organizational adaptation processes (Greenwood & Hinings 1993). Although state-promoted hybridization may also be associated with unintended and conflictual outcomes (Gornitzka & Maassen 2000, p. 284), in this paper we argue that the state can strategically develop hybrid organizations to solve a public policy problem, namely establishing collective governance in new sectors. We thus suggest that the state has the capacity to establish hybrid organizations as a means of system development, for instance, to foster exchanges between relevant stakeholders.
In sum, we analyze a case of bricolage, whereby the state combines associational and state logics to reach an instrumental policy goal, the extension of collective governance to sectors that previously followed a state logic for offering vocational training. As the two institutional logics are combined, this should foster hybrid arrangements. Furthermore, the state has a high capacity to act as bricoleur. Once more, this is likely to promote hybridity because due to its strong position, the state's policy goals are likely to getat least partlyimprinted on the associational logic in the newly integrated sectors. For instance, the state being the bricoleur may imply that societal goals are promoted beyond what would be typically associated with the traditional associational logic. Consequently, the new intermediary organizations (New-Sector-OdA) created with the 2002 VET reform through what we call state-led bricolage should differ from the organizational forms traditionally prevalent in the VET field (Old-Sector-OdA). We expect New-Sector-OdA to represent hybrid arrangements both due to the footprint of the bricoleur on the reform process and the combination of associational and state logics.

Research design: Operationalization, methods, and data
This section introduces an operationalization that enables us to measure the hybridity between the associational and the state-oriented logic at the organizational level. We also describe our methods and data.

Operationalizing hybridity
In the following, we employ our arguments around state-led bricolage and hybrid organizations to develop specific expectations for New-Sector-OdA. As a structuring element, we refer to three well-established dimensions of organizations, which apply to any organization, irrespective of more specific organizational characteristics (Aldrich & Ruef 2006): First, goal direction means that organizations are purposive systems that do not just follow some random path. Second, organizations are maintained through the control of socially constructed boundaries, which implies a distinction between members and non-members. Third, activity system refers to a set of socially constructed routines: "forms, rules, procedures, conventions, strategies, and technologies around which organizations are constructed and through which they operate" (Aldrich & Ruef 2006, p. 6).
For each of these organizational dimensions (see Table 1, column 1), we develop indicators regarding associations in the Swiss VET system (column 2). We then derive theoretical expectations of the differences between New-Sector-OdA and Old-Sector-OdA (column 3). The more each of the indicators reflects hybridity of the associational and state logics, the higher is the overall degree of hybridity of the respective organization. Our core expectation is that while both Old-and New-Sector OdA enable collective governance, they differ as Old-Sector-OdA embrace the associational logic, whereas New-Sector-OdA display significant hybrid characteristics in combining associational and state logics.

Goal direction
Concerning goal direction, we examine four indicators: policy field, membership composition, organization of short-track dual training, and validation of prior learning.

) Membership in Federal Commissions
More often member in Federal Commissions (9) Participation in national VET events More frequently participating in national VET events Activity system (10) Degree of professionalism Higher degree of professionalism (11) Responsibility for collective labor agreements Less often negotiating collective labor agreements (12) Responsibility for generally binding collective VET funds Less often responsible for generally binding collective VET funds †If New-Sector-OdA are hybrid organizations combining state and associational logics, then the statements in the last column should hold. See Appendix S1 for coding and data sources.
Associations develop and formalize their organizational goals in the form of associational aims, which are codified in their statutes. Such associations represent member interests toward public authorities (Schmitter & Streeck 1999). Yet, the policy field (Indicator-1) in which they pursue aims can either cover various areas or be limited to a narrower area of interest representation such as VET. The more heterogeneous the interests of an organizations' members, the more difficult it becomes to create internal consensus and to represent these interests toward external stakeholders (Traxler & Huemer 2007). We expect that New-Sector-OdA mainly focus on the policy field of VET, as this limits the degree of heterogeneity in at least one organizational domain and thus helps to compensate for the heterogeneity already linked to the combination of associational and state logics. Furthermore, the focus on VET would correspond to what the state, as the bricoleur, was envisaging when setting the stage for these new intermediary organizations.
As these associations represent their members' interests, membership composition (Indicator-2) translates into different goal directions. Here, hybridization may be linked to the inclusion of various stakeholders in an organization (Mair et al. 2015). Owing to their role as governance actors before the inclusion in the national VET legislation, we expect that public authorities are included in New-Sector-OdA membership. This, in turn, implies the presence of a state logic next to an associational logic.
As indicated in the previous section, it matters which actor promotes hybridization. The goals of this actor may come to the fore in the goals of the newly created organizations. In our case, the state was the key driver of the reform. The state aims to ensure that VET is inclusiveness-enhancing, which is not always the case for employers (Bonoli & Wilson 2019). The Swiss VET reform introduced two inclusion-enhancing measures: The option to organize short-track training (Indicator-3), which lasts only 2 years and aims to integrate weaker students (Di Maio et al. 2019), and validation of prior learning (Indicator-4), which refers to the possibility for adults to gain a nationally recognized VET diploma without going through an apprenticeship, for example by formally recognizing skills they acquired on-the-job (BBV, Art. 31, al. 1). Thus, owing to state involvementand, thus, the state logicin their creation, new OdA should orient their goals more often toward public goods. We expect them to offer more inclusion-enhancing measures such as short-track dual training and validation of prior learning than the employer-dominated Old-Sector-OdA, which build more exclusively on the associational logic and might be reluctant to implement inclusion-enhancing measures.

Boundary maintenance
We examine five indicators for boundary maintenance: organizational set-up and sponsors, membership in umbrella associations, distance between OdA and firms, membership in Federal Commissions, and participation in national VET events.
Some organizational arrangements (Indicator-5) create new boundaries whereas others allow preserving preexisting organizational boundaries. Crucially, setting up a joint venture (rather than creating a traditional association) allows the constituting organizations, that is, organizational sponsors, to safeguard their boundaries. 3 In contrast, being member of a higher-level umbrella association (Indicator-6) implies being part of a system of interest representation with specific boundaries. Membership in associations affects the boundaries and identity of members, as umbrella associations are expected to be able to influence their members' behavior and contribute to the formulation of common goals (Schmitter & Streeck 1999). The organizational set-up as joint venture and membership in higher-level umbrella associations thus indicate specific organizational boundaries. To allow for high levels of heterogeneity and, thus, to enable the bricolage of two institutional logics (i.e., associational and state-oriented) while preserving organizational boundaries, we expect New-Sector-OdA to be more often set up as joint ventures than Old-Sector-OdA. Moreover, to maintain pre-reform boundaries, we expect New-Sector-OdA to refrain from becoming members of umbrella associations. Becoming members of such traditional national-level umbrella associations would hamper their ability to bring together associational and state logics by including employers, employees, and public authorities.
Another important aspect of boundaries is the distance between OdA and firms (Indicator-7). Firms are key stakeholders in the collective governance model of dual apprenticeship but were not in the other sectors. The ability to influence firms' behavior is crucial for employer associations in skill formation systems. Firms can either be direct members of OdA or only indirectly through another association that is an organizational sponsor of the OdA. Given their roots in a less employer-dominated and more school-based and state-dominated governance logic, we expect New-Sector-OdA not to include firms as direct members. Thereby, New-Sector-OdA can maintain pre-reform organizational boundaries, and, in this way, preserve elements of the state logic.
A further characteristic of associations is that they can negotiate with public authorities. Thus, besides their members, the relationship to public authorities constitutes a decisive organizational boundary. If associations become too closely involved with public authorities, members may lose confidence in the associations' reliability regarding interest representation. Yet, if associations have too little contact with public authorities, they may not enjoy the necessary influence to represent interests toward public authorities (Schmitter & Streeck 1999). Federal Commissions (Indicator-8) are central arenas for steering the VET system. They are composed of a selected group of representatives from all stakeholder groups including state, employer, and employee representatives. Thus, OdA membership in such commissions indicates a close relationship with public authorities and, thus, to the state logic. Given the strong support by the state for New-Sector-OdA, we expect them to be more often represented in Federal Commissions than Old-Sector-OdA in relative terms. Moreover, their participation in national VET events (Indicator-9) indicates proximity to public authorities. National VET events organized by the responsible government agency, the State Secretariat for Education, Research, and Innovation (SBFI), provide crucial platforms for OdA to deliberate on their activities and exchange best practices. Not least due to the need to legitimize the strong support by the state, and to maintain the state logic next to the associational one, we expect New-Sector-OdA to be more regularly present at such events than Old-Sector-OdA.

Activity system
We examine three indicators for activity system: degree of professionalism, collective labor agreements, and VET funds.
The organizational development or degree of professionalism of OdA varies substantially. We assess their degree of professionalism (Indicator-10) by evaluating to which extent they publish and make accessible relevant information on their websites. This information includes visions, missions and aims, statutes and membership regulations, advertising membership, and organizing activities for members or offering them various services (see Appendix S1, Table A3). We argue that the extent to which substantial information is provided on the organizational website reflects professionalism. To create a reasonably stable activity system despite the heterogeneity related to the involved associational and state logics and actors, and given their deliberate design through stateled bricolage rather than traditional evolution, a sustainable hybrid organization is expected to build on a high degree of professionalism. Thus, New-Sector-OdA are likely to display on average a higher degree of professionalism than Old-Sector-OdA.
A specific domain, in which traditional employer associations engage, is negotiating collective labor agreements (Indicator-11) with unions. Although, in Switzerland, VET is not tightly linked to wage coordination and unions play a minor role in VET governance , some OdA are in charge of collective labor agreements. Given that collective labor agreements are a core activity for traditional employer associations but not in the newly integrated sectors, and that these collective agreements might overstretch the organizational capacity to combine associational and state logics, we expect them to be less prevalent among New-Sector-OdA.
In contrast, although sectoral VET funds (Indicator-12) have a long tradition in certain sectors, only the 2002 VET act introduced the possibility to declare them generally binding. If they create generally binding VET funds, OdA rely on the coercive authority of the state, that is, indirect state support (Traxler & Huemer 2007). Thus, generally binding VET funds commit all firms (not only member firms) employing a specific occupation to financially contribute to training activities. Due to the important role of the state and the state logic in their creation, we expect New-Sector-OdA to develop VET funds more often than Old-Sector-OdA, the latter being exclusively based on the associational logic and more reluctant to develop such coercive measures toward firms.

Methods and data
This section describes our mixed-methods approach. We begin by providing a historical-institutionalist account of the relevant VET reform and the development and organizational characteristics of one exemplary OdA, OdASanté, which was created in the framework of the VET reform that led to the extension of the associational logic to the new sectors. This first part of the empirical analysis serves to demonstrate how the state acted as bricoleur to creatively combine associational and state logics, a process that is crystalized in the form of new hybrid intermediary associations. We select OdASanté as it represents a key example of a New-Sector-OdA, which is today responsible for one of the most prominent apprenticeship programs in Switzerland. In 2018, the health care apprenticeship was the third most common initial VET occupation offering dual training (BFS 2019). Similar developments also occurred in very different fields such as farming or social pedagogy. While the details for each of the sectors integrated into the collective VET governance system differ, the general transition pattern is well-illustrated by OdASanté. Our case study is based on information from available secondary literature, document analysis, our extensive new OdA database (details below), and four expert interviews. Interview partners for semi-structured interviews were selected based on their expertise and practical knowledge on VET governance in Switzerland. 4 In the second part of the empirical analysis, we conduct a cross-sectional statistical analysis of all New-and Old-Sector-OdA in Switzerland, which allows us to see whether our argument around New-Sector-OdA being hybrid organizations also holds more generally. Thus, the qualitative and quantitative parts of the analysis complement each other. To compare hybrid features of New-Sector-OdA to Old-Sector-OdA, we analyze the complete population of Swiss OdA in charge of initial VET occupations. This population consists of 146 organizations, which are recognized by public authorities as responsible for an initial VET occupation. We collected data on these organizations' characteristics in 2017-2018 (i.e., approximately 10 years after their creation) from various written sources based on a structured questionnaire. To ensure the quality of the data, we developed a coding manual with decision rules. Our data sources are the register of all federally recognized occupations with their responsible OdA, the organizations' websites, their statutes, published registers of collective labor agreements and binding VET funds, member lists of peak associations and federal commissions as well as participant lists of events. Additional phone interviews were conducted with OdA publishing only limited information online.
We examine differences between New-and Old-Sector-OdA on our 12 indicators. However, a simple comparison of means does not consider the differences in within-group variance and group sizes. We therefore use bivariate regressions to test if the observed differences in means between New-and Old-Sector-OdA are statistically significant despite the rather small number of New-Sector-OdA and heterogeneity within the two groups. Given the properties of our dataset, this is a conservative test of the differences between the two groups. For most indicators, we use logistical regression except for the indicator "participating in national events," where we use poisson regression, and "degree of professionalism," where we use OLS regression. When the outcome is present in under 15 percent of cases, we use logistic regression adapted for rare events (see Appendix S1, Table A4, for the regression results).
Our independent variable is the binary variable "new sector," which is coded as 1 if the OdA is responsible for an occupation in the newly integrated economic sectors. We identified these OdA based on administrative sources. Our classification of New-Sector-OdA was then confirmed by several experts in the field with whom we discussed it independently from each other. Most of our dependent variables are coded as binary outcome variables taking the value 1 if the OdA shows the characteristic. The indicator "participating in national events" is a count variable, which varies from 0 to 5 participations. The indicator "degree of professionalism" is a continuous variable, which varies from 0 to 12 points (see Appendix S1, Table A3, for more information on the variables).

Findings: State-led bricolage and the hybridity of New-Sector-OdA
In this section, we first offer an historical account of the VET reform that led to the extension of apprenticeship training to new sectors and provide an in-depth case study of OdASanté, a sector in which training previously followed a state-oriented logic. We find that the new OdA resulted from a process of state-led bricolage and features a high level of hybridity. Second, our cross-sectional analysis of all Swiss Old-and New-Sector-OdA shows that this finding of hybridity holds more generally for the latter group.

The VET reform process and the creation of OdASanté
The Swiss VET reform in the early 2000s expanded the associational logic to new sectors such as health care. The new VET act included all non-university vocationally oriented education and training rather than only training in the craft, industrial, and commercial sectors (BBT 2000). This state-led reform was broadly supported by the relevant stakeholders both inside and outside of parliament (Bundesrat 2000, p. 5,708). In fact, in December 2002 the new VET act was passed with unanimity in both parliamentary chambers. However, the parliament demanding a collective solution to training in new economic sectors does not solve the problem of creating these collective solutions. The parliament has no means to create effective intermediary associations that can fulfill the required governance tasks. The difficult task to implement the reform, and to assist in the creation of these organizations, was therefore entirely left to the federal state agency responsible for VET (BBT, later renamed SBFI). This bureaucratic actor thus faced the challenge of integrating a diverse group of economic sectors (e.g., agriculture, arts, and health care) into a collectively governed system. In these economic sectors, there had been no work-plus school-based apprenticeships governed by employer associationsrather, training had taken various mainly school-oriented forms.
Regional authorities (cantonal offices) had been the essential stakeholders for providing training, although in some of these sectors, professional associations, for example of farmers and nurses, had been involved (Wettstein 2020). Additionally, in these often para-public sectors, the state tends to be an important employer (even in agriculture and the arts, as public subsidies play an important role). Thus, the state and especially the cantonal offices had been key governance actors in the pre-reform setting of the concerned sectors. As explained earlier, we classify the pre-reform institutional logic in these sectors as state oriented.
Indeed, what unites the newly integrated sectors is that training was not organized in the form of dual apprenticeships but rather based on a state-oriented logic in cooperation with partners. However, otherwise, these new sectors differ significantly. With the major 2002 VET reform, the federal state aimed to align training in these sectors to the industrial-craft apprenticeship model that is based on the associational logic. For implementing apprenticeships in these newly integrated sectors, the responsible state agency promoted the creation of new training occupations. Moreover, given the key role of employer associations in the collective governance of dual apprenticeship training (Culpepper 2003), the state provided the legal framework and assisted in founding associations responsible for these apprenticeships. In this context, one of the state agency's main goals was to prominently involve employers in these new associations.
Although the federal state aimed to integrate these sectors into the collectively governed apprenticeship system, it was aware that necessary structures, in particular intermediary organizations in charge of VET governance, were lacking (Bundesrat 2000, p. 5,709). Yet rather than setting up organizations similar to the traditional associations described earlier, the state agency initiated and supported the creation of new organizational forms broad enough to combine institutional features of the state logic and the associational logic. This choice reflected the difficulty for the state to create regular employer associations (Culpepper 2003), but also the need to involve already existing actors into the newly created organizations. Importantly, to describe these new organizational forms, the VET lawadopted by parliament but drafted by the state agencyintroduced the deliberately broad term "Organizations of the World of Work" (OdA) that encompasses the variety of old and new associations in charge of VET. Once the relevant OdA were created in the new sectors, the state bureaucracy stepped back from its pro-active steering role to leave the space to private actorsthus contributing to the strengthening of the associational logic. Overall, we observe that in the creation of these new intermediary associationswhich fulfill collective governance tasks usually delegated to employer associations in the traditional sectorsthe state bureaucracy played a critical steering role in combining state and associational logics. For instance, the state agency was in charge of identifying the respective stakeholders, engaging them in a regular exchange, and moderating their dialog. In the following, our deep dive into the case of OdASanté serves to further substantiate our analysis of this reform process. We begin by tracing the pre-reform governance context for training in the health sector.
Historically, nursing training in Switzerland was primarily school-based and the most prominent entrance point for health occupations (Maurer 2013). The respective programs were mainly focused on school diplomas, while the employers' side played a very limited role in their governance (Flury 2004, p. 25). Cantonal health offices were responsible for securing public health, including the availability of health professionals. Together with the cantons, the Swiss Red Cross (SRC) played a relevant role in developing health care diplomas in Switzerland (Wettstein 2020, pp. 118-119). All cantons recognized SRC diplomas and SRC monitoring ensured that training followed nation-wide standards, but training was not clearly positioned within the educational system (Spitzer & Perrenoud 2006;Bender 2011). However, with the growing importance of health care training, cantons were increasingly taking over responsibilities from the SRC (Bender 2011). For instance, the intercantonal conference of health offices established a public-private partnership with the SRC. 5 The SRC's competencies included the definition of training curricula, the monitoring of the schools, and the recognition of foreign diplomas. Yet, cantonal health offices organized and financed the health care training schools (BBT 2000). Overall, the SRC had to carry out the will of the cantons regarding the training (Flury 2004, p. 5). The nurses' association was involved in questions of training by cooperating with the SRC, while employers remained sidelined (Bender 2011). However, the influence of professional associations was rather marginal: training in the health sector prior to the reform mainly involved cantonal health offices, the SRC, and schools (Flury 2004, p. 7). Training in the health sector was thus following a school-and state-oriented logic, while employer associations played virtually no role.
In the late 1990s, the Federal Parliament decided to integrate all non-university-based training into the VET system. Hence, the BBT was commissioned to align training in the health sector to the industrial-craft apprenticeship model that is based on the associational logic (Bundesrat 2000). This required a genuine effort on the side of the state agency to nudge the relevant private actors to work together and, at the same time, handing over significant responsibility for training to the employers (Flury 2004, p. 33). While the SRC did not see another option than to agree to the integration of health care training into the regular education system (Bender 2011), it tried to limit the damage to their own organization and staff by trying to transfer their training section to the BBT, but this initiative failed. The nursing association supported the reform because it hoped to attain their aim of professionalizing nursing training (Kaufmann 2010).
In terms of transition process, first, to anticipate the integration of health care training in the VET system, various cantonsi.e., the respective regional state agenciesdeveloped pilot projects and created the first forms of the future health care apprenticeship (Interview-CH3). Public funding destined to counter the lack of apprenticeship places was made available to finance these local projects. In these projects, the employers, such as hospitals, were involved at the local level, while the SRC was also involved and issued diploma (Interview-CH3). In a second phase, based on these local projects, a national health care apprenticeship was developed. In 2000, the relevant state agency, BBT, created a project structure for integrating this sector into the national VET framework (Project-Group-OdASanté 2004, p. 4). The BBT fulfilled the overall steering role in this process as responsible agency for implementing the VET reform. It guided the major reform process that was to transfer the associational logic to a sector previously dominated by a state logic (Flury 2004). However, representatives of the intercantonal health office conference and the intercantonal education office conference were also involved as members of the project's steering group (BBT 2000).
Once the new health care apprenticeship had been created, the state-led project steering group had the task of creating a new organization that would fulfill employer associations' traditional role in the Swiss VET system. This resulted in a new intermediary organization, namely OdASanté. While the project steering group was composed of both public and private actors, the Swiss Health Directors Conference, that is, the political coordination body of the cantons in health policy, was chairing the group (Project-Group-OdASanté 2004, p. 5). The state actors' lead position in the transition process is also illustrated by the availability of public start-up funds for OdASanté (which otherwise is fully funded by membership fees) in this development phase (Project-Group-OdASanté 2004, p. 3). While previously the SRC on behalf of the cantons had been in charge, during this transformation process other stakeholders, especially representatives of the employers in the health sector, discovered VET as an important policy field and became more strongly involved (Flury 2004, p. 32). In this context, we can observe that the state agency fulfilling its task to align the new sectors into a unified national framework acted as a bricoleur in bringing together these actors and in combining the pre-reform state logic of the newly integrated sectors with the associational logic otherwise dominant in apprenticeship training.
In line with this process of state-led bricolage, we find that OdASanté differs from the average OdA in traditional VET sectors in all three core organizational dimensions (goal direction, boundary maintenance, activity system). In the following, our application of the 12 indicators (see Section 3.1) establishes that OdASanté displays organizational features reflecting hybridity.
Regarding goal direction, our analysis shows that OdASanté's policy field is restricted to training (Indicator-1), whereas traditional employer associations represent interests in a broad policy domain: "OdASanté is a training organization, not a strong employer or occupational association" (Interview-CH1). The project team in charge for the transition process decided that this organization should include various stakeholders with diverging interests (Wettstein 2020, p. 122): employers and occupational associations (Interview-CH4), but crucially also cantonal authorities in the health domain . This hints at the continued relevance of the pre-reform state-oriented logic. The nurses' association is included in OdASanté and thus continues representing nurses' interests, especially regarding a clear distinction between the (tertiary educated) nurses and the newly established health care workers (Interview-CH3)but was a relatively weak actor in the OdA's establishment relative to the intercantonal health office conference. The SRC was excluded from governance, as there is no role foreseen for an NGO in the VET system. In contrast, employershospitals, home care nursing organizations, and nursing homeshave received a prominent role in collective governance. Their associations are included in OdASanté and are also represented in the cantonal sections of OdASanté together with cantonal VET and health offices (Interview-CH4). In addition, the goal direction of OdASanté is characterized by it organizing inclusiveness-enhancing short-track training (Indicator-3) and validation of prior learning .
Moving on to boundary maintenance, we find that OdASanté is a joint venture that allows bringing together different pre-reform stakeholders as sponsoring organizations . Typically, in a joint venture, the sponsoring organizations have a broader domain of activity than the joint venture, which has a more specific task (Interview-CH4). OdASanté is not member of a national-level employer association (Indicator-6). Although employer associations have gained an important role in governance, firms are only indirect members of OdASanté through their associations (Indicator-7). However, OdASanté was included as member in prestigious federal commissionsbased on state invitationalongside the national umbrella employer and employee associations (Indicator-8). Furthermore, OdASanté is an active participant in national VET events, again based on state invitation (Indicator-9).
Finally, regarding its activity system, OdASanté displays a high degree of professionalism (Indicator-10), which is illustrated, for instance, by its professional outreach activities and communication strategy. In contrast to traditional employer associations, it is not in charge of a collective labor agreement (Indicator-11) but OdASanté has developed a generally binding VET fund, which, as mentioned above, is a coercive measure to commit all firms to training investments backed up by state authority .
In sum, the case study of the creation of this major New-Sector-OdA has provided insights into both the state-led change process and the ensuing hybridity characteristics of this organizational formespecially regarding the state as a bricoleur in the reform process and as an important pre-reform player in the integrated sectors. Next, we turn to the statistical analysis of all New-and Old-Sector-OdA, which allows us to see whether our argument around New-Sector-OdA being hybrid organizations also holds more generally.

Cross-sectional comparison of New-and Old-Sector-OdA
Our statistical analysis shows that New-and Old-Sector-OdA are clearly distinct regarding our indicators of hybridity. Figure 1 summarizes our statistical findings for the 12 indicators listed in Table 1, showing group averages in percentage points for the binary variables and the means of the two groups' participation and professionalism as well as the results of our tests for statistical significance. To increase readability, the indicators are organized in a way that higher values always point toward the features associated with higher levels of hybridity.
Across almost all indicators, we find a strong confirmation for our expectation that New-Sector-OdA (outer circle) represent hybrid features (see also Appendix S1, Table A4). Concerning the organizational dimension of goal direction, the comparison of New-and Old-Sector-OdA shows that New-Sector-OdA limit their domain of activity more often to the policy field of VET than Old-Sector-OdA (Indicator-1). This, as we argue, is in line with what the state was envisaging when developing the National VET Act of 2002, which sets the stage for the creation of the new intermediary organizations. This leads us to the next point, membership composition. New-Sector-OdA include public authorities in their membership more often than their counterparts (Indicator-2). Public authorities were important stakeholders in the pre-reform setting, and, thus, transferred their influence into the new governance arrangement. Furthermore, our analysis suggests that New-Sector-OdA are oriented toward public goals and, hence, offer more inclusiveness-enhancing VET, but the difference is only statistically significant for short-track training (Indicator-3) and not for validation of prior learning (Indicator-4).
More generally, these insights suggest that these hybrids govern new apprenticeships, for instance, by uniting the stakeholders typically found in traditional associations (i.e. employers) and stakeholders that had an important role in the previous institutional arrangement for training (i.e. public authorities).
Regarding the second dimension, boundary maintenance, New-Sector-OdA are more often organized as joint ventures than as membership-based associations (Indicator-5). Their constituents are more often composed of organizational sponsors than in the case of Old-Sector-OdA. This organizational arrangement allows bringing together organizations of very different types, such as intercantonal conferences, employer associations, or employee organizations. In line with this organizational form, firms are less often direct members of New-Sector-OdA than Old-Sector-OdA (Indicator-6). This points toward the persistence of pre-reform boundaries and highlights that New-Sector-OdA are clearly distinct from the ideal-typical associations in VET. Moreover, none of the New-Sector-OdA is member of a traditional umbrella association of employers, while 43.9 percent of the Old-Sector-OdA are members of such umbrella associations (Indicator-7). Thus, New-Sector-OdA remain partly outside the existing system of interest representation in VET. Concerning the relationship with public authorities, New-Sector-OdA are more closely involved in federal commissions than Old-Sector-OdA (Indicator-8), which indicates the proximity of New-Sector-OdA to state actors. New-Sector-OdA also participate more often in national VET events than Old-Sector-OdA (Indicator-9).
These results show that New-Sector-OdA have developed organizational forms that allow them to maintain pre-reform boundaries. Moreover, New-Sector-OdA are in closer contact to public authorities than Old-Sector-OdA. Thus, a mutual approximation takes place but certain boundaries also persist. Concerning their activity system, New-Sector-OdA have a slightly higher degree of professionalism than Old-Sector-OdA but this difference is not statistically significant (Indicator-10). Yet, an additional analysis shows that if professionalization items associated with their role in VET (6 items) are distinguished from items associated with general organizational development (7 items), New-Sector-OdA score significantly higher on VET-specific professionalization and lower on organization-specific professionalization. This is in line with our expectation that New-Sector-OdA have Figure 1 Comparison of New-Sector-OdA and Old-Sector-OdA (high values = more hybrid). Notes: ** p < 0.01, * p < 0.05, perfect match. The higher the score or percentage, the more salient are the hybrid features. Indicators 1-8, 11, 12 are binary variables: values represent group averages in %. Indicator 9 (count variable taking values from 0-5): average number of participations in national events, multiplied by ten to increase readability. Indicator 10: continuous variable; average score on professionalism index multiplied by ten to increase readability. a strong need to become professionalized and to create a reasonably stable activity system despite the heterogeneity of the involved logics and actors (including state actors). However, Old-Sector-OdA may be more developed as organizations due to their longer existence and broader domains of activity. Further, while 28.1 percent of the Old-Sector-OdA are direct partners in a collective labor agreement, New-Sector-OdA never engage in this activity . This shows that these organizations limit their tasks to a specific field. Finally, New-Sector-OdA are more often responsible for collective VET funds. This difference is statistically significant (Indicator-12).
The results indicate that New-Sector-OdA create an activity system that enables them to fulfill their role in VET governance, but do not develop some of the typical, broader activities of traditional employer associations such as negotiating collective labor agreements. Moreover, the traditional employer associations appear more reluctant to introduce binding collective VET funds, which may be due to the associational logic and members' resistance to coercive measures.
Summing up, in terms of the hybridity characteristics, the organizational form of New-Sector-OdA allows for heterogeneous organizational arrangements that include employers and, crucially, state actors. They focus on only one policy field and are set-up as joint ventures rather than traditional membership-based associations. Moreover, pre-reform stakeholders are included into the new organizations, which do not become integrated in traditional umbrella associations, do rarely include firms directly, and do not get directly involved in negotiating collective labor agreements. More generally, our indicators point toward the strong role of the state in the process of integrating new sectors, for instance, because New-Sector-OdA contribute to inclusiveness-enhancing measures, participate in national events, and are members of federal commissions. Moreover, they rely on the state's coercive authority to overcome free-rider problems, for instance through generally binding VET funds.

Conclusion
This paper examined a major reform around a collectively governed policy field in which the provision of collective goods by private actors is promoted through an associational institutional logic. Collectively governed VET is a key policy area in which the associational logic prevails. While the organizational form of associations and their functions are well known, their core tasks are increasingly demanding. In particular, the deindustrialization of the economy challenges established associational systems, which have their strongholds in the industry and crafts sectors . How can collective governance be extended to sectors without collective governance tradition and what does this transfer of the associational logic of traditional apprenticeships imply for collective organizing in the new sectors?
We addressed these questions using the case of a Swiss VET reform. Whereas traditional dual VET systems are increasingly under pressure, the Swiss dual VET system remains surprisingly stable, not least because the state managed to extend the system of apprenticeship training to several new and rapidly growing economic sectors. Our analysis focused on the establishment of new forms of collective organizing that support this development. Using a mixed-method design, we have traced how the associational logic was transferred to the health sector through the creation of a new, hybrid organization. We then forwarded a novel way to measure hybridity at the level of organizations and compared all New-Sector-OdA to their older counterparts. We found that the new hybrid organizations provide platforms through which employers, employees, and state actors collaborate to govern apprenticeships. Thus, the successful transfer of apprenticeship to new sectors hinges on the ability to develop innovative organizational features to bridge institutional logics, for example, by creating joint ventures of stakeholders with diverging interests and limiting their policy domain to VET. While only future research can explore the long-term effects of this reform, these hybrid intermediary organizations represent a promising approach to transfer collective governance to the increasingly important service sector. Indeed, New-Sector-OdAenvisaged by policy-makers as a permanent solution from the starthave enhanced the stability of apprenticeship training in the last 20 years since their establishment. Therefore, our findings support the argument in recent organizational scholarship that hybrids have the potential to handle tensions and accommodate heterogeneity in an enduring way.
To explain the establishment of the new forms of collective governance, we introduced the concept of state-led bricolage. While we acknowledge that there also more bottom-up pathways to bricolage, this paper focused on exploring state-led bricolage as a strategy to promote change. It thus contributes to ongoing discussions about possible driving forces for bricolage and the emergence of hybrids. Further research could explore in greater detail the conditions for either state-led or more bottom-up bricolage. Moving beyond studies of hybridity lacking a clear agent, we identified a state-sponsored process of bricolage. The critical institutional variety available to the bricoleur was the state-oriented logic in the VET system in the new sectors to be integrated in apprenticeship training and the associational logic in the traditional industry and crafts sectors. The state-led combination of these two logics within the newly created associations led to hybrid organizational forms fulfilling the governance tasks that are typically dealt with by traditional employer associations. Thus, we demonstrated that the state as bricoleur can combine different institutional logics to reach an instrumental policy goalhere the stabilization of the VET system. Furthermore, we found that the bricoleur successfully imprinted some of its own core policy goals on the newly created organizations (e.g., societal goals that go beyond what would typically be associated with the associational logic).
Structural economic change poses a challenge to institutions that foster the provision of collective goods by private actors (Thelen 2007;Strebel et al. 2020). This is true for diverse policy fields in CMEs such as wage bargaining and policy concertation (Hall & Soskice 2001) but to some extent also for other economies (Fortwengel et al. 2019). Using the case of apprenticeship training in Switzerland, we derived general insights on how state-led bricolage has the potential to create new forms of collective organizing that take over central governance tasks. In some ways, our case is both a most likely and least likely case for state-led bricolage. On the one hand, in Switzerland, there was a large political consensus that collective governance should be extended to new economic sectors even though these sectors lacked the necessary intermediary associations. This political will cannot be taken for granted, as other countries might opt for a more direct state provision to deal with an apprenticeship crisis, which was however at no point seriously considered in Switzerland. On the other hand, in collective skill formation systems more generally, private actors have a strong sense of ownership and thus resist sustained state intervention (Culpepper 2003;Emmenegger et al. 2019). Therefore, the state can rarely resort to a strongly regulatory approach. Our analysis shows that in systems based on the associational logic, an enabling state may use bricolage to foster private actors' organizational capacity to fulfill collective governance functions before resorting to its relatively passive enabling role, leaving collective good provision again (mainly) to private actors. Bricolage thus allows states to strengthen systems of collective governance, which calls for a reassessment of the state's role in CMEs.