Managing Employment Relationships in Flexible Labour Markets: The Case of German Repertory Theatres

In theatres, `new' forms of employment are rather old. Based on qualitative case study research, this article analyses policies for managing human resources in a German non-profit repertory theatre. Referring to Marsden's theory of employment systems, the article suggests regarding these policies as being embedded in an interorganizational employment system, which comprises rules of job design and task assignment, the labour market, inter-firm institutions and the education system. This employment system for German theatre artists is marked by a high labour mobility and contingent work arrangements, but is also characterized by an ensemble structure providing (temporary) stability of the workforce. By studying how employment relationships are `managed' in theatres and how the organizational level is linked to the field's labour market characteristics, this article aims at contributing to the exploration of institutional prerequisites and organizational consequences of contingent work arrangements. In doing so, the article continues recent efforts to link studies on careers, labour markets and work arrangements in the cultural industries to the `future of work' debate.

Contingent work, i.e. temporary work, self-employment and involuntary part-time work (Barker & Christensen, 1998;Kalleberg, 2000), has become more prevalent. For most ballet dancers, actors and actresses, singers, directors, stage designers, writers, painters and many musicians, these 'new' forms of employment are rather old. They are not employed on a regular basis, but work as temporary employees, on a project basis or are self-employed. In recent years, there have been many studies on labour markets, careers and forms of employment in the creative and entertainment industries (see, e.g., Benhamou, 2000;Christopherson & Storper, 1989;Faulkner & Anderson, 1987;Jones, 1996;Menger, 1999;Peiperl et al., 2002; see also Caves, 2000). It has been argued that analysing these industries in which contingent work arrangements are widespread can help us to understand future labour regulation (Windeler et al., 2001), open labour markets (Haak & Schmid, 1999), individual consequences of contingent work (Dex et al., 2000) and new, project-based forms of inter-firm organizing (DeFillippi & Arthur, 1998;Jones & Walsh, 1997).
This article analyses human resource policies in a public German nonprofit repertory theatre with respect to its embeddedness in an employment system marked by contingent work. With very few exceptions, (German) actors and other theatre artists work under temporary contracts (one to three years) as ensemble members or under contracts for certain plays or periods. Employment in theatres meets both criteria for contingent work suggested by Polivka and Nardone (1989): there is no explicit or implicit contract for long-term employment, and the minimum hours worked can vary in a nonsystematic manner.
There are two main reasons why such an analysis can contribute to the literature on contingent work and to the literature on cultural industries. First, in comparison with the film industry and TV production, there are fewer studies dealing with employment in theatres (exceptions are Caves, 2000;Karhunen, 1996; see also Benhamou, 2000;Meredith, 1996). This also holds true for personnel management issues that are an altogether neglected research topic in the cultural industries literature. Second, there are very few studies linking the industry or field level of cultural industries to the organizational or firm level (see, however, DeFillippi & Arthur, 1998;Jones & Walsh, 1997). Besides these more quantitative shortcomings, there are also some characteristics of employment in public German theatres that make them a fruitful field of research. Rather than equating contingent work with a peripheral workforce, in the theatre even the employment contracts of the core workforce are based on contingent work arrangements. Remarkably, employment relationships in these theatres combine aspects of inter-firm mobility and project-based work with team structures (ensembles or theatre companies) that remain stable for at least some years and that are supplemented by so-called guests. Theatre ensembles are therefore located between the basic principles of organizing work in the arts that Howard Becker (1982) has distinguished, organization and the freelance system. Theatre thus provides a new (historically old) and noteworthy example of established flexible work arrangements.
By studying how contingent employment relationships are 'managed' in theatres and how the organizational level is linked to the field's labour market characteristics, this article attempts to contribute to the exploration of institutional prerequisites and the organizational consequences of contingent work arrangements. Other important aspects of open labour markets and contingent work arrangements, such as the individual consequences of contingent work (Dex et al., 2000;Garsten, 1999;Sennett, 1998), or labour market details such as under-and unemployment issues, and also problems relevant to theatrical work such as conflicts between artists and more craftoriented theatre workers are not dealt with in this article.
The article first uses Caves' (2000) basic economic properties of the creative industries to characterize the theatre as an 'industry' and then lays out the terms of employment in theatres. As each theatre company does not design theatrical work relations in isolation, the article then investigates the institutional framework that enables the dominance of contingent work arrangements. Referring to Marsden's (1999Marsden's ( , 2000 theory of employment systems, it suggests regarding theatres as one interorganizational employment system, which comprises rules of job design and task assignment, the labour market, inter-firm institutions and the education system. The influence of an interorganizational occupational community that enables 'boundaryless careers' (Arthur, 1994;Tolbert, 1996) and within which artists achieve and sustain a reputation raises the question how theatres as employers cope with the organizational consequences of contingent work. For organizations, flexible labour markets particularly involve problems of predictability and quality of labour supply as well as of skill development, recognition and transmission (Marsden, 1999). In addition, temporary work relations and high labour mobility might restrict the development of workplace community and commitment to the organization.
The article explores how these problems are solved in theatres. This study draws on qualitative case study research. It is based on semi-structured interviews with theatre artists in a public German repertory theatre and with representatives of inter-firm institutions. Proceeding in this way, the article takes up suggestions deduced from studies of careers in the US film industry (Hollywood), namely, to overcome firm-centred formulations of strategic activity and to regard human resource and training issues as well as inter-firm institutions in their industrial community context (DeFillippi & Arthur, 1998;Jones & Walsh;1997). Finally, the article discusses possible implications for the 'future of work' debate. As the German theatre field is relatively stable, it might provide a model of future work systems. But this, of course, raises the question as to what extent the findings of this article can be generalized. The analysis suggests that theatres provide examples of how institutional frameworks and organizational strategies might be interlinked to provide a basis for stability in spite of the predominance of contingent work arrangements.

Characteristics of work and employment in theatres
The theatre as a cultural industry According to Caves, the creative industries comprise 'industries supplying goods and services that we broadly associate with cultural, artistic, or simply entertainment value' (2000: 1). Among these creative industries are the performing arts: theatre, opera, concerts and dance. Several scholars of the creative arts industries have tried to elaborate the central characteristics of these industries (H.S. Becker, 1982;Benhamou, 2000;Caves, 2000;Faulkner & Anderson, 1987;Menger, 1999;Throsby, 2001). In the following, the seven 'basic economic properties of creative industries' given by Caves (2000) are used to illustrate some particularities of theatres.
As with other creative products, in the theatre it is uncertain how the audience will value a newly produced play (Nobody knows property); there is also a large variety of dramas and their interpretations which requires specialists (e.g. critics) to analyse them (Infinite variety property). As theatre plays are only in exceptional cases marketed as videotapes or on TV, durable products and rents (Ars longa property) are not that relevant in the theatre compared with, for example, the film industry, the music industry or the visual arts. Production and consumption are therefore strongly interlinked in theatres, which reduces the need for an 'industry' mediating between artists and 'consumers'.
The fact that (most) actors care about their products (Art for art's sake property) can help us to understand why theatre artists accept work relations that are seen as precarious by 'normal' employees.
In contrast to those already mentioned, the remaining three properties give a more specific picture of theatres. Caves differentiates simple and complex creative goods: whereas in the first case, single creative inputs are relevant, in the second case complex teams of creative and 'humdrum' inputs are required (Motley crew property). Although there are exceptions (e.g. oneperson plays directed by the actress herself), the theatre certainly belongs to this second group. Even when there are no technical and craft workers required in the production of a play, there is a team of actors with different skills and talents, and in most cases there is a director. In Germany, for example, this property together with historical circumstances (see later) has led to the emergence of numerous resident ensembles. The inputs provided by theatre artists differ with respect to 'technical' skills, talent and originality, and are ranked from and within the occupational community (A list/B list property). As acting (rehearsals and performances) is always observable (by colleagues, audience and critics), evaluation and ranking is ubiquitous.
Because of the Motley crew property, together with the Time flies property, which results from the need to finish projects at a certain time ('the curtain has to be lifted'), a strict temporal coordination of inputs is necessary. A theatre can be described as an extreme form of a project organization. Owing to the labour market for freelance actors, directors, stage designers, etc., there is a need for, in relative terms, long-range planning (1-2 years) concerning the plays to be performed and the casting. The project organization also makes it necessary to synchronize activities in an extreme way. Everyone engaged in a play has to adhere to a tight time regime.
But theatres do not constitute the typical form of a network organization in which single self-employed workers and firms structure and restructure themselves around (temporary) projects (R.E. Miles & Snow, 1986) as characteristic of the US film industry (Faulkner & Anderson, 1987;Jones & Walsh, 1997) or the German television industry (Windeler et al., 2001). Because German repertory theatres have an ensemble, and each theatre (i.e. its theatre manager with his or her team) creates its repertoire (as work of art and as a product) by bringing together ensemble members and guests, the notion of the (dynamic) network organization is inappropriate. Rather, the way of organizing activities and the allocation of people to tasks in this system can be described as a fixed network organization with theatre companies and their theatre managers as relatively stable focal points.

Terms of employment in German repertory theatres
The German theatre scene is unique. Historically grounded in the decentralized structure of princedoms and electorates, a large number of local court theatres, each with their own buildings and ensembles, at the end of the 18th century have bequeathed about 150 state-supported theatres (Stadttheater, Staatstheater, Landestheater). These theatres have groups of about 12 to sometimes 50 actors and actresses (ensembles) employed on a temporary basis (Deutscher Bühnenverein, 2002). In contrast to state-owned theatres in many other countries, and in contrast to commercial theatres, these theatres are repertory theatres (which means that the performances of each play included in the repertoire are spread over the theatre season, i.e. usually no play is performed for more than one or two days running).
Ensemble contracts in Germany are temporary contracts of one, two or, occasionally, three years' duration. The 'Normalvertrag Solo' (NV Solo), which arose from multi-employer collective bargaining, provides a mutually accepted framework for these contracts and also applies to employed directors, stage designers, dramaturges etc. 1 The NV Solo fixes a minimum salary, the minimum length of ensemble contracts (one season), the conditions for not prolonging this temporary employment, and the time span after which an artist gets tenure (after 15 years of continuous employment with one theatre). There are no further restrictions concerning the arrangement of particular tasks or the circumstances of job assignments.
In addition to contracts for ensemble members, there are other work arrangements in use for theatre artists. There is usually a large group of artists, including directors, stage designers, costume advisers, musicians, etc., who are engaged for single plays, and of actors who are engaged for single plays, the whole season or a part of the season. The terms of these contracts are subject to mutual negotiation.

Research setting
This article draws on qualitative case study research in a public German repertory theatre (Staatstheater), situated in a town with approximately 500,000 inhabitants. This theatre (in the following: theatre Y) also includes an opera house with an orchestra and a ballet employing about 900 people, of which about one-third are artistic staff. The study focuses on the theatre company. At the beginning of the research period (2000)(2001)(2002) this company comprised about 30 employed actors, 2 directors, the theatre manager, 3 dramaturges and about 40 guests and fixed-term contractors.
The aim of the field research was two-fold. First, at the organizational level, it aimed at exploring the way in which theatre artists' work is contracted, how 'the organization' copes with the consequences of contingent work relations, and how workers consider their employer's human resource policies. Second, it aimed at taking a broader look at how these human resource policies are embedded in a wider institutional context of inter-firm institutions. According to these research areas, data were collected in a two-step process.
In the first part of the research project, ten interviews, mainly with theatre artists who were associated with the focal theatre, were conducted. Interview partners here include the theatre manager (artistic director), a senior administration manager, a director, an author/director, a project coordinator/dramaturge and five actors linked to the theatre by different contractual arrangements. After initial informal conversations with theatre artists, the interview with the theatre manager was the starting point for finding new interview partners. In order to provide a representative cross-section, theatre artists with different job experiences and of both sexes were interviewed ( Table 1 gives an overview of the interviewed artists in this part of the study).
The theatre was chosen because it is representative of the group of 35 state theatres in cities with 200,000 to one million inhabitants with respect to financial subsidies, number of ensemble members, and artistic reputation. Another reason for selecting theatre Y was that, at the time of analysis, the theatre manager was about to move to another theatre in another (larger) city (theatre Z). This (normal but always critical) event gave the opportunity to learn more about this change process.
In the second part of the project, five interviews with representatives of inter-firm institutions were carried out. Interview participants were representatives of the national employers' association (Deutscher Bühnenverein), the national employees' association or union (Genossenschaft Deutscher Bühnen-Angehöriger, GDBA), the state-run work agency for actors (ZBF) and a state-run drama school (university). All relevant inter-firm institutions were analysed to explore their function within the employment system. Results of these interviews were also used to validate some of the claims of the respondents in the first part of the study.
Altogether, 15 in-depth interviews were carried out applying a semistructured interview technique. Interviews were open-ended in nature and lasted at least one and a half hours. The questionnaire used in the interviews included a core set of questions to gather information on the individual respondent and on the research questions presented above. The write-ups of the interviews were analysed using descriptive and interpretative codes (M.B. Miles & Huberman, 1994). Based on previously conducted and analysed interviews, specific questions or topics were adapted or added, particularly when representatives of inter-firm institutions were interviewed. Hence, there was a strong link between the two levels of data collection and an overlap between data collection and data analysis. This is also because insights gained from interviews with those theatre artists who had job experiences in different theatres and theatrical working environments (see Table 1) revealed that there are community-wide rules of organizing work and managing employment relationships that lead to organizational isomorphism among German repertory theatres. This interplay between the organizational level and the industry or employment system level is discussed later. In addition to the collection of primary data, other sources of evidence were used (Yin, 1994). These include labour statistics, documents provided by interview respondents (e.g. employment contracts, brochures), statistical reports and information given on theatres' and inter-firm institutions' Internet sites, site visits (audition, rehearsal), and interviews with theatre artists in newspapers and specialist journals. Furthermore, informal discussions mainly with self-employed theatre artists provided further information on theatre employment.
The empirical approach chosen is well suited to deal with 'how' and 'why' questions (Yin, 1994), but it has its own limitations. Among these are problems concerning the generalizability of the findings (discussed in more detail in the final sections of this article), and a lack of reliability of the kind provided by quantitative studies. But as the article addresses the question of how organizational and inter-firm practices constitute and react to the employment rules characteristic of (German) theatres, a qualitative case study approach seemed to be appropriate.

Institutional context: The theatre as an employment system
In this section, I first describe the labour market for German theatre artists. I then introduce the concept of 'employment system' (Marsden, 1999), which conceptualizes employment relations and their institutional context.

The labour market for German theatre artists
Being an actor does not equate to having a 'protected' occupation in Germany, i.e. in contrast to most other occupations, calling oneself an actor does not require a certain certificate or examination. Qualified actors will have attended one of the 19 state drama schools (university degree) or one of the many private schools. Altogether, there are about 9000 professional (i.e. qualified and active) actors. This number is an estimate taking into account information provided by the Interessenverband deutscher Schauspieler (ids), an informal association of actors, and by a representative of the ZBF, the state-run work agency for actors, which belongs to the national employment office and has local branches in five large German cities. Both institutions attach importance to the fact that the number includes only 'professional' actors. The number of those who call themselves actors or who work as actors but are not accepted or seen as professionals is unknown. In the season 2000/2001 only 2413 actors worked as ensemble members (NV Solo) in one of the approximately 150 German public repertory theatres (Deutscher Bühnenverein, 2002). All other actors worked as guests engaged for a certain play or for a part of the season, in private, commercial or independent theatres, not as theatre actors but within their profession (film, TV, dubbing and radio productions), or they were unemployed. For all performing artists (which also includes singers, directors, dancers, etc.), an unemployment rate of 22.6 percent in 2000 is given by the Bundesanstalt für Arbeit. In 2000, in total, 7314 actors worked as ensemble members and/or as guests for all German theatres (private and public), 2904 of them were women (about 40%; Genossenschaft Deutscher Bühnenangehöriger, 2000). 2 But given that at drama school the proportion of women is larger (information given by the ZBF representative), the likelihood that actresses leave the labour market or become unemployed is higher than for men. Owing to changes between unemployment and work and to multiple job holding in the arts (Haak & Schmid, 1999;Menger, 1999), unemployment statistics for artistic occupations are problematic, however.
Although there is no quantitative evidence available, the interviews indicated that there is segmentation of the labour market. Those who manage to sign an ensemble contract in one of the renowned theatres seem to be separated from those who are involuntarily self-employed whilst seeking contracts with theatres for single plays or seasonal engagements or joining independent theatre companies (see also the Finnish data analysed by Karhunen, 1996). None of the ensemble members of theatre Y was freelance before becoming engaged in the ensemble. In theatre Z, where the theatre manager and some of the actors moved, the same holds true.
Another important feature of the labour market for theatrical artists is interorganizational mobility. In addition to Table 1, which contains the career experiences of interview respondents, Figure 1 summarizes career moves on the occasion of the theatre manager's job move from theatre Y to Z (at the same time another theatre manager moved to Y). As rule of thumb, the theatre manager usually takes one-third of the ensemble members with him/her, one-third of the actors stay in the ensemble, and one-third is given notice that their contract will not be renewed.
As most German cities have just one larger state theatre (regional clusters of theatres exist in Berlin and, perhaps, in Hamburg and Munich only), changing the ensemble usually involves spatial mobility. All theatre artists interviewed regard this mobility as being essential for creative and innovative theatrical work. As the labour market for actors can be characterized by movements across the boundaries of separate employers, careers in the theatre are typical of what has been coined 'boundaryless careers' (Arthur, 1994). This mobility is enforced by the norm of temporary contracts which results from the artistic freedom of theatre managers deviated from the German constitution and the status of theatres as a 'tendentious organization'. Factors that enable and support mobility are a more or less standardized basic qualification of actors, a good observability of this qualification, and a social network that makes spatial moves easier.

The theatre as an employment system
The high mobility inherent in the theatrical labour market and the 'nature' of creative theatre work, such as the described temporal characteristics of rehearsals and performances, underline the experience of all interviewees that there are only slight differences between work environments and conditions in different theatre companies. Therefore, there are good reasons for trying to conceptualize employment in theatres on a macro-level. To do this, I introduce the concept of the 'employment system', which deals more directly with employment issues than do the more general concepts of 'field' (Bourdieu, 1979(Bourdieu, /1984DiMaggio, 1986) or 'industry' (Caves, 2000).
Usually, the term 'employment system' is used to describe a holistic choice of firms or employers concerning the terms of employment and the characteristics of human resource policies (cf. Osterman, 1987). As Marsden (1999Marsden ( , 2000 has shown, the scope for these decisions is limited by national (or other macro-level) characteristics. Marsden argues that if the employment relationship were to have any appeal to workers and firms, there would have to be robust and stable rules delimiting managerial authority under the condition of being productively efficient and enforceable (Marsden, 1999). Criteria for task assignments solving the problem of enforceability can be based on tasks or inputs, or on functions or outputs. In the so-called training approach (according to Marsden prevalent in Great Britain and Germany), grouping tasks according to complementarities in training needs ensures efficiency. Qualified workers are available on an occupational labour market in this approach, whereas in the production approach, where production needs determine how tasks are organized into jobs and matched to worker competencies, on-the-job-training dominates. Marsden, thus, links qualification and labour market to job design. But he also emphasizes the roles of inter-firm institutions such as unions and employers' associations: [E]mployment transaction rules, and the inter-firm institutions that support occupational skills and job and pay classification systems, are constituted into employment systems in which the inter-firm and joblevel factors are mutually supporting. (Marsden, 1999: 139) Marsden does not restrict the term employment system to the firm level. Rather, the institutional environments in which firms organize employment relations constitute an employment system. Referring to the theory of employment systems, this article suggests regarding theatres as one interorganizational employment system, which comprises rules of job design and task assignment, the labour market, inter-firm institutions, and the education system. Marsden mainly contrasts open market relations with open-ended relationships (employment contracts) and focuses on the industrial sector. But with respect to 'boundaryless careers' (and taking Hollywood and Silicon Valley as examples), he highlights that labour mobility requires employment rules based on the training approach and some form or a functional equivalent of an occupational labour market. In the theatre, this substitute is provided by an occupational or industrial community (Tolbert, 1996) that is heavily based on personal connections (see later) and that -due to language restrictions and to the gate-keeping function of drama schools and theatre managers -consists of a fairly closed circle of people. A rather homogeneous education with a manageable number of schools, a comparatively uniform design of jobs, and a system of inter-firm regulations form the institutional framework in which German theatres are embedded. 3 Taken together, this section has presented a characterization of theatres on an interorganizational level. In the case of theatre artists, the employment system comprises the German inter-firm institutions and the labour market. Marsden concludes that 'the firm lies at the centre of a complex of labour market institutions ' (1999: 249), and he emphasizes that, therefore, firms' personnel management (e.g. skill development, labour deployment policies, performance management) has to fit into the firm's institutional environment.

Organizational level: Managing a non-permanent workforce
This section analyses personnel management practices in a German repertory theatre in the light of the employment system of which it is part. It concentrates on some key personnel management areas that are intensely affected by labour mobility: recruitment, job assignment/staffing, performance management and employability/training. Before considering these areas, both the ensemble as a specific team structure and the ties between theatre artists constituting a network of social relations deserve attention. This social network does not just tie theatre artists together, but is (hence) also crucial for human resource management in theatres. For the reasons given earlier, there is strong evidence that this case study is representative of most German repertory theatres.

Ensemble and guests
Caves (2000: 366) makes the point that 'the more strongly does effective performance depend on the close cooperation of different artists, the stronger the case for organizing long-hired teams.' But, as outlined in the previous section, there are different national ways of organizing this cooperation. The German solution, the ensemble, provides a stable 'island' within a mobile and flexible occupational community.
An ensemble is a valuable, with respect to experience, age and talents, heterogeneous group of people. It is the mainstay of the theatre.
(Theatre manager) The theatre manager uses team sport (football) as a metaphor to illustrate his relation to the ensemble. As team coach he has to take care that this group of people stays in a good mood and develops collective energy.
A theatre cannot be managed abstractedly. Instead of management, I would prefer the word care. Actors who are used to one another are essential for good work, so not just the short-term output should be evaluated.
(Theatre manager) Many things go faster when you know each other well; each ensemble has its own language, stories and jokes.
(Actor 1) But as several of the respondents remarked, knowing each other makes it easier to produce a play -but may also restrict creativity.
The more artists move and change their environment, the more development becomes possible. Routines and well-worn work relations limit creativity. It's good not to know everything about each other. Changes are good; theatre is all about broken biographies. Some actors love to live like nomads; others stay at one theatre company for a long time. To me, the latter have a bad handicap, they do not make me as curious as the first.
(Director 1) There are two ways to enhance innovation and creativity by bringing in new artists. First, the ensemble 'passes away' and a new theatre manager builds a new ensemble. As building a team takes time and requires cooperation, it is comprehensible that, usually, theatre managers stay in a theatre for at least some years. Second, guest actors and directors are hired for certain plays.
In general, employing temporary workers and contractors might give the organization numerical and financial flexibility, but it may also lead to loss of control and deterioration in the quality of the labour supplied (Legge, 1995). Also, as several authors have observed, the heavy use of contingent employees can adversely affect regular employees by violating norms of fairness, trust and equity, which can also destabilize the core workforce (Davis-Blake & Uzzi, 1993;Feldman et al., 1994;Legge, 1995;Pfeffer, 1998). In theatre Y, however, the problems resulting from a workforce consisting of ensemble members with mainly temporary contracts, as well as with guests and contractors, were not perceived as crucial. From the perspective of ensemble members, guests bring in new ideas, and they make work easier because they play roles that ensemble members would otherwise have to play. From the perspective of self-employed guests, project-based jobs are opportunities to work for public theatres (Actor 3; Actor 5).
At first sight, the types of work arrangements prevalent in theatres seem to constitute a typical core-periphery structure (Atkinson, 1984). But it is not always easy to determine what the core and what the periphery are (see also Legge, 1995). There are, for example, good reasons for regarding freelance directors who stage most of the plays in a season as forming the core of a theatre. Whereas technicians and craftsmen are hired on a regular basis, theatre artists have contingent work arrangements; guests working on single projects are mostly hired recurrently and can be found in listings of ensemble members, marked as (permanent) guests. In addition, most guests in fact strive for an ensemble contract, but they belong to the periphery because of the hours worked and not because of the lower skill level characteristic of secondary labour markets (Tolbert, 1996). Work arrangements of ensemble members are less contingent than those of guests, but as Menger has noted with respect to artistic labour markets, a 'paradoxical picture emerges, in which highly skilled and quite differentiated workers maintain weak employer attachments ' (1999: 546).

Hierarchy, task assignment and industrial relations
The size of the ensemble and the extent to which guests are hired depends on the policy of the theatre manager. He or she is responsible for the repertoire and for staffing. At theatre Y, it was the (common) policy of the theatre manager to cast plays using mainly ensemble members, but to work with 'external' directors who bring 'their' costume and stage designers. The theatre manager, together with his core team of (employed) directors and dramaturges, casts the plays. Sometimes (and in practice depending on their status) actors can express a request, but the cast remains the decision of the theatre manager.
The theatre manager is a little king. In our view even a very powerful king.
(Actor 5) It makes no sense to cast a play against the ideas of the director. The way in which casting decisions are communicated depends on the management team. In theatre Y casting lists were published four to six weeks before the beginning of rehearsals, in some theatres these lists are published at the beginning of the season for all plays. These lists have a tremendous significance for the actors because they are interpreted as (public!) signals of how their talent and their performance have been valued. Altogether, the theatre manager has to ensure that the distribution of work is perceived as fair, which is particularly complicated because each project (the roles played and the team gathered) implicates the opportunity to achieve reputation and to build social ties, e.g. with a well-known director. Actors therefore stress that they often have more roles to play than they would like, but that the real problem is more being assigned to the right projects (directors, other actors) and the right roles. In her interview, the union representative remarked that one important type of case heard at the Bühnenschiedsgericht (a special court for conflicts between employers and performing artists) deals with the problem of actors who feel they have been given a role that does not correspond to their talent and skills. As in other formal organizations, actors cannot choose their supervisor and -although a unique and creative artist -no actor is irreplaceable. Of course, director and actors do interact when a new play is developed, but the director (as project leader) has all rights to determine how the job is carried out.
The director is the boss, this is not up for negotiation. If I want to do my own thing, I have to work as a freelance musician.
(Actor 5) The actors need a strong hand.
(Director 1) Only in exceptional cases does the theatre manager intervene and influence the production or even nominate a new director. Because there are no explicit contractual or institutional regulations that prescribe criteria for assigning tasks to jobs (which was common practice some decades ago when actors' contracts contained a clause concerning the specific character the actor was engaged for), the 'area of acceptance' (Simon, 1951) is fairly large and subject to artistic reasoning. Although most students strive for an ensemble contract, for some of the actors interviewed, working as an ensemble member is sobering. They are not usually granted a say in the casting of plays, they feel dependent on directors and are restricted in their possibilities to accept other jobs (in other plays or movies). The ensemble contract includes clauses that force actors to be always available during the theatre season.
Ensemble contracts are 'oppressive, feudalistic contracts making actors dependent on arbitrary acts'.
(Director 1) Many theatre artists find employment contracts in the theatre too rigid and too restricting. This account makes clear that seeing these contracts as precarious because of their temporal limitation is ambiguous. The artistic freedoms of theatre managers and directors, as well as temporary employment contracts, provide employers with opportunities for opportunism. As production in the theatre requires cooperation among actors, directors and the theatre manager, the framework negotiated by the inter-firm institutions helps to make the 'game' of beginning and ending work contracts more acceptable despite the uncertainties involved. The regulative frame for personnel decisions on the occasion of a departing theatre manager can illustrate this. In these special (but also normal) situations, both the new and the old theatre manager can terminate contracts with ensemble members for so-called artistic reasons. Under certain conditions (for those employed for more than four years) a pay-off has to be made and the affected actor can demand a hearing. In theatre Y, eight actors who had been in the ensemble for two to six years had to leave (see Figure 1). During the negotiations between the old theatre manager and his new employer, uncertainty was ever-present in the ensemble. Whereas three or four actors knew quite early that they would go with the theatre manager, many others were unsure if they could join the old theatre manager, if the new theatre manager would keep them or if they would have to leave the theatre. At these times the casting lists had even more significance because actors interpreted them as signals of the likelihood of being considered to join the departing theatre manager. Under these circumstances, norms fixed in the NV Solo guarantee that at least some basic rules are followed that limit, for example, the period of notice required when terminating a contract. This rule ensures that actors whose contracts end have the opportunity to present themselves to potential employers.

Social networks, community and organizational social capital
The mobility of theatre artists implies work relations that are not limited to members of one organization or its customers. All interviewees emphasized that social networks that span organizational boundaries are crucial for making a (boundaryless) career and even for staying in business (see also Jones' [1996] corresponding data on the US film industry). The concept of social capital addresses social relations as resources in which investment can be made and that facilitate the achievement of certain ends (Bourdieu, 1979(Bourdieu, /1984(Bourdieu, , 1986Coleman, 1988Coleman, , 1990. I use this concept to explain the roles social relations play in the theatrical employment system. In addition, social networks of theatre artists can, as an example of flexible labour markets, illustrate: (i) limits in applying either internal or external definitions of social capital (Adler & Kwon, 2002) to social networks of flexible employees, and (ii) problems of conceptualizing organizational benefits of social capital (Leana & Van Buren III, 1999;Leenders & Gabbay, 1999;Nahapiet & Goshal, 1998;Tomer, 1998) when careers are 'boundaryless'. Whereas internal definitions of social capital emphasize the benefits of social relations for a (more or less) closed group of actors (Coleman, 1988(Coleman, , 1990Putnam, 2000), the focus of external definitions rests on the individual benefits resulting from a focal actor's social ties (Bourdieu, 1986;Raider & Burt, 1996).
Starting with the latter perspective, the individual actor, two main functions of social capital seem to be relevant in theatres. Owing to their often nomadic life and their unusual working hours it is difficult for theatre artists to build up intense relations with people outside the theatre. The friends and acquaintances of actors are mostly actors or other artists themselves, and it is no problem to form new networks or to be integrated into social networks after changing location. Here, the family metaphor is suitable, which describes a theatre as a relatively closed system (with only the theatre manager as an interface to the world of budgets and politics). In addition to providing a social background that enables mobility, social ties are crucial for attaining reputation and, thus, for making a career (see also Raider & Burt, 1996). These two functions of social capital can explain the tremendous importance of premiere celebrations and the theatre canteen as central locations where these networks of contacts are created and sustained, i.e. where investments in social capital are made.
Employing an internal definition of social capital calls for a decision on the level of analysis: the occupational community, the theatre as an organization or the ensemble. The family metaphor can be applied to all these levels: the team of actors working together for some time in an ensemble, all artists, 'core' and 'periphery', working together on certain projects belonging to the repertoire of a theatre, or all theatre artists as an occupational community. Respondents (ensemble members) agreed on that they work within an extended family. The theatre manager, however, commented: Social relations in theatres rather constitute a virtual family with virtually intense relations. Spatial mobility and the large number of freelance and temporary workers result in sometimes transitory work relations and many former colleagues are forgotten fast. Altogether, most relations remain work relations, and one has to be able to let people go. Nevertheless, I try to keep this in balance with the development of more than functional relationships and social responsibility.
(Theatre manager) At the community level, the family metaphor is even more questionable.
Theatre artists belong to a small, socially interconnected network, but, as highlighted, this community is segmented. For those having an ensemble contract with one of the less recognized theatres or working as a guest for these theatres, family ties mean something different than for ensemble members of well-recognized theatres. From the organizational perspective, the social network (or the virtual family) helps to solve problems resulting from the flexible labour market. Because -at least for the artistic workforce -nearly all work contracts are temporary and there are no contractual limitations for assigning actors to projects, theatres are, within a given financial budget, provided with a relatively high degree of functional and numerical flexibility. However, stability is essential for planning and running projects. Turnover of already assigned actors causes financial and artistic problems and can, in extreme cases, lead to the cancellation of rehearsals or performances. This stability is guaranteed by maintaining the ensemble and by interorganizational work relations -mostly between directors, stage and costume designers, and certain actors. These strong networks provide the continuity of relations (time closure), which, prospectively, encourages investment in relations and, retrospectively, gives relations a past history that can provide the basis for placing trust (Coleman, 1990), and is therefore necessary for intense symbiotic work relations. Although time closure cannot be established by the organization, the integrating activities of the theatre manager can be interpreted as investment in (his) social capital and as the provision of prerequisites for (others) investing in social capital. The importance of social relations in the theatre reveals that contingent work does not generally result in an erosion or lack of social capital (as Sennett's, 1998, analysis suggests). But can this actually be defined as organizational (social) capital (Leana & Van Buren III, 1999;Leenders & Gabbay, 1999;Tomer, 1998)? Whereas the commitment to the profession, to a network of strong ties to colleagues and directors, and, ideally, to the projects actors currently work on is high, the commitment to a certain theatre is altogether low. Thus, stable and tight personal work relations often outlast and transcend intra-organizational work relations; loyalty is expected to people and the profession, not to particular organizations. Also, the organization, the public repertory theatre, benefits from the reputation the theatre manager achieves (see later). This reputation is strongly affected by the people the theatre manager -based on his or her social capital -has brought together. This is why a move by the theatre manager to another theatre (Intendantenwechsel) is a pivotal event, not only for the actors, but also for other stakeholders in the organization (the town, the state, the local authority district, the audience).
The remainder of this section concentrates on (1) performance appraisal, (2) recruitment and (3) training/employability as key areas of personnel management. These areas are strongly related to the employment system level, and activities on these areas are to a wide extent based on the social network of theatre artists.

Performance and reputation
Actors are continuously observed by other actors, directors, dramaturges, the audience and critics. According to Caves (2000), for creative workers deciding whether a performance is consistent with the worker's terms of employment is doubly complex, because how the job is done is also based on the worker's tastes as an artist. Also, because theatre artists work in teams, not only are talent and perceived skills relevant in performance appraisal, but so too are reliability and adaptability (Caves, 2000; see also Faulkner & Anderson, 1987).
The performance is just one part of the job. You have to integrate yourself and you have to cooperate with the other ensemble members.
(Actor 4) All respondents highlighted what Caves (2000) has called the A list/B list property of the creative industries.
At the beginning of the season all are happy, they are busy and play their roles. But then competition begins: 'How good am I?' 'How good are the others?' 'What are my next roles?' (Actor 3) 80 per cent of the energy goes into 'non-productive' activities, e.g., to achieve a place in the ensemble, to care about the roles I do and will play etc.
(University teacher/director) In comparison with business organizations, it is striking that there are no explicit performance appraisals in theatres. This might be because performance is visible anyhow and mobility requires indicators of how an individual is ranked within the community. But how do these rankings take place? Caves' (2000) observation that profits and revenues do not exclusively control rankings of artists holds especially true for German repertory theatres (see also Anheier et al., 1995). Of course, the success of an actor in public has an impact on his/her reputation, but this is not a causal connection. In contrast to commercial theatres (e.g. light theatres), there is always a demand on performing arts theatres to perform (artistically) good, sophisticated theatre. Thus, in addition to theatre critics from respected newspapers and theatre journals, the network of colleagues has an enormous impact. In particular, the theatre manager, directors and dramaturges, as those who have an impact on career development, give indirect feedback by enabling, supporting or preventing project or role assignments and 'hire or fire' decisions. Thus, feedback information on where the actor stands is given indirectly rather than in face-to-face talks.
What an actor can accumulate is job experience ('interesting' projects, ensembles, directors, theatre managers, colleagues). Although teamwork is essential, in the end, when it comes to personnel decisions, individuals are assessed. 'Reputation is widely shared information (via third parties) about a person's character, skills, reliability, and other attributes important to exchanges' (Jones, 2002: 218, my emphasis). The ways in which these assessments are achieved within the community are, however, supported by other rankings, namely those of schools and of theatres (see above). As H.S. Becker (1982: 351) has noted, 'every aspect of art world activities and organizations contributes to and affects the making of reputations and their results'. Awards and prizes (theatre or actor/actress of the year, etc.) assist this process of generating community-wide signals of the reputation achieved (see also K.H. Becker & Haunschild, 2003;Jones, 2002). In the theatre, reputation is even more dependent on networking activities within the community than it is in the film industry because it is more difficult to prove the product after production, and because box office revenues are seen as a less important indicator of project success. For theatre artists it is also difficult to assess their position in the community because 'there are fads and fashions con-cerning the type of actors perceived as being interesting' (university teacher/director). The way reputation is gained in the theatre is self-enforcing.
When you play in a small theatre, not a soul sees you. Theatre managers go to Berlin, Hamburg and Munich to look for new talent, but not to the provinces.
(Actor 3) The chance to be seen arises when the school or theatre where the actor studies or works at that moment has a high reputation. Because there are many ensembles, and the economic success of productions has no direct influence on the reputation of the actors, the 'money seeks money' and 'performance seeks performance' effects (Faulkner & Anderson, 1987: 893) are less strong than they are in the film industry. This also results in a distribution of earnings that is fairly narrow (about 1500 to about 5500 Euro). Theatre actors also working in the German film industry can increase their income significantly. Film acting, however, does not necessarily amplify reputation in the theatrical community. From the perspective of a focal organization, prestige attained, i.e. the position within the stratification (see also Menger, 1999), depends drastically on how the work of the theatre manager is appreciated. This depends on the artistic repertoire (Taalas, 2000), which, in turn, results from the overall performance of the ensemble, the directors engaged, and the young authors and actors supported. As mentioned earlier, critics and audience (box office success, number of subscribers) play a part in assessing the overall quality of the theatre in a certain period, but -as with the artists -it is the business of the professional community to do the ranking (which exists in the heads of the community's members only).

Recruitment
The way in which actors are recruited strongly reflects the features of the theatrical employment system. Beginners graduate from state or private schools. The state schools accept only a small number of applicants, about 1 in 20-200. Being a graduate of a drama school can be seen as a first proof of artistic talent (Karhunen, 1996), a mechanism to 'sort out artists from nonartists' (H.S. Becker, 1982: 16). Giving the chance of a formal qualification (in a state school) to such a small number of students means, hence, a first gateway to be passed. But as 'budding artists in each art world undergo an apprenticeship to prepare for encounters with gatekeepers' (Caves, 2000: 365) a second gateway follows. Members of the management team travel to (some of the) drama schools' auditions to watch the graduates' presentations. Here they look for talent, but also for personalities who would fit into their ensemble. Occasionally, a theatre works together with local drama schools and uses these projects as a trial period during which young actors can be assessed. The more popular theatres also receive applications from many graduates; sometimes applicants are invited to an audition.
The recruitment of experienced actors takes a different path. Theatre manager and dramaturges 'screen' the population by going to performances in other (selected!) cities. But information from extra-organizational networks may support (or even replace) own impressions.
Top theatres recruit from other top theatres. Or you have to have connections to relevant directors. Look at the girlfriends and spouses of directors. They all have played in the theatres the director works for.
(Actor 3) In theatre Y, which has a fairly good reputation, networks of connections were seen as the crucial way of recruitment. Because on-the-job experience is highly relevant for an actor's reputation, personal connections can, as intermediaries, replace labour market institutions which set skill standards and certificates (Baumann, 2002;Marsden, 1999). Interestingly, professional intermediate agents are of little importance in the German theatrical employment system. In contrast to film actors (see also Christopherson & Storper, 1989), and, for example, opera singers, theatre artists rarely have agents for their theatre jobs. But there is the ZBF, the governmental work agency for actors. Local ZBF representatives provide free help to both actors and theatres in the matching and recruitment process. The ZBF is contacted by theatre Y mostly when guests actors are needed who have characteristics (type of character, certain physical characteristics) currently not available within the ensemble. But there are small theatres, not attractive to young actors, that have problems in hiring qualified actors. For these theatres, but also for self-employed actors with a lower reputation or following a career break, the ZBF is essential in bringing workers and jobs together.
The minor importance of professional agencies in the theatrical employment system might be grounded on: (i) the low standardization of jobs, which results from team production; and (ii) a rather small monetary volume of contracts and a low economic impact of the cast of a play (in contrast to Hollywood productions; cf. Caves, 2000).
In summary, the recruitment of theatre artists is heavily based on personal ties that are extra-organizational and span the professional community. The theatre manager (as the agent of the organization!) has the decision-making powers to hire and fire. One of the key human resource management areas is, therefore, extremely personalized, and is to a large extent delegated to processes external to the organization.

Employability and training
Features of the training and education system are at the core of an employment system because they govern the way jobs are designed, and they affect incentives to develop and transmit skills, how skills are recognized and to what extent skills are transferable (Marsden, 1999).
As pointed out earlier, formal training (in state schools) provides relatively homogeneous basic qualifications and the entrance ticket into the labour market for theatre artists (whereas in the film industry 'learning-bywatching' as a 'runner' is widespread; DeFillippi & Arthur, 1998). For further career steps after initial training, however, formal qualification is irrelevant (see also the data on income levels of Finnish actors by Karhunen, 1996).
Actors develop their skills mainly through on-the-job experience. Apart from learning on-the-job there are -with the exception of voluntary singing lessons or speech training -no explicit employee development activities in the theatre. Furthermore, none of the actors interviewed regarded the director as a trainer or coach. There are directors who try to teach actors during rehearsals, and there are sometimes unique cases of master-student relationships among directors and actors, but usually directors concentrate on fulfilling their project; this may even include limiting others' creative contributions: The director demands things and enforces his or her ideas. Actors have to cope with this; the director does not contribute consciously to their learning processes.
(University teacher/director) Although discovering talented actors and fostering them can enhance the reputation of a director and theatre manager, the responsibility for skill development is left to the actors. Having passed the second gateway after their initial training, actors must continuously gain new job experience, achieve recognition, and build a network of ties to colleagues, directors, dramaturges and theatre managers. Thus, they alone are held responsible for their employability on the labour market, i.e. for developing marketable skills. For theatre artists this is especially precarious because: (i) what these skills are very much depends on tastes rather than on measurable characteristics, (ii) modes of skill recognition are uncodified, and (iii) strength and weaknesses perceived by others principally concern the whole personality of any given actor. In view of the high labour mobility and the orientation towards professional (artistic) standards and reputation in the theatrical employment system, the demand for employability appears to be an economically reasonable way of encouraging (enforcing) continuous efforts in gaining relevant job experience and, thus, developing transferable skills. But this 'solution', which leaves all risks to the actors, is certainly supported, if not enabled, by the Art for art's sake property and a significant oversupply on the labour market.
Theatres as employers are attractive for actors when they provide learning opportunities (see also Menger, 1999) and opportunities to contact important members of the community. In particular, beginners prefer ensemble contracts because of the range of job and team experiences they offer. But, these experiences have to be valued by the community, and this calls for a certain reputation of the employing theatre.
To conclude, in an employment system in which (i) interorganizational mobility is a precondition for 'career success', (ii) initial off-the-job training endows workers with a basic qualification only, and (iii) in which the responsibility for additional skill development and learning is allocated to the workers, workers will, if they can, choose employers who provide good learning opportunities.
This section has shown how theatres cope with the problems resulting from the flexible labour market they are embedded in. Apart from a few administrative personnel functions, no explicit institutionalized personnel management could be found in theatres. But there are practices that fulfil the functions of recruitment, performance appraisal, skill development, etc. The case study has shown that these practices do correspond to features of the employment system.
If so, the way in which theatres cope with the problems resulting from boundaryless careers and contingent work arrangements might give us an idea of upcoming ways of managing employment relationships in flexible labour markets.

Conclusion: Is the theatre a model of future employment relationships?
On an organizational level, this article has shown how theatres manage contingent employment relationships. Several interorganizational mechanisms or institutionalized practices, such as dividing artists from non-artists by gatekeeping, a competitive, interorganizational struggle for reputation, a pool of qualified and motivated workers, and a community-wide network of ties provide a (consistent) framework that enables organizations to handle boundaryless careers. At the same time, this historically institutionalized framework sets a corridor for viable human resource strategies. Also, multiemployer collective bargaining restricts the scope for German theatres to make use of flexible employment and to assign tasks to actors. Recruitment, training or employee development, performance management and compensation, task assignment and redundancies are therefore embedded in an employment system that provides stability in the face of contingent work relations. The organizational level is, therefore, not sufficient to explain employers' choices and management of contingent work relations exhaustively. Rather, this article has demonstrated that interorganizational ties between artists are crucial for recruitment and for achieving recognition. There is an ensemble constituting, as a (temporary) closing mechanism (Baumann, 2002), strong ties between artists, but there is also a network of weak and strong ties. The great importance of interorganizational networks for bringing people together in an ensemble and in certain projects, the boundaryless careers of actors and directors, and the central roles of the theatre managers (and their social capital) in this community illustrate that to regard theatres as isolated firms does not allow for understanding of the structure of their employment relations. Furthermore, the function of single theatres or playhouses within the employment system, mainly as (humdrum) resource providers for individual artists and their teams, sheds new light on the roles of organizations in flexible labour markets. DeFillippi and Arthur (1998) have made a similar point with respect to the dynamic network in the US film industry where temporary enterprises sustain a permanent industry. But the (German) theatrical employment system has more similarities with team sport because in both 'industries' relatively stable teams play a far more important role than in the film industry. In theatres, however, this team is also reinforced with guest actors.
The rules, institutions and organizational practices discussed so far constitute a well-functioning and established employment system, so why should this not serve as a perfect model of contingent work systems in general? There are some features that might limit the generalizability of the theatre case.

Product and production
Although the ensemble stays together for some time, its duration is not openended. As the products are team products with a high degree of synchronization, this requires production processes with a definite beginning and end. The products themselves -performances and the repertoire as a whole -are works of art and this may lead to the conclusion that they are incomparable with industrial or service goods. However, there are non-theatrical products in which the production process comes close to the creative development of a play in rehearsals, such as consulting in a team, and the development of a marketing campaign or an Internet strategy.

Nature of tasks
An important difference between business organizations and theatres could be the roles of internal or external customers in comparison to the audience and the peers. The fact that the most important part of an actor's work takes place in public, improves the possibility of interorganizational assessments and careers. In all cases in which outputs are harder to measure and tasks are performed which are invisible to outsiders, information asymmetries are larger and therefore interorganizational mobility causes higher transaction costs.
Size of the employment system The employment system presented in this article is small with respect to the number of workers and the number of jobs. Some problems resulting from a flexible labour market might therefore be solved more easily by a network of ties than would be the case in a larger employment system.

Characteristics of the labour market
Actors are qualified, dedicated professionals who stick to their profession even when unemployment is imminent. This constitutes specific forms of power asymmetries on the labour market that make certain contingent forms of employment more 'accepted'.

Hierarchy and artistic freedom
The considerable authority of theatre managers (and directors), which is derived from the right to artistic freedom, facilitates task assignments and hire and fire decisions according to individual tastes. This increases the importance of individuals' networks or social capital.

National context: Public subsidies and collective bargaining
The 'habitat' of German repertory theatres is characterized by a large number of decentralized theatres and significant public subsidies (Krebs & Pommerehne, 1995). Even though artists and critics compare theatres, there is no direct (economic) competition between theatres. Because in other countries (e.g. UK, USA) neither public subsidies nor stable ensembles are as prevalent as in Germany, the German employment system might be an incomparable case. Also, the roles of inter-firm institutions could be seen as results of the particular German context of co-determination and regulation (Marsden, 2000;Muller, 1999). The important function of the unions and guilds in the US film industry (Caves, 2000;Christopherson & Storper, 1989;Paul & Kleingartner, 1994), however, supports Marsden's (2000) account that inter-firm institutions which provide a basis of mutual trust and fairness are more required in the training approach and when task assignments leave considerable room for interpretation and negotiation, regardless of country.
Despite national idiosyncrasies and differences between the theatre and other industries, the theatre could be an interesting model of future work systems in those cases in which a constant stream of outputs (of whatever goods or services) is required, a team of professionals and intense social ties are important, and where mobility is desirable (for whatever reason). The case study of a German repertory theatre has portrayed a way of organizing work and employment that differs from the merely project-based, marketmediated work relations of freelance or self-employed artists because the ensemble provides a stable team structure despite contingent work arrangements.
Even though one path towards future work relations seems to lead to contingent work arrangements in large-scale taylorist industrial environments, another path leads to more craft-oriented, knowledge-based ways of production (consulting, services, media industries, IT experts, interim managers). 4 This is not the place to discuss whether one of these paths will be more relevant than the other. For the latter, however, the theatre can provide indications of possible characteristics of the employment system, and of how organizational or firms' human resource strategies are embedded in this employment system. 1 In January 2003, the NV Bühne has superseded the NV Solo (which was in force when the empirical analysis was carried out). The NV Bühne covers a wider range of theatre workers than the NV Solo, but for the groups of workers and the issues regarded in this article, there are no significant changes to be considered. 2 Austrian and German-speaking Swiss theatres and drama schools are strongly interlinked with the German labour market but can be neglected in the context of this article. 3 As one of the reviewers has rightly remarked, this raises the question as to what extent the characteristics of the German theatre labour market are due to it being 'German' as opposed to it being a theatre. Exploring this issue in detail would require a historical and comparative analysis which is beyond the scope of this article. Although not being able to provide further evidence here, I assume that the training approach prevalent in Germany has had a significant influence on the emergence of regulation and formal training in the theatrical labour market. 4 Interim managers have a similar position as guest directors who direct one play and then move to another project (cf. Inkson et al., 2001).