Week 5 - Assess an Organization's Structure and Propose Changes

r Academy of Management Journal 2015, Vol. 58, No. 6, 1658–1685. http://dx.doi.org/10.5465/amj.2013.0903

HARNESSING PRODUCTIVE TENSIONS IN HYBRID ORGANIZATIONS: THE CASE OF WORK INTEGRATION

SOCIAL ENTERPRISES

JULIE BATTILANA Harvard Business School

METIN SENGUL Boston College

ANNE-CLAIRE PACHE ESSEC Business School

JACOB MODEL Stanford University

We examine the factors that influence the social performance of hybrid organizations that pursue a social mission and sustain their operations through commercial activities by studying work integration social enterprises (WISEs). We argue that both social im- printing, defined as a founding team’s early emphasis on accomplishing the organiza- tion’s social mission, and economic productivity are important drivers of a WISE’s social performance. However, there is a paradox inherent in the social imprinting of WISEs: Although social imprinting directly enhances a WISE’s social performance, social imprinting also indirectly weakens social performance by negatively affecting economic productivity. Results based on panel data of French WISEs gathered between 2003 and 2007 are congruent with our predictions. To understand how socially im- printed WISEs may mitigate this negative relationship between social imprinting and economic productivity, we also conduct a comparative analysis of case studies. We find that one effective approach is to assign responsibility for social and economic activities to distinct groups while creating “spaces of negotiation”—arenas of interaction that allow members of each group to discuss the trade-offs that they face. We conclude by highlighting the conditions under which spaces of negotiation can effectively be used to maintain a productive tension in hybrid organizations.

Over the last 30 years, we have witnessed an un- precedented increase in the number of organizations that operate at the intersection of the social and commercial sectors. These organizations, often called “social enterprises,” primarily pursue a social mis- sion while also engaging in commercial activities to

sustain their operations through sales of products and/or services (see, e.g., Battilana & Dorado, 2010; Galaskiewicz & Barringer, 2012; Hoffman, Gullo, & Haigh, 2012; Pache & Santos, 2013). They straddle the well-established categories of business and charity (see Austin, Wei-Skillern, & Stevenson, 2006; Mair &

We are grateful to Michel Anteby, Rodrigo Canales, Tina Dacin, Marie-Laure Djelic, Robin Ely, Mary Ann Glynn, Boris Groysberg, Martine Haas, Johanna Mair, Joshua Margolis, Charles Perrow, Woody Powell, Michael Pratt, Michael Tushman, and three anonymous reviewers for their detailed comments on earlier versions of this paper, and to Stefan Dimitriadis, Ting Wang, and the Harvard Business School European Research Center for research assistance. This research also benefited from comments and suggestions made by Steve Barley, Alnoor Ebrahim, Javier Gimeno, Ranjay Gulati, Rebecca Henderson,

Matthew Lee, Christopher Marquis, Kathleen McGinn, Rosabeth Moss Kanter, Tomasz Obloj, Jeff Polzer, Filipe Santos, Sandra Waddock, and seminar participants at Boston College, Cornell University, Harvard Business School, INSEAD, London Business School, McGill Uni- versity, Stanford University, University of California Davis, University of Michigan, University of Pennsylvania, and University of Toronto. Finally, we are indebted to Jean-Marie Huguesat CNEIfor accessto theannualmember survey data and to the members of the WISEs who shared their time and experience with us.

1658

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express written permission. Users may print, download, or email articles for individual use only.

Marti, 2006), and are thus “hybrid” organizations, combining aspects of multiple organizational forms (Battilana & Lee, 2014; Haveman & Rao, 2006; Padgett & Powell, 2012). A large subset of these social enter- prises face the challenge of serving two categories of constituents: the customers of their commercial ac- tivities and the beneficiaries of their social activities.

Organization theorists have long argued that or- ganizations that serve multiple constituencies com- plymore readily with demandsfrom the constituents on whom they depend for access to key resources (Oliver, 1991; Pfeffer & Salancik, 1978; Wry, Cobb, & Aldrich, 2013). A direct implication of this pre- diction is that social enterprises that serve distinct groups of beneficiaries and customers are likely to focus on their customers, on whom they depend for financial resources, and potentially to neglect their beneficiaries. Such neglect, however, would jeopar- dize the ability of social enterprises to achieve their social mission and call into question their raison d’être. Consequently, these hybrid organizations (or simply “hybrids”) pose an interesting question to organizational theory: Can they sustainably achieve their social mission in spite of the risk that they may prioritize customers over beneficiaries?

In order to address this question, we study one type of socialenterprisethatisprevalent aroundthe world: the work integration social enterprise (WISE). Prom- inent examples include the magazine The Big Issue in the United Kingdom, the recycling network ENVIE in France, and Goodwill stores in the United States. The primary goal of these organizations is to help the long- term unemployed to transition back into the labor market. To accomplish this goal, WISEs hire un- employedpeople,towhomwereferinthispaperasthe “beneficiaries” of the WISEs’ social mission. These beneficiaries work for WISEs to produce goods or ser- vices, which are then sold on the commercial market.

These enterprises face a dilemma in deciding how to allocate their resources. They need to provide their beneficiaries with job training as well as in- dividualized social counseling, which requires the dedication of resources to social activities. Mean- while, WISE customers expect goods and services at a competitive price and quality, which requires the dedication of resources to commercial activities. The ensuing resource allocation is likely to prioritize customers over beneficiaries because WISEs are de- pendent on their customers for revenue generation. However, this prioritization presents a challenge because it diverts resources away from counseling and other social support activities and thus puts the fulfillment of a WISE’s social mission at risk. In

order to understand how WISEs may overcome this challenge, we examine the factors that enable them to achieve high levels of social performance. In this context, “social performance” corresponds to a WISE’s effectiveness in enhancing the job market prospects of its beneficiaries.

Building on research on the enduring influence of founders’ imprinting on organizations (for a review, see Marquis & Tilcsik, 2013), we argue that social imprinting—defined as the founding team’s early emphasis on accomplishing the organization’s social mission—plays a critical role in enabling these hy- brids to maintain their focus on serving their bene- ficiaries and achieving their social mission. Social imprinting promotes the recruitment of permanent staff with a background in social work as well as the design of social-mission-oriented systems and pro- cesses. In these ways, social imprinting helps hy- brids to sustain their focus on their social mission and offsets the risk of neglecting beneficiaries; hence, it is positively associated with social performance.

However, a heightened social emphasis stemming from social imprinting may come at a cost to eco- nomic productivity (that is, the organization’s over- allefficiencyinturninginputsintoeconomicoutputs). This is because permanent staff with a social work background may be less experienced in managing commercial operations. Furthermore, their creation, and subsequent institutionalization, of social-mission- orientedorganizationalprocessesandsystemsmaylead these staff to prioritize social activities over commercial ones. Therefore, social imprinting is likely to be asso- ciated with lower levels of economic productivity. Lower economic productivity, in turn, decreases the resources available with which to pursue the social mission and may signal lower human capital to po- tential employers of beneficiaries. This would reduce beneficiaries’ employment opportunities and thereby diminish a WISE’s social performance. We thus hy- pothesize, echoing Smith and Lewis (2011) and Jay (2013), that there is a paradox inherent in the social imprinting of social enterprises that serve separate groupsofcustomersandbeneficiaries:Althoughsocial imprinting directly enhances a hybrid’s social perfor- mance, social imprinting also indirectly weakens so- cial performance by reducing economic productivity.

In this paper, we test our predictions regarding the paradoxical nature of the relationship between so- cial imprinting and social performance in WISEs using data from a detailed survey administered an- nually between 2003 and 2007, inclusive, to a large panel of WISEs operating in France. The regression results support our predictions and show that social

2015 1659Battilana, Sengul, Pache, and Model

imprinting and economic productivity are positively associated with a WISE’s social performance. Also, in line with our predictions, economic productivity partially mediates the relationship between social imprinting and social performance: Social imprint- ing is negatively associated with economic pro- ductivity and consequently has a negative indirect effect on social performance. Our regression results thus confirm the paradoxical relationship between social imprinting and social performance.

Second, we extend our findings by exploring how socially imprinted WISEs can resolve this paradox. Todoso,weconductedanin-depthlongitudinalqual- itative comparative case study of two WISEs: one that successfully attenuated this relationship and one that was unsuccessful in addressing the paradox. The comparative analysis, based on data that we collected from 35 interviews and a wide range of archival ma- terial for each case, suggests that a structural differ- entiation approach that assigns responsibility for social and commercial activities to distinct groups may allow hybrids to mitigate this relationship. Im- portantly, in order to be successful, this approach needs to be accompanied by “spaces of negotiation,” which we define as arenas of interaction that allow all staff members to discuss and agree on how to handle the daily trade-offs that they face across social and commercial activities. These spaces of negotiation maintain a productive tension between the staff members in charge of each of these activities.

WORK INTEGRATION SOCIAL ENTERPRISES

WISEs emerged in the late 1970s as an attempt to address the rise of structural unemployment in de- veloped countries (Bode, Evers, & Schulz, 2006). Their founders recognized that a growing number of people were out of work for long periods of time, duringwhichtheirconfidenceandskillswereeroded. These long-term unemployed people risked entering a downward spiral of isolation, overindebtedness, and substance abuse that would make it extremely difficult for them to re-enter the job market. The founders of WISEs developed a simple model to ad- dress these issues: offering the long-term unemployed work opportunities to help them to rebuild their hu- man capital and ultimately to reintegrate into the job market. Accordingly, WISEs hire the long-term un- employed to produce goods or services in low-skilled industries such as construction, catering, gardening, or recycling, which are then sold at market prices.

WISEs can be found today around the world, in Ireland, the United Kingdom, and the United States,

and as Beschäftigungsgesellschaften in Germany, empresas de inserção in Portugal, and entreprises d’insertion in France (Defourny & Kim, 2011; Spear & Bidet, 2005). For instance, in the United States, the Delancey Street Foundation developed a range of WISEs, including a restaurant, a removals company, andalandscapingcompany.InSwitzerland,Velosfür Afrika recruits unemployed people to recycle old bikes, thensells them inSwitzerland and West Africa.

The benefits of the WISE approach lie in its poten- tial to provide the long-term unemployed with the confidenceandskillsthattheyneedtoreintegrateinto the workforce (Cooney, 2011). As they work on the production line, workers acquire soft skills such as attendance, workplace socialization, and discipline as well as more job-specific skills such as dismantling a refrigerator or building a wall. In addition to pro- fessional training, WISEs also provide beneficiaries with individualized social support to enhance their job readiness. This includes counseling to help them to address the personal issues (such as health or housing) that are often significant barriers to em- ployment. It also includes training to help beneficia- ries to acquire basic aptitudes (such as literacy and self-esteem) and job search skills (including writing a resume or handling job interviews).

DRIVERS OF SOCIAL PERFORMANCE IN WISES

Social Imprinting

We argue that a WISE’s ability to achieve high levels of social performance depends in part on its social imprinting, which we define as the founding team’s early emphasis on accomplishing the orga- nization’s social mission. All WISEs, by definition, are founded to pursue the social mission of helping the long-term unemployed to reintegrate into the workforce. However, some founders may emphasize the importance of the social mission at the time of founding while others may instead emphasize the establishment of effective commercial operations in order to ensure sustainable revenues. For instance, in one of our interviews, the founder of a (socially imprinted) WISE operating in the recycling industry stated that the goal of his organization has always been “work integration, clearly work integration: we could be producing peas or whatever; it would make no difference.” In contrast, the founder of another (commercially imprinted) WISE operating in the same industry explained that commercial activities have taken precedence from the start: “A WISE that focuses on its social mission against all odds is likely

1660 DecemberAcademy of Management Journal

to close down and, when it does, there is no social impact anymore. That’s it. That’s why I always say: first the business, then the social mission.”

Prior research has demonstrated that founding conditions have enduring consequences for organiza- tions. Most of the early work explored Stinchcombe’s (1965) notion that environmental conditions during founding affect organizational outcomes (such as sur- vival) by creating structures and routines that are dif- ficult to change (for a review, see Marquis & Tilcsik, 2013).Buildingonthisearlywork,amorerecentstream of research has looked inside organizations and ex- plored how founding teams’ early decisions have effects that endure far beyond an organization’s found- ing stage (see, e.g., Baron, Hannan, & Burton, 1999; Beckman & Burton, 2008; Eisenhardt & Schoonhoven, 1990). In line with this more recent stream of research, and echoing Whetten, Mackey, and Poly (2011), we contend that social imprinting has a lasting influence on the organization’s commitment to its social mission that continues to affect a WISE’s social performance long af- ter its founding.

This lasting influence works through two related mechanisms. First, WISE founders who make a strong commitment to the achievement of social wel- fare goals early in the life of the organization are likely to embed this commitment in the stated goals and values of the organization (Ruef, Aldrich, & Carter, 2003). They are further likely to consider competencies related to accomplishing the organi- zation’s social mission as critical when hiring full- time staff. For example, in the socially imprinted WISE mentioned earlier, production supervisors were hired based on their relational and social work skills. In contrast, founders who emphasize creating effective economic operations are likely to consider technical and business competencies to be the more critical part of their hiring process for full-time staff. For example, in the commercially imprinted WISE mentioned earlier, industry experience was sought when hiring permanent staff, “because [they] knew the work [of repairing appliances], and that is what mattered.” Homophilic processes, whereby organi- zational members hire candidates that resemble themselves, thus lead socially imprinted organiza- tions to attract, select, and retain permanent staff who adhere to similar social goals and values, and who demonstrate mastery of similar skills (Burton & Beckman, 2007). An important consequence of this pattern is that the permanent staff in these organi- zations are more likely to focus their attention on the accomplishment of their social mission—that is, on serving beneficiaries.

Second, WISE founders with a strong commitment to social goals are likely to set up processes and systems that are aligned with these goals. For in- stance, they are more likely to go beyond training beneficiaries only for their specific tasks at the WISE. They are likely to adopt training policies designed to enhance the general employability of their bene- ficiaries, to design specific skills assessment pro- cedures to help beneficiaries to market their skills to prospective employers, or to organize meetings with local employers to facilitate beneficiaries’ placement. As these processes and systems are repeatedly enacted by the staff, they become routines (Feldman, 2003) that contribute to enhancing beneficiaries’ job prospects. Therefore, we hypothesize that:

Hypothesis 1. In WISEs, social imprinting at founding is positively associated with social performance.

Economic Productivity

We argue that a WISE’s ability to achieve high levels of social performance depends also in part on its economic productivity, which we define, build- ing on Caves, Christensen, and Diewert (1982), as a WISE’s overall efficiency in turning inputs into economic outputs. Economic productivity is likely to be positively associated with the social perfor- mance of WISEs in three main ways. First, a more economically productive WISE, by definition, is able to produce higher levels of output for any given level of input than its less productive counterparts. As a result, it will have relatively higher margins, prof- itability, and capacity to innovate (Bourgeois, 1981). The resulting slack reduces the pressure to achieve sufficient revenues to ensure its survival and allows a WISE to focus more on the attainment of social objectives. For example, in a WISE with high eco- nomic productivity, permanent staff can spend more time training beneficiaries to help them to acquire broader professional skills.

Second, an economically productive WISE is more likely to be perceived as legitimate by key external market constituents such as customers or investors. In the eyes of these constituents, achieving high economic productivity signals that the WISE is well managed and has adopted business practices that are taken for granted in competing corporations. Legiti- macy increases the likelihood that these constituents will support the organization (D’Aunno, Sutton, & Price, 1991; Suchman, 1995). For example, a more productive WISE may have an easier time attracting

2015 1661Battilana, Sengul, Pache, and Model

new contracts from customers who are satisfied with its products or services. In turn, a more productive WISE may be able to mobilize additional resources for supporting its beneficiaries (for example, through training or mentoring) and thereby better prepare them to find a job.

Third, the benefits of legitimacy are likely to spill over to a WISE’s beneficiaries as well. The level of economic productivity of a WISE is likely to reflect the degree to which the organization has been able to en- hance the capacity of its beneficiaries to transform inputs (such as raw material) into economic outputs. Byimplication,beingamemberofanorganizationthat has the reputation of being highly productive signals higher human capital (Morrison & Wilhelm Jr, 2004). As a result, a more productive WISE will be more likely to be able to place its beneficiaries in jobs once they leave the WISE. Therefore, we hypothesize that:

Hypothesis 2. In WISEs, economic productivity is positively associated with social performance.

Relationship between Social Imprinting and Economic Productivity

Adhering to the demands of both their customers and beneficiaries is challenging for WISEs because the activities that serve their beneficiaries are not fully aligned with those that serve their customers. With constrained resources, a decision to attend to one constituency may carry a significant opportunity cost. Because they depend on customers for their survival, WISEs run the risk of placing too much fo- cus on their customers at the cost of neglecting their beneficiaries. One advantage of social imprinting, as we highlight in the discussion of Hypothesis 1, is that it counteracts this tendency through its influence on workforce composition, organizational processes, and systems. Yet, early social imprinting may also affect how theWISE attendsto commercial activities.

Given the tendency of socially imprinted WISEs to hire permanent staff with social work backgrounds, their staff are likely to have skills and beliefs that fit the demands of the social sector rather than those of the commercial sector (Bourdieu, 1977). This is be- cause these staff members have been socialized in the social sector through their training and/or work experience, and have thus become imbued with the values and work practices associated with that sector (Louis, 1980; Van Maanen & Schein, 1979). Their prior socialization makes these staff likely to priori- tize systems and processes that help beneficiaries over systems and processes that improve economic

productivity. For example, they may introduce more flexibility in production processes in order to allow beneficiaries to attend individualized counseling sessions, or they may be lenient vis-à-vis some ben- eficiaries’ unprofessional behaviors that are harmful to economic productivity, such as absenteeism or lateness. Moreover, permanent staff members with social work backgrounds are less likely to have de- veloped skills and experience in establishing and running commercial operations than those who have been trained to work, and/or who have worked, in commercial enterprises. For example, they are less likely to be familiar with best practices associated with operational efficiency, marketing, or account- ing. Therefore, we hypothesize that:

Hypothesis 3. In WISEs, social imprinting is neg- atively associated with economic productivity.

We summarize the relationships between social im- printing, economic productivity, and social per- formance in WISEs in Figure 1. According to our hypotheses, both social imprinting and economic productivity are positively associated with a WISE’s social performance. However, socially imprinted WISEs face a paradox because social imprinting also indirectly weakens social performance through its negative relationship with economic productivity.

METHODS

To fully understand the relationship between social imprinting and social performance, we adopt a “mixed methodology”approach(seeEdmondson&McManus, 2007). We first test the existence of the hypothesized paradox by conducting regression analyses based on our quantitative data. We then turn to our qualitative data and conduct an in-depth comparative case anal- ysis to explore how socially imprinted WISEs may be able to resolve the paradox of social imprinting by mitigating the negative relationship between their so- cial imprinting and economic productivity.

FIGURE 1 Antecedents of Social Performance in WISEs

Social imprinting Social performance

Economic

productivity

– +

+

1662 DecemberAcademy of Management Journal

Setting

We study WISEs operating in France (that is, entreprises d’insertion).1 To operate as a WISE in France, an organization is required to obtain accredi- tation from the Ministry of Labor. This accreditation entitles it to a public subsidy intended to offset the opportunity cost of employing less-productive people who require extra supervision and training. This is a fixed amount per beneficiary (V9,681 annually in 2007) that is the same across all WISEs, and WISE beneficiaries are limited by law to two years of em- ployment under this program. A yearly evaluation process allows the state to monitor this accreditation and towithdraw itif evidence is found that a WISE has abused the subsidy (that is, the WISE has received it without complying with its workforce development commitment). This subsidy, together with additional grants, accounts on average for less than a quarter of a WISE’s revenues, with the remaining revenues com- ing from the sale of products and/or services.

The public accreditation also requires French WISEs to hire their beneficiaries from a pool of in- dividuals designated by Pole Emploi (the national agency for employment) as “deserving” access to work integration programs. These individuals have all been unemployed for a prolonged period of time (typically at least two years) and have been assessed by Pole Emploi as “experiencing specific social and professional difficulties which make it impossible for them to access the regular job market” (DGEFP, 2003: 7). This requirement limits the pool of poten- tial beneficiaries to individuals who face multiple obstacles to work, including low qualifications, low levels of self-confidence, and a lack of professional skills. It further ensures that WISEs recruit those who are really in need of support.

Despite the role of the state in providing accredita- tion and subsidies, WISEs in France operate as private entities. Half of them operate under a for-profit legal status while the other half operate under a not-for- profit legal status. However, given the commercial character of their activity, not-for-profit WISEs cannot

benefit from the tax exemptions traditionally granted to not-for-profits. This is because not-for-profits benefit from these exemptions only if they do not compete in commercial markets with similar products and prices as for-profit companies. In turn, if for-profit WISEs abandon their social mission and pursue profit gener- ation as their only goal, they will have their accredi- tation withdrawn by the state. Thus, regardless of the legal status that they adopt, WISEs face similar con- straints in the French context.

Entreprises d’insertion constitute an ideal setting in which to test the arguments developed in this paper because, like their counterparts around the world, they are hybrid organizations caught between the distinct and potentially competing demands of their beneficiaries and their customers. Although they pursue a social goal, commercial activities are important to them because they depend on sales of products and services to generate the majority of their revenues. Furthermore, WISEs have a signifi- cant and growing economic presence in France. First established in the late 1970s, the number of French WISEs increased significantly in the 2000s: At the end of our observation period in 2007, there were 1,178 WISEs in France, employing more than 22,000 beneficiaries, and with a combined sales volume of more than V800 million (US$1.17 billion).2

Quantitative Data

The quantitative data used in this study are from an annual survey administered by the Comité National des Entreprises d’Insertion (CNEI)—that is,the French National Federation for Work Integration Social Enterprises—to its members since 2003.3 The survey collects detailed establishment-level information about the WISEs, including their field of activity, sales, wages, composition of human resources, and place- ment of beneficiaries. The CNEI invests significant resources in the administration of the survey in order

1 Although we focus exclusively on entreprises d’inser- tion in this study, there are two additional organizational forms, associations intermédiaires (AI) and associations chantier d’insertion (ACI), which are part of the broader workforce development sector in France. These organiza- tions, however, operate outside of the commercial market since they rely mainly on public support and benefit from advantageous labor law exemptions. They therefore di- verge from the hybrid organizational forms that are the focus of this study.

2 The sales figure is the authors’ estimate based on data from DARES, the research and statistics unit of the French Ministry of Labor.

3 About half of French WISEs are affiliated with the CNEI. Those that are not are either too small to be able to pay the required membership fees or are affiliated with two other social federations: Fédération Nationale des Asso- ciations d’Accueil et de Réinsertion Sociale (FNARS), the main federation for homeless shelters; or Fédération des Comités et Organismes d’Aide aux Chômeurs par l’Emploi (CORAACE), another federation for social service organi- zations. The CNEI is the largest of these three federations and is the only one exclusively open to WISEs.

2015 1663Battilana, Sengul, Pache, and Model

to use the data collected in its lobbying efforts (Hugues, 2007). As a result, the response rate is ex- tremely high (for example, 98% in 2007), covering nearly the entire population of CNEI members.

Using this data source, we gathered a panel for years between 2003 and 2007, inclusive. We then imposed three criteria in the construction of our working sample. First, we excluded from our ana- lyses WISEs that had fewer than 10 employees (that is, permanent staff and beneficiaries). Data from very small WISEs are relatively less reliable and less sta- ble than data from larger ones because small WISEs are less likely to have the time and resources to col- lect and provide accurate data for the CNEI survey every year. Second, we focused exclusively on ob- servations that covered a full year of data and hence excluded the data from the founding year of WISEs (if it was founded within our observation period). Third, we excluded WISES operating in the tem- porary work industry because they do not directly hire beneficiaries on a full-time basis but operate as intermediaries by placing beneficiaries in spe- cific short-to-medium-term assignments with cli- ent companies. As a result, beneficiaries have only limited interactions with their temporary work WISE.

Applying these criteria and eliminating observa- tions with incomplete or missing information for one or more variables of interest, we obtained 641 establishment-year observations with complete information.4

Dependent Variable: Social Performance

Building on Scott (1977), we define social perfor- mance as the degree to which an organization is ef- fective at producing positive social outcomes. Within the WISE field, a consensus has emerged that social performance can be assessed by the proportion of beneficiaries that are able to find regular jobs at the end of their employment with the WISE (DGEFP, 2003). This rate, known in the field as the “positive gradua- tion rate,” is reported yearly by all WISEs to the state. Accordingly, we measure a WISE’s social performance as the percentage of beneficiaries completing their

term at the WISE in a given year who found a regular job with a contract lasting more than six months.

Independent Variables

Social imprinting. We used the industry classifi- cation of WISEs as a proxy for their social imprinting. In France, at the time of incorporation, WISE founders (like the founders of any organization) are required to file a statement, called an objet social, describing their organization’s purpose and its branch of activity. The French National Institute of Statistics and Economic Studies (INSEE) assigns this information to an Activité Principale de l’Entreprise (APE) code, which corre- sponds to the main activity of an organization. If founders do not agree with their assigned APE code, they can contest and ask INSEE for another. However, such requests are very rare, as are changes requested later in the life of the organization.5 Therefore APE codes reflect the set of activities that the founder em- phasized at the time of founding.

Qualitatively, a WISE founder can primarily de- scribe his or her organization’s main activity in one of two ways: either as performing a social activity (such as social services) or as operating in a specific industry (such as recycling, gardening, or catering). Following the terminology used in the WISE field, we refer to the former organizations as having a “so- cial APE,” and we refer to the latter as having an “industry-specific APE.” Two field experts whom we interviewed strongly supported the notion that APE codes reflect early imprinting. As one of these experts explained:

It tells you something when [a WISE] has a [social APE]: that [the founder] viewed his role as being mainly social. For these guys, the commercial activity

4 Observations that were eliminated as a result of in- complete or missing information were identical to those kept in our final working sample in most attributes and demonstrated negligible differences in others. The only noticeable difference was a higher percentage of female beneficiaries at 42%, compared to 33% in our final work- ing sample.

5 Such code changes are very rare, mainly because APE codes are assigned by INSEE only for statistical purposes and do not have any fiscal, financial, or legal implications at the time of founding or later (see Mercadal, Janin, Charvériat, & Couret, 2004). Furthermore, once filed, an objet social can be modified only by an extraordinary de- cision of the board through a qualified majority. In our sample, only two WISEs (corresponding to less than 1% of unique WISEs) changed their APE codes between 2003 and 2007; both of these changes were from an industry-specific APE to a social one in 2005 andback to the original industry- specific APE in 2006. According to our best assessment and cross-checks, these two cases reflect data entry errors in 2005. Even though the results were insensitive to inclusion or exclusion of these two WISEs, we took a conservative approach and excluded them from our analyses.

1664 DecemberAcademy of Management Journal

is only a medium for workforce development. Those who [have] an industry-specific APE are in a different state of mind. They know that they are here for the social mission, but they think that their main role is to develop their commercial activity and that the social impact will follow from it.

Accordingly, we coded social imprinting as “1” if the WISE had a social APE code and “0” if it had an industry-specific APE code. The WISEs in our data- set had 56 different APE codes. We classified the following APE codes as social: 85.3H, aide par le travail et ateliers protégés (sheltered workshops); 85.3K, autres formes d’action sociale (other types of social service); and 91.3E, organizations associatives n.c.a. (associations). In our sample, roughly 27% of CNEI members had a social APE code. The rest were industry-specific APE codes, such as 55.5D, traiteurs et organization de receptions (catering services), or 52.7D, réparation d’appareils électroménagers (ap- pliance repair).

Economic productivity. Following Huselid (1995) and Rangan and Sengul (2009), we measured the economic productivity of a WISE as the ratio of total annual sales to the number of employees, including both permanent staff and beneficiaries. To account for productivity differences across sectors, we com- puted standardized economic productivity scores that capture by how many standard deviations the economic productivity of each WISE lies above or below the median score in its sector of activity. (As we describe in “Control variables” below, in all models we also included sector dummies.) We use these median-standardized scores in all regres- sions.6 The sales per employee measure is particu- larly well suited to examining economic productivity differences across organizations that operate in dif- ferent sectors and, when controlling for mean differ- ences across sectors, reflects a combination of the managerial (sales, training, mentoring, etc.) capabil- ity of permanent staff and production capability of beneficiaries.

Control Variables

In terms of organizational characteristics, we controlled for organization size (measured by the

log of total number of full-time equivalents, or FTEs, working in the WISE) and organization age (mea- sured by the number of years since its founding, scaled by 1/10 to help the demonstration of co- efficients). These factors might affect a WISE’s so- cial performance because the extent of available organizational resources may vary by organization size (Huber, Sutcliffe, Miller, & Glick, 1993) and because an organization’s ability to convert re- sources to organizational outcomes is likely to vary as a function of its age (Miller & Shamsie, 2001). Another organizational characteristic of WISEs is their legal status. As we noted earlier, WISEs face similar constraints in the French context regardless of their legal status. Still, we took a conservative approach and included a control for the not-for- profit legal status, coded as “1” if not-for-profit and “0” otherwise.

In terms of human resources, we first controlled for the level of supervision (measured by the log of the ratio of the total number of permanent staff involved in the mentoring and training of beneficiaries to the total number of beneficiaries). Higher levels of su- pervision may reflect higher levels of on-the-job training and/or individualized support and may, in turn, enhance beneficiaries’ employment prospects. We also controlled for the demographic profile of beneficiaries. Although WISEs hire all of their ben- eficiaries from the pool of individuals listed by Pole Emploi as experiencing social and professional dif- ficulties, prior studies and anecdotal evidence have shown that women and/or individuals who are deemed to be too young (that is, too inexperienced) or too old are particularly disadvantaged and find it more challenging to find a job (D’Autume, Bethbèze, & Hairault, 2006; Margaret, 2006). Accordingly, we include the percentage of total beneficiaries who are female, under 26 years old, and over 50 years old as control variables.

In terms of financial structure, we first controlled for subsidies received (measured by the ratio of the total amount of all subsidies that the WISE received in tens of thousands of euros to the number of its employees). This variable captures the extent of or- ganizational financial resources that do not directly depend on economic productivity on a per employee basis. Because effort allocation and marginal returns to effort can be expected to be a function of costs (Miller & Shamsie, 2001), we also controlled for av- erage wage cost, which we measured by the log of the ratio of total wage cost of the WISE to the number of employees. Controlling for region and sector effects, wage costs may account for some underlying quality

6 This measure is highly correlated with the raw sales per employee values, as well as an alternative measure standardizing at the sector mean (0.919 and 0.994, re- spectively). Importantly, the results were qualitatively identical when using raw, mean-standardized, or median- standardized sales per employee values.

2015 1665Battilana, Sengul, Pache, and Model

of organizational members, which could affect the WISE’s social performance.

Finally, we included founding period dummies to control for the possible influence of environmental imprints and historical conditions at the time of founding (Stinchcombe, 1965). The WISE field in France has evolved through four distinct stages: the experimentation phase prior to 1985, the contesta- tion phase between 1986 and 1990, the institution- alization phase between 1991 and 1997, and the professionalization phase starting in 1998 (see Ap- pendix A). Accordingly, leaving founded prior to 1985 as the benchmark category, we include foun- ded in period 1986–1990, founded in period 1991– 1997, and founded in period after 1997 dummies as control variables.

We report the summary statistics and bivariate zero-order correlations in Table 1. A prototypical (that is, average) WISE establishment had about 28 permanent staff and beneficiaries. Legal statuses of WISEs were roughly equally split, with half having a not-for-profit status and half having a for-profit status. Of beneficiaries, 18% were under 26 years old, 12% were over 50 years old, and 33% were female. Average sales were V24,270 per FTE, with a corresponding average wage cost of V20,537. On average, WISEs were able to place 38% of their graduating beneficiaries in a job.

Estimation

In choosing our estimation method, we took into consideration the cross-section and time-series nature of our data and ran all regressions using a generalized least squares (GLS) (random effects) estimator. The random effects estimator assumes that average values for each establishment differ from the population average (that is, that there are establishment-specific differences) and that differ- ences across establishments have an underlying distribution (that is, that each establishment is ran- domly selected from a larger population of estab- lishments). Accordingly, we report regressions with random establishment effects.7 We also took into account that factors unobservable to us (but observ- able to WISEs) might affect the levels of both the dependent variable and the right-hand-side vari- ables. If this were the case, it would result in si- multaneity: A shock that affects a WISE’s social performance might also affect its size, wage costs,

and so on in the same year. To attenuate the potential issue of simultaneity, and keeping with convention, we lag all right-hand-side variables by one year. Addi- tionally,toaccountforunobservedheterogeneityin the data, we included year, region, and sector dummies in all regressions. For regions, we used the official French regional classification, which comprises 27 administrative regions. For sectors, using the clas- sification in the CNEI dataset, we coded WISEs into six sectors: business services (services aux entre- prises); construction (bâtiment, travaux publics); en- vironment (environnement, espaces verts); garbage (déchets:collecte,tri,déconstruction,dépollution);recy- cling (récuperation, recyclage et commerce d’occasion); and others (all remaining sectors). We report robust (that is, heteroskedasticity-corrected) standard errors clus- tered by WISE for all models.

REGRESSION RESULTS

We hypothesized that both social imprinting (Hy- pothesis 1) and economic productivity (Hypothesis 2) are positively associated with WISEs’ social per- formance. The regression results reported in Table 2 strongly support these hypotheses: The coefficients of social imprinting (Model 2) and economic pro- ductivity (Model 3) are both positive and statisti- cally significant, as predicted. The results are substantively significant as well. Holding all other variables at their mean values, organizations with social imprinting have, on average, 9.9% (and a 3.7 percentage-point) higher social performance than those without it. Similarly, a 1SD increase in economic productivity is associated with a 5.8% (and a 2.2 percentage-point) increase in social performance.8

Table 3 reports the regressions explaining the drivers of economic productivity in WISEs. In line with Hypothesis 3, the coefficient of social imprint- ing is negative and statistically significant (Model 2). Holding all other variables at their mean values,

7 Pooled ordinary least squares (OLS) yielded qualita- tively identical results.

8 We also checked for a curvilinear relationship between economic productivity and social performance. If achiev- ing very high levels of economic productivity requires WISEs to devote more and more resources to production activities, this may result in diverting resources away from activities that directly contribute to social performance. Simply put, there might be an optimal level of economic productivity. We did not find support for this argument in supplementary regressions including a squared term for economic productivity, which was insignificant in all models.

1666 DecemberAcademy of Management Journal

organizations withsocialimprinting have, on average, 13.3% lower economic productivity (corresponding to V3,342 lower sales per employee on an annual basis) than those without it. It is important to note, however, that economic productivity only par- tially mediates the relationship between social imprinting and social performance. Even though social imprinting weakens social performance through the negative relationship that it has with economic productivity, it still has a positive and significant net effect on social performance (see Table 2, Models 2 and 4). To obtain more conclu- sive evidence, we conducted Sobel, Aorian, and Goodman tests of mediation, which enable an as- sessment of indirect effects of the independent variable on the dependent variable via the medi- ating variable (see Iacobucci, 2008: ch. 2). The results of these tests (which are all significant at p , .05) indicate that social imprinting has a nega- tive indirect (that is, partially mediated) effect on social performance through its relationship with economic productivity.

In supplementary regressions, we also explored whether the observed relationships between social

imprinting and social performance, and between social imprinting and economic productivity, persist over time. To do so, we included an interaction term between our measure of social imprinting and organization age. The interaction term was negative and statistically significant on both social perfor- mance and economic productivity, indicating im- print decay. We then calculated differences in predicted values of social performance across so- cially imprinted and commercially imprinted WISEs over time. A year after its incorporation, a socially imprinted WISE has an 8 percentage-point higher positive graduation rate compared to a commercially imprinted one. The difference goes down to a 5 percentage-point spread in the eighth year and is halved, after a decade, to a 4.1 percentage-point dif- ference. It is still roughly at 2 percentage points 15 years after incorporation. These results imply that the relationship between social imprinting at founding and organizational outcomes decreases in magnitude over time but persists long after founding.

Lastly, some brief observations are in order re- garding our control variables. First, organizational characteristics—organization size and age—explain

TABLE 1 Means, Standard Deviations, and Bivariate Zero-Order Correlations

Variables Mean SD 1 2 3 4 5 6 7

1 Social performance 0.38 0.19 2 Social imprinting 0.27 0.45 0.07 3 Economic productivity 0.04 1.05 0.19 20.14 4 Organization size (log) 3.33 0.54 20.10 0.01 20.05 5 Organization age (410) 1.02 0.56 20.03 0.16 0.08 0.19 6 Not-for-profit legal status 0.54 0.50 20.08 0.44 20.29 20.11 0.36 7 Level of supervision (log) 20.99 0.77 0.17 20.05 0.40 0.05 0.13 20.18 8 Beneficiaries (%)—Female 0.33 0.30 20.08 20.04 20.28 0.06 20.02 0.10 20.16 9 Beneficiaries (%)—Under 26 years old 0.18 0.15 0.10 0.02 0.12 0.00 0.09 20.08 0.19 10 Beneficiaries (%)—Over 50 years old 0.12 0.10 20.01 0.09 20.04 20.04 0.02 0.16 20.13 11 Subsidies received 0.85 0.33 20.10 0.12 20.17 20.13 0.09 0.25 20.41 12 Average wage cost (log) 9.93 0.31 0.13 0.04 0.47 20.03 0.10 20.07 0.22 13 Founded in period: 1986–1990 0.12 0.32 20.08 20.06 0.03 0.20 0.38 20.02 0.18 14 Founded in period: 1991–1997 0.46 0.50 20.09 0.03 20.09 20.06 0.07 0.18 20.06 15 Founded in period: After 1997 0.33 0.47 0.05 20.12 0.02 20.10 20.74 20.34 20.09

Variables Mean SD 8 9 10 11 12 13 14

8 Beneficiaries (%)—Female 0.33 0.30 9 Beneficiaries (%)—Under 26 years old 0.18 0.15 20.09 10 Beneficiaries (%)—Over 50 years old 0.12 0.10 0.05 20.26 11 Subsidies received 0.85 0.33 0.02 20.11 0.06 12 Average wage cost (log) 9.93 0.31 20.19 0.15 20.04 0.12 13 Founded in period: 1986–1990 0.12 0.32 0.05 0.13 20.05 20.09 20.01 14 Founded in period: 1991–1997 0.46 0.50 20.13 20.14 0.04 0.09 20.07 20.33 15 Founded in period: After 1997 0.33 0.47 0.09 0.01 20.02 20.10 20.01 20.25 20.65

Note: n 5 641; all independent and control variables are one-year lagged.

2015 1667Battilana, Sengul, Pache, and Model

productivity differences across WISEs but not social performance differences across them. Economic productivity tends to be higher in WISEs that are more experienced and relatively smaller.

Second, not-for-profit legal status does not have a significant effect on social performance. Yet eco- nomic productivity tends to be lower in not-for- profit WISEs.

Third, supervision helps both economic pro- ductivity and social performance. This reflects that better trained and guided beneficiaries are more productive, and that productive beneficiaries are more likely to find jobs after their time at WISEs.

Fourth, the percentage of female beneficiaries is negatively associated with economic productivity.

Fifth, average wage cost is positively associated with economic productivity, in line with efficiency wage models in labor economics (see, e.g., Yellen, 1984).

Sixth, WISEs founded in the 1986–1990 period have lower social performance compared to WISEs founded in other periods. This period was marked by the withdrawal of state support from WISEs; hence, the WISEs founded in this period faced significant challenges in obtaining the required resources in their early years (see Appendix A). This result, al- though at a different level of analysis, resonates with Tilcsik (2014), who showed that early career re- source environment has a lasting influence on career trajectories of individuals.

TABLE 2 Random-Effects Regressions Explaining the Social Performance of French Work Integration Social Enterprises, 2003–2007

Variables 1 2 3 4

Social imprinting 0.031† 0.037* (0.022) (0.022)

Economic productivity 0.020* 0.022* (0.011) (0.012)

Organization size 20.021 20.023 20.015 20.017 (0.019) (0.019) (0.019) (0.019)

Organization age 20.006 20.008 20.022 20.024 (0.047) (0.047) (0.047) (0.047)

Not-for-profit legal status 20.022 20.035 20.010 20.025 (0.021) (0.024) (0.021) (0.024)

Level of supervision 0.040** 0.040** 0.034* 0.034* (0.014) (0.014) (0.014) (0.014)

Beneficiaries Female 20.009 20.006 20.008 0.014

(0.037) (0.037) (0.038) (0.038) Under 26 years old 20.005 20.007 20.006 20.008

(0.060) (0.059) (0.060) (0.059) Over 50 years old 0.042 0.038 0.034 0.029

(0.075) (0.075) (0.075) (0.075) Subsidies received 20.012 20.014 20.008 20.010

(0.023) (0.023) (0.023) (0.023) Average wage cost 0.013 0.015 0.009 20.008

(0.024) (0.025) (0.026) (0.027) Founded in period 1986–1990 20.083† 20.079† 20.083† 20.078†

(0.044) (0.043) (0.044) (0.043) 1991–1997 20.048 20.047 20.055 20.054

(0.058) (0.058) (0.058) (0.058) After 1997 20.041 20.042 20.058 20.060

(0.087) (0.086) (0.087) (0.086) R2 (overall) 0.180 0.184 0.187 0.191 Wald x2 99.24** 100.82** 107.04** 110.20**

Note: n 5 641; robust standard errors, clustered by WISE, in parentheses; constant, year, sector, and region dummies included in all models; all independent and control variables are one-year lagged.

† p , .10 * p , .05

** p , .01 (two-tailed tests; one-tailed tests, when hypothesized)

1668 DecemberAcademy of Management Journal

Finally, unlike year and sector dummies, most re- gion dummies (not reported) were statistically signifi- cant, reflecting the influence of economic, social, and demographic differences across regions on WISEs’ social performance.

Robustness Checks and Supplementary Analyses

We conducted a set of supplementary empirical analyses using the available data to explore the ro- bustness of the regression results to a number of po- tential concerns. First, to verify the validity of APE codes as a proxy for social imprinting, we collected additional data and created two alternative measures of social imprinting for the WISEs in our sample: one

based on our coding of original objet social state- ments9 and another based on the assessment of a former CNEI president.10 These two alternative measures of social imprinting were highly similar to the social vs. industry-specific APE codes that we use: the former measure identical to APE codes in 91% of the cases and the latter, in 84%. When we re- estimated our models using these two alternative measures, social imprinting was still positively and significantly associated with social performance, and negatively and statistically associated with eco- nomic productivity. Therefore, we are confident that our measure (which is available for the full sample) reliably captures social imprinting at founding.

We also checked whether the measurement of so- cial performance as a ratio bounded between 0 and 1 impose any bias in the reported results. Tobit is the suggested estimation method for dependent

TABLE 3 Random-Effects Regressions Explaining the Economic

Productivity of French Work Integration Social Enterprises, 2003–2007

Variable 1 2

Social imprinting 20.334** (0.144)

Organization size 20.540** 20.526** (0.093) (0.092)

Organization age 0.829** 0.833** (0.317) (0.310)

Not-for-profit legal status 20.687** 20.541** (0.106) (0.126)

Level of supervision 0.146** 0.149** (0.060) (0.057)

Beneficiaries Female 20.753** 20.793**

(0.184) (0.183) Under 26 years old 0.121 0.122

(0.215) (0.216) Over 50 years old 0.408 0.243

(0.264) (0.262) Subsidies received 0.086 0.088

(0.139) (0.139) Average wage cost 0.687** 0.686**

(0.182) (0.184) Founded in period 1986–1990 0.056 0.019

(0.308) (0.308) 1991–1997 0.296 0.277

(0.381) (0.380) After 1997 0.736 0.739

(0.559) (0.554) R2 (overall) 0.408 0.416 Wald x2 246.67** 250.39**

Note: n 5 641; robust standard errors, clustered by WISE, in parentheses; constant, year, sector, and region dummies included in all models; all independent and control variables are one-year lagged.

** p , .01 (two-tailed tests; one-tailed tests, when hypothesized)

9 These statements are kept at the local préfectures (i.e., local government administrative centers) within which each organization was originally incorporated. Consequently, we filed official requests for each WISE that we had in our final working sample to local préfectures. Out of 230 individual requests that we made, we received 128 objet socialstatements inprint bymail at the end of this process, corresponding to a total of 350 observations. Then, two of the coauthors independently coded the social im- printing of each WISE based on the information provided in its objet social. Specifically, they coded whether the social mission was emphasized at the time of founding. A WISE with social imprint would, for example, have an objet social such as “providing work for people facing so- cial and professional challenges and offering them tailored training in order to reintegrate them into the regular job market,” and a WISE with an commercial imprint would, for example, have an objet social such as “refurbishing and recycling of used computer and electronic appli- ances; sale of these products and associated services.” Interrater reliability, as assessed by the k correlation co- efficient, was 0.89, suggesting a high degree of agreement between the two coders. The few remaining disagree- ments were resolved through discussions until consensus was reached.

10 We contacted a former CNEI president, who had wit- nessed the rise of the field over 30 years, and asked him to code the imprints of the organizations about which he felt he had enough information to assess their early imprint. We provided him the names and addresses of all of the WISEs in our sample but did not share their APE codes or objet social statements. He delivered his personal assess- ment of whether a given WISE had an early social imprint for 96 WISEs in our sample, corresponding to a total of 265 observations.

2015 1669Battilana, Sengul, Pache, and Model

variables that are bounded and have limit observa- tions.(In our dataset, there were only a total of 46 limit observations, with 26 occasions on which a WISE was unable to place any of its participants in a job upon graduation and 20 occasions on which it was able to place all of them.) However, Tobit estimation is based on an underlying latent variable, which is meaningful when the latent variable could exist beyond the limits—hence implying truncation—but not when this is not feasible. When this is the case, frac- tional logit, using generalized linear modeling (GLM) or generalized estimating equations (GEE), is the suggested estimation method (McDonald, 2009; Papke & Wooldridge, 1996). The results are qualitatively identical across all different esti- mations (that is, GLS, Tobit, GLM, and GEE), fur- ther increasing our confidence in the reported results.

Third, one might be concerned about a potential survivorship bias if a nontrivial number of the WISEs were to fail and those WISEs were to be qualitatively different from our sample. To verify, we checked for mortality in the CNEI dataset during our observation period. In line with our knowledge of the field and the information that we collected from field experts, the mortality rate was low. Furthermore, the few WISEs that ceased operations did not differ from those in our working sample in observables, imply- ing that there is no significant survival bias in the reported results.

A fourth concern relates to the possibility that the relationship between economic productivity and social performance may be picking up the effect of some of the causes of economic productivity and hence may be spurious. To check, we obtained fitted (that is, predicted) economic productivity values and residuals from Model 2 of Table 3 (the fully specified model on economic productivity); then, as a second step, we re-estimated the models in Table 2 (on social performance) using these fitted values and residuals. In these regressions, the fitted values and the residuals were both positive and statistically significant, implying the presence of an independent effect of economic productivity on social performance in addition to those attributable to potential predictors in common, such as quality of management.

A fifth concern relates to the possibility that the error terms may be correlated across the models predicting social performance and economic pro- ductivity. If that were the case and the correlation were large, we could expect gains in efficiency by estimating the equations together rather than

estimating each equation separately (Baltagi, 1998). Accordingly, we reran all regressions using seem- ingly unrelated regressions (SUR) and using struc- tural equation modeling (SEM). The results of these analyses (see Table 4) are highly similar to each other, as well as to the GLS results that we reported in Tables 2 and 3. Importantly, all hypothesized re- lationships remain unchanged in terms of sign and significance.

Finally, in terms of model specification, we also checked the robustness of the results to serial corre- lation and multicollinearity. To check the robust- ness of the results to potential serial correlation, we re-estimated all of the models using two separate methods that account for first-order autocorrelation in panel data: the Baltagi-Wu autoregressive esti- mator and GEE regressions imposing a within-panel correlation of the order AR(1). In both checks, social imprinting and economic productivity were still positively and (slightly more) significantly related to social performance. Separately, to check the robust- ness of the results to potential multicollinearity (even though there are no critically collinear vari- ables), we reran the regressions by dropping mod- erately correlated control variables (such as average wage cost, founding period dummies), both in- dividually and in combinations. Regression results were qualitatively insensitive to the inclusion or ex- clusion of these variables (but exclusions typically resulted in lower model fit).11

UNPACKING THE PARADOXICAL RELATIONSHIP BETWEEN SOCIAL

IMPRINTING AND SOCIAL PERFORMANCE IN WISES

The results of the quantitative analyses confirm the paradoxical relationship between social im- printing and social performance in WISEs: Although social imprinting enhances a WISE’s social perfor- mance by keeping it from neglecting its beneficiaries, social imprinting also indirectly weakens social performance through the negative relationship that social imprinting has with economic productivity. To unpack this paradox, we now examine the mechanisms that underlie it and explore whether, and how, socially imprinted WISEs can resolve it

11 We also calculated variance inflation factors (VIFs) on both pooled and WISE-demeaned data. All calculated VIFs were much lower than the critical value of 10, indicating no serious multicollinearity.

1670 DecemberAcademy of Management Journal

through a longitudinal comparative analysis of two WISEs, referred to as “ALPHA” and “BETA.”

Qualitative Data Collection and Analysis

Relying on a theoretical sampling approach (Eisenhardt, 1989), we analyzed two cases of socially imprinted WISEs operating in the recycling industry: “ALPHA” and “BETA.” We selected these two or- ganizations from our sample because they are both socially imprinted WISEs (that is, they both have a social APE code), yet achieved different levels

of economic productivity when compared to the other socially imprinted WISEs operating in the recycling sector: ALPHA’s economic productivity was below average (at V17,460 of sales revenue per employee) and BETA’s, above average (at V25,867), in the period that followed an economic crisis that both organizations coincidentally experienced in 2003.12 Beyond this stark difference in postcrisis

TABLE 4 Seemingly Unrelated Regressions (SUR) and Structural Equation Modeling (SEM) Regressions Explaining the Economic

Productivity and Social Performance of French Work Integration Social Enterprises, 2003–2007

SUR SEM

Economic productivity Social performance Economic productivity Social performance

Social imprinting 20.257** 0.038* 20.255** 0.039* (0.085) (0.020) (0.085) (0.020)

Economic productivity 0.022** 0.023** (0.009) (0.009)

Organization size 20.252** 20.016 20.257** 20.019 (0.066) (0.016) (0.066) (0.016)

Organization age 0.804** 20.018 0.802** 20.020 (0.175) (0.041) (0.175) (0.041)

Not-for-profit legal status 20.429** 20.026 20.435** 20.029 (0.082) (0.020) (0.082) (0.020)

Level of supervision 0.296** 0.033** 0.291** 0.030* (0.051) (0.012) (0.050) (0.012)

Beneficiaries Female 20.816** 0.011 20.813** 0.013

(0.135) (0.032) (0.135) (0.032) Under 26 years old 0.125 0.006 0.117 0.002

(0.230) (0.053) (0.230) (0.053) Over 50 years old 0.461 0.036 0.471 0.042

(0.328) (0.076) (0.329) (0.077) Subsidies received 20.235* 20.008 20.237* 20.009

(0.117) (0.027) (0.117) (0.027) Average wage cost 1.189** 20.005 1.184** 20.008

(0.115) (0.029) (0.115) (0.029) Founded in period 1986–1990 20.032 20.075† 20.032 20.075†

(0.165) (0.038) (0.165) (0.038) 1991–1997 0.397† 20.045 0.393† 20.049

(0.212) (0.049) (0.212) (0.049) After 1997 0.804** 20.049 0.881** 20.051

(0.309) (0.072) (0.309) (0.073) R2 0.489 0.192 0.489 0.186 x 2 606.26** 152.23**

AIC 1258.15

Note: n 5 641; robust standard errors, clustered by WISE, in parentheses; constant, year, sector, and region dummies included in all models; all independent and control variables are one-year lagged.

† p , .10 * p , .05

** p , .01 (two-tailed tests; one-tailed tests, when hypothesized)

12 In comparison, the average productivity of the socially imprinted WISEs that operate in the recycling industry was V23,153 in the same time period.

2015 1671Battilana, Sengul, Pache, and Model

economic productivity levels, ALPHA and BETA do not demonstrate significant differences from each other (with the exception of a relatively higher per- centage of female employees at ALPHA) and are similar to other WISEs in the recycling industry. They are both midsized WISEs founded roughly around the same time; they both operate in the recycling sector under a not-for-profit legal status. Therefore, the comparison between these two cases allows us to gain valuable insights into the strategies that socially imprinted organizations may use to re- solve the social imprinting paradox.

We collected a broad range of data about both or- ganizations. One of the authors conducted semi- structured interviews with members of each WISE, including founders, board members, executive di- rectors, production supervisors, and social coun- selors. A total of 20 interviews were conducted onsite, each lasting between 30 minutes and 3 hours. Four follow-up phone interviews were conducted with case informants. In addition to case-level in- terviews, 11 more interviews were conducted with members of the broader WISE community (such as CNEI executives, workforce development consul- tants) in order to explore the field-level stakes and trends. All interviews were conducted in French, then later transcribed and translated to English.

During the case-level interviews, all informants were asked about the founding and growth of their WISE, its goals and values, and its interactions with its environment, as well as their own perception of the determinant of the WISE’s social performance. The interview guide was adapted to the role of inter- viewees (for example, executive director vs. social counselor).Whileallinterviewsstartedwithquestions related to the profile and background of interviewees, founders were specifically asked about their motiva- tion to create the WISE and about the process through which they created, and later managed, it. Executive directors were asked about the specifics of the WISE’s functioning, including economic and social opera- tions, organizational structure, governance, and hu- man resources practices. Production supervisors and social counselors were asked to describe in detail their activities, as well as their interactions with the rest of the organization. For eachcase, we asked interviewees to provide us with archival material that they thought might help us to understand the functioning of the WISE. Through thisprocess, wecollected awiderange of archival material, including by-laws, annual re- ports, brochures, anniversary booklets, etc.

In analyzing the qualitative data, we adopted a comparative case study design. Following Eisenhardt

(1989), we first focused on within-case analysis and then turned to cross-case analysis. (See Appendix B for a detailed description of the steps that we followed in analyzing our qualitative data.) The analyses that we conducted were longitudinal, ex- amining the evolution of both organizations from their founding until 2008.

We now present the results of these analyses in two steps. We first analyze the evolution of both organi- zations from their founding to the financial crisis that both, coincidentally, experienced in 2003. We then turn to the postcrisis period, during which BETA was able to execute a successful turnaround whereas ALPHA was not.

Social Imprinting at ALPHA and BETA (from Founding to 2003)

Our within-case analyses suggest a general pat- tern that supports our regression results and pro- vides a more fine-grained understanding of the mechanisms underlying the hypothesized relation- ships that we tested in prior sections: In both ALPHA and BETA, the founders’ goals and values had a lasting effect on subsequent organizational prac- tices and routines. The comparison of the lasting influence of social imprinting on ALPHA and BETA, which we present in more detail below, is summa- rized in Table 5.

Founders’ imprints. A group of social workers founded ALPHA in 1990. Drawing upon their expe- rience of “sheltered workshops” for disabled workers, its founders conceived ALPHA as a place dedicated to teaching working skills to long-term unemployed people. They chose recycling as their commercial activity because it allowed them to cre- ate an in-house workshop within which work could be organized to fit the capacities of the long-term unemployed. ALPHA’s founders defined the mis- sion of the organization as “supporting individuals with social and professional difficulties through mentoring, tailored trainings and on-the-job practice as a means to help them access or reintegrate into the workforce.” The board of directors of ALPHA was composed of a group of social workers that actively promoted a focus on accomplishing ALPHA’s social mission. Thus, ALPHA experienced a strong social imprint from its founders, who emphasized from the very beginning the achievement of the social mission.

BETA was founded in 1993 by a former school- teacher who was a board member of the community center of his small town. BETA’s founder recalls:

1672 DecemberAcademy of Management Journal

After a few years, I realized that the workshops that we organized in the community center were okay, but it was not enough for the long-term unemployed . . . So I got interested in the WISE model. When we decided to create one, we were initially not sure what our com- mercial activity was going to be.

Like ALPHA’s founders, he chose recycling as the main activity of BETA because it was a simple ac- tivity that beneficiaries without many skills could learn. He translated his deep commitment to BETA’s social mission into an organizational charter that claimed to “fit the company to its beneficiaries, rather than fit beneficiaries to the company,” and emphasized values such as “humanism” and “em- pathy.” He reinforced his commitment to the social mission by recruiting board members from fellow activists from the local community center. Like AL- PHA, BETA’s founder stamped the new organization with a strong social imprint.

Hiring patterns. At ALPHA, the first executive director was one of the founding social workers and the two subsequent executive directors also had backgrounds in social work. Their profile had im- portant consequences on the skills and values that they brought to the organization. They hired staff based on their care for beneficiaries and on their knowledge of counseling. The first executive di- rector created—and his successors perpetuated—a permanent social counseling position, staffed by a social worker. Three production supervisor posi- tions were staffed either through the director’s social work networks or by recruiting former beneficiaries.

Similar hiring patterns were found at BETA. The strong social focus of the founder led him to recruit an executive director with a social work background. BETA’s founder explained: “Our core job is work- force development. . . . So we needed to recruit an executive director who could manage the workforce

TABLE 5 Social Imprinting at ALPHA and BETA (from Founding to 2003)

ALPHA BETA

Founders’ imprints 1990 1993 Motivation: ALPHA’s founders were a group of social workers frustrated with the limited ability of social work to address the needs of people with social and professional difficulties. The founders created ALPHA to provide them with mentoring and on-the-job training to facilitate their (re)integration into the workforce.

Motivation: The motivation of BETA’s founder, a social activist involved in the local community center, was to address the needs of the jobless members of the community center on the board of which he sat.

Activity emphasized: The founding team was influenced by the notion of a “sheltered workshop,” conceived as a place in which work would be conducted in a way that protects fragile workers. They chose recycling as their commercial activity because it allowed them to create an in-house workshop in which work could be organized to fit the capacities of the long-term unemployed.

Activity emphasized: The focus of the founding team was to create an organization that was able to meet the needs of youth at risk. The commercial activity (the management of waste reception centers) was later chosen to provide development opportunities to young people without qualification.

The lasting influence of founders’ imprints

1990–2003 1993–2003 Hiring patterns: Recruitment of socially oriented staff, including:

Hiring patterns: Recruitment of socially oriented staff, including:

Three successive executive directors with social work profile

Executive director with social work profile

Up to five production supervisors with social work profile

Up to five production supervisors recruited among beneficiaries

Up to three social counselors with social work profile

Up to three social counselors with social work profile

Organizational features: Implementation of socially oriented processes and systems, including:

Organizational features: Implementation of socially oriented processes and systems

Creation of a training center Development of services in: Creation of a housing shelter Psychological support

Social support Job readiness support

Development of reporting tool (“work integration passport”) to track beneficiaries’ progress

Development of reporting tool (“monitoring scorecard”) to track beneficiaries’ progress

2015 1673Battilana, Sengul, Pache, and Model

development part of the job.” Subsequent recruitments of permanent staff were also driven by a willingness to help people in need. Overall, permanent staff at BETA exhibited a combination of care and skills geared to- ward the advancement of beneficiaries. In line with the mechanisms that we presented in Hypothesis 1, our qualitative analysis thus suggests that, in both organi- zations, social imprinting led to the hiring of social- mission-oriented staff members.

Organizational systems and processes. Social imprinting also influenced the development of activities that embedded social-mission-oriented routines within the fabric of both organizations. ALPHA’spermanent staff, for instance, createdan in- house trainingcentertoprovidebeneficiarieswithjob readiness and literacy training. They also created an in-house shelter to provide accommodation for ben- eficiaries who faced housing difficulties. In addition, they designed a tool, which they called a “work in- tegration passport,” to track beneficiaries’ perfor- mance and help them to enhance their employability during the course of their time at ALPHA.

In BETA, like ALPHA, the founder’s early emphasis on the accomplishment of the social mission led the organization to invest in the support of beneficiaries. BETA developed specific processes in three distinct areas: psychological support, social support, and job- readiness support. On each of these themes, beneficia- ries were offered specific counseling sessions that were designed to help them to address various obstacles to work. A scorecard, similar to ALPHA’s work in- tegrationpassport,wascreatedtomonitorandsharethe progress of beneficiaries in each of these dimensions.

Negative relationship between social imprint- ing and economic productivity. In line with the mechanisms that we presented in Hypothesis 3, our qualitative analysis revealed that, in both organiza- tions, staff members with a social work background had developed skills and beliefs that fit the demands of the social sector but did not fit with the com- mercial sector. Between 1990 and 2003, ALPHA’s successive executive directors, all with social backgrounds, did not have experience and knowl- edge about business and production processes in the recycling industry. None of them had ever worked in commercial operations, let alone in a for- profit company, prior to joining ALPHA. Because they had spent their entire career in social work organizations, these former executive directors did not have well-developed business skills and, as the current executive director explained, “prioritized beneficiaries’ needs over everything else.” Although the issue of poor productivity had been lingering for

years, its persistence finally led to a major organiza- tional crisis in a period of economic hardship in France in the early 2000s.13 With debts increasing and in- sufficient cash flow to pay staff salaries, ALPHA was in a dire financial situation by 2003. The board of di- rectors, still composed of social workers, was forced to reconsider established organizational practices and searched for alternative approaches.

Like ALPHA, BETA’s board and permanent staff had a strong commitment and deep expertise in the social realm but lacked business skills and experi- ence. The current executive director described the situation when she joined BETA in 2004, after it had been managed for seven years by an executive di- rector with a social work background and no prior business experience, as follows: “Contracts were developed orally, conventions were not signed. Charges were not monitored and controlled, and as a result, costs were inevitably too high.” Further- more, because of the permanent staff’s extreme care and empathy toward beneficiaries, they tended to be lenient when beneficiaries were unprofessional. In the current executive director’s words, “beneficia- ries knew that they would never get in big trouble.” As a result of this approach, BETA’s economic pro- ductivity was recurrently low. Like ALPHA, BETA also experienced a financial crisis in 2003, when many years of poor productivity led to unsustainably high levels of debt. This problematic situation forced the board to embrace change and to take economic demands more seriously into account.

After the Crisis: ALPHA and BETA’s Different Courses of Action

The crises experienced by both ALPHA and BETA played an important role in challenging the status quo and unfreezing socially imprinted patterns in both organizations. As a result, both organizations changed their recruitment approach and hired an executive director with a combination of social sec- tor and business experience; both organizations also relied on structural differentiation, with one group of people (that is, production supervisors) focusing on the needs of clients and another group (that is, social counselors) focusing on the needs of beneficiaries. However, ALPHA and BETA adopted different ap- proaches to ensuring coordination between these two

13 Growth in gross domestic product (GDP) was only 0.9% in both 2002 and 2003—the lowest level in the five decades (with the exceptions of 1975 and 1993) prior to the 2007–2008 global financial crisis (World Bank, 2013).

1674 DecemberAcademy of Management Journal

groups. Whereas, at ALPHA, the executive director was the single “integrator” (Galbraith, 1973; Lawrence & Lorsch, 1967), responsible for coordinating both groups, BETA relied on “spaces of negotiation” that engaged staff members throughout the organizational ladder to ensure coordination among them. These “spaces of negotiation” were arenas of interaction that allowed staff members in charge of different (that is, social vs. economic) activities to discuss and agree on how to handle the trade-offs that they faced.

We now compare ALPHA and BETA’s approaches after the crises that they both experienced.

Organizational structure and staffing. Follow- ing the crisis, ALPHA’s board decided to change the leadership of the organization and recruited—for the first time—a new executive director with both social sector and business experience. This director had five years of experience in the banking industry and had previously worked for two years at ALPHA as part of his civic service. With this diverse experi- ence, he brought to the organization a much better understanding of business processes. The board made him responsible for both economic produc- tivity and social performance.

When he took over, there were five production supervisors, who had been recruited on the basis of their social work experience and were responsible for supervising all production processes, and two social counselors with social work background, who were in charge of beneficiaries’ individualized social support. Over the course of his first three years at ALPHA, the new executive director brought in five new production supervisors with industry (that is, technical and commercial) experience to replace all of the existing ones and made the new supervi- sors responsible for customer satisfaction on a daily basis. The executive director explained: “These production supervisors are in direct contact with clients and pressured to fill orders . . . So they are re- ally focused on production.” In turn, social coun- selors were in charge of managing social issues and beneficiaries’ performance appraisal, which focused mostly on their social skills. Thus, ALPHA adopted a structurally differentiated approach, with social counselors and production supervisors specializing, respectively, in social or commercial activities in which they had experience and/or training.

After the crisis that it experienced, BETA also en- gaged in a series of changes. Like ALPHA, the board of directors decided to hire a new executive director with a background in both social work and business. The new executive director had been trained as an engineer, had worked in a multinational corporation

for six years, and had recently obtained a master’s degree in social enterprise and community devel- opment. Her personal values, which spurred her career shift from the commercial to the social sector, aligned her with the board of directors, whose main concern, like ALPHA’s board, was to ensure that the organization’s social goal would be achieved. Im- portantly, for the first time in BETA’s history, the board gave a clear mandate to the new executive di- rector to improve economic productivity.

In a deliberate attempt to counterbalance her own social focus, the new executive director created a dep- uty director position to take responsibility for BETA’s commercial activities and hired a seasoned profes- sional with expertise in production. The executive di- rector explained: “We work as a pair. I understand the business, but I am more attuned to social issues, [the deputy director] is more attuned to economic issues. He is really committed to the business logic.” This executive pair set up a differentiated structure similar to ALPHA. Five production supervisors, who were selected based on their technical and commercial backgrounds, were in charge of activities and de- cisions pertaining to customers’ orders and the day- to-day organization of production, while three social counselors with social work backgrounds were in charge of activities and decisions pertaining to the personal development plan of beneficiaries.

Managing tensions. In both organizations, struc- tural differentiation led to internal tensions between production supervisors and social counselors, who quarreled over decisions such as the time spent by beneficiaries in production vs. social activities. In- terestingly, whereas both organizations experienced tensions, they managed them in different ways. At ALPHA, the executive director managed the ten- sions between social and commercial activities as they arose. For instance, he arbitrated when tensions emerged regarding the need for beneficiaries to take time off from the production line to attend counsel- ing sessions or trainings. These ad hoc interventions allowed him to relieve tensions temporarily and provided fixes to immediate issues. However, they did not equip staff members with the necessary tools and skills to manage these tensions themselves. As a result, tensions escalated into interpersonal con- flicts between social counselors and production su- pervisors, who were focused on the activities for which they were responsible and remained blind to others. Each group resented the other for keeping it from achieving its own goal. This resentment accu- mulated over time and ultimately led to ongoing in- ternal conflicts. Despite the clear efforts made by

2015 1675Battilana, Sengul, Pache, and Model

ALPHA to enhance its economic productivity, it remained below the average level of productivity of socially imprinted WISEs in the recycling sector.

In contrast, BETA did not rely solely on the exec- utive director to ensure coordination but instead created “spaces of negotiation,” as we detail in the next section. Similarly to ALPHA, roles and decision- making responsibilities were clearly assigned, with production supervisors in charge of technical training, production,andorders, andsocial counselorsincharge of the personal and professional development of ben- eficiaries. In contrast to ALPHA, BETA set up spaces of negotiation so that each group had to engage and con- sult with the other group before making a decision. This ensured that the entire staff was responsible for handling the day-to-day trade-offs and remained at- tentive to the demands of both social and commercial activities. If tensions between the two groups could not be resolved within the spaces of negotiation, the executive director would ultimately make the final decision. However, she and other staff mem- bers reported that she very rarely had to intervene.

Understanding the Role of Spaces of Negotiation

As this analysis reveals, ALPHA was not able to manage tensions between its social and economic activities effectively, and continued to suffer from below-average economic productivity during the time frame of our study. In contrast, BETA was able to manage these tensions effectively through the use of spaces of negotiation that enabled sustained co- ordination across structurally differentiated groups. Table 6 summarizes the approaches adopted re- spectively by ALPHA and BETA. Next, we examine BETA’sapproach in detail to uncover the origin, role, and the conditions that ensured the effective func- tioning of spaces of negotiation.

Creating and delineating spaces of negotiation. Spaces of negotiation at BETA were created through a combination of mandatory meetings and formal processes. Mandatory meetings were held to ensure the coordination of social and commercial activities. Every trimester, all social counselors and production supervisors met together to jointly assess the prog- ress of each beneficiary and to discuss any issues related to the coordination of their respective activ- ities. These meetings were internally referred to as “regulation meetings,” because they were meant to “regulate” tensions among staff by allowing each group to listen to the concerns of the other. In be- tween these regulation meetings, social counselors met every week to discuss issues related to benefi- ciaries. Similarly, production supervisors met every

week with the deputy director to organize production matters. Both groups shared minutes of these meet- ings with the executive director, keeping her in- formed of issues being discussed.

In addition to these mandatory meetings, formal processes were used to foster coordination between social and commercial activities throughout the orga- nizational ladder. In particular, work plan scheduling and beneficiaries’ performance appraisal played a crit- ical role. At the beginning of each month, the social counselorspreparedareport about all ofthecounseling activities organized for the following month. This re- port was then circulated to production supervisors, whowereaskedtoreportonpotentialclasheswiththeir ownactivities.Discussionsensuedtoresolveconflicts, such as the scheduling of training sessions during a peak production period. Staff members had three weeks within which to agree on a final schedule. The executive director explained:

With this process, social counselors are able to inform production supervisors about their own demands for training or mentoring. In turn, production supervisors are able to inform social counselors about production peak periods. It took us hours to make [this planning process] work,but itultimatelyreallyhelped. Once the planning is agreed on, people have to work according to it.

In addition to forcing production supervisors and social counselors to coordinate with each other, this process madeeach group publicly commit to specific work-planning arrangements.

Beneficiaries’ performance appraisal was another formal process that forced social counselors and technical supervisors to interact regularly. Each beneficiary met monthly with his or her assigned production supervisor and social counselor to jointly discuss his or her progress on technical, behavioral, social, and professional dimensions. Every meeting participant, including the benefi- ciary, filled out a performance appraisal document prior to the meeting. Assessments were compared during the meeting and differences discussed. Par- ticipants then jointly developed and committed to a plan to ensure future progress. Performance ap- praisal documents were then sent to the executive director.

In summary, mandatory “regulation” meetings, com- bined with the use of formal processes, created spaces of negotiation at BETA in which both social counselors and production supervisors had to engage before making any of the decisions for which they were responsible.Thearenasweredelineatedby:(a)theclear

1676 DecemberAcademy of Management Journal

definition and scope of the issues being addressed in meetings; (b) decision-making rules that defined who was responsible for what type of decision, who should participate in that decision, and what would happen if an agreement was not reached; and (c) temporal patterns for meetings and clear deadlines for decision making. Within these boundaries, the use of formal processes (such as work plan schedules and performance appraisal grids) helped members with various profiles and priorities to express their views and to develop a shared understanding of a given issue.

Conditions enabling the effective functioning of spaces of negotiation. The creation, and subsequent use, of spaces of negotiation was associated with higher economic productivity at BETA but did not make the tensions between production supervisors and social counselors disappear; rather, the inter- actions between production supervisors and social counselors came to be seen as “positive confronta- tions.” The executive director explained:

It is important that [social counselors and production supervisors] continue to voice their respective con- cerns. All I ask them to do is to listen to each other and understand each other. It means that they have to be ready to defend their position while being able to listen to the other group’s position. . . . This is all about positive confrontation.

Similarly, reflecting on the instances of high ten- sion between social counselors and production su- pervisors, a social counselor reported that: “We never avoid confrontation. When doubts are raised and the tension is high, we just have to remind ourselves why we are here. This helps us put things in perspective.” Thus,theuseof spacesofnegotiation enabled BETA to facilitate coordination between social counselors and technical supervisors, while maintaining a “pro- ductive tension” between them (Murray, 2010; Stark, 2009). By empowering both social workers and production supervisors, this approach enabled BETA to maintain a social focus while better at- tending to commercial imperatives.

TABLE 6 ALPHA and BETA’s Approaches to Managing Commercial and Social Activities after the Crisis

ALPHA BETA Single integrator model Spaces of negotiation model

Spaces of negotiation

Note: Shaded areas highlight differences between ALPHA and BETA’s approaches after the crisis.

2015 1677Battilana, Sengul, Pache, and Model

Our analysis revealed two conditions that enabled the effective functioning of spaces of negotiation at BETA. First, organizational members across the organizational ladder had a common understanding of their superor- dinate goal. This common understanding facilitated negotiation and thereby prevented tensions from esca- lating into intractable conflicts (Fiol, Pratt, & O’Connor, 2009). Second, social counselors were aware that they needed technical supervisors in order to achieve the organization’s goal and vice versa. This mutual aware- ness of interdependence fostered mutual respect, which facilitated the management of tensions and trade-offs. Because each group recognized the important contri- bution of the other group to their common goal, they were more likely to search for, and achieve, a common ground. These two enabling conditions were fostered and sustained in the organization by means of sociali- zation processes (Louis, 1980; Van Maanen & Schein, 1979) that the executive director put in place, including biannual retreats, the annual general assembly, internal communication, training, and job shadowing.

Socialization processes. To ensure that all staff membershadacommonunderstandingofBETA’sgoal (that is, fostering professional integration of the long- term unemployed), the executive director referred to it on a daily basis as “the compass of all organizational decisions.” Furthermore, all permanent staff members got together twice a year for a retreat to nurture con- viviality. These retreats were meant to allow them to find joint solutions to a current organizational chal- lenge (for example, how to improve mentoring). Staff members were also asked to participate in the annual general assembly at which all key stakeholders of BETA met with the board to discuss the objectives of the organization, how well the WISE did in the past year, and what could be done to achieve its objectives in the future. The executive director explained that she used “the General Assembly as a key moment to re- mind staff members of our mission and to celebrate it.”

To make production supervisors and social coun- selors aware of their mutual interdependence, the ex- ecutive director relied on communication, training, and mandatory job shadowing. Highlighting the importance of internal communication, she explained: “I always remind staff members that we all depend on each other to reach our goal. Social counselors and production supervisors are in charge of different aspects of the work that are equally important to our success.” In addition, various training interventions were orga- nized to increase everyone’s awareness of the im- portance of each group’s contribution toward the achievement of organizational objectives. For ex- ample, all production supervisors attended external

training sessions designed to help them to better understand the challenges faced by beneficiaries and to appreciate the importance of social counseling in helping beneficiaries to become ready to reintegrate into the workforce. In turn, social counselors were asked to attend meetings at the local chamber of commerce so that they could become more aware of the demands and norms of the corporate world.

Finally, and most importantly, the executive di- rector made internal “job shadowing” mandatory. Each social counselor was required to spend two hours every week observing the work of production supervisors to better understand its logic and con- straints, and vice versa. As the executive director explained: “Job shadowing made specialized groups aware of their interdependence. It fostered mutual understanding and respect across functional groups, which was a necessary condition to maintain a pro- ductive tension between them.”

Taken together, at BETA structural differentiation, combined with spaces of negotiation, enabled staff members in charge of economic and social activi- ties, respectively, to jointly find solutions to the trade-offs that they faced. Spaces of negotiation did not make tensions disappear; rather, they trans- formed potential internal conflict into a productive tension between production supervisors and social counselors. The intense socialization and ongoing routines that shaped staff members’ understanding of the organization’s superordinate goal, and of their mutual interdependencies, ensured the effec- tive functioning of spaces of negotiation.

DISCUSSION AND CONCLUSION

We hypothesized, and validated through regression analyses, that the relationship between social im- printing and social performance is paradoxical: While social imprinting is positively associated with social performance, social imprinting may also indirectly weaken social performance through the negative as- sociation that social imprinting has with economic productivity. We also uncovered, through our lon- gitudinal comparative case analysis, that socially imprinted WISEs may resolve this paradox when: (a) they face a situation (in our case, a severe financial crisis) that enables organizational members to par- tially unfreeze the organizations’ social imprint; and (b) they introduce new structural and procedural ele- ments that create spaces of negotiation, which facili- tate coordination between structurally differentiated groups of organizational members throughout the or- ganizational ladder.

1678 DecemberAcademy of Management Journal

Contributions

Our study contributes to the growing body of work onhybridorganizing(forareview,seeBattilana&Lee, 2014). In organization theory, multiple streams of re- search have examined the nature and functioning of hybrid organizations that combine dissimilar, and potentially conflicting, aspects of different organiza- tional forms (Haveman & Rao, 2006; Padgett & Powell, 2012). One stream of research has examined how hybrids deal with the internal identity challenges that they face (see, e.g., Albert & Whetten, 1985; Glynn, 2000; Golden-Biddle & Rao, 1997; Pratt & Foreman, 2000). Another stream of research has examined how hybrids deal with the different and potentially conflicting institutional demands that stem from their environment, which prescribe how they should operate (see, e.g., Greenwood, Raynard, Kodeih, Micelotta, & Lounsbury, 2011; Kraatz & Block, 2008; Pache & Santos, 2013; Smets, Jarzabkowski, Burke, & Spee, 2015; Thornton, Ocasio, & Lounsbury, 2012). While these studies have considerably enriched our understanding of the internal and external chal- lenges that hybrids face, they remain silent about the factors that influence hybrids’ ability to achieve their main goals. We address this gap by identifying and providing the first (to our knowledge) quantitative evidence for the factors that are associated with the ability of one type of social enterprise (that is, the WISE) to achieve high levels of social performance.

Second, our study uncovers spaces of negotiation as important mechanisms that hybrids may use to successfully coordinate structurally differentiated staff with potentially competing interests. Spaces ofnegotiationsharesimilaritieswith“relationalspaces” (Kellogg, 2009) in that they both foster coordination among diverse participants. However, they perform different functions. Kellogg’s (2009) conceptualiza- tion of relational spaces—like Mair and Hehenberger’s (2014) extension of the concept to activists in organi- zational fields—emphasizes the role that they play in facilitating the activities of change agents. In the case of our study, rather than facilitating change, spaces of negotiation provide a mechanism to resolve the tension that structurally differentiated staff with potentially different interests will face. The combination of struc- tural differentiation and spaces of negotiation fa- cilitate ongoing coordination.

Third, and relatedly, our study contributes to a long traditionoforganizationalscholarshipthathasstudied the challenge of insuring coordination within organi- zations (for a review, see Okhuysen & Bechky, 2009). Of particular relevance to our study, research on

organizational ambidexterity has emphasized the role of leaders in coordinating potentially contradictory sets of activities in corporations (O’Reilly & Tushman, 2008; Tushman, Smith, Wood, Westerman, & O’Reilly, 2010). Going beyond the existing literature, our study shows that coordination in hybrids may be better achieved by means of formal structures and pro- cesses that sustain spaces of negotiation, in which staff members throughout the organization ladder jointly find solutions to the trade-offs that they face, instead of relying on a subset of managers or emer- gent informal mechanisms to ensure coordination.

Additionally, we uncovered two critical conditions that enable the effective functioning of spaces of nego- tiation: (a) staff members’ common understanding of the organization’s superordinate goal, and (b) aware- ness of their mutual interdependencies. We also found that creating and fostering such conditions requires intense socialization processes, including internal communication, training, and job shadowing. Overall, our study suggests that the creation and maintenance of spaces of negotiation, while consuming organizational resources, may be necessary if hybrids are to sustain a balance of focus while pursuing multiple objectives.

Finally, our study contributes to the imprinting literature (for a review, see Marquis & Tilcsik, 2013)—especially to the study of the lasting influence of founders (see, e.g., Almandoz, 2012). We argue, and empirically show, that the relationship between so- cial imprinting and social performance in hybrids that serve distinct groups of beneficiaries, such as WISEs, is made of contradictory yet interrelated effects: Although social imprinting enhances a hybrid’s social performance, social imprinting also indirectly weakens social performance through the negative relationship that social imprinting has with economic productivity. Furthermore, our qualitative findings suggest that a critical condition to overcome this paradox is the partial unfreezing of the hybrid’s social imprint, which opens the possibility for it to manage its im- printing legacy (Cooper, Hinings, Greenwood, & Brown, 1996; Johnson, 2007). While most of the literature on imprinting views it as “an ecological force devoid of agency” (Marquis & Tilcsik, 2013: 222), we show that, after founding, subsequent generations of organizational members may be able to depart from imprinted practices and introduce new ones during sensitive periods of time, such as crises.

Future Research Directions

Future work should explore whether our results hold in other contexts. Within our institutional

2015 1679Battilana, Sengul, Pache, and Model

context (France), our rich data enabled us to tease out heterogeneity across organizations. Subsequent studies of WISEs will need to explore the role of public authorities across different contexts and, more generally, the heterogeneity across institu- tional settings. In our research setting, the state played a role as a gatekeeper for the field by granting WISEs the authorization to operate. In other contexts with less active external monitoring and/or less sup- port from publicauthorities,WISEsmaybemorelikely to invest in economic productivity at the expense of activities that enhance social performance.

Second, future research will need to explore in more detail how hybrids like social enterprises, which combine aspects of business and charity at their core, can sustain a focus on both the accom- plishment of their social mission and the establish- ment of productive operations. In this paper, we provide foundational work on the understanding of how a subset of these hybrid organizations can suc- cessfully achieve their social mission, even when they serve distinct groups of beneficiaries and cus- tomers. Much remains to be explored about the way in which these organizations function. Social enter- prises demand a deeper exploration and new theo- rizing because they are neither typical businesses nor typical charities; rather, they combine aspects of both (Battilana & Lee, 2014; Besharov & Smith, 2014; Dacin, Dacin, & Tracey, 2011).

Future research will also need to examine the in- fluence of other factors on a hybrid’s social perfor- mance. In particular, certain aspects of organizational design—including allocation of decision rights, pro- vision incentives, and financial structure—may affect how hybrids manage intraorganizational negative spill- overs (Sengul & Gimeno, 2013; Sengul, Gimeno, & Dial, 2012). In addition, aspects of organizational governance (such as board composition) may also play an important role (Ebrahim, Battilana, & Mair, 2014). Furthermore, as governments are implementing new types of incorpo- ration to better fit the needs of social enterprises, future research will need to explore the influence of these new legal statuses—such as the community interest com- pany (CIC) in the United Kingdom, or benefit corpora- tion in the United States—on the ability of hybrids to achieve their objectives.

Finally, while our study focuses on one type of hybrid organization, future research will need to explore the extent to which our findings apply to other kinds of hybrid, such as hospitals or universi- ties. In particular, future research should explore the generalizability of our findings on spaces of nego- tiation and the conditions under which they can

effectively be used to maintain a productive tension in hybrids. Our study shows that creating and maintaining spaces of negotiation is an expensive endeavor; ongoing socialization processes such as job shadowing have significant opportunity costs for organizations, and these costs may exceed their benefits. More work should be done to understand the relationship between the coordination benefits achieved from various implementations of spaces of negotiation and their relative costs to the organiza- tion. For instance, we expect organizational size to be a boundary condition because effectively engaging staff throughout the organizational ladder in spaces of negotiation is likely to become increasingly chal- lenging and costly as the size increases. That being said, larger hybrid organizations may still be able to create, maintain, and effectively use spaces of ne- gotiation in a more bounded way by engaging a sub- set of their employees.

In conclusion, hybrids face the distinct challenge of trying to optimize performance on multiple di- mensions as they pursue multiple objectives at once. Doing so requires distinct organizational arrange- ments. Importantly, practices that may be perceived as inefficient in the pursuit of a single objective may be effective when it comes to the joint pursuit of multiple organizational objectives. For example, al- though spaces of negotiation may be costly to create and maintain, in certain situations they may be a necessary condition if hybrids are to achieve high levels of performance in the objectives that they pur- sue. Future research in this area will have profound implications for the study not only of hybrids but also, more broadly, of contemporary organizations that are increasingly straddling the boundaries of multiple sectors as they pursue multiple objectives.

REFERENCES

Albert, S., & Whetten, D. A. 1985. Organizational identity. Research in Organizational Behavior, 7: 263–295.

Almandoz, J. 2012. Arriving at the starting line: The impact of community and financial logics on new banking ventures. Academy of Management Journal, 55: 1381–1406.

Austin, J., Wei-Skillern, J., & Stevenson, H. 2006. Social and commercial entrepreneurship: Same, different, or both? Entrepreneurship Theory and Practice, 30: 1–22.

Baltagi, B. H. 1998. Econometrics. Berlin: Springer-Verlag.

Baron, J. N., Hannan, M. T., & Burton, M. D. 1999. Building the iron cage: Determinants of managerial intensity in

1680 DecemberAcademy of Management Journal

the early years of organizations. American Sociolog- ical Review, 64: 527–547.

Battilana, J., & Dorado, S. 2010. Building sustainable hy- brid organizations: The case of commercial micro- finance organizations. Academy of Management Journal, 53: 1419–1440.

Battilana, J., & Lee, M. 2014. Advancing research on hybrid organizing: Insights from the study of social enterprises. The Academy of Management Annals, 8: 397–441.

Beckman, C., & Burton, M. D. 2008. Founding the future: The evolution of top management teams from found- ing to IPO. Organization Science, 19: 3–24.

Besharov, M. L., & Smith, W. K. 2014. Multiple in- stitutional logics in organizations: Explaining their varied nature and implications. Academy of Man- agement Review, 39: 364–381.

Bode, I., Evers, A., & Schulz, A. 2006. Work integration social enterprises in Europe: Can hybridization be sustainable? In M. Nyssens (Ed.), Social enterprise: At the crossroads of market, public policies and civil society: 237–258. London: Routledge.

Bourdieu, P. 1977. Outline of a theory of practice. Cam- bridge: Cambridge University Press.

Bourgeois, L. J. 1981. On the measurement of organiza- tionalslack.AcademyofManagementReview,6:29–39.

Burton, M. D., & Beckman, C. M. 2007. Leaving a legacy: Position imprints and successor turnover in young firms. American Sociological Review, 72: 239–266.

Caves, D. W., Christensen, L. R., & Diewert, W. E. 1982. The economic theory of index numbers and the measure- ment of input, output, and productivity. Econo- metrica, 50: 1393–1414.

Cooney, K. 2011. An exploratory study of social purpose business models in the United States. Nonprofit and Voluntary Sector Quarterly, 40: 185–196.

Cooper, D. J., Hinings, B., Greenwood, R., & Brown, J. L. 1996. Sedimentation and transformation in organiza- tional change: The case of Canadian law firms. Orga- nization Studies, 17: 623–647.

D’Aunno, T., Sutton, R. I., & Price, R. H. 1991. Isomorphism and external support in conflicting institutional en- vironments: A study of drug abuse treatment units. Academy of Management Journal, 34: 636–661.

D’Autume, A., Bethbèze, J. P., & Hairault, J. O. 2006. L’emploi des séniors en France. Paris: Conseil d’Analyse Economique.

Dacin, M. T., Dacin, P. A., & Tracey, P. 2011. Social en- trepreneurship: A critique and future directions. Or- ganization Science, 22: 1203–1213.

Defourny, J., & Kim, S. Y. 2011. Emerging models of social enterprise in Eastern Asia: A cross-country analysis. Social Enterprise Journal, 7: 86–111.

Délégation Générale à l’Emploi et à la Formation Pro- fessionnelle (DGEFP). 2003. Circulaire n�2003-24 du 3 octobre 2003 relative à l’aménagementde la procédure d’agrément par l’ANPE et au suivi des personnes embauchées dans une structure d’insertion par l’activité économique. Accessed online at https://www.cnle. gouv.fr/IMG/pdf/Circulaire_dgefp-dgas_n2003-24_ 3octobre2003.pdf.

Ebrahim, A., Battilana, J., & Mair, J. 2014. The governance of social enterprises: Mission drift and accountability challenges in hybrid organizations. Research in Or- ganizational Behavior, 34: 81–100.

Edmondson, A. C., & McManus, S. E. 2007. Methodological fit in management field research. Academy of Man- agement Review, 32: 1155–1179.

Eisenhardt, K. M. 1989. Building theories from case study research. Academy of Management Review, 14: 532– 550.

Eisenhardt, K. M., & Graebner, M. E. 2007. Theory building from cases: Opportunities and challenges. Academy of Management Journal, 50: 25–32.

Eisenhardt, K. M., & Schoonhoven, C. B. 1990. Organiza- tional growth: Linking founding team, strategy, envi- ronment, and growth among U.S. semiconductor ventures, 1978–1988. Administrative Science Quar- terly, 35: 504–529.

Feldman, M. S. 2003. A performative perspective on sta- bility and change in organizational routines. In- dustrial and Corporate Change, 12: 727–752.

Fiol, C. M., Pratt, M. G., & O’Connor, E. J. 2009. Managing intractable identity conflicts. Academy of Manage- ment Review, 34: 32–55.

Galaskiewicz, J., & Barringer, S. N. 2012. Social enterprises and social categories. In B. Gidron & Y. Hasenfeld (Eds.), Social enterprises: An organizational perspec- tive: 47–70. New York: Palgrave Macmillan.

Galbraith, J. R. 1973. Designing complex organizations. Reading, MA: Addison-Wesley.

Glynn, M. A. 2000. When cymbals become symbols: Con- flict over organizational identity within a symphony orchestra. Organization Science, 11: 285–298.

Golden-Biddle, K., & Rao, H. 1997. Breaches in the board- room: Organizational identity and conflicts of com- mitment in a nonprofit organization. Organization Science, 8: 593–611.

Greenwood, R., Raynard, M., Kodeih, F., Micelotta, E., & Lounsbury, M. 2011. Institutional complexity & orga- nizational responses. The Academy of Management Annals, 5: 317–371.

Haveman, H., & Rao, H. 2006. Hybrid forms and the evo- lution of thrifts. The American Behavioral Scientist, 49: 974–986.

2015 1681Battilana, Sengul, Pache, and Model

Hoffman, A. J., Gullo, K., & Haigh, N. 2012. Hybrid orga- nizations and positive social change: Bridging the for- profit and non-profit domains. In K. Golden-Biddle & J. E. Dutton (Eds.), Using a positive lens to explore social change and organizations: Building a theoreti- cal and research foundation: 131–150. New York: Routledge.

Huber, P., Sutcliffe, K., Miller, C. C., & Glick, W. H. 1993. Understanding and predicting organizational change. In P.Huber&W.H.Glick(Eds.),Organizationalchangeand redesign: 215–265. Oxford: Oxford University Press.

Hugues, J. M. 2007. Observatoire 2006 des entreprises d’insertion. Paris: CNEI.

Huselid, M. A. 1995. The impact of human resource man- agement practices on turnover, productivity, and corporate financial performance. Academy of Man- agement Journal, 38: 635–672.

Iacobucci, D. 2008. Mediation analysis. Los Angeles, CA: Sage.

Jay, J. 2013. Navigating paradox as a mechanism of change and innovation in hybrid organizations. Academy of Management Journal, 56: 137–159.

Johnson, V. 2007. What is organizational imprinting? Cultural entrepreneurship in the founding of the Paris Opera. American Journal of Sociology, 113: 97–127.

Kellogg, K. C. 2009. Operating room: Relational spaces and microinstitutional change in surgery. American Journal of Sociology, 115: 657–711.

Kraatz, M. S., & Block, E. S. 2008. Organizational impli- cations of institutional pluralism. In R. Greenwood, C. Oliver, R. Suddaby & K. Sahlin-Andresson (Eds.), The Sage handbook of organizational institutionalism: 243–275. London: Sage.

Lawrence, P. R., & Lorsch, J. W. 1967. Organization and environment: Managing differentiation and inte- gration. Boston, MA: Division of Research, Harvard Business School.

Louis, M. R. 1980. Surprise and sense making: What newcomers experience in entering unfamiliar orga- nizational settings. Administrative Science Quar- terly, 25: 226–251.

Mair, J., & Hehenberger, L. 2014. Front-stage and backstage convening: The transition from opposition to mutu- alistic coexistence in organizational philanthropy. Academy of Management Journal, 57: 1174–1200.

Mair, J., & Marti, I. 2006. Social entrepreneurship research: A source of explanation, prediction, and delight. Journal of World Business, 41: 36–44.

Margaret, M. 2006. Travail et emploi des femmes. Paris: La Découverte.

Marquis, C., & Tilcsik, A. 2013. Imprinting: Toward a multilevel theory. The Academy of Management Annals, 7: 193–243.

McDonald, J. 2009. Using least squares and Tobit in second stage DEA efficiency analyses. European Journal of Operational Research, 197: 792–798.

Mercadal, B., Janin, P., Charvériat, A., & Couret, A. 2004. Droit des affaires: Sociétés commerciales, 2004. Levallois: Éditions Francis Lefebvre.

Miles, M. B., & Huberman, A. M. 1994. Qualitative data analysis: An expanded sourcebook. Thousand Oaks, CA: Sage.

Miller, D., & Shamsie, J. 2001. Learning across the life cy- cle: Experimentation and performance among the Hollywood studio heads. Strategic Management Journal, 22: 725–745.

Morrison, A. D., & Wilhelm, W. J., Jr. 2004. Partnership firms, reputation, and human capital. The American Economic Review, 94: 1682–1692.

Murray, F. 2010. The Oncomouse that roared: Hybrid exchange strategies as a source of distinction at the boundary of overlapping institutions. American Journal of Sociology, 116: 341–388.

O’Reilly, C. A., & Tushman, M. L. 2008. Ambidexterity as a dynamic capability: Resolving the innovator’s di- lemma. Research in Organizational Behavior, 28: 185–206.

Okhuysen, G. A., & Bechky, B. A. 2009. Coordination in organizations: An integrative perspective. The Acad- emy of Management Annals, 3: 463–502.

Oliver, C. 1991. Strategic responses to institutional processes. Academy of Management Review, 16: 145–179.

Pache, A. C., & Santos, F. 2013. Inside the hybrid organi- zation: Selective coupling as a response to competing institutional logics. Academy of Management Jour- nal, 56: 972–1001.

Padgett, J. F., & Powell, W. W. 2012. The emergence of organizations and markets. Princeton, NJ: Princeton University Press.

Papke, L. E., & Wooldridge, J. M. 1996. Econometric methods for fractional response variables with an ap- plication to 401(k) plan participation rates. Journal of Applied Econometrics, 11: 619–632.

Pfeffer, J., & Salancik, G. R. 1978. The external control of organizations: A resource dependence perspective. New York: Harper & Row.

Pratt, M. G., & Foreman, P. O. 2000. Classifying managerial responses to multiple organizational identities. Academy of Management Review, 25: 18–42.

Ragin, C. C. 1994. Introduction to qualitative comparative analysis. In T. Janoski & A. Hicks (Eds.), The compar- ative political economy of the welfare state: 299–319. Cambridge: Cambridge University Press.

Rangan, S., & Sengul, M. 2009. The influence of macro structure on the foreign market performance of

1682 DecemberAcademy of Management Journal

transnational firms: The value of IGO connections, export dependence, and immigration links. Ad- ministrative Science Quarterly, 54: 229–267.

Ruef, M., Aldrich, H. E., & Carter, N. M. 2003. The structure of founding teams: Homophily, strong ties and iso- lation among U.S. entrepreneurs. American Socio- logical Review, 68: 195–222.

Scott, W. R. 1977. Effectiveness of organizational effec- tiveness studies. In S. P. Goodman & J. Pennings (Eds.), New perspectives on organizational effectiveness: 63–95. San Francisco, CA: Jossey-Bass.

Sengul, M., & Gimeno, J. 2013. Constrained delegation: Limiting subsidiaries’ decision rights and resources in firms that compete across multiple industries. Ad- ministrative Science Quarterly, 58: 420–471.

Sengul, M., Gimeno, J., & Dial, J. 2012. Strategic delegation: A review, theoretical integration, and research agenda. Journal of Management, 38: 375–414.

Smets, M., Jarzabkowski, P., Burke, G., & Spee, P. 2015. Reinsurance Trading in Lloyd’s of London: Balancing conflicting-yet-complementary logics in practice. Academy of Management Journal, 58: 932–970.

Smith, W. K., & Lewis, M. W. 2011. Toward a theory of paradox: A dynamic equilibrium model of organizing. Academy of Management Review, 36: 381–403.

Spear, R., & Bidet, E. 2005. Social enterprise for work integration in 12 European countries: A descriptive analysis. Annals of Public and Cooperative Eco- nomics, 76: 195–231.

Stark, D. 2009. The sense of dissonance: Accounts of worth in economic life. Princeton, NJ: Princeton University Press.

Stinchcombe, A. L. 1965. Social structure and organiza- tions. In J. G. March (Ed.), Handbook of organizations: 142–193. Chicago, IL: Rand–McNally.

Strauss, A., & Corbin, J. 1998. Basics of qualitative re- search: Techniques and procedures for developing grounded theory (2nd ed.). Thousand Oaks, CA: Sage.

Suchman, M. C. 1995. Managing legitimacy: Strategic and institutional approaches. Academy of Management Review, 20: 571–610.

Thornton, P. H., Ocasio, W., & Lounsbury, M. 2012. The institutional logics perspective: A new approach to culture, structure, and process. Oxford: Oxford Uni- versity Press.

Tilcsik, A. 2014. Imprint–environment fit and perfor- mance: How organizational munificence at the time of hire affects subsequent job performance. Adminis- trative Science Quarterly, 59: 639–668.

Tushman, M., Smith, W. K., Wood, R. C., Westerman, G., & O’Reilly,C.2010.Organizational designs andinnovation

streams. Industrial and Corporate Change, 19: 1331–1366.

Van Maanen, J., & Schein, E. H. 1979. Toward a theory of organizational socialization. Research in Organiza- tional Behavior, 1: 209–264.

Whetten, D., Mackey, A., & Poly, C. 2011. Applying the concept of organizational identity to the study of per- sistently distinctive organizational practices. Working Paper, Brigham Young University.

World Bank. 2013. World development indicators 2013. Washington, DC: World Bank Publications.

Wry, T., Cobb, J. A., & Aldrich, H. E. 2013. More than a metaphor: Assessing the historical legacy of resource dependence and its contemporary promise as a theory of environmental complexity. The Academy of Man- agement Annals, 7: 441–488.

Yellen, J. L. 1984. Efficiency wage models of unemployment. The American Economic Review, 74: 200–205.

Yin, R. K. 2003. Case study research: Design and methods. Thousand Oaks, CA: Sage.

Julie Battilana ([email protected]) is an associate pro- fessor in the organizational behavior unit at Harvard Business School. Her research examines the process by which organizations or individuals initiate and imple- ment changes that diverge from the taken-for-granted practices in a field of activity. She is particularly in- terested in hybrid organizations that diverge from the archetypes of both typical corporations and typical charities by combining aspects of both at their core. She holds a joint Ph.D. in organizational behavior from INSEAD and in management and economics from École Normale Supérieure de Cachan.

Metin Sengul ([email protected]) is an associate pro- fessor at the Carroll School of Management at Boston Col- lege. His research examines those internal organizational choices that structure the organization, define the nature of power and influence, and impact the firm’s competitive stance. He received his Ph.D. in strategy from INSEAD.

Anne-Claire Pache ([email protected]) is a professor of social entrepreneurship at ESSEC Business School and holder of the ESSEC Chair in Philanthropy. Her research lies at the intersection of organizational theory and social entrepreneurship, with a particular emphasis on pluralis- tic environments, hybrid organizations, and scaling-up processes in organizations. She received her doctorate in organizational behavior from INSEAD.

Jacob Model ([email protected]) is a doctoral candi- date in Organizational Behavior at the Stanford Gradu- ate School of Business. His research includes work in

2015 1683Battilana, Sengul, Pache, and Model

organizational theory, social movements and economic sociology.

APPENDIX A

Phases of Development of the WISE Field in France

The experimentation phase (before 1986). Dur- ing this phase, the first WISEs emerged, scattered across France, invented by social workers who found typical interventions inefficient in addressing structural un- employment. They created WISEs to provide the un- employed with work experience as a way in which they could acquire the skills necessary for them to reintegrate within traditional companies. Some government repre- sentatives identified and recognized the contribution of these initiatives to the fight against unemployment. Two ministerial circulars, issued in 1979 and 1985, not only recognized the value of these experiments but also pro- vided them with some ad hoc financial support. Over 87% of the WISEs created during this period were incorporated as not-for-profits because it was the legal status most fa- miliar to their founders, given their social work back- grounds. The first regional union of WISEs was created in Lyon in 1983 and soon replicated in other regions.

The contestation phase (1986–1990). During this phase, the WISE model was challenged. A change in the political balance at the National Assembly forced socialist President Mitterrand to appoint a right-wing prime min- ister, who formed a right-wing government. Its new Min- ister of Employment interrupted all state support to WISEs. The withdrawal of state financial support forced WISE entrepreneurs to become financially self-sustainable. The rate at which WISEs were being founded dropped during this period, and the proportion of WISEs incorporated as not-for-profits declined to 58%. The existing regional unions federated into a National Union of WISEs—the Comité National des Entreprises d’Insertion (CNEI) in 1988—to represent their vision more forcefully to politi- cians and public officials. Over time, the CNEI also en- gaged in building partnerships with other professional organizations to enhance the visibility and performance of WISEs and provided training and networking opportuni- ties to its members.

The institutionalization phase (1991–1997). Dur- ing this phase, political changes and the growing visibility of the WISE movement led the new socialist government to institutionalize the WISE model. In 1991, a law was passed instituting WISEs as an important building block of em- ployment policies; once accredited, WISEs were entitled to systematicfinancialsupportfromthestatethatcompensated for the “social service” that they delivered. They were, however, required to comply fully with business and em- ployment regulations, regardless of their legal status. This

official recognition created momentum for WISEs: The number of incorporations increased. There was also a slight increase in the proportion incorporated as not-for-profit (to 61%), as WISEs became more visible tools for social workers in search of models with which to promote workforce development.

The professionalization phase (after 1997). This phase was characterized by the combined influence of the state (which wanted to make WISEs more accountable) and normative pressures from the CNEI (which encouraged WISEs to operate as “real businesses” and to become more “professional”). In 1998, the socialist government passed a new law, which strongly encouraged the development of WISEs and required beneficiaries first to be filtered by Pole Emploi (the national agency for employment). It further required WISEs to report to the state their positive gradu- ation rate (that is, the proportion of beneficiaries able to find regular jobs at the end of their employment at the WISE) and also changed the administrative attachment of WISEs from the Ministry of Social Affairs to the Ministry of Labor, thus reinforcing their economic character. Also in 1998, following a resolution passed during its 1997 as- sembly, the CNEI published a manual entitled Promoting WISEs’ Social Mission with a Commercial Legal Status, which was distributed through regional unions to potential WISE founders. During that period, the number of WISEs created rose. Yet, as a result of the strong normative pres- sure imposed by the CNEI, the rate of incorporation as not- for-profit dropped drastically to 30%.

APPENDIX B

Qualitative Data Analysis

Within-case analysis. Relying on interview and ar- chival data on ALPHA and BETA, we first conducted a within-case analysis, with the goal of allowing the unique patterns of each case to emerge (Strauss & Corbin, 1998). To this end, we developed a longitudinal case report of about 40 pages for each case, which synthesized and organized the material collected. The case reports covered the evo- lution of both organizations from their founding until 2008 and were organized around nine topics that emerged from our reading of the interviews: founding and growth, goals and values, people and human resources, organizational structure and governance, business model and commercial strategy, social integration strategy, performance, stake- holders, and miscellaneous. For each topic, we triangulated interview and archival data to ensure the reliability of our accounts (Yin, 2003). We paid specific attention to tracking the evolution of practices over time. These case reports provided us with a comprehensive longitudinal under- standing of the functioning of each WISE.

In the second step, we explored how each WISE had or- ganized its operations over time to deal with the various demands that it faced. To do so, we analyzed, for each case, how responsibilities were assigned within the WISE and

1684 DecemberAcademy of Management Journal

how work was organized among organizational mem- bers. We also analyzed the rules, processes, and routines implemented. Finally, we analyzed the governance pro- cesses at play in each WISE to understand how decisions were made within the organization. We paid attention to tracking major changes in these processes over time. We developed tables and graphs for each case to describe core features in a systematic fashion (Miles & Huberman, 1994). As we traced these different organizational fea- tures, we were able to identify and describe the approach designed by each WISE to satisfy customers’ and bene- ficiaries’ demands.

Cross-case analysis. In the third step, we turned to cross-case analysis, in which we compared the patterns identified in one case with those from the other case to identify consistent patterns, as well as differences (Eisenhardt & Graebner, 2007). We systematically com- pared the composition of workforce, human resources policy, task design, and organizational design over time.

We paid particular attention to how social activities and commercial activities were organized and coordinated.

A fourth step in our analysis involved follow-up in- terviews with ALPHA and BETA’s executive directors to confirm our preliminary findings. We described to them the approach that we had identified and askedthem tocomment on our description. Both confirmed our analysis, and com- plemented it with details and anecdotes, which we used to refine our final descriptions, analyses, and findings.

Finally, we compared the emergent findings of our data analysis with the extant literature, thereby entering into a dialogue between theory and data (Ragin, 1994). In par- ticular, since our analysis revealed that coordination pro- cesses played an important role in allowing BETA to mitigate the negative relationship between social im- printing and economic productivity, we leveraged the lit- erature on coordination in organizations (for a review, see Okhuysen & Bechky, 2009) to explore how it may illumi- nate our understanding of the patterns uncovered.

2015 1685Battilana, Sengul, Pache, and Model

Copyright of Academy of Management Journal is the property of Academy of Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.