Making the most of group relationships: The role of gender and boundary effects in microcredit groups

Journal of Business Venturing 30 (2015) 822–838

Contents lists available at ScienceDirect

Journal of Business Venturing

Making the most of group relationships: The role of gender and boundary effects in microcredit groups☆ Hana Milanov a,1, Rachida Justo b,⁎, Steven W. Bradley c,2 a b c

Technische Universität München, TUM School of Management, Arcisstr. 21, 80333 Munich, Germany IE Business School, Maria de Molina 11, 13, 15, 28006 Madrid, Spain Hankamer Business School, Baylor University, One Bear Place #98006, Waco, TX 76798-8006, United States

a r t i c l e

i n f o

Article history: Received 6 May 2014 Received in revised form 30 March 2015 Accepted 3 April 2015 Available online 23 April 2015 Field Editor: J. Jennings Keywords: Gender Status Microfinance Networks Performance

a b s t r a c t Relationships and networks are important for a range of entrepreneurial outcomes. However, gender scholars' efforts to compare networks across genders rarely extend to provide empirical evidence for the link between networks and performance. Building on expectation states theory and network perspectives, we examine between- and within-gender differences in the network size–performance relationship, highlighting the conditions under which some females leverage their relationships for firm performance better than others. Using data collected from microcredit entrepreneurs in Kenya, we find that the number of within-group ties positively influences firm performance but more positively for male entrepreneurs. For female entrepreneurs, this relationship is contingent on both their individual and their group's characteristics. We discuss implications and future research directions for the gender, networks, and microcredit literatures. © 2015 Elsevier Inc. All rights reserved.

1. Executive summary Personal networks are important for a range of entrepreneurial outcomes. To understand the role of gender in networks, scholars' efforts have largely focused on examining between-gender differences in network activities and characteristics while only implicitly assuming that such differences ultimately translate into differences in performance. Rather than furthering our examination of differences in network structures, the noted absence of evidence motivates us to first understand why and when the same network characteristics (e.g., network size) may differ in their effects on the performance of male as compared to female entrepreneurs. Further, we deepen our understanding of the gendered nature of networks and examine within-gender differences by looking at the conditions under which some females better leverage their network ties for performance compared to others. We examine network ties in the context of microcredit groups in Nairobi, Kenya. While these groups' main purpose is to facilitate loan repayment mechanisms for banks, for microcredit entrepreneurs, network ties within such groups can also be important sources of support, information, and advice and, more broadly, serve as drivers of stronger sales performance. However, the value of a network tie for an entrepreneur is under other group members' discretion: the extent to which a group member is willing to engage in or commit to a relationship is likely dependent on the focal entrepreneur's status. In considering entrepreneurship as a typically male-typed ☆ Dr. Rachida Justo gratefully acknowledges financial support from the Spanish Ministry of Economy and Competitiveness, Grant No. ECO2012-33099. ⁎ Corresponding author. Tel.: +34 91 5689728. E-mail addresses: [email protected] (H. Milanov), [email protected] (R. Justo), [email protected] (S.W. Bradley). 1 Tel: +49 89 289 26748. 2 Tel.: +1 254 710 3921.

http://dx.doi.org/10.1016/j.jbusvent.2015.04.001 0883-9026/© 2015 Elsevier Inc. All rights reserved.

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

823

occupation as well as the role of females in Kenya, we build on expectations states theory (EST) and posit that females have lower status than males in this context. This in turn implies that the value of network ties for entrepreneurs' performance may be delimited by their status in the group such that even with the same number of ties, male entrepreneurs will have higher performance compared to their female counterparts. Building on this logic, we further theorize that under certain conditions, the status deficit of female entrepreneurs may be partially offset or compensated for by entrepreneurs' individual experience and group characteristics, such as group gender composition and loan officer gender. To test our hypotheses, we collected data in Nairobi, Kenya, relying on help from local surveyors and collected questionnaires from 237 entrepreneurs across 25 microcredit groups (153 females). Consistent with our theoretical expectations, we found that the number of within-group ties positively influenced microcredit entrepreneurs' performance, but the relationship was more positive for male entrepreneurs. In addition, we found that among females, both individual and group characteristics created nuances in this relationship. On the individual level, prior industry experience reinforced the relationship and at least partially compensated for the status deficit of experienced females compared to their non-experienced counterparts. On the group level, the presence of a female loan officer positively reinforced the effect of the number of within-group ties on performance, lending credence to the argument that the presence of highly ranked female staff may offset the importance of gender status in network dynamics. We observed a similar effect in groups in which females constituted more than 80% (i.e., represented the dominant majority in groups) though the effect lacked significance when entered alone in the model. Our study suggests that status lens is a powerful tool to help us understand both between- and within-gender differences in entrepreneurial networks' effects on performance. We provide an enriched understanding of the status mechanisms in microcredit groups by demonstrating their effect on performance differences across genders when it comes to the entrepreneur's number of within-group ties as well as by delineating the conditions under which status may become less relevant or even compensated for by individual characteristics. By their nature, entrepreneurial ties are inseparable from gender dynamics. However, our study points to the importance of considering within-gender variance, or else, we may miss opportunities to uncover important factors that lurk behind the network size–performance relationship among female entrepreneurs. In that sense, we hope that this study opens doors for a broader inquiry into a range of factors that may allow entrepreneurs—male and female alike—to better leverage their networks for performance. Related to practice, our work suggests that microcredit agencies could consider both individual entrepreneurs' and group characteristics when thinking about the composition mechanisms of groups as well as about loan officer assignment. This, of course, invites considerations of the extent to which microcredit agencies should actively shape and manage this process. More broadly, our results suggest to entrepreneurs that not all ties are equally valuable depending on one's own status and the characteristics of the microcontext within which they are embedded. To the extent that individual status shapes the potential that comes from relationship cultivation, female entrepreneurs are particularly advised to think about actively communicating their experiences, which may at least partially compensate for their status deficit. They should also carefully consider the settings in which they choose to invest in, selecting only those that will yield the highest returns. 2. Introduction A well-established literature supports the importance of networks for entrepreneurship (e.g., Hoang and Antoncic, 2003; Semrau and Werner, 2013; Stam et al., 2013). Personal networks are particularly important for the development of small firms (Maurer and Ebers, 2006). In the early stages after firm formation, personal networks build legitimacy and can facilitate opportunity recognition (Bhagavatula et al., 2010). In later stages, networks support entrepreneurs' efforts to mobilize resources (Batjargal, 2003). While the body of work on entrepreneurial networks cumulatively suggests that personal relationships positively influence entrepreneurs' venture performance, the contingencies of this relationship are still an important area of inquiry (Stam et al., 2013). Scholarly understanding of the link between personal networks and venture performance is especially in need of further empirical investigation in women's entrepreneurship research (Hanson and Blake, 2009; Watson, 2012). Here, efforts have largely focused on assessing (a lack of) differences in males' and females' networking activities and network characteristics while only implicitly assuming that any differences indeed translate to differences in venture performance (e.g., Aldrich et al., 1989). However, before continuing our efforts to study gender differences in network structure and use, it would be helpful to understand and explicitly test why and when the same network characteristics (e.g., network size) produce similar or different effects on the performance of male and female entrepreneurs. Indeed, early empirical findings from the organizational network literature lead us to expect the impact of ties on performance to differ by gender (e.g., Brass, 1985; Burt, 1992; Ibarra, 1992). Moreover, given the growing recognition in the entrepreneurship literature that networks have contingent value for different entrepreneurs (Maurer and Ebers, 2006; Stam et al., 2013), it is important to understand why and when some females are better able to leverage the same network resources for performance than others. To pursue the above research goals, we draw on EST (Berger et al., 1972; Ridgeway and Berger, 1986) and empirically examine our hypotheses in the context of microcredit groups. In essence, EST explains how status beliefs in task groups impact the nature of interactions among group members. The theory conceptualizes gender as an important status characteristic that has the potential to influence perceptions of an individual's value in a task group, which subsequently affects the extent of the benefits stemming from group interactions. The theory is especially potent for studying gender in entrepreneurial networks given that entrepreneurship is frequently conceptualized as a male-typed occupation (Baughn et al., 2006; Bruni et al., 2004; Gupta et al., 2009). Moreover, EST enables the exploration of the circumstances under which female entrepreneurs are able to reduce barriers posed by status issues in their ties. Specifically, EST emphasizes two types of boundary conditions: (1) the group context—as determining the relevance of

824

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

gender for assessing the individual's value in a group (Ridgeway, 2011)—and (2) the availability of additional status cues—as these could also be relevant for the task (Webster and Hysom, 1998). Importantly, EST is interesting as it provides a theoretical basis that explains potential differences in entrepreneurial networks for performance both between and within gender groups. If female entrepreneurs have lower status than males, this can help us explain between-gender differences and their lower returns on relationships given the same number of contacts. At the same time, if status issues are not relevant in a male comparison group—as is likely the case with males in male-typed occupations, such as entrepreneurship (Ridgeway, 2011)—EST points us to studying within-gender differences and conditions that can partially reduce or compensate for the relevance of gender as a status marker for some females more than others. Our theoretical model is empirically tested in a Kenyan microcredit setting. Microcredit groups emerged as a means for banks to manage risk when clients have little or no physical assets to secure loans. Beyond the benefit of “social collateral” for banks, the formation of social ties in groups is viewed as an important driver of microcredit clients' success (Armendariz de Aghion and Morduch, 2005; Bruton et al., 2011). Microcredit groups are also a natural setting for examining females' status due to the tension between intended female empowerment and concurrent challenges to its achievement (Strier, 2010). Our consideration of microcredit is also timely and relevant. Indeed, recent work has called for more in-depth examination of lending group dynamics (Bruton et al., 2011). Theoretically, given EST's focus on group status dynamics (Ridgeway and Berger, 1986) and microcredit institution's structuration of clients into groups, there is an intimate connection between theory and the empirical setting. This study offers three primary contributions. First, in addressing the connection between networks, entrepreneurial outcomes, and gender, we contribute to an important area of research on women entrepreneurship (Hanson and Blake, 2009; Loscocco et al., 2009). Acknowledging that network characteristics may be a necessary but insufficient condition for economic success (Loscocco et al., 2009), we provide empirical evidence on contingencies in the network–performance relationship for female entrepreneurs (Hanson and Blake, 2009; Watson, 2012). Second, we address a gap in the literature by exploring the heterogeneity evident among female entrepreneurs using an EST framework (de Bruin et al., 2006; Hughes and Jennings, 2012; Hughes et al., 2012). Specifically, we examine how female entrepreneurs' status is shaped by their individual characteristics as well as characteristics of the context in which entrepreneurs and their network ties are embedded (Hughes et al., 2012). In this regard, our sample has the important added benefit of extending our understanding of entrepreneurial networks in a developing African economy (Khayesi et al., 2014). Third, our work offers insights to the microfinance literature, where consideration of group characteristics is limited to date. While this research area recognizes the potential benefits of group ties, scholars also warn that support between group members is not a given (Bruton et al., 2011). Our study highlights that leveraging the network potential of microcredit groups for females' business performance is a function of both group and individual characteristics. In doing so, this study also answers the call to understand when microcredit group relationships benefit individual members (Anthony, 2005; Bruton et al., 2011).

3. Background and theory development 3.1. Expectation states theory, gender, and entrepreneurship In order to develop our understanding of why, how, and when within-group ties influence female entrepreneurs' performance, we rely on EST (Berger et al., 1972). The origins of EST can be traced to the inquiry of why some members in group interactions “have more opportunities to speak, their ideas are taken more seriously, and they have more influence over other group members” (Correll and Ridgeway, 2006, p. 29). The central thesis of EST is that status characteristics (i.e., in principle, any attribute on which people differ) are relevant to the extent that they shape individuals' access to action opportunities and also affect group status hierarchies (Berger et al., 1972). Status dynamics emerge from societal beliefs and norms that associate status characteristics to inferences about an individual's capacity or worthiness (Berger et al., 1972). Societal beliefs materialize as either diffuse or specific status characteristics. As their name suggests, specific status characteristics carry performance expectations for a well-defined, specific range of tasks (e.g., accounting expertise), whereas diffuse status characteristics (e.g., gender, age, or race) carry broad and generalized expectations of competency. In terms of gender, in many countries, there are common beliefs about males' stronger diffuse competencies across a range of tasks, with some variance for particular tasks (e.g., females are perceived to be better at nurturing tasks) (Ridgeway, 2011). Although EST was previously not extensively used in the entrepreneurship field, it is becoming increasingly appealing to entrepreneurship scholars (Saparito et al., 2012; Thébaud, 2010; Yang and Aldrich, 2014) as it provides a rich theory about gender as a relevant factor in the organization of social life. Indeed, gender scholars have often used EST as a surrogate for a “network theory of gender” (Chafetz, 1997) because status explanations are a powerful means to uncover the gendered nature of interactions as well as the differences that result from them (Ridgeway and Diekema, 1992; Ruef et al., 2003). In the entrepreneurship literature, gender is commonly a relevant status characteristic (Baker et al., 1997; Baughn et al., 2006; Yang and Aldrich, 2014) mainly because entrepreneurship is predominantly considered a masculine concept (Ahl, 2006; Bruni et al., 2004; Gupta et al., 2009). As a result, female entrepreneurs often face status-related hurdles when interacting with different stakeholders ranging from financial providers (Amine and Staub, 2009; Iakovleva and Kickul, 2011; Murphy et al., 2007) to customers, suppliers, and employees (Fagenson and Marcus, 1991) to co-founders (Yang and Aldrich, 2014). For example, financial providers are known to make the project/entrepreneur-assessment process more stringent for female applicants (Buttner and Rosen, 1988; Carter et al., 2007; Saparito et al., 2012) and offer females less favorable terms for accessing capital (Coleman, 2000). Accordingly, EST is an appropriate paradigm to understand the value female entrepreneurs can obtain from their within-group ties. Additionally, in claiming that “no status characteristic advantages or disadvantages an actor in all settings” (Correll and Ridgeway, 2006, p. 33), EST is especially

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

825

appropriate in this study as it encourages the examination of settings and factors that allow some female entrepreneurs to better utilize their ties for firm performance than others. 3.2. Microcredit in Kenya and lending groups In order to assess the role of gender-status differences in the relationship between network size and entrepreneurial firm performance, we focus on the microcredit context in Kenya. Microcredit institutions vary in their form across (and within) countries, but their defining feature is issuing microloans to support business generation or development among the poor, many of whom do not have access to traditional banking services. While, historically, various institutions have tried to tackle problems of providing credit access to the poor, it was the novel group-lending microcredit model championed by Mohammed Yunus that provided a reliable way to assist entrepreneurs in poverty (Khavul, 2010). Group structures in microcredit emerged from banks' needs to tackle adverse selection and moral-hazard problems in securing loan repayments when physical assets are unavailable (Bruton et al., 2011). Accordingly, microcredit agencies rely on groups as a form of “social collateral” that provides discipline in the loan-repayment process and minimizes defaults. While groups typically form through member self-selection, it is also common that a microcredit organization participates in suggesting members to groups (Feigenberg et al., 2014; Sharma and Zeller, 1997). This is the case in our study, which implies that while some members have prior acquaintances in groups, lending groups are also places where new ties can be formed among members (Feigenberg et al., 2014).3 In that sense, this context is especially well suited to our study given the focus of EST on group structures and respective interactions between group members. Members of microcredit groups meet on a schedule prescribed by the microcredit agency. From the agency's perspective, the group meeting has the primary purpose of organizing loan repayments and/or dealing with defaults. However, these meetings are also opportunities for participant exchanges on social and business topics (Feigenberg et al., 2014) as evidenced in our sample (see Methods section for details). Depending on the microcredit organization, a group's frequency of meetings can vary from a weekly (as in our study) to monthly. Meeting frequency is important for social-capital development and for greater collaboration in newly formed microcredit groups (Feigenberg et al., 2014). The relevance of group meetings corresponds to the importance of relationships in the African context as much of the entrepreneurial activity in Kenya operates within an informal economy (Khavul et al., 2009). When considering gender dynamics among microcredit group members, it is important to note that although some microcredit agencies traditionally focused exclusively on female clients, we observe a steady increase in the proportion of males (Underwood, 2007).4 In that sense, while some microcredit agencies originated with the aim of contributing toward gender equality (Downing, 1990), thus far, available “evidence indicates that all the assumed linkages between microcredit and women's empowerment must be questioned” (Mayoux, 2007, p. 45). Indeed, an exhaustive report on microcredit and gender suggests that microcredit can be a good instrument to initiate changes with respect to female empowerment; however, microcredit agency's efforts to profoundly change deeply embedded social norms (including female's lower status) have yet to be evaluated as successful (Guerin and Palier, 2006). In that sense, even in gender-sensitized contexts like microfinance, gender issues and resulting status dynamics are very much in line with those posited in EST. Gender issues in microcredit are likewise echoed in the Kenyan context, where “compared to males, females in Kenya face more severe legal, regulatory, and administrative barriers” when it comes to running a business (Ellis et al., 2007, p. 1). In Kenya's culture, “the natural order is that a woman comes after the man” (Ellis et al., 2007: 22). Further evidence of this disparity comes from a case study of Kenyan microcredit, which reports that “Due to existing socio-cultural values and practices, women in most parts of Kenya are perceived, and to a large extent, perceive themselves, as being subordinate to men …. Limited access to, and control over resources of production, illiteracy and limited exposure and low participation in decision-making and leadership positions, all contribute to their low image in society” (Hospes et al., 2002, p: 79). Hence, despite important contributions of MFIs toward gender equality (Downing, 1990), gender can still be considered a potent status attribute that acts as a frame of reference in microcredit groups' member interactions. In conclusion, microcredit groups in Kenya enable the study of a meso-environment for which the structure of social arrangements explains differences among females (Ahl, 2006). In what follows, we explore between- and within-gender differences regarding leveraging within-group ties for performance. 4. Hypotheses 4.1. Microcredit group ties The network literature in entrepreneurship has repeatedly found that the number of relationships has an important influence on new firm development (Aldrich et al., 1987; Lerner et al., 1997), which explains scholars' continued interest in including this variable in their investigations of entrepreneurial phenomena (Stam et al., 2013). Personal ties are an important aspect of entrepreneurs' social capital (Greve and Salaff, 2003) that provides access to information and advice on improving business. The microfinance literature also suggests that entrepreneurs join microcredit groups not only for access to financial capital but also for building valuable networks (Velasco and Marconi, 2004). For these entrepreneurs, group membership is important for building network size that in turn supports their ability to generate income (Kabeer, 2001). 3 4

To address this issue empirically, in our model, we control for any prior acquaintances entrepreneurs had in a group prior to its formation. In our sample, the percentage of males (38%) largely surpasses the average of 16% in developing countries (Underwood, 2007).

826

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

Network ties between group members offer value-added relationships (Bruton et al., 2011) that can positively influence entrepreneurs' performance and well-being (Gomez and Santor, 2001). Within-group ties can convey information about local market conditions or trusted suppliers and serve as good referral points for new customers, which can be “worth far more than the loan” (Maclean, 2010). Case studies from Kenya report the importance of group meetings for information sharing (Hospes et al., 2002), preserving collective resources, and boosting productivity (Pretty and Ward, 2001). Additionally, relationships provide social support for microcredit entrepreneurs, as evidenced in Kenyan studies (Akoten et al., 2006; Hospes et al., 2002). Social support through within-group ties helps entrepreneurs deal with handling loan-repayment challenges as well as managing different types of external shocks (Velasco and Marconi, 2004). Building on prior evidence, we propose the following baseline hypothesis: H1a. In microcredit groups, the number of within-group ties an entrepreneur has is positively related to business performance. While network theory and evidence from microcredit indicate the potential benefits of group relationships for access to referrals, advice, and support, EST highlights how perceived status based on characteristics like gender can alter the realization of these benefits (Loscocco et al., 2009; Ridgeway and Diekema, 1992). For example, EST scholars have found that a female's lower status can result in comparatively higher difficulties to translate work networks into professional benefits (Ridgeway and Smith-Lovin, 1999). More generally, we know that “a given contact can vary in their accessibility to others depending on who the seeker is to others”; a “classic example is the person who has time for high-status people, but not for others” (Borgatti and Cross, 2003, p. 435). Beyond the time devoted to information sharing, recent reviews of gender and entrepreneurial networks also suggest that the quality of information is likely to vary because people use status as a “filter for information” (Hanson and Blake, 2009, p:139). Given the qualitative evidence from Kenya (Ellis et al., 2007) and the microcredit case studies presented earlier (Hospes et al., 2002), it is plausible to argue that gender acts as a diffuse status characteristic in microcredit groups. As a result, we expect that group members will consciously or unconsciously use status as a filter in the exchange of quality and timely information, support, or referrals, leading to differences in tie effects for female versus male entrepreneurs' performance. Thus, we propose the following: H1b. In microcredit groups, the number of within-group ties an entrepreneur has is more positively related to business performance for male than for female entrepreneurs.

4.2. Moderating effects of group characteristics In addition to helping us understand how status shapes differences in the realization of the benefits from group relationships for male and female entrepreneurs, EST is also advantageous in offering us mechanisms to understand the factors that may allow some females (more than others) to partially unlock the benefits embedded in their current relationships. While the theory suggests that gender can be a source of social disadvantage in entrepreneurship, it similarly emphasizes the importance of understanding both the local setting as well as individuals' other characteristics for the relevance of gender as a key organizing mechanism in the group (Correll and Ridgeway, 2006). The group-level context can be an important trigger in shaping the relevance of gender (Bunderson, 2003; Tyler, 2006) because dynamics operating at the group level can influence the microstructure of status in groups and corresponding performance expectations (Ridgeway and Berger, 1986). In this way, group-level dynamics may reduce the relevance of gender effects that otherwise operate at a more societal level. In what follows, we use EST and the microcredit literature to examine two group characteristics that are recognized as important in group status dynamics: group leadership (Bruton et al., 2011; Stewart and Stasser, 1995) and group gender composition (Mayoux, 2001). 4.2.1. Group leadership In the context of microcredit groups, a loan officer oversees each group. Loan officers represent the most critical interface of microcredit agencies with their clients (Ahmad, 2002; Goetz, 2001). Their responsibilities often go beyond administrative tasks of screening clients and collecting loan payments. Loan officers may also participate in group formation and training, thus acting as genuinely participative “facilitators” (Siwale and Ritchie, 2012). As such, loan officers can play a “transformational role” in empowering clients (Dixon et al., 2007). While little research exists on how loan officers' characteristics influence microcredit entrepreneurs' performance, literature in the field of traditional banking offers insights for our study. For example, loan officer gender plays an important role in clients' access to credit (Carter et al., 2007), with anecdotal evidence suggesting that female loan officers tend to mitigate existing gender biases in loan assessment (Agier and Szafarz, 2010). In our context, loan officer gender is relevant for shaping the group status dynamics in important ways. External assignment of a leadership role can influence other members' perceptions of that individual's task-relevant expertise (Stewart and Stasser, 1995). In our context, if a microcredit agency appoints a female as a loan officer, this likely serves as an endorsement of her expertise in financial and entrepreneurial tasks. Accordingly, in groups for which the loan officer is female, existing status structures and social norms may be challenged (Tyler, 2006), which may influence group interactions (Major, 1994). Hence, in groups with female loan officers, gender as a status characteristic may be less relevant in shaping group members' interactions due to diminished perceptions of status distance between genders (than for groups for which the loan officer is male). Additionally, a female attaining a responsible position, such as becoming a loan officer in a microcredit agency, raises the perception of females' task abilities (Hogue and Yoder, 2003; Yoder, 2001). This may result in a “transfer effect” such that other female group members may also be perceived as having higher task abilities (even if the loan officer was not selected for a specific attribute, such as

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

827

gender) (Beaman et al., 2009; De Kelaita et al., 2001; Pugh and Wahrman, 1983).5 For example, prior work has shown that exposure to female political leaders transfers to reducing voters' biases against other female candidates running for elections (Beaman et al., 2009). Applying the concept of transfer effect to our context, it follows that the presence of a female loan officer in a group is likely to reduce the relevance of gender as a framing device for assessing members' competence within that group. This carries important implications for interactions among group members and the corresponding value that female entrepreneurs derive from withingroup ties. Indeed, given that the benefits of relationships hinge heavily on the actual value that group members are willing to provide, reducing the relevance of gender in a group as a result of having a female loan officer would diminish members' hesitation in providing such value to the focal female entrepreneur. The same is not expected for females in groups with male loan officers. Accordingly, we propose the following: H2. In microcredit groups, the relationship between the number of within-group ties and business performance is stronger (i.e., more positive) for female entrepreneurs in groups working with female loan officers than in groups working with male loan officers.

4.2.2. Group composition EST suggests that group composition is important because group heterogeneity determines the relevance of status characteristics and hence the extent to which these are likely to be used in competency assessments and member interactions (Smith-Lovin and Brody, 1989). Similarly, the microcredit literature finds group composition to be important to the extent that it influences members' tendency to support each other (Paxton et al., 2000). Extending this to studying females' status characteristics in leveraging withingroup ties, we analyze groups' gender composition as an important contingency factor.6 Since Kanter's (1977) seminal work, much of the organizational research on gender has focused on how females' proportional representation in the workplace affects their professional experience. In the context of work groups, female underrepresentation is associated with their alienation from social and professional networks (Ely, 1995; Sackett et al., 1991; Yoder, 2001) because minority representation of females in groups leads both males and females to adopt stereotyped expectations about females (Ely, 1995; Lockheed, 1985). While organizational research points toward a generally positive association between an increasing proportion of females and their work experience, any such theorizing needs to consider that there may be a threshold at which status perceptions shift. Moreover, it is important to understand that the level of such a threshold is highly context dependent (Tolbert et al., 1999). In contexts in which the presence of females is fairly unusual (e.g., female partners in law firms), already small increases in the proportion of females (e.g., crossing the threshold of 15%)—theoretically considered to constitute a female minority (Kanter, 1977)—have been argued to be sufficient for a change in gender-status dynamics and accordingly have been found to be associated with less typecasting on a gender basis among both males and females (Ely, 1995). While research generally points to the development of a favorable environment for females when females represent a majority in a group (see Tolbert et al., 1999 for a review), in the context of microcredit, where females are relatively well represented compared to other work environments, a significant shift in status beliefs is unlikely to take place in a lending group unless females become a dominant majority (that is, more than 80%).7 Indeed, in female-dominant settings, being female becomes more of the norm and, as such, likely reduces the relevance of gender as a statusorganizing principle in the group. This should lower barriers for females to benefit from group relationships compared to females with the same number of within-group ties that belong to groups where the proportion of females is not dominant (smaller than 80%). Therefore, we expect the following: H3. In microcredit groups, the relationship between the number of within-group ties and business performance is stronger (i.e., more positive) for female entrepreneurs with dominant proportions of females in their group (higher than 80%).

4.3. Moderating effect of individual expertise The EST literature demonstrates that the degree to which females' lower status impacts members' conduct depends not only on the context of the interaction (i.e., group characteristics) but also on other characteristics of the focal agent. An important individual-level factor that has been found to have the potential to compensate for gender status is the focal agent's perceived expertise (Carli, 2001). Observed as a status characteristic, expertise can help us understand social processes, such as people's influence in groups or work-group outcomes (e.g., Bonner et al., 2007; Bottger, 1984). In our study, the significance of expertise as a status characteristic is even more important given that female entrepreneurs' lower status stems from occupation-related beliefs about their lack of competence (Correll and Ridgeway, 2006). 5 Specifically, transfer effect refers to a situation when a status attribute is transferred from one person (i.e., a referent) to another member who was not perceived to possess that attribute before, thus producing a possible status generalization across several members in a group. 6 Here, it is important to highlight that task groups do not have to be invariably heterogeneous for status beliefs to become relevant. For example, Ridgeway explains that “gender may be effectively salient in same-sex settings too, if the context is gender linked to the culture” (2011, p. 71). Indeed, a status characteristic is activated if either group members differ on the characteristic in question or if they consider it relevant for the task at hand (Correll and Ridgeway, 2006). Given that entrepreneurial and financial activities—which make up the daily tasks of microfinance borrowers—are highly gender-typed tasks, it is likely that lending group members will be mindful of gender as a status marker even in the absence of mixed-sex groups. 7 As explained in the methods section, 80% was taken as a cutoff for classifying groups into female-dominant majorities following Kanter's (1977) categorization of female's proportional representation in work groups.

828

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

Because expertise is a latent construct that is often difficult to directly assess, group members generally form their perceptions of expertise based on any potentially relevant cue that is available to them (Bunderson, 2003). These cues often include some aspects of past experience (Wittenbaum, 2000), with indicators of prior task experience being particularly powerful in informing a female's status in a work or entrepreneurial group (Yang and Aldrich, 2014). In entrepreneurship, prior industry experience is a key indicator of expertise that positively influences venture performance (Cooper et al., 1994; Delmar and Shane, 2006) with similar results reported in studies of female entrepreneurs (Lerner et al., 1997). Prior industry experience captures tacit business knowledge about the nature of customer tastes and demand, production processes, as well as prior investments in relationships specific to the industry (Cooper et al., 1994; Delmar and Shane, 2006). To that end, prior industry experience can supply entrepreneurs with credibility that facilitates the development of relationships and sales (Delmar and Shane, 2006). We expect that mechanisms surrounding prior industry experience would also apply in microcredit. Indeed, in the context of microfinance, prior experience was found to be more important than general education as a predictor of business success (Bradley et al., 2012). Following EST, prior industry experience might be a potent specific status marker for expertise in microcredit. A recent study of entrepreneurial teams (Yang and Aldrich, 2014) supports this premise by showing that industry experience is a stronger predictor of merit-based status than other indicators of expertise, such as education level or managerial experience. Accordingly, a female entrepreneur's prior industry experience is likely to be a specific status marker that may compensate for any reduced performance expectations based on a more diffuse status characteristic, such as gender. H4. In microcredit groups, the relationship between the number of within-group ties and business performance will be stronger (i.e., more positive) for female entrepreneurs with previous industry experience.

5. Data and methods 5.1. Population and sample We collected the data for this study in the spring of 2011 as a part of a larger data-collection effort seeking to understand microcredit entrepreneurs. Data was collected with the assistance of a microcredit institution operating in Nairobi, Kenya. The organization began as a coordinated relief effort by the National Council of Churches of Kenya to the slums in Nairobi in 1975. It became a microcredit organization in 1999 with the help of US Agency for International Development. The microcredit bank had 54,000 total active borrowers (2012) across Kenya and approximately 60% of its total loans were to females. The bank reported a number of stated core values, including “justice, fairness and equal opportunity and participation” along with “gender balance for equitable distribution of management responsibilities.” This was reflected in our sample, in which five out of nine loan officers were female. Most of the bank's loan products were in the form of group loans, with members being selected by groups or assigned to groups with the assistance of the bank's loan officer. It is important to note that the described importance of microlending groups for business development was evident in our sample. When respondents were asked about the topics discussed during lending group meetings, the majority listed “business ideas” among the three key topics they discussed (the other two topics were “loan repayments” and “personal issues”). Similarly, a large majority (72%) of the respondents found their group members' behaviors helpful for their business development. Examples of this helpful behavior could be roughly grouped under “sharing of business ideas,” “advice exchange,” and “customer growth” (e.g., sharing customers or referring them to each other's shops). The bank requested we work with their Nairobi Regional Office for easier coordination due to the sizeable number of surveys. Given that most of the group members in the city center of Nairobi are able to speak and write in English, we proceeded to use it as the survey instrument's language. Prior to final data collection, a pilot test was conducted to ensure the questions were well understood and also to assess the time required to complete the questionnaire. To collect the data, four trained surveyors followed the nine loan officers in their weekly group visits. The loan officers and surveyors requested that clients participate in this important and voluntary survey with a reward offered to those who completed the questionnaire. They additionally highlighted the anonymity of all responses and encouraged the entrepreneurs to approach the interviewer in case of any doubts. While members of the visited groups largely agreed to participate in the survey, microcredit entrepreneurs' time is scarce, which sometimes resulted in them rushing to leave the meeting or them being absent from the meeting altogether (with a delegate sent to pay the loan). With some groups, the interviewers were not allowed to perform the interview at the group meeting premises. Instead, they were encouraged to visit individuals at their home or work location. These circumstances often resulted in missing responses for one or more members per group. While within-group non-responses would not be problematic for studies focused on individual entrepreneurs, in a network study such as ours, response rates are important (Wasserman and Faust, 1994) in order to be able to map respondents' network activities reliably. Accordingly, we followed prior network literature on groups by adhering to a criterion to have a minimum of 80% of respondents in a group (Sparrowe et al., 2001). Our final sample consists of 249 respondents (83 males and 153 females) across 25 groups. Within these groups, the response rates averaged 94%. 5.2. Variables 5.2.1. Dependent variable—business performance In order to capture the performance of microcredit entrepreneurs, we asked them to report their current monthly sales. This value was converted from Kenyan shilling to USD using the exchange rate at the time of data collection. Sales are a key and straightforward

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

829

indicator that entrepreneurs use to evaluate their performance (Delmar and Shane, 2006; Jenssen and Greve, 2002; Sine et al., 2006), and this is particularly true in the African context (Khayesi et al., 2014). Moreover, sales are closely related to profits in small businesses (Collins-Dodd et al., 2004). For entrepreneurs in the microcredit context, this measure is commonly collected by Kenyan microcredit agencies (Hospes et al., 2002) and one that entrepreneurs can understand and more accurately report (e.g., Al-hassan et al., 2011; Karlan and Valdivia, 2011). Given the high skewness of the variable, we employed a natural logarithm transformation, which is commonly used in entrepreneurship and network studies (e.g., Delmar and Shane, 2006; Greve and Salaff, 2003).8 5.2.2. Independent variable—number of within-group ties In order to assess entrepreneurs' within-group ties, we asked them to first list all members of their groups. Then, we asked respondents to identify members of their groups who they talk to about information and gain business advice from to improve their business, effectively capturing entrepreneurs' discussion networks—an important aspect of their social capital (Greve and Salaff, 2003). This procedure allowed us to create an N x N matrix of relationships within each group (where N is the number of respondents in the group), which we created by recording a tie if both group members reported a relationship for this question. This procedure is advantageous for two reasons. First, it does not suffer from a limitation of self-report data as every self-reported tie is verified as being reported by both respondents (i.e., the focal respondent and the identified group member). Second, by eliminating non-reciprocated exchanges, this procedure effectively captures more stable and recurring ties (Friedkin, 1990).9 5.2.3. Moderators (our models include four moderators) In order to test the between-gender differences for the effect of within-group ties on entrepreneurial performance, we coded gender as 1 when the entrepreneur was female and interacted this variable with the number of within-group ties in the full sample. In investigating within-gender differences among female entrepreneurs, our first moderator is loan officer gender. The variable takes on a value of 1 if the loan officer working with the group was a woman and 0 otherwise (in our sample, female loan officers worked with 45% of the sample respondents of fairly equal gender distribution). Our second moderator is the proportion of females in entrepreneurs' groups. We counted the number of female respondents in the group and divided it by the number of members (Ruef et al., 2003). To operationalize groups in which the proportion of females is such that it likely reduces the relevance of status, we considered both the context of microcredit groups and our sample. With an above-average proportion of females being the norm in this context (in our sample, the average was .63, and the standard deviation was .29) and following Kanter's (1977) categorization of “skewed groups,” we coded groups for which the female proportion was dominant (more than 80%) as 1. Our last moderator captures entrepreneurs' industry experience. We asked entrepreneurs whether they had prior experience in the same industry of their current venture and entered a dummy variable (1 = yes). 5.2.4. Controls We introduced a series of control variables to improve the robustness of our findings. First, we account for potential sector differences in performance with three dummy variables (i.e., manufacturing, wholesale, and retail) and services as a reference sector. We also included controls for characteristics at the business and individual level. In the early years after founding, business sales are closely related to age (Evans, 1987; Mitchell, 1994) and growing legitimacy (Zimmerman and Zeitz, 2002). While we do not have historical sales, we controlled for these related features to attenuate selection (see further robustness checks in Section 5.2). We controlled for business age as the difference of founding year from year of data collection, recognizing that younger organizations often struggle in their performance (Stinchcombe, 1965). We also accounted for business legitimacy by asking entrepreneurs whether they registered their business with the government. In East Africa, especially for females, registration with the government creates greater legitimacy with stakeholders (Khavul et al., 2009). We also controlled for the group leader's gender with a dummy variable, with 1 representing females. In microcredit, group leaders perform administrative tasks and take care of the overall well-being of the group (Bruton et al., 2011; Paxton et al., 2000). Given that group leaders' identity has been found to influence subtle behaviors in the group (Hermes et al., 2005), leaders' gender could be an important control. At the individual level, we controlled for entrepreneurs' age as it reflects potential credibility when exchanging information with others. Next, we controlled for entrepreneurs' education as it may serve as a status marker (Berger et al., 1972). We also controlled for family business background with a dummy variable because entrepreneurs with prior exposure to business could have acquired business skills to exploit opportunities (Delmar and Davidsson, 2000). We introduced a dummy variable to control for the marital status of entrepreneurs because in the African context, having a spouse influences female entrepreneurs' performance (Khavul et al., 2009). Our last set of individual controls is particularly relevant for studying networks. We controlled for entrepreneurs' perceived relationship-management skills with a five-item scale inspired by prior research (Chandler and Jansen, 1992; Starr and MacMillan, 1990) that captures entrepreneurs' tendency to rely on relationships to support their business and their confidence in doing so (Cronbach alpha = 0.71). Scale items include the following: “I am good at managing and organizing people,” “I often coordinate with 8 Log linear transformation preserves the linearity of the overall model while allowing for specific relationships to be non-linear. In such a model, the interpretation of the estimated beta coefficient is that a one-unit change in x results in an expected change in the natural log of variable y. Hence, in interpreting interaction plots, for every unit of x, the expected value of y is multiplied by eβ. In order to calculate any other change in x on y, it is necessary to include a value of such change c in the exponent. Hence, multiplying the expected value of y by ecβ would reflect the effect for a c-unit increase in X. 9 We also tried operationalizing the independent variable as asymmetric (i.e., not accounting for reciprocal ties). Interestingly, the results were not significant with this measure, which possibly suggests the importance of accounting for more stable two-way exchanges in microfinance groups when it comes to their impact for venture performance.

830

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

others to help my business grow,” “I often know of someone in my business or personal contacts who I can ask when I need outside help,” “I am good at finding money and people to build my business,” and “I often prefer to do a job myself rather than delegate to others (R)”. Next, we controlled for external advice by following a similar approach to recent research on entrepreneurial networks (Semrau and Werner, 2013). We asked respondents whether they had any established relationships with people outside their lending group from whom they can draw information or advice to improve their business. Next, although our arguments do not make any assumptions about the origins of within-group ties, it was important to capture any “inherited” social capital that entrepreneurs may have brought into the group to tease out the effects of the within-group ties above and beyond any past acquaintances in the group. Indeed, because microcredit groups are often largely self-selected, we asked entrepreneurs to check off any group members they had met prior to group formation. We added up these prior acquaintances and divided them by the number of group members to capture how familiar the entrepreneur was with the group. Lastly, we controlled for entrepreneurs' network centrality given that prior research has suggested that positions in networks can have important influences on entrepreneurial performance (Stam et al., 2013). We operationalized centrality using the betweenness centrality measure (Wasserman and Faust, 1994) because it captures the extent to which central entrepreneurs can broaden their perspective (Reagans and McEvily, 2003) and use such diverse perspectives to solve problems efficiently (Cross and Cummings, 2004). Betweenness centrality effectively measures the number of times an actor is positioned between otherwise disconnected actors (Freeman, 1979). We used Ucinet software to calculate it (Borgatti et al., 2002). 6. Results Given the continuous nature of our dependent variable, we used ordinary least squares (OLS) regression to test the hypotheses. In order to account for the fact that more than one entrepreneur belonged to the same microcredit group—and thus for possible non-independence of observations within each group—we used a robust estimation of the standard errors adjusted for clustering on microcredit group (Wooldridge, 2003). This approach uses the Huber–White standard error option, which adjusts for the intraclass correlation and non-independence of observations by effectively inflating standard errors. This approach has also been used previously in studies with nested observations given its methodological rigor yet intuitive appeal compared with more familiar regression models (e.g., Damanpour et al., 2009; Dimov, 2007; Katila and Shane, 2005). In Table 1, we present the descriptive statistics and correlations for all entrepreneurs in our sample with a breakdown by gender for comparison of the key variables. The majority of the sample (70% of entrepreneurs) had monthly sales under USD500 at the time of the study, and we found no significant differences in average sales between males and females. While the extant literature largely suggests that female entrepreneurs underperform compared to their male peers (see Jennings and Brush, 2013 for a literature review), gender differences became more apparent when we looked at within-industry differences. In particular, female entrepreneurs had significantly lower average sales in the retail (USD 594 for males versus 381 for females) and services sectors (USD 808 for males versus USD 570 for males). The results of multivariate regression with robust standard errors clustered by group membership are presented in Table 2 for the complete sample and Table 3 for the female subsample (Models 4–8) with the male subsample for comparison with the full model (in Model 9). We calculated the mean variance inflation factor (VIF), which was 1.57 in the full sample (Table 1, Model 3)—well below the recommended threshold of 10 (Neter et al., 1996). The control Model 1 from Table 2 reveals a strong relationship between legitimacy and firm sales. On the individual level, entrepreneurs' age, education, past acquaintances with group members, and centrality are significantly related to sales, and on the group level, a dominant proportion of females in the group is also positively related to entrepreneurs' sales. We derived our baseline hypotheses (H1a) by building on arguments in the entrepreneurship and microfinance literatures suggesting that the number of within-group ties positively influences entrepreneurs' sales. Table 2 reports these results in Model 2, where the coefficient for the number of within-group ties is positive and significant (0.13, p b 0.01), thus supporting H1a. In order to further inspect differences among males and females (H1b), we introduced an interaction effect of the number of within-group ties and the entrepreneurs' gender. The coefficient for this interaction effect in Model 3 of Table 2 is negative and significant (β = −0.06, p b 0.05), thus confirming H1b. In order to inquire deeper into the nature of the interaction, we plotted the results following established methods (Aiken and West, 1991) in Fig. 1. In this figure, the two lines represent relationships between the number of within-group ties and sales performance for female and male entrepreneurs in the group. To interpret this effect, we converted predicted ln(sales) back to sales and examined the percent change. For example, moving from one within-group tie (mean) to three ties (+1SD) increased predicted sales 19.6% for females and 29.9% for males. In our Hypotheses H2–H4, we proposed a series of moderators that would condition the positive relationship between the number of within-group ties and sales performance for female entrepreneurs. The results are found in Table 3, Models 5–8. In Hypothesis 2 (H2), we argued that the presence of a female loan officer would positively moderate the relationship between the number of within-group ties and performance for female entrepreneurs. The interaction effect in Model 5 is positive and significant (β = 0.16, p b 0.05) and remains so in the full model (Model 8) in the presence of other interactions (β = 0.20, p b 0.05). We illustrate this interaction in Fig. 2. The steeper slope illustrates the effect of the number of within-group ties on sales performance for female entrepreneurs in groups with female loan officers versus female entrepreneurs in groups with male loan officers. In terms of the effect on sales, a change from one to three within-group ties increases sales 50.1% for females in groups with female loan officers versus females in groups with male loan officers, for which change in sales was negligible. Overall, H2 is supported. Next, we hypothesized that the dominance of females in microcredit group composition would positively moderate the direct relationship between the number of within-group ties and sales performance. The interaction effect in Model 6 is positive and but not significant (β = 0.01, p N 0.10). The interaction effect becomes significant in Model 8 when all interactions are entered (β = 0.12,

Table 1 Descriptive statistics and correlations.

Gender (female = 1) Sales (in 100s USD) Industry manufacturing Industry wholesale Industry retail Business age Business registered Group — female leader Entrepreneur age Entrepreneur education Family business background Entrepreneur married Relationship management skills Reliance on out-of-group advice Past acquaintance of group members Entrepreneur's centrality Female loan officer Proportion of women (N0.80) Prior industry experience Number of within-group ties

Women Men

0.65 5.21 0.18 0.08 0.49 6.41 0.48 0.53 35.73 11.41 0.49 0.88 5.76 0.32 0.67 0.06 0.45 0.425 0.53 1.20

5.15 0.11 0.06 0.56 5.77 0.49 0.69 35.57 11.64 0.49 0.86 5.77 0.28 0.68 0.06 0.38 0.62 0.48 1.23

(0.48) (8.56) (0.38) (0.27) (0.50) (4.30) (0.50) (0.50) (8.09) (2.66) (0.50) (0.33) (1.12) (0.47) (0.36) (0.13) (0.50) (0.50) (0.50) (2.23)

5.33 0.29⁎⁎⁎ 0.12⁎ 0.37⁎⁎ 7.55⁎⁎⁎ 0.47 0.27⁎⁎⁎ 36.01 11.02† 0.49 0.90 5.74 0.38† 0.65 0.06 0.57⁎⁎ 0.07⁎⁎⁎ 0.63⁎ 1.17

n = 237 (153 females, 84 males); standard deviations in parentheses. Correlations above 0.125 are significant at p b 0.05. † p b 0.10. ⁎ p b 0.05. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

1

2

3

4

5

6

7

8

9

10

0.04 −0.22 −0.12 0.19 −0.20 0.02 0.40 −0.03 0.11 0.00 −0.06 0.02 −0.10 0.03 0.00 −0.18 0.53 −0.15 0.01

−0.11 0.07 −0.13 0.10 0.22 −0.06 0.23 0.19 0.04 0.02 0.08 0.19 0.33 −0.12 −0.16 0.28 0.12 0.17

−0.14 −0.45 0.09 −0.13 −0.16 0.00 −0.10 0.01 −0.02 0.00 −0.02 −0.02 0.05 0.22 −0.21 0.05 0.06

−0.29 0.10 −0.02 −0.05 0.04 0.08 −0.02 −0.07 −0.05 −0.07 0.01 0.10 −0.06 0.01 −0.02 0.07

−0.02 −0.03 0.08 0.01 0.00 0.05 0.03 −0.02 0.07 −0.09 0.07 −0.05 0.08 −0.05 −0.12

−0.04 −0.08 0.20 −0.18 0.00 −0.01 0.01 0.14 0.08 0.05 −0.03 −0.12 0.03 0.10

0.01 −0.05 0.07 0.10 0.02 0.02 0.19 −0.06 0.00 −0.22 −0.02 0.08 −0.16

0.10 0.03 0.01 −0.01 −0.07 −0.11 −0.05 −0.03 0.15 0.49 −0.10 0.23

−0.01 −0.06 0.23 0.10 0.10 0.00 0.18 −0.04 0.22 0.12 0.05 0.0577 0.06 0.07 −0.01 0.10 0.12 −0.01 0.06 −0.03 −0.15

11

12

13

14

15

16

0.05 0.15 0.04 0.00 −0.04 0.00 −0.02 0.30 0.10

−0.09 0.05 0.08 −0.05 0.05 −0.02 0.06 0.02

0.20 0.18 −0.02 −0.06 −0.03 0.25 −0.09

0.00 −0.02 0.03 −0.09 −0.12 0.07 0.02 0.15 −0.02 0.01 0.22 −0.11 −0.14 0.26 −0.02

17

18

−0.08 −0.04 0.02 −0.02 0.04

19

0.13

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Full sample

831

832

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

Table 2 Results of OLS regression (full sample), dependent variable is entrepreneur's sales. Model 1 Control variables Industry manufacturing Industry wholesale Industry retail Business age Business registered Group — female leader Entrepreneur age Entrepreneur education Family business background Entrepreneur married Relationship management skills Reliance on out-of-group advice Past acquaintance of group members Entrepreneur's centrality Female loan officer Proportion of females (N0.80) Prior industry experience Gender (Female = 1) Hypothesized relationships Number of within-group ties Interaction of gender and within-group ties Constant Adjusted R2 Change in adjusted R2 F value Mean VIF (max VIF)

−0.31† −0.11 −0.30† 0.03† 0.47⁎⁎⁎ −0.46 0.02⁎⁎ 0.05⁎ −0.01 −0.12 −0.03 0.20 0.72⁎ −1.12⁎ 0.00 0.70⁎⁎ 0.02 −0.09

Model 2 (0.15) (0.30) (0.15) (0.01) (0.12) (0.27) (0.01) (0.02) (0.14) (0.18) (0.06) (0.14) (0.30) (0.53) (0.20) (0.25) (0.13) (0.11)

−0.33⁎ −0.19 −0.25† 0.02† 0.57⁎⁎⁎ −0.68⁎⁎ 0.03 0.07⁎⁎⁎ −0.10 −0.16 −0.01 0.21 0.48† −1.12⁎ 0.06 0.78⁎⁎⁎ −0.03 −0.09 0.13⁎⁎

−0.66 0.33 18.95⁎⁎⁎ 1.35 (1.8)

(0.53)

−1.08⁎ 0.38 0.05⁎ 19.95⁎⁎⁎ 1.38 (1.83)

Model 3 (0.15) (0.29) (0.14) (0.01) (0.11) (0.21) (0.01) (0.02) (0.11) (0.16) (0.05) (0.15) (0.28) (0.51) (0.19) (0.20) (0.11) (0.09)

−0.34⁎ −0.19 −0.27† 0.02† 0.57⁎⁎⁎ −0.71⁎⁎ 0.03⁎⁎⁎ 0.07⁎⁎⁎

(0.04)

0.17⁎⁎⁎ −0.06⁎ −1.07⁎ 0.39 0.01⁎⁎ 65.10⁎⁎⁎

(0.52)

−0.11 −0.16 −0.01 0.21 0.48† −1.08⁎ 0.04 0.80⁎⁎⁎ −0.03 −0.01

(0.16) (0.29) (0.14) (0.01) (0.11) (0.21) (0.01) (0.02) (0.11) (0.16) (0.05) (0.15) (0.27) (0.52) (0.19) (0.21) (0.11) (0.11) (0.03) (0.03) (0.51)

1.57 (3.08)

Note: dependent variable (sales) transformed using natural logarithm; n = 237; robust standard errors (adjusted for group membership) in parentheses. Two-tailed tests. † p b 0.10. ⁎ p b 0.05. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

p b 0.05; figure not shown for space-conservation purposes). While the interaction alone is not significant, our results are in line with what we expected. Noting the recommendation to include all interaction effects in order to appropriately interpret the results (Aguinis et al., 2013), we consider our third hypothesis as being partially supported. In Hypothesis 4, we argued that entrepreneurs' prior industry experience would positively moderate the relationship between the number of within-group ties and sales performance such that the relationship is more positive for entrepreneurs with higher industry experience. In Model 7, the interaction effect between prior industry experience and ties is positive and significant (β = 0.21, p b 0.05) and remains so in the full model (β = 0.16, p N 0.05). We illustrate this interaction in Fig. 3. The steeper slope illustrates the effect of the number of within-group ties on sales performance for female entrepreneurs with previous industry experience versus female entrepreneurs without it. The effect size of industry experience for females increases sales 29.7% as within-group ties increase from one to three, while the increase in within-group ties for female entrepreneurs without experience shows a negligible sales change. 6.1. Supplementary analyses In discussing findings about the consequences of gender as a diffuse status characteristic that shapes females' ability to leverage within-group ties for performance, we wanted to validate two underlying assumptions: whether the results hold for the male subsample and whether the status dynamics hold for other network characteristics of female entrepreneurs. The first assumption we wanted to validate was that any factors theorized to neutralize the relevance of gender as a status marker for females do not make the same impact on males' ability to leverage ties for performance. In Model 9 of Table 3, analyses from Model 8 are replicated and the results are presented for the male subsample. None of the interactions are significant, with the exception of a positive interaction effect for the proportion of females (β = 0.65, p b 0.05). While scholars should not attempt to draw implications from non-significant findings, it is comforting that findings in Table 3 largely do not contradict our theoretical expectations.10 Additionally, because there were fewer males than females in our sample, we also checked whether smaller sample size influenced the male subsample results. To do this, we performed a bootstrap of the models (Adèr and Mellenbergh, 2008). This approach develops 10 The positive and significant effect observed with group composition could be explained by social identity theory, an alternative theoretical perspective on the effect of gender composition of work groups. The theory posits that in-group support increases among members of a higher-status group (in our case, males) when the lowerstatus group expands. Applied to our context, this support could take the form of the exchange of privileged information or high-quality advice among males in the group, whose subgroup identity has become more salient as a result of the numerical dominance of females in the microcredit group.

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

833

Table 3 Results of OLS regression with robust standard errors (female sample); dependent variable is entrepreneur's sales. Model 4 Control variables Industry manufacturing Industry wholesale Industry retail Business age Business registered Group — female leader Entrepreneur age Entrepreneur education Family business background Entrepreneur married Relationship management skills Reliance on out-of-group advice Past acquaintance of group members Entrepreneur's centrality Female loan officer Proportion of women (N0.80) Prior industry experience Hypothesized relationships Number of within-group ties Interaction of number of within-group ties with … Female loan officer Proportion of females (N0.80) Prior industry experience Constant Adjusted R2 Δ adjusted R2 (compared to Model 4) F value

Model 5

Model 6

−0.35 0.30 −0.19 0.02 0.47 ⁎⁎ −0.53⁎ 0.02⁎ 0.08⁎ −0.10 −0.29 −0.05 0.04 0.29 −1.08† −0.06 0.71⁎⁎ 0.09

(0.25) (0.32) (0.15) (0.01) (0.13) (0.22) (0.01) (0.03) (0.15) (0.19) (0.06) (0.17) (0.29) (0.59) (0.18) (0.24) (0.16)

−0.25 0.39 −0.12 0.02 0.52⁎⁎⁎ −0.56⁎ 0.03⁎ 0.07⁎ −0.12 −0.30 −0.03 0.00 0.25 −1.07† −0.24 0.84⁎⁎ 0.04

(0.25) (0.31) (0.14) (0.01) (0.13) (0.22) (0.01) (0.03) (0.14) (0.19) (0.07) (0.17) (0.28) (0.60) (0.17) (0.26) (0.14)

−0.34 0.31 −0.18 0.02 0.48⁎⁎⁎ −0.54⁎ 0.02⁎ 0.08⁎ −0.10 −0.29 −0.05 0.04 0.29 −1.08† −0.07 0.73⁎⁎ 0.09

0.09⁎

(0.04)

0.04

(0.04)

0.10

0.16⁎

(0.06) −0.01

−0.62 0.31

8.48⁎⁎⁎

(0.59) −0.73 0.33 0.02⁎ 9.21⁎⁎⁎

(0.60) −0.63 0.31 0.00 8.15⁎⁎⁎

Model 7

(0.26) (0.31) (0.14) (0.01) (0.13) (0.21) (0.01) (0.03) (0.15) (0.18) (0.06) (0.17) (0.29) (0.58) (0.20) (0.24) (0.16)

−0.33 0.31 −0.14 0.02 0.50⁎⁎⁎ −0.64⁎⁎ 0.03⁎ 0.07⁎ −0.12 −0.31 0.00 0.07 0.30 −0.97 −0.11 0.79⁎⁎⁎ −0.18

(0.09) −0.05

Model 8

(0.23) (0.32) (0.12) (0.01) (0.13) (0.21) (0.01) (0.03) (0.15) (0.20) (0.05) (0.16) (0.27) (0.61) (0.15) (0.20) (0.16)

(0.07) −0.15⁎⁎

(0.10)

0.21⁎ (0.61) −0.87 0.35 0.04⁎⁎

−0.30 0.33 −0.15 0.02 0.52⁎⁎⁎ −0.59⁎⁎ 0.03⁎ 0.07⁎ −0.15 −0.30 0.01 0.05 0.25 −1.02 −0.26 0.76⁎⁎ −0.17

0.20⁎ 0.12⁎ 0.16⁎

(0.10) (0.61) −0.85 0.36 0.05⁎⁎⁎

9.36⁎⁎⁎

10.83⁎⁎⁎

Model 9 (male sample) (0.25) (0.31) (0.12) (0.01) (0.13) (0.20) (0.01) (0.03) (0.15) (0.20) (0.06) (0.16) (0.27) (0.60) (0.18) (0.23) (0.17)

−0.37 −0.61† −0.40 0.03 0.74⁎⁎⁎ −0.84⁎

0.03† 0.07⁎⁎ −0.09 0.15 0.04 0.45 0.43 −0.77 0.12 0.39 −0.37

(0.27) (0.34) (0.30) (0.03) (0.19) (0.36) (0.02) (0.02) (0.18) (0.22) (0.15) (0.32) (0.43) (0.57) (0.35) (0.38) (0.33)

(0.05) −0.14

(0.12)

(0.08) 0.15 (0.06) 0.65⁎ (0.08) 0.21 (0.63) −1.56 0.51

(0.17) (0.31) (0.16) (1.11)

5.14⁎⁎⁎

Note: dependent variable (sales) transformed using natural logarithm; for models 4 to 8 n = 153; for model 9, n = 84. Robust standard errors (adjusted for group membership) in parentheses. Two-tailed tests. † p b 0.10. ⁎ p b 0.05. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

estimates by drawing new samples from the original data. Bootstrapping can account for distortions caused by a specific sample that is not representative of a larger population. The results with 500 bootstrap replications were similar in significance, suggesting that sample-size differences did not affect the results.

Fig. 1. Interaction of number of within-group ties and entrepreneur's gender.

834

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

Fig. 2. Interaction of number of within-group ties and loan officer's gender on sales (female sample).

The second validation of status effects in entrepreneurial microcredit networks stems from finding gender differences for other network characteristics.11 Considering that the large majority of the network literature points to the importance of accounting for network structure and position, we inspected two variables respectively capturing these network features—ego network density and betweenness centrality—both operationalized using standard network measures (Wasserman and Faust, 1994). Interestingly, in both cases, we found a negative interaction effect of the respective network variable and gender though the interaction with betweenness centrality was not significant.

6.2. Robustness checks In order to show the robustness of our results, we ran a number of additional analyses. First, to account for the number of possible ties in the group, we included an additional control for group size. The results remained unchanged, and the effect of group size on entrepreneurs' sales was positive yet not significant. In another analysis, we added a control for group-realized density of ties (dividing the number of ties in the group with the number of possible ties) and found that this variable was not significant (nor did it change our results). Second, our data included several female-only groups. While gender-status dynamics are theorized to still be relevant even in gender-homophilous groups (Ridgeway, 2011), we reanalyzed our models by excluding those groups and found our results to be consistent. Given the nested nature of our data, we also reran our models using a multilevel approach (Stata: mixed). The likelihood ratio test that compares whether the random coefficients model is superior to an OLS model was rejected by the data (Likelihood Ratio test versus linear regression: chi2(3) = 1.63 Prob N chi2 = 0.6536). We also performed several steps to address potential issues of selection. First, it is possible that firm performance is driven by prior performance or that the key indicators are a result of unobserved factors prior to joining the group. Attempting to rule out these alternative explanations, we tested an approximation-dependent variable as a lag. This is common when appropriate instruments are unavailable (Keele and Kelly, 2006). Respondents were asked if their sales increased, decreased, or stayed the same from the year before. Including this as an approximation for lagged sales performance did not alter the significance or direction of the results, and the Durbin–Wu–Hausman test was not significant. Second, we used matching techniques to address potential selection issues. According to Davidsson and Delmar (2009:29), “Post-matching techniques rest on the basic idea that an almost experimental design with a control group and a treatment group can be created while still using observational data if we can eliminate differences in individuals prior to being exposed to the treatment. This alleviates the problem of self-selection and it is based on the conditioning on the back door variables. Through matching the estimation of the causal effect of the said treatment is improved.” In our study, the “treatment” is within-group ties, and the outcome is firm performance. The challenge is that there are several observable and unobservable variables that may affect within-group ties, and those might also have an effect on sales income. We isolated and controlled for these disturbance variables with propensity score matching (Rosenbaum and Rubin, 1983). This approach uses a series of observed variables to predict the probability of experiencing the treatment to create a counterfactual group. We can then estimate the effect of a treatment (i.e., within-group ties) on those with varying experience with within-group ties in the observational data through matching. This approach is commonly applied in areas like medicine and economics (Heckman and Navarro-Lozano, 2004). Because our within-group-ties treatment is non-dichotomous, we use generalized propensity score matching (Stata gpscore). Following Cuong's (2013) recommendation from Monte Carlo simulations, all available observed variables in the outcome equation are included in the estimation of propensity scores. Including the propensity score as a covariate in our models was non-significant and did not

11

We thank the Editor for this suggestion.

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

835

Fig. 3. Interaction of number of within-group ties and entrepreneur's prior industry experiences on sales (female sample).

alter the significance of our hypothesized relationships. In summary, while longitudinal data would be ideal, our arguments and robustness checks give us additional confidence in the results. 7. Discussion Relationships are important in supporting entrepreneurial activity for males and females alike. However, an important qualifier in assessing their potential is the consideration of entrepreneurs' competency assessment. In this paper, we observed microcredit entrepreneurs' within-group ties through an EST lens to investigate both between- and within-gender differences in the extent to which entrepreneurs' group ties relate to their ventures' performance. In an effort to go beyond examining differences in network characteristics and behaviors between male and female entrepreneurs, we add to this inquiry an empirical test of the network-performance link with a deeper understanding of within-gender differences among female entrepreneurs. Consistent with the EST perspective, we found that while both females and males benefit from within-group ties, males are able to extract more from their relationships (indeed, the effect held for other network characteristics, such as network density). Moreover we found that for females, both loan officer gender (and in part group gender composition) and aspects of their expertise can positively enforce the relationship between within-group ties and performance, while the same was largely not observed with male group members. Our study answers calls for empirical testing of oft-implied gender differences in network effects for performance (Watson, 2012) and builds a model of contingency factors that may shape female entrepreneurs' ability to unlock more of the value of their withingroup ties for venture performance. These results reestablish network size as an important attribute of entrepreneurial networks (Hansen, 1995; Hoang and Antoncic, 2003)—one that holds above and beyond other network characteristics (e.g., entrepreneurs' centrality and prior acquaintances with group members)—yet importantly highlight the between-gender differences of these ties for venture performance. To that end, we echo Hanson and Blake's (2009) call for incorporating a status lens for understanding female entrepreneurial networks. Taken together, our results highlight more than gender differences. The EST framework helps us highlight individual and group contextual factors as important in understanding how female entrepreneurs' within-group ties influence performance. This offers both important theoretical and practical implications. Theoretically, our results reestablish the value of EST in examining network dynamics from a gender perspective (Chafetz, 1997) and extend its applicability to an entrepreneurial setting. In doing so, we not only establish whether a status deficit operates in how microcredit ties translate to entrepreneurial firm performance but also go beyond male–female comparisons to examine factors that shape status in ways that are idiosyncratic to females—among females. We contribute to the EST literature by refining our understanding of how individual status markers (e.g., expertise) and microcontext factors (e.g., group characteristics) shape entrepreneurs' competency expectations. Specifically, our results confirm the importance of industry experience in shaping the competency expectations of female entrepreneurs, attesting to the importance of this type of experience in the entrepreneurship literature (Cooper et al., 1994; Delmar and Shane, 2006). Microcredit agencies' could accordingly potentially consider prior industry experience as an important client-assessment criterion. Further, microcredit agencies could proactively share members' background expertise in microcredit group meetings to reduce status issues early on and improve group networking between all members. At the group level, our results point to the importance of considering the context in which female entrepreneurs and their relationships are embedded (Hughes et al., 2012). Group-level characteristics appear to have potential for facilitating group member outcomes even where society-level expectations may be a hindrance to female network usage. Our results regarding group loan officer gender generalize the importance of status-transfer effects to the microcredit context and, more generally, to the entrepreneurship context. We also found that groups composed predominantly of females had a more positive impact on female entrepreneurs' ability to translate within-group ties into venture performance (though this was only significant in the full model). This finding echoes

836

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

research from organizational settings (Kanter, 1977) and helps extend their generalizability in the entrepreneurship microcredit context. Moving forward, we encourage gender scholars to pay attention to the relevant attributes of the microcontext to better understand how these shape potential interactions and value from ties. On a practical note, informed by our findings, microcredit agencies might carefully consider the match between loan officer and group composition to support group and member outcomes. Our results also contribute to the previously discussed role of microcredit agencies in group formation and the extent to which they should actively manage it (Bruton et al., 2011; Mayoux, 2001). While our results cannot be prescriptive to the agency's role in this process, they do speak to relevant factors for consideration. More broadly, by placing the female entrepreneur rather than the microcredit agency at the center of this study, we provide an integration of management and development economics logics for microcredit (Bruton et al., 2011; Khavul et al., 2009). We also move beyond common assumptions about automatic contributions of group formation to females' success and uncover some of the boundary conditions under which group-based microcredit can realize its potential (Mayoux, 2001). As microcredit agencies engage with clients in different cultural and social settings (e.g., developed versus developing, rural versus urban, etc.), the context of our study offers an interesting platform for diversifying research settings (Bruton et al., 2011), which is especially relevant given the calls to understand female entrepreneurship beyond Western countries and high-growth industries (Aldrich et al., 1989; Loscocco et al., 2009; McGuire, 2002). This study furthers the inquiry into female entrepreneurs in developing countries, which have traditionally received less scholarly attention despite higher prevalence of gender inequality (WEF, 2013). 7.1. Limitations and future research The contributions of our study come with potential limitations that may offer opportunities for future research. Along with many network studies and those collecting data in challenging contexts, we are limited in establishing causality by the cross-sectional nature of our data (Hoang and Antoncic, 2003; Khayesi et al., 2014; Zaheer et al., 2010). The cross-sectional design also has implications for status, which is captured at a point in time and treated as a static construct. While this is the case for much of the literature drawing on EST (e.g., Saparito et al., 2012), it is quite possible that individuals can experience shifts in status over time (De Kelaita et al., 2001). Future research is invited to examine the possibility of such shifts using longitudinal studies. Finally, while our results speak to understanding heterogeneity among females in leveraging within-group ties, we cannot know whether the boundary conditions we discovered make them equal to their male counterparts. Such an investigation would involve a three-way interaction test between gender, within-group ties, and respective moderators, for which our sample size was not powerful enough. We encourage future research to assess the extent to which individual and group factors can offset status-related challenges in leveraging ties for females. 7.2. Conclusion This study has simultaneously established the importance and the contingent effect of within-group ties on female entrepreneurs' business performance. Using the context of microcredit groups in Kenya, we further theoretical explanations of the effect of microcredit groups' ties on female members' outcomes. Our findings indicate that firm performance for clients is a function of increasing the number of ties within the group and that for females, the moderating effects of prior industry experience, loan officer gender, and group gender composition need to be considered when assessing the value of within-group ties for venture performance. This study provides a number of future avenues for entrepreneurial research in the gender, EST, and microfinance fields. References Adèr, H.J., Mellenbergh, G.J., 2008. Advising on Research Methods: A Consultant's Companion. Johannes van Kessel Publishing. Agier, I., Szafarz, A., 2010. Microfinance and gender: is there a glass ceiling on loan size? World Dev. 42, 165–181. Aguinis, H.H., Gottfredson, R.K., Culpepper, S.A., 2013. Best-practice recommendations for estimating cross-level interaction effects using multilevel modeling. J. Manag. 39, 1490–1528. Ahl, H., 2006. Why research on women entrepreneurs needs new directions. Enterp. Theory Pract. 30, 595–621. Ahmad, M.M., 2002. Who cares? The personal and professional problems of NGO fieldworkers in Bangladesh. Dev. Pract. 12, 177–191. Aiken, L.S., West, S.G., 1991. Multiple Regression: Testing and Interpreting Interactions. Sage. Akoten, J.E., Sawada, Y., Otsuka, K., 2006. The determinants of credit access and its impacts on micro and small enterprises: the case of garment producers in Kenya. Econ. Dev. Cult. Chang. 54, 927–944. Aldrich, H., Rosen, B., Woodward, W., 1987. The impact of social networks on business foundings and profit: a longitudinal study. Front. Entrep. Res. 7, 68. Aldrich, H., Reese, P.R., Dubini, P., 1989. Women on the verge of a breakthrough: networking among entrepreneurs in the United States and Italy. Entrep. Reg. Dev. 1, 339–356. Al-hassan, S., Abdul-Malik, A., Andani, A., 2011. The role of Grameen Ghana in improving income of women shea butter processors. J. Dev. Agric. Econ. 3, 537–544. Amine, L.S., Staub, K.M., 2009. Women entrepreneurs in sub-Saharan Africa: an institutional theory analysis from a social marketing point of view. Entrep. Reg. Dev. 21, 183–211. Anthony, D., 2005. Cooperation in microcredit borrowing groups: Identity, sanctions, and reciprocity in the production of collective goods. Am. Sociol. Rev. 70, 496–515. Armendariz de Aghion, B.A., Morduch, J., 2005. The Economics of Microfinance. MIT Press, Cambridge, MA. Baker, T., Aldrich, H., Nina, l., 1997. Invisible entrepreneurs: the neglect of women business owners by mass media and scholarly journals in the USA. Entrep. Reg. Dev. 9, 221–238. Batjargal, B., 2003. Social capital and entrepreneurial performance in Russia: A longitudinal study. Organ. Stud. 24, 535–556. Baughn, C.C., Chua, B.-L., Neupert, K.E., 2006. The normative context for women's participation in entrepreneurship: a multicountry study. Enterp. Theory Pract. 30, 687–708. Beaman, L., Chattopadhyay, R., Duflo, E., Pande, R., Topalova, P., 2009. Powerful women: does exposure reduce bias? Q. J. Econ. 124, 1497–1540. Berger, J., Cohen, B.P., Zelditch Jr., M., 1972. Status characteristics and social interaction. Am. Sociol. Rev. 37, 241–255.

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

837

Bhagavatula, S., Elfring, T., van Tilburg, A., van de Bunt, G.G., 2010. How social and human capital influence opportunity recognition and resource mobilization in India's handloom industry. J. Bus. Ventur. 25, 245–260. Bonner, B.L., Sillito, S.D., Baumann, M.R., 2007. Collective estimation: accuracy, expertise, and extroversion as sources of intra-group influence. Organ. Behav. Hum. Decis. Process. 103, 121–133. Borgatti, S.P., Cross, R., 2003. A relational view of information seeking and learning in social networks. Manag. Sci. 49, 432–445. Borgatti, S.P., Everett, M.G., Freeman, L.C., 2002. Ucinet for Windows: Software for Social Network Analysis. Analytic Technologies, MA. Bottger, R., 1984. Expertise and air time as bases of actual and perceived influence in problem-solving groups. J. Appl. Psychol. 69, 214–221. Bradley, S.W., McMullen, J.S., Artz, K., Simiyu, E.M., 2012. Capital is not enough: innovation in developing economies. J. Manag. Stud. 49, 684–717. Brass, D.J., 1985. Men's and women's networks: a study of interaction patterns and influence in an organization. Acad. Manag. J. 28, 327–343. Bruni, A., Gherardi, S., Poggio, B., 2004. Doing gender, doing entrepreneurship: an ethnographic account of intertwined practices. Gend. Work. Organ. 11, 406–429. Bruton, G.D., Khavul, S., Chavez, H., 2011. Microlending in emerging economies: building a new line of inquiry from the ground up. J. Int. Bus. Stud. 42, 718–739. Bunderson, J.S., 2003. Recognizing and utilizing expertise in work groups: a status characteristics perspective. Adm. Sci. Q. 48, 557–591. Burt, R.S., 1992. Structural Holes: the Social Structure of Competition. Harvard University Press, Cambrige, MA. Buttner, E.H., Rosen, B., 1988. Bank loan officers' perceptions of the characteristics of men, women, and successful entrepreneurs. J. Bus. Ventur. 3, 249–258. Carli, L.L., 2001. Gender and social influence. J. Soc. Issues 57, 725–741. Carter, S., Shaw, E., Lam, W., Wilson, F., 2007. Gender, entrepreneurship, and bank lending: the criteria and processes used by bank loan officers in assessing applications. Enterp. Theory Pract. 31, 427–444. Chafetz, J.S., 1997. Feminist theory and sociology: underutilized contributions for mainstream theory. Annu. Rev. Sociol. 23, 97–120. Chandler, G.N., Jansen, E., 1992. The founder's self-assessed competence and venture performance. J. Bus. Ventur. 7, 223–236. Coleman, S., 2000. Access to capital and terms of credit: a comparison of men- and women-owned small businesses. J. Small Bus. Manag. 38, 37–52. Collins-Dodd, C., Gordon, I.M., Smart, C., 2004. Further evidence on the role of gender in financial performance. J. Small Bus. Manag. 42, 395–417. Cooper, A.C., Gimeno-Gascon, F.J., Woo, C.Y., 1994. Initial human and financial capital as predictors of new venture performance. J. Bus. Ventur. 9, 371–395. Correll, S.J., Ridgeway, C.L., 2006. Expectation states theory. Handbook of Social Psychology. Springer, pp. 29–51. Cross, R., Cummings, J.N., 2004. Tie and network correlates of individual performance in knowledge-intensive work. Acad. Manag. J. 47, 928–937. Cuong, N.V., 2013. Which covariates should be controlled in propensity score matching? Evidence from a simulation study. Statistica Neerlandica 67, 169–180. Damanpour, F., Walker, R.M., Avellaneda, C.N., 2009. Combinative effects of innovation types and organizational performance: a longitudinal study of service organizations. J. Manag. Stud. 46, 650–675. Davidsson, P., Delmar, F., 2009. Dealing with Heterogeneity Problems and Causal Effect Estimation in Entrepreneurship Research (Available at SSRN 2210320). De Bruin, A., Brush, C.G., Welter, F., 2006. Introduction to the special issue: towards building cumulative knowledge on women's entrepreneurship. Enterp. Theory Pract. 30, 585–593. De Kelaita, R., Munroe, P.T., Tootell, G., 2001. Self-initiated status transfer: a theory of status gain and status loss. Small Group Res. 32, 406–425. Delmar, F., Davidsson, P., 2000. Where do they come from? Prevalence and characteristics of nascent entrepreneurs. Entrep. Reg. Dev. 12, 1–23. Delmar, F., Shane, S., 2006. Does experience matter? The effect of founding team experience on the survival and sales of newly founded ventures. Strateg. Organ. 4, 215–247. Dimov, D., 2007. From opportunity insight to opportunity intention: the importance of person-situation learning match. Enterp. Theory Pract. 31, 561–583. Dixon, R., Ritchie, J., Siwale, J., 2007. Loan officers and loan delinquency in microfinance: a Zambian case. Accounting Forum. Elsevier, pp. 47–71. Downing, J., 1990. Gender and the Growth and Dynamics of Microenterprises. US Agency for International Development, Private Enterprise Bureau, Washington, D.C. Ellis, A., Coutura, J., Dione, N., Gillson, J., Manuel, C., et al., 2007. Gender and Economic Growth in Kenya: Unleashing the Power of Women. The World Bank, Washington D.C. Ely, R.J., 1995. The power in demography: women's social constructions of gender identity at work. Acad. Manag. J. 38, 589–634. Evans, D.S., 1987. The relationship between firm growth, size, and age: estimates for 100 manufacturing industries. J. Ind. Econ. 567–581. Fagenson, E.A., Marcus, E.C., 1991. Perceptions of the sex-role stereotypic characteristics of entrepreneurs: women's evaluations. Enterp. Theory Pract. 15, 33–47. Feigenberg, B., Field, E., Pande, R., Rigol, N., Sarkar, S., 2014. Do group dynamics influence social capital gains among microfinance clients? Evidence from a randomized experiment in urban India. J. Policy Anal. Manag. 33, 932–949. Freeman, L.C., 1979. Centrality in social networks conceptual clarification. Soc. Networks 1, 215–239. Friedkin, N.E., 1990. A Guttman scale for the strength of an interpersonal tie. Soc. Networks 12, 239–252. Goetz, A.M., 2001. Women Development Workers: Implementing Rural Credit Programmes in Bangladesh. SAGE Publications. Gomez, R., Santor, E., 2001. Membership has its privileges: the effect of social capital and neighbourhood characteristics on the earnings of microfinance borrowers. Can. J. Econ. 943–966. Greve, A., Salaff, J.W., 2003. Social networks and entrepreneurship. Enterp. Theory Pract. 28, 1–22. Guerin, I., Palier, J., 2006. Microfinance and the empowerment of women: will the silent revolution take place? In: A.D.A. (Ed.), Microfinance and Gender: New Contributions to an Old Issue. DTP — Schildgen Communication, Luxembourg, pp. 27–34 Gupta, V.K., Turban, D.B., Wasti, S.A., Sikdar, A., 2009. The role of gender stereotypes in perceptions of entrepreneurs and intentions to become an entrepreneur. Enterp. Theory Pract. 33, 397–417. Hansen, E.L., 1995. Entrepreneurial networks and new organization growth. Enterp. Theory Pract. 19, 7–20. Hanson, S., Blake, M., 2009. Gender and entrepreneurial networks. Reg. Stud. 43, 135–149. Heckman, J., Navarro-Lozano, S., 2004. Using matching, instrumental variables, and control functions to estimate economic choice models. Rev. Econ. Stat. 86, 30–57. Hermes, N., Lensink, R., Mehrteab, H.T., 2005. Peer monitoring, social ties and moral hazard in group lending programs: evidence from Eritrea. World Dev. 33, 149–169. Hoang, H., Antoncic, B., 2003. Network-based research in entrepreneurship: a critical review. J. Bus. Ventur. 18, 165–187. Hogue, M., Yoder, J.D., 2003. The role of status in producing depressed entitlement in women's and men's pay allocations. Psychol. Women Q. 27, 330–337. Hospes, O., Musinga, M., Ong'ayo, M., 2002. An evaluation of microfinance programmes in Kenya as supported through the Dutch co-financing programme. Steering Committee for the Evaluation of the Netherlands Co-financing Programme. Hughes, K.D., Jennings, J.E., 2012. Global Women's Entrepreneurship Research: Diverse Settings, Questions and Approaches. Edward Elgar, Cheltenham/Northampton, U.K. Hughes, K.D., Jennings, J.E., Brush, C., Carter, S., Welter, F., 2012. Extending women's entrepreneurship research in new directions. Enterp. Theory Pract. 36, 429–442. Iakovleva, T., Kickul, J., 2011. Beyond social capital: the role of perceived legitimacy and entrepreneurial intensity in achieving funding success and superior venture performance in women-led Russian SMEs. Int. J. Entrep. Small Bus. 14, 13–38. Ibarra, H., 1992. Homophily and differential returns: sex differences in network structure and access in an advertising firm. Adm. Sci. Q. 422–447. Jennings, J.E., Brush, C.G., 2013. Research on women entrepreneurs: challenges to (and from) the broader entrepreneurship literature? Acad. Manag. Ann. 7, 663–715. Jenssen, J.I., Greve, A., 2002. Does the degree of redundancy in social networks influence the success of business start-ups? Int. J. Entrep. Behav. Res. 8, 254–267. Kabeer, N., 2001. Conflicts over credit: re-evaluating the empowerment potential of loans to women in rural Bangladesh. World Dev. 29, 63–84. Kanter, R.M., 1977. Men and Women of the Corporation. Basic Books, New York. Karlan, D., Valdivia, M., 2011. Teaching entrepreneurship: impact of business training on microfinance clients and institutions. Rev. Econ. Stat. 93, 510–527. Katila, R., Shane, S., 2005. When does lack of resources make new firms innovative? Acad. Manag. J. 48, 814–829. Keele, L.J., Kelly, N.J., 2006. Dynamic Models for Dynamic Theories: The Ins and Outs of LDVs. Polit. Anal. 14 (2), 186–205. Khavul, S., 2010. Microfinance: creating opportunities for the poor? Acad. Manag. Perspect. 24, 58–72. Khavul, S., Bruton, G.D., Wood, E., 2009. Informal family business in Africa. Enterp. Theory Pract. 33, 1219–1238. Khayesi, J.N.O., George, G., Antonakis, J., 2014. Kinship in entrepreneur networks: performance effects of resource assembly in Africa. Enterp. Theory Pract. 38, 1323–1342. Lerner, M., Brush, C., Hisrich, R., 1997. Israeli women entrepreneurs: an examination of factors affecting performance. J. Bus. Ventur. 12, 315–339.

838

H. Milanov et al. / Journal of Business Venturing 30 (2015) 822–838

Lockheed, M.E., 1985. Sex and social influence: a meta-analysis guided by theory. In: Berger, J., Zelditch, J.M. (Eds.), Status, Rewards, and Influence: How Expectations Organize Behavior. Jossey-Bass, San Francisco, pp. 406–429. Loscocco, K., Monnat, S.M., Moore, G., Lauber, K.B., 2009. Enterprising women: a comparison of women's and men's small business networks. Gend. Soc. 23, 388–411. Maclean, K., 2010. Capitalizing on women's social capital? Women-targeted microfinance in Bolivia. Dev. Chang. 41, 495–515. Major, B., 1994. From social inequality to personal entitlement: the role of social comparisons, legitimacy appraisals, and group membership. Adv. Exp. Soc. Psychol. 26, 293–355. Maurer, I., Ebers, M., 2006. Dynamics of social capital and their performance implications: lessons from biotechnology start-ups. Adm. Sci. Q. 51, 262–292. Mayoux, L., 2001. Tackling the Down Side: Social Capital, Women's Empowerment and Micro-Finance in Cameroon. Dev. Chang. 32. Mayoux, L., 2007. Not only reaching, but also empowering women: ways forward for the next microfinance decade. In: Faber, V. (Ed.), Microfinance and Gender: New Contributions to an Old Issue. ADA, Canada, p. 35. McGuire, G.M., 2002. Gender, race, and the shadow structure a study of informal networks and inequality in a work organization. Gend. Soc. 16, 303–322. Mitchell, W., 1994. The dynamics of evolving markets: the effects of business sales and age on dissolutions and divestitures. Adm. Sci. Q. 575–602. Murphy, P.J., Kickul, J., Barbosa, S.D., Titus, L., 2007. Expert capital and perceived legitimacy: female-run entrepreneurial venture signaling and performance. Int. J. Entrep. Innov. 8, 127–138. Neter, J., Wasserman, W., Kutner, M.H., 1996. Applied Linear Statistical Models. Irwin, Chicago. Paxton, J., Graham, D., Thraen, C., 2000. Modeling group loan repayment behavior: new insights from Burkina Faso. Econ. Dev. Cult. Chang. 48, 639–655. Pretty, J., Ward, H., 2001. Social capital and the environment. World Dev. 29, 209–227. Pugh, M.D., Wahrman, R., 1983. Neutralizing sexism in mixed-sex groups: do women have to be better than men? Am. J. Sociol. 746–762. Reagans, R., McEvily, B., 2003. Network structure and knowledge transfer: the effects of cohesion and range. Adm. Sci. Q. 48, 240–267. Ridgeway, C.L., 2011. Framed by Gender: How Gender Inequality Persists in the Modern World. Oxford University Press, New York. Ridgeway, C.L., Berger, J., 1986. Expectations, legitimation, and dominance behavior in task groups. Am. Sociol. Rev. 51, 603–617. Ridgeway, C.L., Diekema, D., 1992. Are Gender Differences Status Differences? Gender, Interaction, and Inequality. Springer, pp. 157–180. Ridgeway, C.L., Smith-Lovin, L., 1999. The gender system and interaction. Annu. Rev. Sociol. 191–216. Rosenbaum, P.R., Rubin, D.B., 1983. The central role of the propensity score in observational studies for causal effects. Biometrika 70, 41–55. Ruef, M., Aldrich, H.E., Carter, N.M., 2003. The structure of founding teams: homophily, strong ties, and isolation among US entrepreneurs. Am. Sociol. Rev. 68, 195–222. Sackett, P.R., DuBois, C.L., Noe, A.W., 1991. Tokenism in performance evaluation: the effects of work group representation on male–female and white–black differences in performance ratings. J. Appl. Psychol. 76, 263–267. Saparito, P., Elam, A., Brush, C., 2012. Bank–firm relationships: do perceptions vary by gender? Enterp. Theory Pract. 37, 837–858. Semrau, T., Werner, A., 2013. How exactly do network relationships pay off? The effects of network size and relationship quality on access to start-up resources. Enterp. Theory Pract. 38, 501–525. Sharma, M., Zeller, M., 1997. Repayment performance in group-based credit programs in Bangladesh: an empirical analysis. World Dev. 25, 1731–1742. Sine, W.D., Mitsuhashi, H., Kirsch, D.A., 2006. Revisiting Burns and Stalker: formal structure and new venture performance in emerging economic sectors. Acad. Manag. J. 49, 121–132. Siwale, J.N., Ritchie, J., 2012. Disclosing the loan officer's role in microfinance development. Int. Small Bus. J. 30, 432–450. Smith-Lovin, L., Brody, C., 1989. Interruptions in group discussions: the effects of gender and group composition. Am. Sociol. Rev. 54, 424–435. Sparrowe, R.T., Liden, R.C., Wayne, S.J., Kraimer, M.L., 2001. Social networks and the performance of individuals and groups. Acad. Manag. J. 44, 316–325. Stam, W., Arzlanian, S., Elfring, T., 2013. Social capital of entrepreneurs and small firm performance: a meta-analysis of contextual and methodological moderators. J. Bus. Ventur. 29, 152–173. Starr, J.A., MacMillan, I., 1990. Resource cooptation via social contracting: resource acquisition strategies for new ventures. Strateg. Manag. J. 11, 79–92. Stewart, D.D., Stasser, G., 1995. Expert role assignment and information sampling during collective recall and decision making. J. Pers. Soc. Psychol. 69, 619–628. Stinchcombe, A., 1965. Social structure and organizations. In: March, J.G. (Ed.), Handbook of Organizations. Rand McNally, Chicago. Strier, R., 2010. Women, poverty, and the microenterprise: context and discourse. Gend. Work. Organ. 17, 195–218. Thébaud, S., 2010. Gender and entrepreneurship as a career choice: do self-assessments of ability matter? Soc. Psychol. Q. 73, 288–304. Tolbert, P.S., Graham, M.E., Andrews, A.O., 1999. Group gender composition and work group relations: theories, evidence, and issues. In: Powell, G. (Ed.), Handbook of Gender and Work. SAGE, Thousand Oaks, CA, pp. 179–202. Tyler, T.R., 2006. Psychological perspectives on legitimacy and legitimation. Annu. Rev. Psychol. 57, 375–400. Underwood, T., 2007. Women and microlending in Western Europe. In: A.D.A. (Ed.), Microfinance and Gender: New Contributions to an Old Issue. DTP — Schildgen, Communication, Luxembourg, pp. 145–150. Velasco, C., Marconi, R., 2004. Group dynamics, gender and microfinance in Bolivia. J. Int. Dev. 16, 519–528. Wasserman, S., Faust, K., 1994. Social Network Analysis: Methods and Applications. Cambridge University Press, Cambridge. Watson, J., 2012. Networking: gender differences and the association with firm performance. Int. Small Bus. J. 30, 536–558. Webster Jr., M., Hysom, S.J., 1998. Creating status characteristics. Am. Sociol. Rev. 63, 351–378. WEF, 2013. Global Gender Gap Report. World Economic Forum, Geneva, Switzerland. Wittenbaum, G.M., 2000. The bias toward discussing shared information: why are high-status group members immune. Commun. Res. 27, 379–401. Wooldridge, J.M., 2003. Cluster-sample methods in applied econometrics. Am. Econ. Rev. 93, 133–138. Yang, T., Aldrich, H.E., 2014. Who's the boss? Explaining gender inequality in entrepreneurial teams. Am. Sociol. Rev. 1–25. Yoder, J.D., 2001. Making leadership work more effectively for women. J. Soc. Issues 57, 815–828. Zaheer, A., Gözübüyük, R., Milanov, H., 2010. It's the connections: the network perspective in interorganizational research. Acad. Manag. Perspect. 24, 62–77. Zimmerman, M.A., Zeitz, G.J., 2002. Beyond survival: achieving new venture growth by building legitimacy. Acad. Manag. Rev. 27, 414–431.