Failure or voluntary exit? Reassessing the female underperformance hypothesis ☆

We reevaluate the female underperformance hypothesis by challenging the assumption that female-owned ventures are more likely to fail. Instead of equating exit with failure, we draw on exit literature and feminist theories to argue that female entrepreneurs are actually more likely than males to exit voluntarily . We argue for further gender differences by using an even more ﬁ ne-grained conceptualization of entrepreneurial exit (failure, exit for personal reasons, and exit for other professional/ ﬁ nancial opportunities). Post-hoc analyses also point to within-gender heterogeneity depending on family status. A sample probe of 219 Spanish entrepreneurs who had exited their business supports our overall reasoning.


Introduction
The survival prospects of female-owned businesses have been the subject of considerable debate over the last two decades (for a review, see Jennings and Brush, 2013;and Klapper and Parker, 2011).While a few studies reveal no gender differences in business survival (Coleman and Robb, 2012;Kalleberg and Leicht, 1991;Robb and Watson, 2012)-and some show female-owned ventures outperforming male-owned ventures in specific industries (e.g., education, clothing) and geographic areas (e.g., large cities) (Kalnins and Williams, 2014)-the bulk of the extant research generally assumes that female entrepreneurs are more likely to fail than their male peers, thereby reinforcing the dominant "female underperformance hypothesis" in entrepreneurship (Du Rietz and Henrekson, 2000;Rosa et al., 1996).This hypothesis refers to a body of research which has found (at least at the aggregate level) that "female entrepreneurs tend to underperform relative to their male counterparts" (Du Rietz and Henrekson, 2000, p. 1).The majority of studies focusing on survival come to this conclusion based on evidence about higher exit rates for female-headed firms (e.g., Allen et al., 2008;Boden and Nucci, 2000;Bosma et al., 2004;Fairlie and Robb, 2009;Robb, 2002).For example, Taylor (1999) notes that in Britain, for all spells of self-employment at the 5-year mark, male-owned ventures had a 58% survival rate while female-owned ventures had a 51% survival rate.Thus, the prevailing notion is that female entrepreneurs tend to underperform and do not "measure up" to their male peers.
We argue, however, that this notion is erroneous because it equates exit with failure.Recent entrepreneurship articles have demonstrated that the concepts of failure and exit derive from different theoretical perspectives and are driven by different factors (e.g., Bates, 2005;Wennberg et al., 2010).Yet, the dominant perspective from strategic management and organizational research is that an entrepreneur's primary goal is to develop a competitive advantage and long-term sustainability.Thus, exit is often viewed as failure and a negative outcome, while survival and continuation are viewed as successful outcomes (Wennberg and DeTienne, 2014).This dichotomous view does not take into account individual volition and decision-making autonomy, wherein an entrepreneur may view exit as a specific goal and positive outcome (Ryan and Power, 2012).The argument for disentangling failure and exit constructs contends that entrepreneurs voluntarily exit ventures for many reasons, including strong firm performance (e.g., acquisitions) (Cumming, 2008), personal reasons (e.g., retirement or full-time education) (Harada, 2007), alternative jobs (Taylor, 1999), or even as a risk-reduction strategy in which entrepreneurs abandon ventures with limited upside potential (e.g., fail quickly) (McGrath, 1999).The obvious implication of this perspective to the current gender literature is that additional research is needed to understand if females actually "fail" more often or simply "exit" more often.
Research provides some preliminary evidence supporting the notion that failure may be only one of many explanations for female entrepreneurs' higher exit rate.For example, Taylor's (1999) study of self-employment duration contains descriptive statistics offering initial evidence that males, on average, are more likely than females to exit a venture for reasons of alternative employment (47.9% vs. 35.8%),bankruptcy (18% vs. 9.9%), and retirement (9.2% vs. 7.8%).Females are more likely to exit for other specified and unspecified reasons which include health issues, family care, moving home, and full-time education (46.6% vs. 24.9%).Taylor's study is important because it is the first (that we are aware of) to provide preliminary evidence that females may exit ventures for different reasons.His work has largely gone under the radar of entrepreneurship and gender scholars, partly because it lacked explicit focus and theorization on the gendered nature of entrepreneurial exit and focused instead on self-employment (SE) duration, which is a prominent theme in economics and labor economics literature (e.g., Boden, 1996;Georgellis et al., 2007).Yet, if female entrepreneurs are indeed more likely than males to exit otherwise healthy businesses, current research would greatly benefit from further inquiry on gendered exit patterns, and-more importantly-from a theoretical grounding to understand the factors that drive such differences.More generally, additional research on the exit decisions of both male and female entrepreneurs is needed because of the importance of including gender as an explanatory variable in entrepreneurship research (Jennings and Brush, 2013;Verheul et al., 2005), the increasing scholarly interest in entrepreneurial exit (Bates, 2005;Dehlen et al., 2014;Fackler et al., 2013;Gimeno et al., 1997), and the importance of entrepreneurial exit to the founder, the firm, the industry, and the global economy (DeTienne, 2010).
Motivated by the above theoretical gaps, and drawing upon the exit literature and feminist theories, we hypothesize a gender effect in the comparison of exit for business failure (involuntary exit) and voluntary exit.Then, we move to an even more fine-grained conceptualization of exit motivations to uncover the rationales that underlie two distinct types of voluntary exit (exit for personal reasons and exit for other professional/financial opportunities).We test our hypotheses on a sample of 219 Spanish entrepreneurs who had exited their business in the previous 12 months and reveal important entrepreneurial exit-related differences across gender.For instance, we find that females are more likely than males to exit their business voluntarily than for business failure.We also find that females are more likely than their male counterparts to exit for personal reasons than for business failure.
On the one hand, these results reinforce the work of Taylor (1999) in that our empirical tests demonstrate that the motivation for exit varies between male and female entrepreneurs.On the other hand, we move beyond extant studies in many important ways, which allow us to make several valuable contributions to entrepreneurship literature.More precisely, we extend Taylor's (1999) work because (1) we draw upon theoretical perspectives to anticipate gender differences and develop testable hypotheses, (2) we draw on entrepreneurship literature to develop an inclusive conceptualization of both involuntary exit (including poor performance as well as bankruptcy) and voluntary exit (including personal reasons and divestment as well as (re)entering paid employment), (3) we use multivariate analyses to control for potential confounding associated with gender, and (4) we escalate our results (post-hoc analyses) by drawing upon Social Constructionist Feminism Theory to further examine the heterogeneity among female and male entrepreneurs (de Bruin et al., 2006;Hughes and Jennings, 2012;Hughes et al., 2012).Specifically, in response to calls for greater attention to the inclusion of family embeddedness in entrepreneurship and female entrepreneurship research (Aldrich and Cliff, 2003;Brush et al., 2014;Gras and Nason, 2014), we investigate the effect of two family-context characteristics-being in a spousal relationship, and number of children-on within-gender differences in exit patterns.We find heterogeneity between familycontext variables and the two voluntary motivations for exit.
Our study makes several main contributions to different streams of literature.We contribute to the gender and exit/failure debate by comparing exit for failure to voluntary exit, and also by using a more fine-grained conceptualization of voluntary exit to explicitly investigate differences in entrepreneurial exit patterns across (and within) gender.This research is the first in the entrepreneurship field (that we are aware of) that empirically demonstrates, and theoretically explains, how females' individual volition and decisionmaking autonomy impact firm exit.We add to existing research on female entrepreneurship by illustrating that high female exit rates may be partially explained by a personal choice to leave the firm rather than by being forced to close due to poor performance.This challenges the predominant view that female entrepreneurs do not "measure up" to their male peers (cf.Ahl, 2006;Jennings and Brush, 2013) and provides a critical explanation for presumed high failure rates among female-owned ventures.These findings also support early entrepreneurial exit research-which has argued for a clear delineation between voluntary exit and failure-as we show empirical and theoretical differences between both events.This suggests that previous research may have greatly overstated failure rates.Also, by delineating between two types of voluntary exit-exit for personal reasons and exit for other professional/financial opportunities-both in our theorizing and in our empirical analyses, we move beyond the basic failure versus voluntary exit distinction.
Moreover, we add to the understanding of both females and males as heterogeneous groups with differing roles, motivations, and options (Hughes et al., 2012) by enriching current explanations of the gender-exit relationship.The finding that two contingency factors related to their family context impact males' and females' exit motivations differently is important because our knowledge of the factors that influence each type of exit remains rather limited.In fact, although literature recognizes the importance of contextual factors in determining entrepreneurial activity (Hughes et al., 2012;Welter, 2010), with few exceptions (Kalnins and Williams, 2014), their specific effect on the decision to exit a business has rarely been explicitly tested or stated.

Entrepreneurial exit
Early entrepreneurship research tended to assume that exit was the result of poor performance (Boden and Nucci, 2000;Caves, 1998); thus, equating firm survival with entrepreneurial success.However, recent studies have shown that entrepreneurs can also make volitional decisions to leave their firm, or exit the firm from the market altogether, for a multitude of non-pecuniary reasons (DeTienne and Cardon, 2012;Mayer and Goldstein, 1961;McGrath, 2006;Ronstadt, 1986;Sarasvathy et al., 2013;Taylor, 1999).As outlined below, the rationales given for entrepreneurial exit appear to cluster around two main categories that cover three specific areas: 1) involuntary exit due to poor firm performance (failure), and 2) voluntary exit that includes a) exit for personal reasons and b) exit for other professional/financial opportunities.
Involuntary exit due to poor performance is well addressed in the current literature (see Ucbasaran et al., 2013 for a recent review), and has been defined and operationalized in many ways.Bankruptcy (e.g., Balcaen et al., 2012;Taylor, 1999) is a narrow definition of failure that counts only those firms that have completed and recorded the bankruptcy event (Ucbasaran et al., 2013).Failure (Lussier, 1995;Zacharakis et al., 1999) is a broader term which includes "the cessation of involvement in a venture because it has not met a minimum threshold for economic viability as stipulated by the entrepreneur" (Ucbasaran et al., 2013, p. 175).This broader definition allows us to account for firms in which the exits are due to poor performance, and we therefore refer to these exits as "failures." In the voluntary exit category, researchers have noted a significant number of successful business closures (Bates, 2005;Wennberg et al., 2010).Headd (2003) states: "It appears that many owners may have executed a planned exit strategy, closed a business without excess debt, sold a viable business, or retired from the work force" (p.51).Taylor (1999) uses the term "voluntary terminations" to refer to those "who exit self-employment to (re)enter paid employment" (p.C152).3However, based upon the following extant research, we separate voluntary closures into two areas: exit for personal reasons and exit for other professional/financial opportunities.Exit for personal reasons includes retirement (Harhoff et al., 1998), family and family structure issues (e.g., marriage, divorce, childcare concerns, aging parent care, empty nest syndrome) (Marlow and Swail, 2014), health issues (Harada, 2007), or a change in motivation (Van Praag, 2003).We refer to these exits as voluntary because they are not dependent upon the firm's financial performance. 4ates (2005) contends that another major issue underlying the decision to exit a successful business is "the availability of a more appealing alternative or alternatives" (p.356).Others (e.g., Harada, 2007;Taylor, 1999) found that while a small proportion of selfemployment terminations were due to bankruptcy, a higher percentage of firms closed because the entrepreneur elected to go on to a better or different professional opportunity such as taking a new job or starting a new venture.Also, some entrepreneurs may simply choose to leave a firm rather than play a managerial role (Boeker and Karichalil, 2002;Hall, 1994), while others do so to recapture their initial investment (Watson and Everett, 1996).We refer to these events as exit for other professional/financial opportunities.This conceptual distinction of voluntary exit into two categories is useful to differentiate between behaviors that are driven by non-work related motives (such as those related to family) and motives that are of a professional nature (such as the opportunity costs of business ownership for an individual's career).This, in turn, leverages and integrates one of the main contributions of female entrepreneurship research to the broader literature: the conceptualization of entrepreneurial outcomes in terms of both economic and non-economic indicators (Jennings and Brush, 2013).

Gender differences in exit patterns: failure versus voluntary exit
Several studies have found gender differences in firm survival (e.g., Boden and Nucci, 2000;Bosma et al., 2004;Fairlie and Robb, 2009).With a few notable exceptions (Kalleberg and Leicht, 1991;Kalnins and Williams, 2014), it appears that when examining aggregate comparisons of male-owned and female-owned ventures, male-owned ventures are more likely to survive.For example, in a study of the U.S. Census data, Boden and Nucci (2000) found the mean survival rates of male-owned ventures to be 4-6% higher.In their Dutch sample, Bosma et al. (2004) found male-owned ventures to survive longer and Fairlie and Robb (2009) found femaleowned businesses to have higher closure rates (24.4% vs. 21.6%).
There are many reasons given for why female-owned ventures are more likely to fail.These include the lack of access to resources such as funding (Fairlie and Robb, 2009), lack of prior managerial and employment experience (Boden and Nucci, 2000;DeTienne and Chandler, 2007), and lack of founding strategy (Carter et al., 1997).These findings tend to focus upon "solutions" which include training and education for females to "bring them up to speed."However, more recent research has moved beyond the female detriment approach (e.g., Hughes et al., 2012;Scott et al., 2012) and has begun to focus more upon feminist perspectives (Marlow and Swail, 2014) and on examining how gender socialization (Brush et al., 2014) and social construction influence female entrepreneurs' efforts and choices (Ahl, 2006;Ahl and Marlow, 2012).This suggests that gender expectations and socio-economic implications (such as those we examine in this manuscript) partially explain "failure" in female-owned ventures.
While lacking a specific gender theorizing, Taylor's (1999) work in the field of economics provides initial evidence that focusing on survival rates alone might not provide a complete picture of females' exit patterns.His data suggests that males in Britain are more likely than females to exit a venture for alternative employment, bankruptcy, and retirement while females are more likely to exit for other specified and unspecified reasons which include health issues, family care, moving home, and full-time education.To better understand these patterns, we turn to feminist theories in sociology (Chafetz, 1997)-and in particular, Social Feminist Theory (SFT) (Chodorow, 1978;Gilligan, 1982)-which offer compelling arguments for anticipating gender differences in exit motives.Social Feminist Theory posits that differences in socialization processes and experiential backgrounds cause males and females to differ in their values and way of thinking (Black, 1989;Calas and Smircich, 1992).It suggests that female traits, though equally as valuable as those of men, may not be readily captured by analytical frameworks developed for males (Acker, 1978).In the context of business ownership, SFT indicates that the decision to engage in entrepreneurship is much more complex for females than for males.Instead of considering it as a mere tool for profit generation, females take a holistic view of business ownership in which financial objectives cannot be separated from family and societal ones (Brush, 1992).
Female entrepreneurs' greater emphasis on a variety of nonmonetary issues (Bird and Brush, 2002;Burke et al., 2002) also gives rise to differences in entrepreneurial outcomes (Brush, 1992;Fischer et al., 1993).For example, Cliff (1998) shows that because female entrepreneurs are more concerned about the quality of interpersonal relationships as a measure of business success than about quantitative indicators (such as size), they attach a lower value to business expansion than males.Similarly, Demartino et al. (2006) found that male entrepreneurs demonstrate a higher-than-average career/achievement orientation, whereas females do not.Accordingly, gender differences exist in the value placed on-and the goals pursued from-business ownership (Carter et al., 2003), which will likely contribute to distinctions in the exit motives of male and female entrepreneurs.More specifically, we suggest in this paper that because female entrepreneurs attach a lower relative importance to financial success as compared to other outcomes, their exit from business ownership will also be more often related to non-financial motives-and hence to voluntary exit-than that of males.
Social feminism also has implications on the psychological side of firm ownership (cf.Pierce et al., 2003).On a general level, independent of gender, the extent to which firm ownership affects the attitudes and behaviors of the owner depends on the strength of ownership feelings, labeled psychological ownership (PO) (cf.Pierce and Furo, 1990;Pierce et al., 1991Pierce et al., , (2003))) define PO as "the state in which individuals feel as though the target of ownership or a piece of it is 'theirs'" (p.86).Ownership feelings are associated with enhanced affective commitment toward the ownership target (Avey et al., 2009;Mayhew et al., 2007;Pendleton et al., 1998) and with a stronger intention to stay with it (Avey et al., 2009;Pierce et al., 2003;Wagner et al., 2003).This is because for entrepreneurs with high PO toward their venture, the venture's loss would imply "shrinkage of [their] personality, a partial conversion of [them]selves to nothingness" (James, 1890, p. 178).Hence, when PO is high, entrepreneurs will try to avoid exit unless forced to do so due to financial distress.
Drawing on SFT, we posit that the level of PO that female entrepreneurs have toward their firms is lower than that of males for two main reasons.First, while males tend to identify more strongly with their profession and organization (e.g., Johnson et al., 2006) and with work roles such as entrepreneur (Bruni et al., 2004;Gupta et al., 2009), females are more likely to identify with aspects that are unrelated to their profession (e.g., Essers and Benschop, 2007;Gutek et al., 1991).Hence, we deduce that female's self-identities are less likely to be linked to the firm.Psychological ownership, in turn, is partly grounded in and driven by self-identity (Dittmar, 1992;McCracken, 1986;Pierce et al., 2003); it follows logically that females' PO toward their firm will likely be lower.Second, female entrepreneurs and managers are more inclined to relinquish control and use a democratic managerial style (Buttner, 2001;Eagly and Johnson, 1990;Fondas, 1997) where control is distributed among several individuals.Having control over the target of ownership feelings, however, is one of the main antecedents to PO (Pierce et al., 2003).This also leads us to believe that female entrepreneurs may have lower levels of PO toward their business than male entrepreneurs, and will thus be more likely to exit voluntarily.
Our arguments also find support in recent literature drawing on gender role theory and arguing that males and females differ in the way they manage the interconnections between work and family.Jennings and McDougald (2007) suggest that male entrepreneurs are particularly prone to adopt work-family interface (WFI) strategies aimed at facilitating business growth.Female entrepreneurs tend instead to enact WFI strategies that constrain venture growth and development.Early empirical tests of this framework indicate that, by assuming a disproportionate share of their households' tasks, females engage to a greater extent than males in constraining couple-level strategies (Jennings et al., 2010).Accordingly, we posit that such WFI strategies will also likely play an influential role in shaping other entrepreneurial outcomes such as exit. 5At the couple level, for example, females might decide to prioritize their husbands' careers to the point of disengaging from a potentially successful career as an entrepreneur.Building on this prior evidence in social feminism, WFI and PO literatures, we propose the following baseline hypothesis: H1. Gender impacts exit motives such that female entrepreneurs, compared to male entrepreneurs, are more likely to exit voluntarily than for business failure (involuntary exit).
3.3.Gender differences in voluntary exit patterns: Exit for personal reasons and exit for other financial/professional opportunities versus exit for failure While female entrepreneurs are, on average, more likely to exit voluntarily than their male peers, exit literature highlights that not all voluntary exits are driven by the same factors.Indeed, some instances of voluntary exit can reflect the opportunity costs of business ownership for the professional career of the entrepreneur (e.g., Gimeno et al., 1997).Other cases illustrate the challenges that entrepreneurship poses for the entrepreneur in her personal life (e.g., Marlow and Swail, 2014).For example, Taylor's (1999) descriptive statistics on self-employment duration in Britain show that 19.8% of the females in his sample exited their business for family care, and 13.4% mentioned reasons such as full-time education and moving home.In what follows, we delve more deeply into these different types of voluntary exits and rely on theory to explain how each one reflects important gender differences in entrepreneurial exit behavior.
First, as previously mentioned, many studies drawing upon feminist theories demonstrate that females and males have differing goals and orientations towards entrepreneurship (Bird and Brush, 2002;Burke et al., 2002).Female entrepreneurs are indeed more likely to possess lower growth aspirations and to pursue non-economic goals (Carter et al., 2003;Cliff, 1998) such as balancing work and family roles (Boden, 1999;Bonet et al., 2013;Caputo and Dolinsky, 1998), pursuing personal enjoyment (Brush, 1992), and caring for employees or solving societal issues (Hechavarria et al., 2012;Jennings and Brush, 2013).We argue that these noneconomic goals-which have been proven as important motives for females' entry into business-also weigh in their subsequent voluntary decisions to exit their business.For example, if socioeconomic status is left out of the decision (or is controlled for, as we do in this research), and the personal factor of needing to care for an elderly parent arose, females may be more likely than males to quit self-employment to provide that care.Similarly, female entrepreneurs' emphasis on personal relationships would make them more prone than their male peers to voluntarily leave a financially sustainable business in the face of enduring conflicts with co-founders.It follows that female (compared to male) entrepreneurs will be more likely to invoke personal factors as a voluntary motive for exit compared to exit for failure.
H2a. Gender impacts exit motives such that female entrepreneurs, compared to male entrepreneurs, are more likely to exit for personal reasons than for business failure.
Research also indicates that females are much more likely than males to be "pushed" into entrepreneurship by necessity-driven circumstances such as gender discrimination in the traditional labor market (Buttner and Moore, 1997;Maniero and Sullivan, 2006) and escaping the glass ceiling (Buttner and Moore, 1997;Mattis, 2004).This literature suggests that, for the average female entrepreneur, entrepreneurship represents a less desirable opportunity than a full-time career in the wage sector, as the latter often includes financial stability, health insurance, sick leave, and retirement (Bonet et al., 2013).Such a preference is reinforced by the fact that, in comparison to males, females extract lower net earnings from entrepreneurship (Caliendo et al., 2010;Hundley, 2001) and that the gender differential in earnings is greater among the self-employed than among wage and salary workers (Clain, 2000;Sexton and Robinson, 1989).Hence, we hypothesize that female entrepreneurs will be more likely than males to exit from a current venture to pursue an attractive job opportunity (compared to exiting for failure).
The financial and personal risk-taking that is often associated with entrepreneurship also contributes to a female's likelihood to exit when an attractive job or the opportunity to sell the business at a profit becomes available.Indeed, risk attitudes play a significant role in the decisions to become and remain self-employed (Caliendo et al., 2009(Caliendo et al., , 2010;;Raffiee and Feng, 2014) as entrepreneurs' earnings are much more volatile than those of employees with similar characteristics (Borjas and Bronars, 1989;Heaton and Lucas, 2000).Literature reports that females are, on average, more risk-averse than males (Croson and Gneezy, 2009;Dohmen et al., 2011), which should not only refrain them from self-selecting into entrepreneurial positions (Caliendo et al., 2009), but also from remaining in those positions over time.They are also socially penalized for displaying behaviors that are considered non-feminine, which includes taking risks (Heilman, 2001) and launching a business (Thébaud, 2010).These gender differences in risk attitudes hold even after individuals become business owners (Sexton and Bowman-Upton, 1990), and have been recently linked to female entrepreneurs' higher exit rates in Germany (Fossen, 2012).These arguments lead us to propose: H2b. Gender impacts exit motives such that female entrepreneurs, compared to male entrepreneurs, are more likely to exit for other professional/financial opportunities than for business failure.

Sample
To examine these questions, our study uses data drawn from the Spanish GEM (Global Entrepreneurship Monitor) data set, which tracks entrepreneurs based on a representative telephone survey of the adult population.Using a probability sample, interviewers at Opinometre, the survey vendor in charge of collecting data for the Spanish GEM study, screened the telephone numbers of a total of 27,880 households in the months of January 2007 through April 2007 to ascertain whether the respondent was at that time an entrepreneur or former entrepreneur that had closed or exited a business during the year preceding the survey (the detailed questionnaire can be found in Reynolds et al. (2005)).We identified a total of 276 respondents as former entrepreneurs who had exited their businesses in the previous year.A follow-up survey constructed for the present study to collect additional information was sent to these former entrepreneurs during the summer of 2007, resulting in 219 usable answers. 6Ninety-nine of the respondents (45.2%) are female.

Variables Dependent variable: Exit motive
In order to capture the exit patterns of entrepreneurs, we used a set of items eliciting the reasons for business exit, allowing us to separate performance-laden reasons (failure or involuntary exit) from reasons stemming from personal issues or related to other professional, financial, and career considerations (voluntary exit).Specifically, former entrepreneurs were asked the following question: "What was the most important reason for quitting this business?"They had the possibility of choosing between seven answers that were consistent with previous research on owner-managers' exit (Winter et al., 2004): 1) the business was not profitable, 2) problems getting finance, 3) early retirement 7 or illness, 4) personal reasons, 5) an opportunity to sell the business, 6) found another job opportunity, and 7) other.Accordingly, to test Hypothesis 1 about failure versus voluntary exit, we created a dummy variable that was coded 0 to indicate exit for business failure, meaning if the entrepreneurs declared they were forced to exit the business for performance reasons by choosing answers 1 or 2. This variable was coded 1 to indicate voluntary exit when entrepreneurs chose answers 3, 4, 5, or 6.To test Hypothesis 2, we created a categorical variable coded 0 in case of business failure (answers 1 and 2), coded 1 in case of exit for personal reasons (answers 3 and 4), and coded 2 in case of exit for other professional/financial opportunities (answers 5 and 6).Respondents choosing the Other category were asked to describe the specific reason for exit and cases were re-classified according to the detailed explanation provided (for example, family issues = personal reasons and lack of clients = business failure).
To test the validity of our three-categorical variable, we performed the Hausman specification test for mlogit (Hausman and McFadden, 1984) that allows testing the reverse implication of the independence of irrelevant alternatives (IIA) assumption.Eliminating one alternative from the choice set (exit for other professional/financial opportunities) and estimating the coefficients from both the restricted and the unrestricted choice sets shows that we clearly fail to reject the null hypothesis (p N 0.9).Given that the coefficients hence are approximately the same for both models, the IIA assumption is not violated, and the variable seems to be correctly specified.

Independent variable
Respondents were asked to indicate their gender.The corresponding dummy variable was coded 1 for females and 0 for males.

Control variables
To improve the robustness of our findings, we controlled for the age of the entrepreneur (Taylor, 1999) and the entrepreneur's industry experience at the time of exit (measured in the number of years of previous experience they had had in the same industry, and logged to avoid skewness).Recent research suggests that both factors influence the degree of psychological attachment to the business (Gimeno et al., 1997) as well as exit paths (DeTienne and Cardon, 2012;Wennberg et al., 2010).Similarly, we controlled for higher education with a dummy variable (1 for entrepreneurs with a graduate degree or higher, 0 otherwise) given its impact on voluntary exit decisions (Bates, 2005;Taylor, 1999).Since resource availability likely affects the likelihood of business failure (Boyle and Desai, 1991), the entrepreneur's socioeconomic status (SES) was captured using a basic measure of household income calculated by GEM.Dummy variables controlled for when the respondent's household was in the lower or upper third of the income distribution (the reference category being the middle third).Research has also found that SES affects the relationship between family structure and females' occupational activity (Archbold, 1983;Zuravin, 1987). 8lso, as professional status-professional and managerial on one hand, and non-professional, non-managerial on the other hand-has long been one of the major sources of social and gendered division in the labor market (Budig, 2006;Warren, 2010), we account for this fact with a dummy variable based on ISIC sector classification.Following Budig (2006), we distinguished between entrepreneurs who had been engaged in "professional occupations" such as those related to consulting, law, advertising, or engineering (coded 0) and who had ventured in "non-professional occupations" (coded 1), such as those related to retail, restaurants, or hairdressing.Finally, we introduced several business characteristics that have been related with exit patterns in the entrepreneurship literature (Gimeno et al., 1997;Wennberg et al., 2010).Specifically, we included business age (log of number of months of existence) and business size (log of number of employees).
As our dependent variable to assess Hypothesis 1 is binary in nature, we chose binary logistic regressions as analytical procedure.We use a three-dimensional categorical variable to test Hypothesis 2 and thus conducted a series of multinomial logistic regressions-an appropriate method for this kind of research question and data (Wennberg et al., 2010;Zellweger et al., 2011). 9ere, the effects of the independent variables on each of the outcomes were compared to a base category.We used failure as the comparison baseline since our paper focuses on uncovering the determinants of alternative exit motives.

Descriptives and hypothesis testing
Descriptive statistics and correlations for the variables are displayed in Table 1.The correlations of our gender variable are all below 0.18 in magnitude, and the variance inflation factors (VIF) did not exceed 1.1, which is below the critical cut-off threshold of 10 (Hair et al., 2006).This indicates that multicollinearity is not a concern.Mean values for the three exit motives indicate that out of all entrepreneurs exiting business ownership, 48% did so because of business failure (involuntary exit) while 34% exited for personal reasons.The remaining 18% exited because they were attracted by other professional/financial opportunities.Table 1 also indicates that being female was significantly and negatively correlated with exit for business failure and significantly and positively correlated with exit for personal reasons.
Table 2 reports the results of our logistic regression analysis related to voluntary exit versus exit for business failure (Hypothesis 1).To be better able to assess and compare the magnitudes of our effects, we report the results not only as coefficients, but also as odds ratios (cf.Autio et al., 2013).Odds ratios greater than 1 indicate a positive association between the independent and the dependent variable, and odds ratios of less than 1 indicate a negative association.In Model 1, we only included our control variables, whereby having a higher education showed a positive association with voluntary exit (p b 0.05).In Model 2, we added our gender variable and found a strongly positive and significant association with voluntary exit (coef.= 0.675, OR = 1.965, p b 0.05).10This offers support for Hypothesis 1.To assess incremental model fit, we conducted a pairwise likelihood ratio (LR) test and found that adding the gender variable significantly improved the fit of Model 2 compared to Model 1 (p b 0.05).
In Table 3, we show the results of our multinomial regression analyses to determine how gender influenced different exit motives (Hypotheses 2a/b).Here, we chose to report coefficients as well as relative risk ratios (RRR) to describe the relationships between our variables and to provide meaningful indicators of the magnitudes of our effects.This is the recommended best practice for multinomial logistic regression (cf.Osborne, 2008).An RRR smaller than 1 indicates a negative relationship between two variables, and an RRR larger than 1 shows a positive relationship.In Models 3 and 4, our dependent variable is the occurrence of exit for personal reasons (1) as opposed to failure (0).In Models 5 and 6, the dependent variable is the occurrence of exit for other professional/financial opportunities (2) as opposed to failure (0).As a first step we entered the control variables (Models 3 and 5).Then, we tested the effect of gender as an independent variable as a second step (Models 4 and 6).Being female has a positive and significant relationship with the likelihood of exit for personal reasons in Model 4 (coef.= 1.038,RRR = 2.822, p b 0.01), which confirms Hypothesis 2a. 11The fit of Model 4 is significantly better than the fit of Model 3. A significant relationship between being female and the likelihood of exit for other professional/financial opportunities, however, could not be found (p N 0.05).Hence, we reject Hypothesis 2b.

Further analyses and robustness checks
To provide a richer understanding of our relationships and to confirm the robustness of our findings, we performed additional analyses following Long and Freese (2005).We used the STATA-SPost package to estimate the change in predicted probabilities of the two and three exit motives at different values of the gender variable.We held all other variables at their mean (if continuous) or at 0 (when binary).
Referring to our binary dependent variable, the results show that when gender changed from male (0) to female (1), the predicted probability of voluntary exit increased by 17%.This further supports Hypothesis 1. Testing our dependent variable with three categories reveals that the probability of exit for personal reasons increased by 23% when gender changed to 1 (female), which further supports H2a.To assess the robustness of our findings, we ran several tests.For instance, we estimated our models with different sets of control variables and without any control variables at all (cf.Becker, 2005).Our main results with regard to gender differences in exit patterns remained very stable.

Escalation of results
While our results clearly indicate that female entrepreneurs are, on average, more likely than their male peers to exit voluntarily-and in particular, for personal reasons-gender scholars increasingly caution against making broad generalizations of "essential" gender differences (e.g., Ahl, 2004;Ahl, 2006).Indeed, Social Constructionist Feminism Theory (SCT) portrays gender as something that is "done" rather than something that simply "is" (Fenstermaker and West, 2002;West and Zimmerman, 2009).The theory predicts that expectations embedded in the gender systems of different contexts will differently influence entrepreneurs' behaviors (Ahl, 2004;Ahl, 2006) and will encourage or discourage adherence to gender-appropriate behavior when making businessrelated decisions (Ridgeway, 2011).In other words, examining heterogeneity-especially among female entrepreneurs ( de Bruin et al., 2006;Hughes and Jennings, 2012;Hughes et al., 2012)-is important because some entrepreneurs-depending on their context-might more readily behave in gender-stereotypic ways than others.
In light of this view, the underlying mechanism behind our findings (and non-findings) could be further uncovered by "manipulating" some of the boundary conditions that are known for further accentuating gender differences.In particular, within-gender heterogeneity with regard to entrepreneurial decisions (Kalnins and Williams, 2014;Yang and Aldrich, 2014) is often attributed to family factors (Budig, 2006;Budig and Hodges, 2010;Kifle et al., 2014).As one female entrepreneur noted: "Being a woman per se is probably not as big an issue as being a wife and mother" (Moult and Anderson, 2005, p. 264).Such a statement begs the question of the exit implications of family structure and its potential differing impact across gender.For example, to what extent are family factors really behind females' inclination to exit for personal reasons?Also, are there any family-related explanations behind the apparent absence of gender differences with respect to exit for alternative opportunities?To explore this further, we escalate our findings by examining effects of two commonly-used family structure variables-namely being in a spousal relationship12 and number of children (cf.Hodges and Budig, 2010;Yang and Aldrich, 2014) on the likelihood of different exit options for female and male entrepreneurs.
While some studies suggest that the effect of family is positively related to females' likelihood of entering self-employment (Bonet et al., 2013;Greenhaus and Callanan, 1994), others suggest that it also represents a potential constraint on how much time they can devote to the business and the level of earning they can derive from it (Budig, 2006;Hundley, 2000).Motherhood and having a spousal relationship often increase females' domestic responsibilities (Bianchi et al., 2000;Gupta, 1999) and make their gender role more salient (Justo et al., 2007;Yang and Aldrich, 2014).By doing so, they heighten the probability of experiencing work-family conflict (for a review, see Jennings and McDougald, 2007).Work-family conflict, in turn, increases absenteeism (Duxbury and Higgins, 2001), generally detracts from job performance (Graves et al., 2007), and reduces female entrepreneurs' commitment towards business growth (Jennings and McDougald, 2007).Extending this rationale to the case of entrepreneurial exit, it follows that a spousal relationship and motherhood increase female entrepreneurs' likelihood of exiting the business for personal reasons due to more severe work-family conflict compared to exit for failure.
Males, in turn, have been found to be more likely than females to scale back their psychological commitment toward the family role in response to work-family conflict (Jennings and McDougald, 2007).For example, studies indicate that the time male entrepreneurs devote to housework does not change, or even declines, after marrying (Bianchi et al., 2000;Gupta, 1999).In the contemporary understanding of family, females are cast as "homemakers" while males are cast as "breadwinners" (Gorman, 1999).This cultural norm has proven resilient to social and economic changes (Cha and Thébaud, 2009;Ridgeway, 2011), and its importance is even triggered by a spousal relationship or fatherhood as both increase the salience of gender as a social framing device (Ridgeway, 2011;Yang and Aldrich, 2014).The breadwinner schema implies that male entrepreneurs who marry, or become fathers, will tend to satisfy their strongest responsibility to their family by further investing in work in order to provide increased economic support.It follows that marriage and number of children would positively affect male entrepreneur's likelihood to exit their business for more lucrative professional or financial opportunities compared to exit for failure.A recent entrepreneurship study in fact claims that "fatherhood reinforces men's responsibility for financially supporting their families and compels them to prioritize wage jobs" (Yang and Aldrich, 2014, p. 8).This is because fatherhood often brings a salary premium in wage employment (Hodges and Budig, 2010;Killewald, 2013), while business ownership might make it more difficult to consistently fulfill their breadwinner role.As one entrepreneur puts it: "For me, probably the largest [reason for exit] was the financial security because as a private company, you feel like you're at the poker table, and every hand you put all the chips back in.I mean, every year, you know, it's all in, all in" (Brad Florin, founder of Acartus Software, July 2014). 13n summary, were the preceding arguments correct, we should expect spousal relationship and number of children to impact entrepreneurs' exit motives such that a) females who are married or have more children are more likely than the rest of females to exit for personal reasons than for business failure, and b) males who are married or have more children are more likely than the rest of males to exit for other professional/financial opportunities than for business failure.
We test these post-hoc arguments by performing additional multinomial logistic regressions for males and females using two new moderators: spousal relationship (coded 1 for married or cohabiting with a partner and coded 0 for single, divorced, or widowed) and the number of children living in the entrepreneur's home at the time of exiting the business (Table 4).
We estimate these interactions through separate equations for each group as this is most appropriate for logistic models (Hoetker, 2007) and is in line with previous research analyzing gender differences in entrepreneurial outcomes (e.g., Davis and Shaver, 2012).The effects of our moderators on the likelihood of exit for personal reasons (versus failure) were tested for the subsamples of females (Model 7) and males (Model 9).The effects on the likelihood to exit for other professional/financial opportunities (versus failure) were tested in Model 8 (females) and Model 10 (males).The results lend mixed support for our post-hoc arguments.Spousal relationship has a positive and significant impact on females' likelihood to exit for personal reasons (coef.= 1.172,RRR = 3.227, p b 0.05) as well on males' likelihood to exit for other professional/financial opportunities (coef.= 2.404, RRR = 11.064,p b 0.05).Number of children, however, is positively, but insignificantly, related to females' likelihood to exit for personal reasons (coef.= 0.125, RRR = 1.133, p N 0.05).Interestingly, and in contrast to our expectations, it is negatively and significantly related to males' likelihood to exit for other professional/financial opportunities (coef.= −0.541,RRR = 0.582, p b 0.05).

Discussion
Our in-depth reassessment of the female underperformance hypothesis leads to numerous valuable insights that enrich different streams of literature.In the following, we discuss the corresponding findings and contributions separately.

Reconceptualizing "failure" in female-owned ventures
Much of the previous research has concluded that female entrepreneurs are subject to the female underperformance hypothesis and are therefore more likely to fail.However, our examination of 219 former Spanish entrepreneurs who had recently exited their business finds that females are more likely to leave their venture based on voluntary decision rather than being forced out or "failing."Thus, it is important to include both voluntary and involuntary outcome variables in analyses of gender differences in entrepreneurial exit.Our findings align with a body of previous gender research which suggests that some of the underperformance findings are due to methodological and study issues such as a lack of theoretical grounding; the neglect of structural, historical, cultural, firm size and geographical factors; and the use of measures developed for male respondents (Ahl, 2006;Du Rietz and Henrekson, 2000;Kalnins and Williams, 2014;Marlow and Swail, 2015;Zolin et al., 2013).It also supports recent exit research arguing that failure and exit are two fundamentally distinct concepts (cf.Wennberg et al., 2010) and enriches it by showing that there are distinct patterns when comparing voluntary exit with failure from a gender perspective.
We enrich entrepreneurial exit literature by moving beyond this dichotomous categorization of exit.Both our theorizing and our empirical findings show the appropriateness of using the categories of exit for personal reasons (cf.Marlow and Swail, 2014;Van Praag, 2003) and exit for other professional/financial opportunities (cf.Bates, 2005;Taylor, 1999) and reveal unique gender-related differences when comparing both types of voluntary exit with business failure.For example, we extend (by drawing upon theory to develop testable hypotheses, using inclusive measures of both involuntary and voluntary exit, relying on multivariate analyses, and escalating our results) the initial evidence provided by Taylor's (1999) study of self-employment duration in which he noted that males are more likely than females to exit a venture for alternative employment, bankruptcy, and retirement while females are more likely to exit for other specified and unspecified reasons which include health issues, family care, moving home, and fulltime education.
In doing so, we also add to the emerging research analyzing female entrepreneurs' exit and survival from a social constructionist perspective (Kalnins and Williams, 2014) and provide a more refined understanding of the actual motivations behind exit differences across gender.That is, females' exit motives seem to be guided by personal factors and gender-related expectations to a higher extent than males'.Our theoretical framework, which draws on feminist theories and separates exit from failure, also answers calls for explicitly employing a feminist theorizing as well as avoiding some of the epistemological assumptions underlying gender differences in entrepreneurship (e.g., Ahl and Marlow, 2012;Marlow and McAdam, 2013).
Our findings support the proposition put forth by Marlow and Swail (2015) that the way in which males' and females' life courses are "gendered" will have an impact on females' exit decisions.The "gendering" process may lead to other explanations for the heterogeneity between male and female entrepreneurs in the exit decision.For example, research has demonstrated that escalation of commitment is a risk-taking behavior (e.g., Wong, 2005) and it is well documented that females are more risk-averse than males (see the meta-analysis by Byrnes et al., 1999).In addition (although the research is not conclusive), some studies have shown that females have a greater fear of failure than do males (e.g., McGregor and Elliot, 2005).These perspectives may provide additional robustness to the discussion of why females might choose to exit voluntarily rather than escalate their commitment or risk failure.
Our paper not only challenges the female underperformance hypothesis; it also enriches and qualifies "the paradox of the contented female business owner" (Powell and Eddleston, 2008) which refers to the fact that even though they are less successful (using objective measures of business success), female entrepreneurs are just as satisfied as their male counterparts.In line with this perspective, our theoretical considerations further suggest that females' lower threshold of economic performance might explain not only their higher level of satisfaction, but also their likelihood to out-survive males in some contexts (Kalnins and Williams, 2014).However, our results also point to a "non-contented female paradox": because female entrepreneurs also place higher value on non-economic factors, they are more likely than males to close otherwise financially sustainable businesses that do not meet their intrinsic motivations.The following statement of a female entrepreneur provides an illustrative example if this type of situations: "One of the things that happens in becoming a public company […] is that the focus changes to earnings per share and bottom line and away from maybe what's right to do or putting some money away for a rainy day or doing what's right for people.It was not the same organization that I had joined, nurtured, or cared about, and I think that makes a huge difference.You don't want to work there anymore" (Sue Schell, founder Advanced Energy, 2014).
As a whole, our in-depth, gender-based theorizing goes beyond earlier studies that have taken a rather economics-oriented approach when comparing voluntary to involuntary exit.For instance, those studies have focused on identifying "fit" firms that have been exited voluntarily and on distinguishing them from outright failures that are regarded as being "of more concern for policymakers" (cf.Taylor, 1999, p. C144).While acknowledging that the fittest do not always survive (Gimeno et al., 1997), we also use a more fine-grained and systematic categorization and classification of exit types.In line with recent research (e.g., Balcaen et al., 2012), we account for a broader set of failure reasons beyond pure bankruptcy, such as that the business was not profitable or other financial reasons.Finally, we account for the option to sell the business and delve much deeper into gender-related exit patterns by introducing additional moderating factors.Due to space reasons, control models are not reported.LR tests of model fit performed between control models and full models are always significant at p = 0.05.AIC in full models is always smaller than in the corresponding control models, which denotes improved model fit.
7.2.The heterogeneity of entrepreneurs (both male and female) and exit Our findings demonstrate that entrepreneurs of both genders make individual and volitional decisions to exit their ventures, suggesting that overall failure rates may be overstated.However, we go beyond that simple statement; our post-hoc escalations also indicate that within entrepreneurs of the same sex, being in a spousal relationship and number of children differently contribute to the exit decision.This speaks to recent research emphasizing the importance of examining heterogeneity not only among female entrepreneurs (de Bruin et al., 2006;Hughes and Jennings, 2012;Hughes et al., 2012), but also among males (Davis and Shaver, 2012;Jennings and McDougald, 2007).
The finding that marriage increases the likelihood of males exiting for "other professional/financial opportunities" and drives females to exit for "personal reasons" is consistent with social constructionism and gender role theory, in particular the idea that entrepreneurs' exit patterns are shaped by their gendered roles in spousal relationships-that is, females as "caretakers" and males as "family providers" (Blair-Loy, 2009;Ridgeway, 2011).There is, however, another plausible interpretation of the results for the male subsample.In line with social constructionist claims that people generally tend to comply with gender-prescriptive stereotypes, we could also speculate that married male entrepreneurs whose firms are in financial distress will try to avoid indicating outright failure because this would be perceived as a failure at both the professional and the family level.Instead, they would prefer to invoke exit for other professional/financial opportunities.Certainly, future research would benefit from clarifying which of the two dynamics is driving such exit pattern among males.
Our findings related to number of children are different from those initially conceptualized based on SCT.While our results were non-significant for mothers, they suggest that fathers were less likely to exit for other professional/financial opportunities.The absence of an effect for mothers could be attributed to our operationalization of the "number of children" variable, as it might not capture the whole complexity of family living arrangements and parents-child relationships.Indeed, it is possible that descendants keep living in their parents' house long after childhood (see for example Hank, 2007;Holdsworth, 2004;Manacorda and Moretti, 2006).While evidence of the potential hurdle small children represent for females' professional careers is consistent (Jennings and McDougald, 2007), the impact of young adults is less clear-cut.Older children may be potentially available as household and small business help (Aldrich and Cliff, 2003), and it was found indeed that children could facilitate business growth intentions for females (Davis and Shaver, 2012).Consequently, with regard to entrepreneurial behavior, both the children's and the mother's age might also matter (Davis and Shaver, 2012).Explanations drawing on life course theory (Elder, 1994) could be helpful in providing potential insights on how these relationships might play out.The same rationale could also explain the negative impact of number of children on males' likelihood to exit for other professional/financial opportunities.If their children are old enough to be involved in the business, entrepreneurs might perceive the business as a family legacy and source of jobs for family members.In this case, male entrepreneurs' role as "family providers" will be embodied in their ability to sustain the business and they would be less likely to exit voluntarily.Recent findings about the reluctance of family businesses to close down seem to lend support to this supposition (cf.Gomez-Mejia et al., 2007).Our findings that family issues are important to males when making a decision about exiting for other professional/financial opportunities is also supported by the work of Saridakis et al. (2014) who find that fertility rates have a negative effect on male self-employment.Although their study examines entry into self-employment, and ours exit from entrepreneurial ventures, both of us find that number of children has a negative effect on the decision by males to move to a different career state.It is not surprising that more family responsibility is likely to lead to more risk-adverse behavior; indeed, Dahl et al.'s (2012) study in the field of lifecycle psychology and sociology of the family confirms that the transition to fatherhood has a deep impact on managers' values, and that it will make the financial imperative more salient.However, our unexpected results are interesting for entrepreneurship research because they point to the importance of examining both genders when addressing family issues.As Marlow and Swail (2015) so aptly argue, "It is evident that women are overwhelmingly positioned as the enactment of the gendered subject within entrepreneurial research.As such, it appears that only women engage in gendered behaviours, whereas men occupy the 'default' status" (p.49).That is, both males and females are important to the study of the gendered nature of work-family interface in entrepreneurship.
Gender role theory and expectations stemming from gender stereotypes also provide alternative interpretations to our results.It could be that "nurturing" males might feel "vulnerable to censure for taking on major responsibilities for core caregiving activities that have been culturally essentialized as feminine" (Ridgeway, 2011, p. 130).This is all more the case for fathers given that parents are more likely to internalize traditional gender roles than childless people (Coontz, 2008;Dahl et al., 2012).Consequently, we can indeed speculate that they would consider it more socially acceptable to attribute exit to some objective external factor (i.e., difficulties raising additional funds) and not to "subjective" personal reasons, which are still proscribed in males by society at large.

Limitations and future research directions
As with most research ours is subject to potential limitations, which at the same time may open up avenues for promising future research.Our unique data set provides many advantages; however, we use a single country sample, which may limit generalizability and certainly indicates a need to further test our hypotheses.Despite that, we provide three arguments which lead us to believe that our results will be replicable in other countries and contexts.First, research has shown that stable male-female differences exist across countries (Langowitz and Minniti, 2007;Minniti, 2009).Second, Spain's degree of "masculinity" is relatively modest compared to most Western and developing countries (Hofstede, 2001); hence, using a Spanish sample is a conservative approach and may actually understate results found in countries with a higher degree of masculinity.Also, we note that Taylor's (1999) descriptive statistics from Britain are broadly consistent with our findings from Spain.Third, our results in the Spanish sample are very robust across different model specifications with different sets of control variables and without any control variables (cf.Becker, 2005).Nevertheless, we call for future research that replicates our study in other countries; in general, studies on female entrepreneurship should ideally include the influence of cultural and/or institutional variables (Ahl and Nelson, 2010).
In addition, we rely on self-reported measures of both key independent and dependent variables; thus, we cannot discount the potential of common method bias.We used several different approaches to help mitigate this concern.First, we collected the data at two separate time intervals.The initial data collection was conducted for the GEM study and the second data collection was done as a follow-up.Second, we minimized the time between exit and the survey (within one year), which should alleviate common method and also recall bias (Huber and Power, 1985).Finally, we followed an established protocol by using reverse-coded items and intermixing items.We also note that, notwithstanding the controls introduced in our model, our results might still be subject to unobserved heterogeneity.Thus, our findings cannot be interpreted in a strictly causal way, and could be addressed by studies relying on experimental or quasi experimental methods such as Propensity Score Matching (Caliendo and Kopeinig, 2008), which have been successfully used in entrepreneurship research (e.g., Caliendo and Künn, 2011).
We also acknowledge that the data collected in this study reflects entrepreneurs' attributions of exit, which might differ from the actual determinants of exit.Gatewood et al. (1995) showed that female business owners were more likely to attribute their motivations to internal reasons, while males were more likely to attribute their reasons to external reasons.While their study focused on business creation, it is plausible that exit attributions also vary across genders.The debate around entrepreneurs' attribution biases is indeed far from settled.In particular, while earlier studies showed that attributions of failure tended to conform to entrepreneurs' self-serving tendency to avoid personal failure (e.g., Rogoff et al., 2004), we now know that accounts of exit cannot be merely seen as biased reconstructions of past events.Rather, failure narratives need to be understood as a complex social construction that is driven by the cognitive and emotional needs of entrepreneurs, which are themselves culturally embedded (Mantere et al., 2013).In their examination of accounts of failure experiences in the U.S. press, Cardon et al. (2011) found, for example, that communities in different regions had unique methods of making sense of venture failures.Their results suggest that cultural sensemaking about failure is an important factor explaining whether blame for failure is attributed to mistakes made by the entrepreneurs themselves (internal causes), or to misfortunes deriving from their environments (external causes).Such differences regarding failure highlight the need to examine the impact of sensemaking on exit attribution more broadly, and how and whether it differs across gender.
To the extent that each subculture (including gender) has its own code of failure (Waldman, 1999), gendered perceptions of the exit event may have important implications on how male and female entrepreneurs make sense of exit motives.In other words, it would be useful for research to attempt to explore and explicate how cultural cues and gender stereotypes might be conducing females and males to construct exit in distinctively different ways.Understanding this is important because attributions differences might reflect the extent to which exit is perceived as productive or destructive (McGrath, 1999) and may ultimately affect the incentive of males and females to engage in subsequent entrepreneurial efforts (Cardon et al., 2011;Shepherd, 2003).
Because our study was designed to challenge the assumption that female entrepreneurs lack the abilities and skills to perform at the same level as males, we have labeled non-financial related motives for exit as "voluntary."Nevertheless, and as previously mentioned, it could be argued that quitting a business to care for family members is not all that voluntary.While we have tried to alleviate this concern by controlling for the socio-economic status of entrepreneurs, future research could benefit from delving deeper into this issue.Specifically, scholars should further investigate how family embeddedness impacts entrepreneurial outcomes (Brush et al., 2014) and the circumstances under which exit might be experienced by females as a forced choice.Additionally, while we have focused on the gender of the principal founder, we acknowledge that entrepreneurs often launch new businesses with co-founders (Ruef, 2010;Ruef et al., 2003), and some of them might be involved in a spousal relationship (e.g., Brannon et al., 2013).Consequently, we believe that future research would benefit from leveraging emerging insights on entrepreneurial teams (Ruef, 2010;Ruef et al., 2003;Schjoedt et al., 2013) and the gender dynamics that emerge within these teams (Yang and Aldrich, 2014) to explore whether, and how, gender might affect exit decisions.Also, the small proportion of unmarried parents in our data set prevents further investigation of interaction effects between family variables, such as single parenthood.Given that the number of single-parent households has grown over the past decades (Bianchi, 2011;Fox et al., 2013), and considering the single parents' need to provide structural stability for the household, we expect such a variable to have important implications regarding voluntary exit decisions.Furthermore, in studying the influence of spousal relationships, we followed common understanding in the literature and focused on mixed-sex couples.While data limitations prevent us from exploring whether exit patterns differ for entrepreneurs involved in a same-sex partnership, we believe this is a fruitful area for research because literature points to the reproduction of gender dynamics in contemporary homosexual couples (Ridgeway, 2011).
Moreover, there are numerous research opportunities that do not directly emanate from our potential limitations.For instance, drawing upon the PO literature proved to be very valuable, as it allowed us to theorize about across-gender differences with regard to PO and, ultimately, exit motivations.While several researchers have commented that entrepreneurs refer to their venture as "their baby" (e.g., Cardon et al., 2005;Dodd, 2002), our research is one of the first to address this in more depth from a gender perspective, which extends the very fragmented knowledge about the link between gender and PO (e.g., Chung and Moon, 2011;Sieger et al., 2011;Van Dyne and Pierce, 2004).Building on the small, but growing body of research that explicitly applies PO in the entrepreneurship context (e.g., Sieger et al., 2013;Townsend et al., 2009), our study encourages future research that could measure PO perceptions of female and male business owners and link them empirically to different types of exit decisions.Also, while our choice of exit types is theoretically justified and empirically valid, future studies could examine and compare specific exit types in more detail, such as exits that are harvests and exits due to a return to wage employment.Furthermore, we believe we have only touched the surface of potential within-gender heterogeneity (de Bruin et al., 2006;Hughes and Jennings, 2012;Hughes et al., 2012) with the inclusion of family moderators.Adopting a contextual lens involves acknowledging that entrepreneurial outcomes within gender groups are also shaped by elements such as geographical area (Kalnins and Williams, 2014) or type of business or industry (Anna et al., 2000;Budig, 2006).Hence, future research could further investigate the role of the professional status of the business (which we included as a control variable) in exit patterns-especially across gender (Budig, 2006;Warren, 2010).Here, scholars could draw on arguments pertaining to the role of social class and labor markets.This line of inquiry could also be expanded further by looking into within-gender moderation effects that may also differ in strength between genders.The influence of higher education is a case in point.Interestingly, our results reveal that higher education increases the likelihood to exit for other professional/financial opportunities (as compared to failure).We deduct that with higher education, there are more attractive alternative opportunities in the job market, which makes entrepreneurs more prone to exit their venture in the absence of failure.This effect, in turn, seems to be stronger for females than for males, as the higher education variable is only marginally significant in the male subsample.It might be that education has a stronger effect in determining heterogeneous career paths among females because females face labor market constrains that are idiosyncratic to gender (e.g., glass ceiling).We could speculate that females who are able to transcend those constrains by exhibiting superior-than-average skills, experience, or education will be more likely than the average female entrepreneur to exit for wage employment.Lastly, building on our insight that family considerations have a stronger impact on females than on males, a preliminary test revealed that females are indeed more likely to exit for personal reasons than for professional/financial opportunities.Future research could explore the gender differences between those two types of voluntary exit in more detail.Lastly, building on our insight that family considerations have a stronger impact on females than on males, a preliminary test revealed that females are indeed more likely to exit for personal reasons than for professional/financial opportunities.Future research could explore the gender differences between those two types of voluntary exit in more detail.

Conclusion
Do female entrepreneurs fail more often than male entrepreneurs?Drawing on feminist theories and disentangling exit from failure, we challenge the corresponding female underperformance hypothesis and find that females do not fail more often, but exit voluntarily more often and generally due to personal reasons.Our findings on gender patterns with regard to different types of entrepreneurial exit offer unique and valuable insights to the gender and exit/failure debate and will hopefully inspire other scholars to conduct further studies in this exciting field of research.

Table 1
Descriptives and pairwise correlations.

Table 3
Multinomial logistic regression on entrepreneurs' exit motives.
Significant at level: †p = .10;*p= .05;**p= .01;***p= .001.RRR = relative risk ratios, SE = standard error.N = 219.Baseline category: "failure" (involuntary exit).In a multinomial logistic model the binary comparisons are computed simultaneously, hence there is one set of fit parameters for the complete model and not for each individual binary comparison.As a consequence, the fit parameters of Model 3 are identical to those of Model 5 (and those of Model 4 are identical to those of Model 6). a Akaike's information criterion (2k − 2)*(log likelihood), k = degrees of freedom.Gradually smaller values denote improved model fit.bLR test of model fit performed between models using maximum-likelihood estimates (MLE).