SSRN Werner-Kay Entrepreneurial Image - SSRN papers

1

Entrepreneurial Image, Gender, and the Formation of New Ventures Arndt Werner and Rosemarie Kay Institute for Small and Medium Sized Businesses Bonn (IfM Bonn) Maximilianstr. 20 53111 Bonn

Abstract The decision to become an entrepreneur does not only have economic determinants. Using a unique data set of individuals who visited business start-up exhibitions and applying logit and probit techniques we show that entrepreneurial self-perception has a strong impact on entrepreneurial activity (i.e. the propensity as well as the probability to become an entrepreneur). The addition of entrepreneurial self-perception to a conventional set of control variables significantly increases the predictive accuracy of our estimated equations. But contrary to mainstream theory, entrepreneurial self-perception has no specific gender effect. This finding holds true although statistics presented include both cross-section and time-series variation.

Acknowledgement: We thank Peter Kranzusch for helpful comments and discussion. We are especially grateful for his immense support in collecting the data.

2

Entrepreneurial Image, Gender, and the Formation of New Ventures

1. Introduction In Germany, as in many other countries, much less women than men start new businesses.1 Approximately every third new firm in Germany is founded by a woman.2 Therefore, women are still noticeably underrepresented among both founders and entrepreneurs, although a considerable increase in female firm formation is observable in recent years.3 The reasons why women are less often engaged in entrepreneurial activities than men are still not completely clear,4 despite increasing research efforts in this field in recent years. This is not only due to the lack of a comprehensive theory of women's entrepreneurship at large but also to a lack of (longitudinal) data. Additionally, until the 1990s primarily descriptive and to a lesser extent analytical techniques of statistical analysis were used.5 So far, crosssectional studies dominate,6 that compare the group of self-employed females with the group of self-employed males or the group of wage-employed females. These studies provide valuable evidence, but no conclusions. Longitudinal studies which are more appropriate to identify the causes of a lower entrepreneurial activity rate of women were much less often conducted because the generating of appropriate statistical series requires a considerable amount of time and effort.7 In this case researchers often resort to data sources such as the German SocioEconomic Panel (GSOEP) or the Panel Study of Income Dynamics (PSID), which were created to primarily investigate other subjects. These data sources usually do not include all relevant variables.8 Despite the afore-mentioned limitations the previous research has produced a number of assured explanations for the lower propensity of women to start new businesses. In short, especially human capital deficiencies, liquidity constraints, difficulties in combining work with family responsibilities, and a higher risk aversion are responsible. However, these factors possibly do not explain the issue completely. Recently, a limited capacity of women to identify with a masculine entrepreneurial image is discussed as another possible factor that produces gender differences in the propensity to create new ventures.9

3 This cognitive theory based explanation draws on the effects of (sex)stereotypes. Sex stereotypes are a set of attributes and behaviors ascribed to individual women and men because they belong to the group of women or men.10 Stereotypes are (simplifying) generalizations, that are neither true nor false or neutral.11 Female and male stereotypes differ noticeably. Women and men are often portrayed as polar opposites. Furthermore, the traits associated with women and men are seen as differentially desirable, those associated with men are more highly valued than those associated with women.12 Sex stereotypes contribute to a sex labeling or sex-typing of professions and occupations as "women's work" and "men's work",13 because this sex labeling reflects widely spread beliefs that certain occupations require certain attributes that are typic for one or the other sex. This sex labeling in turn influences individuals' occupational choices: men and women tend to choose such professions/occupations which seems appropriate for their respective sex.14 Applied to the decision to become an entrepreneur one can argue as follows: Entrepreneurship is still perceived as masculinely sex-typed.15 A woman who compares the male-typed traits and behaviors associated with entrepreneurship to her own (female toned) traits and behaviors possibly perceives an incongruity. Consequently, she possibly does not assess herself fit for entrepreneurship and decides against becoming an entrepreneur.16 Moreover, the decision to start a business is usually reached after consulting family and friends. Stereotypes have the same effects on these people as described above, so they rather tend to discourage than encourage a woman to form a business. In this paper we test the hypothesis that a limited capacity of women to identify with a masculine entrepreneurial image contributes, interacting with other determinants, to gender differences in propensity to create a new business. We predict that women have a lower propensity than men to become an entrepreneur. Our analyses are based on a specially generated data set, applying multivariate techniques of statistical analysis. The rest of the paper is structured as follows: A review of the relevant literature on gender differences in the propensity to start a new business is presented in section 2.17 Section 3 provides a description of the data used, whereas the empirical findings are outlined in section 4. The main conclusions and implications of this research appear in section 5.

4

2. Literature Review So far, to simplify matters we spoke of gender differences in the propensity to start a business. But caution is advised. Even though we do not want to resume the extensive discussion on gender equality and difference, one should bear in mind that, not only in Germany, the social reality of men (as a group) and women (as a group) still differs, for example with respect to education, occupational position, work experience or the assigned responsibility for household and childcare. So gender presumably reflects socioculturally imprinted and bequeathed differences rather than biological or "natural" ones. Gender differences that emerge in empirical studies therefore do not result from the affiliation of the respondent to one or the other sex but rather from still existing different living conditions of women (as entirety) and men (as entirety).18 The majority of German longitudinal studies on the entry into self-employment are based on longitudinal data from the German Socio-Economic Panel (GSOEP). Merz and Paic (2004) as well as Lohmann and Luber (2004) found, contrary to Hübler (1991), gender differences in the decision to switch from wage-employment into self-employment.19 Merz and Paic (2004) as well as Lohmann and Luber (2004) show that women are, as expected, less likely to enter into self-employment, even when controlling for individual characteristics.20 But Merz and Paic (2004) can show that gender differences only emerge if the new venture does not belong to the liberal professions ("Freie Berufe"). In order to examine gender differences, Lohmann and Luber (2004) estimated separate models for women and men. These models reveal that the general finding - the propensity to move into self-employment is highest for employees with a tertiary degree - holds true for women as well as for men.21 Nevertheless, some gender differences occur. Women with incomplete or only compulsory education are much less likely to move into self-employment than their male counterparts. Furthermore, in contrast to women, men with vocational education, especially those with a secondary degree, are quite likely to move into self-employment. Tertiary education is therefore of specific relevance for women. Lohmann and Luber (2004) can also show that the employment status of the father has, in contrast to men, no significant effect on the likelihood of women's self-employment. And finally, for both men and women, entry into self-employment is more likely when the partner is self-employed. However, this effect is much stronger and more significant for men.

5 Based on the Regional Entrepreneurship Monitor Germany 2003, Wagner's (2004) crosssectional analysis yields also a significant gender influence on the probability to become a nascent entrepreneur. As expected, men are more likely to step into self-employment than women. For both women and men, fear of failure has a negative influence on the propensity to become a nascent entrepreneur. But this effect is smaller for men than for women. McManus (2001) compared the pathways into self-employment in the United States and Germany.22 Her multivariate estimates reveal not only gender differences but also cross-national differences. The last-mentioned finding is of special importance as it reveals that national results, because of institutional differences, do not necessarily hold true for other countries. With regard to gender differences, in the United States as well in Germany, parents' selfemployment has a significant effect on the likelihood of men's self-employment only.23 For women, in the United States as in Germany, the self-employment of the partner is the most important factor for entry into self-employment. Young children in the household increase the probability to move into self-employment rather for women previously not working than for women already working.24 The latter have already found a way to combine work with family obligations. They do not need to switch into self-employment to gain flexibility and therewith a better opportunity to combine family and work responsibilities. Boden (1996) also shed light on the role of young children in the household on the switch from wage employment to self-employment. Both men and women are more likely to switch to self-employment if they have at least one child under the age of six. However, the influence of this factor upon entry into self-employment is much stronger and more statistically significant for women than it is for men. Based on cross-sectional data, Carr (1996) as well as Lohmann (2004) and Lauxen-Ulbrich et al. (2004) corroborate these finding for the United States and Germany respectively. Women with young children who switched to self-employment substantially reduced the number of working hours, thereafter.25 Boden (1999) can show, also based on cross-sectional data, that the flexibility of schedule in self-employment (and other family-related factors) are indeed reasons for becoming self-employed, especially for selfemployed men and women with young children. Clain's (1996) cross-sectional analysis also reveals several gender differences in full-time selfemployment in the United States. The likelihood of self-employment increases with education for both men and women;26 having a high-school diploma is critical for women, whereas having a college diploma is critical for men. Individuals in white-collar occupations are more

6 likely to be self-employed than individuals in blue-collar occupations. Among men, service workers appear to be the least likely candidates for self-employment, whereas workers in this occupation are the most likely candidates for self-employment among women. Being married raises the likelihood of self-employment for men. For women, the effect of marriage works most strongly through the spouse's income: the greater the income, the greater the likelihood of being self-employed. Carr (1996), in contrast, found an influence of being married only for women. Gender has a significant effect on the likelihood of becoming self-employed in the Netherlands27 and in Sweden28, too. The fact that women are less likely to be self-employed is partly due to direct choice, partly due to the fact that they expect to earn more as an employee than as a self-employed.29 The longitudinal studies presented so far examine the transition from "not self-employed" at a time t-1 to "self-employed" at a time t. This approach reflects the availability of respective data and has without question its advantages over cross-sectional studies. However, this design ignores that the decision to start a business is not made once for all. Starting a business is indeed a process in which the potential entrepreneur reassesses the original decision every now and then. This process remains a black box in this kind of study design. Studies that take this process into account are rare. An exception is the aforementioned RWI Founder Study. This study shows that women are much less inclined than men to consider entrepreneurship as a professional option. 15.9 % of male and 7.4 % of female respondents expressed a (vague) wish to start a business, 8.0 % of males and 3.6 % of females still have a (more precise) intention to start a business, and finally, 2.4 % of males and 1.3 % of females are nascent entrepreneurs.30 Thus, gender differences already exist in the first stage of the start-up process. But afterwards, these gender differences disappear.31 As mentioned above, women and men differ in various respects. These differences are partly responsible for the different likelihoods of men and women to develop a wish to start a new venture. In a multivariate analysis, comprising age, employment status, household income, household size, education, unemployment rate, and East/West Germany, the gender gap of 8,5 percentage points decreases to just under 4 percentage points.32 The questionnaire used by the RWI supposedly did not include all relevant factors that determine the propensity to become

7 an entrepreneur. So it remains unclear what the reasons are for the remaining gap of 4 percentage points. The KfW Foundation Monitor 2002 also shows that much less females (1.9 %) than males (3.6 %) intend to create a new business within the next six months.33 Contrary to the RWI Founder Study, after the expiration of this term, less women (25 %) than men (35 %) have actually carried out their intentions. Also the share of females (19 %) who are still in the realization phase is below the share of males (25 %). This finding indicates that even in this stage prior to the actual start-up some factors exert influence which have a different impact on women and men. In the following we argue that individuals make career choices based upon their perception of and associated fit with a certain profession.34 Or as Chen et al. (1998, S. 297) state it: “…people assess their personal capabilities against the requirements of different occupations.” The choice to engage in entrepreneurial activity is therefore dependent upon whether individuals can identify with the characteristics and behaviours that are associated with entrepreneurship.35 Relating entrepreneurial self-perception on gender differences it can be argued that because entrepreneurship is often associated with masculine characteristics, such as love of independence, perseverance, self-confidence, and decisiveness,36 this might negatively effect women’s entrepreneurial self-perception and in consequence female entrepreneurship activity. In the next section we will statistically analyse if this implication holds true.

3. Research Sample Description The data used in this study were specially generated for this purpose. The idea was to meet individuals who are interested in starting a business before they actually start the business. This approach would enable us to observe and investigate the start-up process. By means of surveys at consecutive points in time we can explore, among other things, the characteristics and attitudes of individuals who tend to start a business, how the start-up process develops, why the start-up effort is possibly abandoned, and which problems potential entrepreneurs face in the start-up process. This study design allows longitudinal analyses.

8 In order to get in touch with individuals who are thinking about creating a new business we visited several business start-up exhibitions. These fairs are staged at regular intervals at various places in Germany. The data set of this study results from surveys conducted at three exhibitions: the "StartMesse" in Essen, the exhibition "KarriereChance" in Dresden, and the "Deutsche Gründer- und Unternehmertage - deGUT" in Berlin. At these exhibitions random selected visitors were interviewed using a standardized questionnaire. As table 1 shows, 2,486 persons in total were interviewed, representing approximately 6 % of all visitors to the fairs. [Table 1 around here] At the time the first surveys were conducted over one fourth of the respondents were already self-employed. These respondents were excluded from subsequent surveys. Those respondents who did not give the address or did not grant the permission to be inquired a second time were excluded as well. Additionally, some of the questionnaires did not reach the interviewee due to an invalid address. The second surveys, conducted some ten months after the respective exhibition, finally included 994 respondents. 505 returned a fully completed questionnaire (50.8 %).37 The data collected in the first and the follow-up surveys were pooled and form the IfM Founder Panel.

4. Data Analyses The literature review has shown that even when controlling for individual characteristics gender differences in the probability to become self-employed exist. Some studies were able to identify factors influencing the probability to become self-employed that have a different impact on women and men. We take up these findings and suggest a further explanation for the shown gender differences. As described above, we assume a different impact of the capacity to identify with an entrepreneurial image on women’s and men’s probability to become an entrepreneur. Thus, we will investigate in the following (1) whether the entrepreneurial selfperception influences the decision to switch to self-employment and if so, (2) whether there are separate gender effects. The corresponding hypotheses are:

9 Hypothesis 1:

Individuals that perceive themselves as more entrepreneurial are more often engaged in entrepreneurial activities than people who do not perceive themselves as entrepreneurial.

Hypothesis 2:

Women are less often engaged in entrepreneurial activities than men because shortfalls in entrepreneurial self-perception have a higher negative impact on entrepreneurial activity for women than for men.

To test these propositions, we use the data set described above. The empirical models in which we regress the observed decision whether to start a new business or not on gender, entrepreneurial self-perception, and a set of personal characteristics, motives and attitudes will be discussed in the next section.

4.1 Description of dependent and independent variables All statistics presented will include both cross-section and time series variation. We estimate two econometrical models with two different dependent variables: (1) The willingness to become self-employed in t-1 on the one hand (a variable measured on a five step scale ranging from 1 (“no, I will definitely not become self-employed in the near future”) to 5 (“yes, I will definitely become self-employed in the near future”)) and (2) the probability of actually switching into self-employment until t on the other hand (a dummy variable taking the value one if the interviewee switched into self-employment in the period of time between t-1 and t). As our major explanatory variable the survey includes a tailor-made question asking if the interviewee perceives him- or herself as an entrepreneur. Following Verheul et al. (2005) entrepreneurial self-perception is recorded as follows: The survey respondents were asked “Do you feel yourself as an entrepreneur?” The question had to be answered on a five step scale ranging from 1 “yes, definitely” to 5 “no, not at all”. This variable is measured in t-1, whereas the probability to switch into self-employment is measured in t. Thus, a possible bias of twoway causality is avoided in this model. Let us now turn to the discussion of the control variables included in the empirical models discussed in the following section. These models include age, entrepreneurial experience,

10 higher education, a variety of entrepreneurial motives, family status, psychological traits, and a set of regional dummies: - Age (measured in years), on the one hand, can be seen as a proxy for human and physical capital requirements which are often unavailable to younger individuals.38 Thus, age should have a significant and positive influence on entrepreneurial activity.39 On the other hand, the shorter the expected life span of the business founder the higher the sunk costs of the new business, i.e. the shorter the period over which these sunk costs can be earned back.40 According to this argument, the coefficient of age should have a negative sign.41 Thus, it should be most attractive to start a new business at middle age when individuals have higher savings to overcome borrowing constraints and lower sunk costs which have to be earned back. To control for such a non-linear influence of age on entrepreneurial activity age is included in squares.42 - Self-employment experience (a dummy variable taking the value one if the interviewee has self-employment experience): According to Jovanovic (1982) entrepreneurs learn about their abilities over time, what they can do only if they have been engaged in entrepreneurship in the past. These implications are consistent with the empirical findings of Evans and Leighton (1989a) who show that self-employment experience has a positive and significant impact on the probability of white male Americans’ self-employment. Yet, Kay et al. (2004) find for Germany that new business founders that went out of business frequently in the past are less successful in their new business. - Industry specific experience (a dummy variable taking the value one if the interviewee has prior job experience in the start-up industry). Based on previous research prior job experience in the start-up industry is one of the most important determinants influencing entrepreneurial decision and start-up success.43 Individuals with industry specific experience have a competitive advantage over others because they are already familiar with potential customers, suppliers and prices for example. - Parent(s) self-employed (a dummy variable taking the value one if one or both parents are/were self-employed): Self-employed parents act as a role model. They offer their offspring instructions on business methods, transfer business experience and provide access to financial capital and equipment, business networks, consultancy and reputation.44

11 - Higher education (dummy variable taking the value one if the interviewee has attained higher education): Higher levels of education can be seen as general human capital resources and may promote entrepreneurship because higher educated people are better informed about business opportunities.45 On the other hand, entrepreneurs have fewer incentives to acquire formal educational qualifications than employees if education is a screening device used chiefly by employers to sort hidden worker types.46 Thus, the influence higher education can have on the entrepreneurial decision is unsettled. - Number of professional degrees (ranging from zero to three): In a recent paper Lazear (2002) presents a model of the choice between self-employment and wage employment which implicates that having a background in a large number of different professions increases the probability of becoming self-employed. The survey collects information about professional degrees after school, i.e. whether or not the individual passed apprenticeship, managed to qualify formally as a master craftsperson, or received a degree from a university. - Entrepreneurial motives: It is commonly suggested in entrepreneurship research that a positive relationship exists between self-employment and (anticipated) unemployment. Furthermore, the probability of self-employment should rise if the individual anticipates relative higher earnings as an entrepreneur.47 The literature review has also revealed a positive correlation, in particular for women, between self-employment and the wish to combine work and family responsibilities. The participation in self-employment may also be motivated by different working conditions between wage and self-employment, for instance autonomy and being one’s own boss.48 In our econometrical models discussed in the next section we test these factors using a set of dummy variables taking the value one if the interviewee agreed that the corresponding motive is important to her or him regarding her or his self-employment decision. - Psychological traits: A large psychological literature has developed which claims that entrepreneurs possess special traits that predispose them to entrepreneurship.49 According to McClelland (1961) for example, the ‘need for achievement’ is the key characteristic of successful entrepreneurs rather than the desire for money. Another important trait which has attracted research is an individual’s innate belief that his or her performance depends largely on her or his own actions rather than external factors (‘internal locus of control’).50 Those with a high internal locus of control might sense a greater urge to become self-employed since selfemployment often offers greater scope for individuals to exercise their own discretion at work

12 than does wage employment.51 Both discussed traits are included in the econometric model as dummies taking on the value one if the survey respondents strongly feel the need of achievement or if they believe that their performance depends largely on their own actions. - Family status (two dummy variables taking the value one if the interviewee has children or a spouse): We learn form research that social relations may increase entrepreneurial activity and success by providing instrumental support, such as cheap labor and capital or psychological support.52 Thus, one would expect that married people and/or those having children are more likely to become self-employed. Furthermore, spouses can provide their income as an insurance against uncertain income in entrepreneurship.53 On the other hand, married people or people with (young) children may be unwilling to take the higher risks associated with selfemployment. Again, the effect of marital status and/or family circumstances is ambiguous. - Second income (a dummy variable taking the value one if the household of the survey respondent disposes of a second labor income): Research results show that individuals sometimes have limited opportunities in self-employment because of difficulties in raising sufficient capital.54 The problem of not having enough financial capital to fund new start-ups is eased if another person in the household (spouse, parents etc.) has income.55 - Regional factors (three dummy variables taking the value one if the interviewee has visited the exhibition in Dresden, Berlin or Essen respectively): While research on micro level tries to identify individual characteristics of entrepreneurial activity there exists also a strand of literature analyzing how characteristics of the economic environment affects the decision to become an entrepreneur.56 In our econometrical models we take these regional factors into account using the aforementioned set of dummy variables. To sum up, the information considered in our empirical models can be seen as close to being comprehensive.57 Omitted variable bias should be of no account in our econometric analyses so that estimation results should come close to the “true” (unbiased) effect of entrepreneurial self-perception on entrepreneurial activity.

4.2 Descriptive and bivariate results

13 This section is devoted to a short presentation of descriptive and bivariate statistics by gender for the variables used in this study. Table 2 compares all (not self-employed) men and women who took part in the survey. Table 3 compares those that switched into self-employment in the period of time between t-1 and t by gender and table 4 compares those that did not switch in the period of time between t-1 and t. [Table 2 around here ] Our analyses yield a number of significant gender differences in mean values. With the exception of ‘children’, ‘parent(s) self-employed’, ‘higher relative earnings’ and ‘(anticipated) unemployment’, men and women differ significantly in all considered characteristics. Most of our expectations derived from theory and/or previous research are supported at bivariate level: Women perceive themselves as less entrepreneurial than men. Furthermore, women have less self-employment or industry specific experience, higher education and a smaller number of professional degrees. Also according to our expectations, the motive combining work and family responsibility by means of self-employment is more important for women than for men. Interestingly, far more men than women are married. With respect to our psychological characteristics, we find that women, on average, have a higher internal locus of control whereas men, on average, show a greater need for achievement. Wanting to be one’s own boss is a slightly more important self-employment motive for women (90.6 %) than for men (87.8 %).58 Next, we will compare the newly self-employed by gender with regard to the variables used in the econometrical model (see table 3): [Table 3 around here] Descriptive evidence in table 3 documents fewer significant differences in characteristics and attitudes between male and female self-employed. This is in line with prior research indicating that most of the factors influencing individuals to participate in entrepreneurship seem to be of the same importance for men and women.59 Yet, some interesting exceptions can be observed: on average, more self-employed men than women have higher education and, again, selfemployed males show a stronger need for achievement than females. In accordance with the flexibility hypothesis60 is the finding that women rather than men switch into self-employment due to the motive ‘combining work and family’. Moreover, the share of self-employed women

14 that emphasized in t-1 that they definitely want to become self-employed in the near future is significantly higher than the share of their male counterparts. To complete our descriptive analyses we compare male and female respondents who have not switched into self-employment in the period of time between t-1 and t and state in t that they do not intend to become self-employed anymore. These individuals will be termed as “quitters” in the following.61 [Table 4 around here] As table 4 reveals, there are more significant gender differences among quitters than among newly self-employed. Interestingly, these differences refer to different characteristics. Neither ‘higher education’, ‘combining work and family’ nor ‘need for achievement’ are crucial anymore. Instead we can observe gender differences with respect to ‘age’, ‘parent’s selfemployed’, ‘number of professional degrees’ and ‘(anticipated) unemployment’. Female quitters are, on average, considerably younger than male quitters. Their parents are much more often self-employed than those of the male quitters. Male quitters have, on average, a larger number of professional degrees than female quitters. And, finally, the unemployment motive is more important to male than to female quitters.62

4.3 Econometrical results 4.3.1 The impact of entrepreneurial self-perception on female entrepreneurial activity As tables 3 and 4 show, more males (35.3 %) than females (32.5 %) actually start a new business (ratio is relating to the number of men and women who did not switch into selfemployment between t-1 and t) whereas more females (43.6 %) than males (31.5 %) are quitters (ratio is relating to the number of men and women who switched into self-employment between t-1 and t or still intend to do so in the near future). Thus, the question arises why females participate in entrepreneurship below average compared to males. The descriptive bivariate evidence above has documented significant differences in characteristics and attitudes between male and females interested in self-employment, between newly self-employed males and females and, finally, between male and female quitters. But with respect to entre-

15 preneurial self-perception, little evidence of gender difference was found. Yet, the descriptive evidence presented does not reveal the extent to which the variables discussed might be interrelated. The bivariate result with regard to gender and entrepreneurial self-perception may be the result of intervening variables such as age, experience or higher education. To reveal the ceteris paribus effect of gender and self-perception multivariate analyses have to be applied. Thus, in the following section we will investigate the ceteris paribus impact of entrepreneurial self-perception on female entrepreneurial activity, i.e. we will analyze the willingness to become self-employed in t-1 (model 1) and the probability of switching into self-employment (model 2) when characteristics and attitudes of respondents are controlled for. The dependent variable in model 1 is an ordinal variable taking the value five if the respondent states that he or she definitely wants to become self-employed in the near future and the value one if she or he definitely does not want to become self-employed. The appropriate econometrical model to use in this case is an ordinal probit model.63 The endogenous variable in model 2 is a dummy variable taking the value one if a person switched into self-employment in the period between t-1 and t, and zero otherwise. The appropriate econometrical model here is a binary logit model.64 In the following econometric investigations three different specifications of the two empirical models are estimated. The first specification consists of gender and the vector of control variables only. In the second specification we additionally include our major explanatory variable: entrepreneurial self-perception. In the third specification we finally include an interaction term of the self-perception variable and the female dummy variable. If this interaction term is statistically different from zero at a conventional level and if the coefficient estimate of the interaction term turns out to be negative65, then shortfalls in entrepreneurial self-perception have a higher negative impact on entrepreneurial activity for women than for men.66 Table 5 displays the results of ordered-probit estimations (with willingness to become selfemployed in t-1 as dependent variable). Table 6 displays estimation results of the logit model (with the probability of switching into self-employment as endogenous variable). [Table 5 around here] [Table 6 around here]

16 With a view to the results, we find that the first hypothesis is confirmed by the data. Specification II in both empirical models shows a significant negative effect of the entrepreneurial self-perception variable.67 In contrast, in specification III of both models the coefficient of the interaction term is not statistically different from zero at a conventional level. Thus, the second hypothesis, that women are less often engaged in entrepreneurial activities than men because of shortfalls in entrepreneurial self-perception, can be rejected. With regard to the control variables for the willingness to become self-employed in the near future, the estimation results for the variables measuring the (non-linear) effect of age, industry specific experience, (anticipated) unemployment, number of professional degrees, combining work with family responsibilities, autonomy, and a high internal locus of control are statistically different from zero at a conventional level and have the expected signs. In addition, we find no evidence for regional effects. Yet, this may be due to the fact that we can control only for rather urban regions (Essen, Dresden and Berlin). The results of the control variables for the probability to switch into self-employment are less in line with our theoretical considerations: Like in the first model, we find a concave relationship between age and the probability to switch. Also, in both models prior industry specific experience has the expected positive impact on entrepreneurial activity. The coefficients of the other control variables in model 2 are not significant at a conventional level. Being male or female has no effect on entrepreneurial activity in both models if our broad set of control variables is included.68

4.3.2 Male and female entrepreneurship: Are there differences? Comparative descriptive statistics gave evidence that males and females may differ in most of the characteristics and attitudes, aside from entrepreneurial self-perception. Multivariate analyses showed that - aside from entrepreneurial self-perception - age (non-linear), industry specific experience, (anticipated) unemployment, second household income are important determinants of entrepreneurship (i.e. statistically significant at any conventional level). Yet, the hypothesis that women are less often engaged in entrepreneurial activities then men because of shortfalls in entrepreneurial self-perception was rejected. But what about the other factors? Even though a comprehensive empirical investigation goes beyond the scope of this paper, in the following we will analyze the gender impact of some of these. Again, we will

17 focus on the willingness to become self-employed and the probability to switch actually into self-employment. For the sake of clarity, only the results of the interaction terms are displayed. The results are given in Table 7. [Table 7 around here] Of the eight examined interaction terms, only the interaction ‘combining work with family responsibilities * female’ is statistically significant at a conventional level in both models. Obviously, male and female self-employed differ strongly in their attitude towards family responsibilities.69 This finding adds to the evidence displayed in the literature review. The prospect of better opportunities to combine work and familiy responsibilities through selfemployment is a specific motive for women to consider self-employment as an occupational alternative or to become actually an entrepreneur. This finding is also evidence for the still prevailing gender-related division of labor. Self-employment provides greater flexibility insofar as individuals can decide on how long and when they work, at least to a certain extent. Table 8 which compares newly founded businesses of women to those of men reveals that self-employed females differ significantly from men in the average weekly working hours. The former work on average more than eight hours less than self-employed males. As reported in table 8 there are some more significant gender differences in start-up characteristics: team start-up and hiring of employees (both rather male start-ups). Furthermore, self-employed females remain over-represented in a few sectors like ‘other services’. Fewer self-employed women are found in construction industry and retail trade. [Table 8 around here]

5. Conclusions The decision to become an entrepreneur does not only have economic determinants. Using a quite unique German data set of individuals visiting business start-up exhibitions we show that entrepreneurial self-perception has a strong ceteris paribus effect on entrepreneurial activ-

18 ity (i.e. the propensity as well as the probability to become an entrepreneur). Yet, contrary to theoretical considerations, entrepreneurial self-perception has no specific gender effect. Shortfalls in entrepreneurial self-perception do not have a higher negative impact on entrepreneurial activity for women than for men all other things equal. Thus, we find no proof for the hypothesis that masculine behaviors and traits commonly ascribed to the entrepreneurial image are negatively influencing female entrepreneurial activity. This finding holds true although statistics presented include both cross-section and time-series variation. Of all possible determinants only the desire to combine work with family responsibilities through selfemployment has turned out to influence female entrepreneurial activity. Contrary to previous research our analyses reveal neither gender differences in the propensity to become an entrepreneur nor in the probability to actually start a new business. And also contrary to previous research we did not find special gender-related effects in the examined factors that influence entrepreneurial activity, aside from the motive ‘combining work with family responsibilities’. These discrepancies with regard to prior research are supposedly due to the applied data sources and therefore the sample design. If women and men have developed a (vague) interest to start a new business (that’s the point when we interviewed the respondents for the first time), in the following start-up process any gender differences in the probability to become an entrepreneur do not occur anymore.70 Thus, the so-called gender gap in entrepreneurial activity emerges in a phase prior to the point at which individuals have already developed an interest in starting a new business. This finding has consequences for the research on the reasons why women are less engaged in entrepreneurial activities, but also for the design of measures fostering female start-up activities.

19 Appendix Table A1: Logit Estimates for the probability to switch into self-employment until t (including the control variable ‘Propensity to become self-employed in t-1’) Coeff. (Std. Dev.) Variables (measured in t-1):

Female

Age

Age (squared)

Self-employment experience

Industry specific experience

Parent(s) self-employed

Higher education

Number of professional degrees

(Anticipated) unemployment

Higher relative earnings

Combining work with family responsibilities

Autonomy

Need for achievement

Internal locus of control

Married

Children

Second household Income

Propensity to become self-employed in t-11

Entrepreneurial self-perception

Log likelihood LR chi2

-0. 1272 (0. 2917) 0. 2159* (0. 1231) -0. 0031** (0. 0016) 0. 3113 (0. 4011) 0. 9304** (0. 4341) -0. 0129 (0. 3242) 0. 1932 (0. 3077) -0. 1908 (0. 2887) 0. 9020*** (0. 3052) -0. 2422 (0. 2831) 0. 0045 (0. 2941) 0. 4584 (0. 4599) 0. 3915 (0. 3265) -0. 3588 (0. 3715) -0. 2263 (0. 3397) 0. 1116 (0. 2568) 1. 2078*** (0. 3501) 1. 7247*** (0. 3909) -0. 3199* (0. 1838) -168. 63 (80. 69)***

Note: Data stem from the IfM Founder Panel. All regressions include regional dummies. 1 A dummy variable taking the value one if survey s state that they want to start a new business in the near future. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent.

20

Table A2: Questionnaire follow-up survey (excerpt) (Erstbefragung)

21

Table A3: Questionnaire first survey (excerpt) (Zweitbefragung)

22

1

Cf. Minniti et al. (2005).

2

In spite of different concepts this share results from both official statistics (cf. Kay et al. 2003) and population surveys such as the Global Entrepreneurship Monitor (e.g. Sternberg/Bergmann 2003, Sternberg/Lückgen 2005) or the KfW Foundation Monitor (cf. Lehnert 2004).

3

Cf. Kay et al. (2003), Leicht/Lauxen-Ulbrich (2004).

4

Cf. Parker (2004), Leicht et al. (2004).

5

Cf. Brush (1992).

6

Cf. Ahl (2004).

7

Cf. Reynolds et al. (2004).

8

Cf. e.g. Merz/Paic (2004). In addition to the GSOEP there are two other German surveys which, at least rudimentarily, facilitate longitudinal analyses: the RWI Founder Study (cf. Welter 2001) and the KfW Foundation Monitor 2000 and 2002 including the follow-up surveys in 2001 and 2003 (cf. Lehnert 2004, Reents et al. 2004). The RWI Founder Study has the drawback that no information is available whether the potential entrepreneurs finally do start a new venture. The data of the KfW Foundation Monitor are not available to the public and corresponding analyses exist only on a small scale. 9

Cf. Welter (2004).

10

Cf. Leyens et al. (1994).

11

Cf. Williams/Best (1982).

12

Cf. Rosenkrantz et al. (1968).

13

Cf. Oppenheimer (1968).

14

Cf. Powell (1987).

15

Cf. Fagenson/Marcus (1991).

16

Cf. Verheul et al. (2005).

17

For lack of space this review is restricted to longitudinal studies and selected multivariate cross-sectional studies. For a more comprehensive review see e.g. Carter et al. (2001) and Leicht et al. (2004). 18 19

Cf. Arum/Müller (2004). Hübler (2004) used the waves of the years 1984 and 1988, Merz and Paic (2004) used the waves of the years 1991 to 2001, and Lohmann and Luber (2004) used the waves of the years 1984 to 1998.

20

Werner (2004) who examined the influence of gender on the likelihood of employees to enter into selfemployment within the next two years yields the same result, based on cross-sectional data of the GSOEP 2001. 21

Based on cross-sectional data of the German Mikrozensus Strohmeyer's analysis (2004) reveals contrary findings. The likelihood to be self-employed is highest for workers with a secondary vocational degree ("Meister, Fachschule der DDR"). This holds true for women and men, even though this degree is of specific relevance for men. On the other hand, tertiary degrees are more important for women than for men. 22

She also used data of the GSOEP (waves 1984-1987) for Germany. For the US she used data of the Panel Study of Income Dynamics (PSID). 23

This holds apparently true for other countries such as France, the Netherlands, Australia, Hungary, Italy and Japan, too (cf. Arum/Mülller 2004).

24

This holds true for Germany as well as for the United States.

25

Cf. Boden (1996).

23

26

Likewise Carr (1996).

27

Cf. de Wit/van Winden (1989).

28

Cf. Gianetti/Simonov (2004).

29

Cf. de Wit/van Winden (1989).

30

Cf. Welter (2001).

31

Cf. MittelstandsMonitor (2004).

32

Cf. MittelstandsMonitor (2004).

33

Lehnert (2004).

34

Cf. Fagenson/Marcus (1991).

35

Theoretically, the causation between environment, perception and behaviour goes back to social learning theory (cf. Bandura 1986). It must be noted however that the opposite perspective can also be taken: Verheul et al. (2005) for example explain entrepreneurial self-perception by way of entrepreneurial activity. Yet, this approach is rather the exception. Based on literature and design of the present study (cross sectional and longitudinal) it is plausible to assume that in most part self-perception influences entrepreneurial activity and not the other way around. 36

Cf. Hisrich/Brush (1983), Chaganti (1986), Verheul et al. (2005).

37

For more details on the dataset, see Kranzusch (2005).

38

Cf. Parker (2004).

39

Older people are more likely to have received an inheritance to overcome borrowing constrains. If a particular type of human capital exists which is productive both in managing an enterprise and in working for others, and which can initially be acquired most effectively by working as an employee older, more experienced people should become entrepreneurs more often (cf. Lucas 1978, Parker 2004). 40

Cf. Wagner (2004).

41

Furthermore, elderly people may be more risk averse than the young and less capable of working long hours. This would also lead to a negative sign of the estimated coefficient of the age variable (cf. Miller 1984). 42

Cf. Evans/Leighton (1989b), Wagner (2004).

43

See for example Brüderl et al. (1996).

44

Cf. Parker (2004).

45

Keeble et al. (1993) for example showed that there are many opportunities for self-employment in knowledgebased industries. 46

Cf. Wolpin (1977).

47

A similar classification is made by the scientific research using data collected as part of the Global Entrepreneurship Monitor (GEM) (Reynolds et al. 2004: 38). The nascent entrepreneurs were asked if they were involved to pursue a business opportunity or because they had ‘no better options for work’. The former are attracted towards an opportunity (opportunity entrepreneurship), while the latter are pushed into the activity out of necessity (necessity entrepreneurship). Wennekers et al. (2005) find a U-shaped relationship in particular between opportunity based entrepreneurial activity and economic development.

48

Cf. Frey/Benz (2003).

49

See Amit et al. (1993) for an overview.

50

Cf. Rotter (1966).

51

Cf. Parker (2004).

52

Cf. Sander/Nee (1996).

53

Moreover, spouses and older children may also be more trustworthy employees, being less likely to shirk (cf. Borjas 1986).

24

54

Cf. Backes-Gellner/Werner (2006).

55

Unfortunately, information on second income was collected in the follow-up survey (t) only.

56

See for example Giannetti/Simonov (2004) or (Sternberg/Wagner 2004).

57

See the list of information that should ideally be included in an empirical model for the decision to become self-employed or not (cf. Parker 2004, ch. 3). 58

Please note that 16 individuals did not state their sex.

59

E.g. Wagner (2004), Lohmann/Luber (2004).

60

Cf. Carr (1996), Boden (1996).

61

Please note: 505 individuals answered the follow-up questionnaire. 167 of these persons switched into selfemployment between t-1 and t, 328 did not switch. 10 respondents were self-employed between t-1 and t but had switched back into paid employment at the time the follow-up survey was conducted and could therefore not be considered. 309 out of the 328 respondents who did not switch into self-employment answered the question if they intend to switch into self-employment in the near future. 209 out of these 309 respondents confirmed their plan to become self-employed in the near future. Yet, 103 interviewees emphasized that they do not intend to become self-employed anymore at any time. 62

We abstain from offering explanations for these gender differences because it is necessary to prove these differences applying multivariate techniques of statistical analysis. Yet, this paper is not focused on explaining the quitting behaviour of males and females. This will be object of future research efforts. 63

As the endogenous variable is ordinal, OLS Regression is inappropriate (cf. Wooldridge 2003).

64

Cf. Parker (2004).

65

Remember, the respondents were asked whether they feel as an entrepreneur. They had to answer on a 5-point scale from 1 = “yes, definitely” to 5 = “no, not at all”. 66

Two independent variables are said to interact when the partial effect of one depends on the value of the other. The most popular way to model this is by introducing a product regressor (multiplicative interaction) (cf. Wooldridge 2003). 67

According to Verheul et al. (2005) entrepreneurship activity effects entrepreneurial self-perception in parts. Thus, the effect of entrepreneurial self-perception on the probability to become self-employed in model 2 may be biased if it is based upon the propensity to become self-employed in t-1. To control for this, we calculated the model 2 once more including this time the variable ‘propensity to self-employment in t-1’. The results give no evidence for the bias. Our major explanatory variable ‘entrepreneurial self-perception in t-1’ is still positive and statistically different from zero at a conventional level (see table A1, appendix). 68

As can be seen the relationship between the independent and endogenous variables remains stable if our major explanatory variable and its interaction with gender is included in the models. 69

Furthermore, the interaction ‘parent(s) self-employed x female’ has a significant impact on the willingness to become self-employed but not for actually switching into self-employment. If we have a look at the descriptive results in table 4 we can see that female quitters having self-employed parent(s) are over-represented. This finding needs further research efforts but goes beyond the scope of this paper. 70

This finding is in line with the afore-mentioned RWI Founder Study.

25 References Ahl, H. (2004): The Scientific Reproduction of Gender Inequality. A Discourse Analysis of Research Texts on Women's Entrepreneurship, Malmö 2004. Amit, R./Glosten, L./Muller, E. (1993): Challenges to Theory Development in Entrepreneurship Research. In: Journal of Mathematical Sociology, Vol. 30 (1993), S. 815-834. Arum, R./Müller, W. (2004): The Reemergence of Self-Employment: Comparative Findings and Empirical Propositions. In: Arum, R./Müller, W. (Hrsg.): The Reemergence of Self-Employment. A Comparative Study of Self-Employment Dynamics and Social Inequality, Princeton/Oxford 2004, S. 426-454. Backes-Gellner, U./Werner, A. (2006): Entrepreneurial Signaling via Education: A Success Factor for Innovative Start-Ups. In: Small Business Economics (forthcoming). Bandura, A. (1986): Social Foundations of Thought and Action: A Social Cognitive Theory, Englewood Cliffs 1986. Boden, R.J. (1999): Flexible Working Hours, Family Responsibilities, and Female SelfEmployment. Gender Differences in Self-Employment Selection. In: American Journal of Economics and Sociology, Vol. 58 (1999), S. 71-84. Boden, R.J. (1996): Gender and Self-Employment Selection: An Empirical Assessment. In: Journal of Socioeconomics, Vol. 25 (1996), S. 671-682. Borjas, G. J. (1986): The Self-Employment Experience of Immigrants. In: Journal of Human Resources, Vol 21 (1986), S. 485-506. Brüderl, J./Preisendörfer, P./Ziegler, R. (1996): Der Erfolg neugegründeter Unternehmen Eine empirische Studie zu den Chancen und Risiken von Unternehmensgründungen, Berlin 1996. Brush, C.G. (1992): Research on Women Business Owners: Past Trends, a New Perspective and Future Directions. In: Entrepreneurship Theory and Practice, Summer 1992, S. 530. Carr, D. (1996): Two Paths to Self-Employment? Women’s and Men’s Self-Employment in the United States 1980. In: Work and Occupations, Vol 23 (1996), S. 26-53. Carter, S./Anderson, S./Shaw, E. (2001): Women’s Business Ownership: A Review of the Academic, Popular and Internet Literature, Sheffield 2001. Chaganti, R. (1986): Management in Women-Owned Enterprises. In: Journal of Small Business Management, Vol. 24 (1986), S. 18-29. Chen, C.C./Greene, P.G./Crick, A. (1998): Does Entrepreneurial Self-efficiacy distinguish Entrepreneurs from Managers? In: Journal of Business Venturing, Vol. 13 (1998), S. 295-316. Clain, S.H. (2000): Gender Differences in Full-Time Self-Employment. In: Journal of Economics and Business, Vol. 52 (2000), S. 499-513. De Wit, G./van Winden, F.A.A.M. (1989): An Empirical Analysis of Self-Employment in the Netherlands. In: Small Business Economics, Vol. 1 (1989), S. 263-272. Evans, D.S./Leighton, L.S. (1989a): The Determinants of Changes in US Self-Employment 1968-1987. In: Small Business Economics, Vol. 1 (1989), S. 111-119.

26 Evans, D.S./Leighton, L.S. (1989b): Some Empirical Aspects of Entrepreneurship. In: American Economic Review, Vol. 79 (1989), S. 519-535. Fagenson, E.A./Marcus, E.C. (1991): Perceptions of the Sex-Role Stereotypic Characteristics of Entrepreneurs: Women's Evaluation. In: Entrepreneurship: Theory and Practice, Summer 1991, S. 33-47. Frey, B.F./Benz, M. (2003): Being Independent is a Great Thing: Subjective Evaluations of Self-Employment and Hierarchy, CESifo Working Paper No. 959, 2003. Gianetti, M./Simonov, A. (2004): Social Interactions and Entrepreneurial Activity, mimeo, Stockholm 2004. Hisrich, R./Brush, C.G. (1983): The Women Entrepreneur: Implications of Family, Educational and Occupational Experience. In: Frontiers of Entrepreneurship Research, Wellesley 1983, S. 255-270. Hübler, O. (1991): Was unterscheidet Freiberufler, Gewerbetreibende und abhängig Beschäftigte? Eine ökonometrische Untersuchung über Gruppenheterogenität, Einkommensdeterminanten und Statuswechsler. In: Mitteilungen aus der Arbeitsmarkt- und Berufsforschung, 24. Jg (1991), S. 101-114. Jovanovic, B. (1982): Selection and Evolution of Industry. In: Econometrica, Vol. 50 (1982), S. 649-670. Kay, R./Günterberg, B./Holz, M./Wolter, H.-J. (2003): Unternehmerinnen in Deutschland, Gutachten im Auftrag des Bundesministeriums für Wirtschaft und Arbeit - Langfassung -, BMWA-Dokumentation Nr. 522, Berlin 2003. Kay, R./Kranzusch, P./Suprinovič, O./Werner, A. (2004): Restart: Eine zweite Chance für gescheiterte Unternehmer?, Wiesbaden 2004. Keeble, D./Walker, S./Robson, M. (1993): New Firm Formation and Small Business Growth: Spatial and Temporal Variations and Determinants in the United Kingdom, Employment Department Research Series, 15, London 1993. Kranzusch, P. (2005): Die Besucher von Gründungsmessen - Ergebnisse aus Besucherbefragungen der Gründermessen in Berlin, Dresden und Essen. In: Institut für Mittelstandsforschung Bonn (Hrsg.): Jahrbuch zur Mittelstandsforschung 1/2005, Wiesbaden 2005, S. 1-46 Lauxen-Ulbrich, M./Leicht, R./Fehrenbach, S. (2004): Flexibel zwischen Familie und Beruf? Zur Lebens- und Arbeitsgestaltung selbständiger Frauen. In: Leicht, R./Welter, F. (Hrsg.): Gründerinnen und selbstständige Frauen. Potenziale, Strukturen und Entwicklungen in Deutschland, Karlsruhe 2004, S. 138-169. Lazear, E.P. (2002): Entrepreneurship, NBER Working Paper, 9109, National Bureau of Economic Research 2002 Lehnert, N. (2004): Gründungsverhalten von Frauen im Spiegel des DtA-Gründungsmonitors. In: KfW Bankengruppe (Hrsg.): Chefinnensache. Frauen in der unternehmerischen Praxis, Heidelberg 2004, S. 71-81. Leicht, R./Lauxen-Ulbrich, M. (2004): Umfang und längerfristige Entwicklung selbstständiger Frauen. In: Leicht, R./Welter, F. (Hrsg.): Gründerinnen und selbstständige Frauen. Potenziale, Strukturen und Entwicklungen in Deutschland, Karlsruhe 2004, S. 41-53. Leicht, R./Welter, F./Fehrenbach, S. (2004): Geschlechterunterschiede in beruflicher Selbstständigkeit: Zum Stand der Forschung. In: Leicht, R./Welter, F. (Hrsg.): Gründerinnen

27 und selbständige Frauen. Potenziale, Strukturen und Entwicklungen in Deutschland, Karlsruhe 2004, S. 10-40. Leyens, J.-P./Yzerbyt, V./Schadron, G. (1994): Stereotypes and Social Cognition, London 1994. Lohmann, H. (2004): Berufliche Selbständigkeit von Frauen und Männern im internationalen Vergleich - Welche Rolle spielt die Vereinbarkeit von Familie und Erwerbstätigkeit? In: Schmid, G./Gangl, M./Kupka, P. (Hrsg.): Arbeitsmarktpolitik und Strukturwandel: Empirische Analysen, Nürnberg 2004, S. 205-226. Lohmann, H./Luber, S. (2004): Trends in Self-Employment in Germany: Different Types, Different Developments? In: Arum, R./Müller, W. (Hrsg.): The Reemergence of SelfEmployment. A Comparative Study of Self-Employment Dynamics and Social Inequality, Princeton 2004, S. 36-74. Lucas, R.E. (1978): On the Size Distribution of Business Firms. In: Bell Journal of Economics, Vol. 9 (1978), S. 508-523. McClelland, D.C. (1961). The Achieving Society, Princeton 1961. McManus, P.A. (2001): Pathways into Self-Employment in the United States and Germany. In: Vierteljahreshefte zur Wirtschaftsforschung, 70. Jg (2001), S. 24-30. Merz, J./Paic, P. (2004): Existenzgründungen von Freiberuflern und Unternehmern - Eine Mikroanalyse mit dem Sozio-oekonomischen Panel. In: Merz, J./Wagner, J. (Hrsg.): Perspektiven der Mittelstandsökonomie. Ökonomische Analysen zu Selbständigkeit, Freien Berufen und KMU, Münster 2004, S. 117-138. Miller, R.A. (1984): Job Matching and Occupational Choice. In: Journal of Political Economy, Vol. 92 (1984), S. 1086-1120. Minniti, M./Arenius, P./Langowitz, N. (2005). Global Entrepreneurship Monitor. 2004 Report on Women and Entrepreneurship, Babson Park 2005. MittelstandsMonitor (2004): Chancen zum Aufschwung nutzen. Jährlicher Bericht zu Konjunktur- und Strukturfragen kleiner und mittlerer Unternehmen. Hrsg. von Verband der Vereine Creditreform, Institut für Mittelstandsforschung Bonn, Zentrum für Europäische Wirtschaftsforschung, Rheinisch-Westfälisches Institut für Wirtschaftsforschung und KfW-Bankengruppe, Frankfurt/M. 2004. Oppenheimer, V.K. (1968): The Sex-Labeling of Jobs. In: Industrial Relations, Vol. 7 (1968), S. 19-234. Parker, S.C. (2004): The Economics of Self-Employment and Entrepreneurship, Cambridge 2004. Powell, G.N. (1987): The Effects of Sex and Gender on Recruitment. In: Academy of Management Review, Vol. 12 (1987), S. 731-743. Reents, N./Bahß, C./Billich, C. (2004): Unternehmer im Gründungsprozess: Zwischen Realisierung und Aufgabe des Gründungsvorhabens. Ergebnisse einer qualitativen Studie und des KfW-Gründungsmonitors. In: KfW-Research, Ausgabe 31 (2004), S. 4-27. Reynolds, P.D./Carter, N.M./Gartner, W.B./Greene, P.G. (2004): The Prevalence of Nascent Entrepreneurs in the United States: Evidence from the Panel Study of Entrepreneurial Dynamics. In: Small Business Economics, Vol. 23 (2004), S. 263-284.

28 Reynolds, P.D./Bygrave, W.D./Autio, E. (2004): Global Entrepreneurship Monitor – 2003 Executive Report. Babson College. Rosenkrantz, P.S./Vogel, S.R./Bee, H./Broverman, I.K./Broverman, D.M. (1968): Sex-Role Stereotypes and Self-Concepts in College Students. In: Journal of Consulting and Clinical Psychology, Vol. 32 (1968), S. 287-29. Rotter, J.B. (1982): The Development and Applications of Social Learning Theory: Selected Papers, New York 1982. Sanders, J.M./Nee, V. (1996): Immigrant Self-Employed: The Family as Social Capital and the Value of Human Capital. In: American Sociological Review, Vol. 61 (1996), S. 231-249. Sternberg, R./Bergmann, H./Lückgen, I. (2004): Global Entrepreneurship Monitor. Länderbericht Deutschland 2003, Köln 2004. Sternberg, R./Lückgen, I. (2005): Global Entrepreneurship Monitor. Länderbericht Deutschland 2004, Köln 2005. Sternberg, R./Wagner, J. (2004): The Decision to Start a New Firm: Personal and Regional Determinants. Empirical Evidence from the Regional Entrepreneurship Monitor. In: Fritsch, M./Niese, M. (Hrsg.): Gründungsprozess und Gründungserfolg - Interdisziplinäre Beiträge zum Entrepreneurship Research, Heidelberg 2004, S. 19-38. Strohmeyer, R. (2004): Berufliche Ausbildung und Gründungsaktivitäten im Geschlechtervergleich. In: Leicht, R./Welter, F. (Hrsg.): Gründerinnen und selbständige Frauen. Potenziale, Strukturen und Entwicklungen in Deutschland, Karlsruhe 2004, S. 97-118. Verheul, I./Uhlaner, L./Thurik, A.R. (2005): Business Accomplishments, Gender and Entrepreneurial Sel-Image. In: Journal of Business Venturing, Vol. 20 (2005), S. 483-518. Wagner, J. (2004): What a Difference a Y Makes. Female and Male Nascent Entrepreneurs in Germany, IZA-Discussion Paper No. 1134, Bonn 2004. Welter, F. (2004): Institutionelle Einflüsse auf Gründerinnen und Unternehmerinnen. In: KfW Bankengruppe (Hrsg.): Chefinnensache. Frauen in der unternehmerischen Praxis, Heidelberg 2004, , S. 33-69. Welter, F. (2001): Nascent Entrepreneurship in Germany, Schriften und Materialien zu Handwerk und Mittelstand, Heft 11, Essen 2001. Wennekers, S./Van Stel, A/Thurik, R./Reynolds, P.D. (2005): Nascent Entrepreneurship and the Level of Economiv Development, Small Business Economics, Vol. 24(2005), S. 293-309. Werner, A. (2004): Arbeitsbedingungen in KMU - Eine mulitvariate Anaylse. In: Institut für Mittelstandsforschung Bonn (Hrsg.): Jahrbuch zur Mittelstandsforschung 1/2004, Wiesbaden 2004, S. 1-20. Williams, J.E./Best, D.L. (1982): Measuring Sex Stereotypes. A Thirty-Nation Study, Beverly Hills, 1982. Wooldridge, J.M. (2003): Introductory Econometrics. A Modern Approach, Mason 2003. Wolpin, K. (1977): Education and Screening. In: American Economic Review, Vol. 67 (1977), S. 949-958.

29 Table 1: Details of the business start-up exhibitions surveys Essen START-Messe

Dresden KarriereStart

Berlin deGUT

Total

26.-28.9.2003

23.-25.1.2004

23.-25.4.2004

2003/2004

First survey Survey period Visitors

12,000

18,000

10,500

40,500

Respondents

1,364

417

705

2,486

Sample in %

11.4

2.3

6.7

6.1

1,025

307

439

1,771

Respondents not yet self-employed Follow-up survey Survey period

24.6.-25.7.2004

23.10.23.12.2004

21.1.-18.3.2004

2004/2005

Posted questionnaires

648

152

194

994

Returned questionnaires

322

85

98

505

Response rate in %

49.7

55.9

50.5

50.8

30 Table 2: Descriptive statistics (Part I): All (not self-employed) males and females in the first survey (t-1) All males

All females

Mean (Std. Dev.)

Mean (Std. Dev.)

Test of H0: Difference in means = 0 (t-value)

Variables (measured in t-1):


Age

Age (squared)




Higher education




Combining work and family

Autonomy



Married

Children

Number of cases1 (ratio)

2. 23589509 (0. ) 36. 4332 (9. 9216) 14250. 7151 (760. 8568) 0. 1560 (0. 3630) 0. 7304 (0. 4439) 0. 2441 (0. 4297) 0. 5309 (0. 4993) 1. 1504 (0. 5435) 0. 5535 (0. 4974) 0. 5754 (0. 4946) 0. 3547 (0. 4786) 0. 8783 (0. 3271) 0. 2625 (0. 4402) 0. 7159 (0. 4513) 0. 4467 (0. 4974) 0. 4676 (0. 4992) 1. 028 (0. 5857)

2. 3978 (1. 0076) 35. 5249 (10. 1309) 1364. 5083 (740. 2299) 0. 0991 (0. 2990) 0. 6497 (0. 4775) 0. 2323 (0. 4226) 0. 4362 (0. 4963) 1. 0842 (0. 5013) 0. 5813 (0. 4938) 0. 5622 (0. 4956) 0. 4684 (0. 4995) 0. 9055 (0. 2928) 0. 2140 (0. 4105) 0. 7860 (0. 4105) 0. 3653 (0. 4818) 0. 4634 (0. 4991)

3.039*** 1.869* 1.676* 3.441*** 3.347*** 0.557 3.886*** 2.605*** 1.085 0.498 4.256*** 1.668* 2.158** 3.096*** 3.429*** 0.173

727 (0. 4143)

Note: Data stem from the IfM Founder Panel. 1 Note: 16 individuals did not state their sex. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent. Some variables have missing values.

31 Table 3: Descriptive statistics (Part II): Newly self-employed males and females in the followup survey (t) Self-employed in t (male) Mean (Std. Dev.)

Self-employed in t (female) Mean (Std. Dev.)




Age

Age (squared)




Higher education





Autonomy



Married

Children

Number of cases (ratio)1

1. 9042 (0. 8684) 38. 8144 (8. 9237) 1585. 3711 (716. 6259) 0. 1477 (0. 3569) 0. 9247 (0. 2653) 0. 2581 (0. 4399) 0. 5833 (0. 4956) 1. 2500 (0. 6156) 0. 6000 (0. 4925) 0. 5638 (0. 4986) 0. 3034 (0. 4623) 0. 8925 (0. 3115) 0. 2979 (0. 4598) 0. 8261 (0. 3811) 0. 5408 (0. 5009) 0. 5208 (0. 5022)

1. 8621 (0. 8045) 37. 9275 (9. 1077) 1520. 2464 (708. 0412) 0. 1364 (0. 3458) 0. 8657 (0. 3436) 0. 1912 (0. 3962) 0. 4265 (0. 4982) 1. 1688 (0. 4442) 0. 6769 (0. 4713) 0. 5714 (0. 4989) 0. 5254 (0. 5036) 0. 9091 (0. 2897) 0. 1515 (0. 3613) 0. 7813 (0. 4167) 0. 4203 (0. 4972) 0. 5652 (0. 4934)

98 (0. 3537)

69 (0. 3251)

0. 305 0. 624 0. 562 0. 199 1. 227 1. 011 1. 993** 1. 066 0. 996 0. 094 2. 713*** 0. 345 2. 251** 0. 684 1. 536 0. 562

Note: Data stem from the IfM Founder Panel. 1 Relating to the number of men and women which did not switch into self-employment between t-1 and t. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent. Some variables have missing values.

32 Table 4: Descriptive statistics (Part III): Male and female quitters in the follow-up survey (t) Male quitters

Female quitters

Mean (Std. Dev.)

Mean (Std. Dev.)

2. 6279 (1. 0241) 38. 5577 (11. 0105) 1605. 5962 (865. 5280) 0. 0851 (0. 2821) 0. 6512 (0. 4822) 0. 2157 (0. 4154) 0. 6346 (0. 4862) 1. 2308 (0. 5813) 0. 5122 (0. 5061) 0. 5476 (0. 5038) 0. 3000 (0. 4641) 0. 8780 (0. 3313) 0. 2000 (0. 4045) 0. 7333 (0. 4472) 0. 5192 (0. 5045) 0. 5385 (0. 5034)

2. 4706 (0. 8611) 34. 2200 (10. 3713) 1276. 4200 (720. 6670) 0. 0816 (0. 2766) 0. 5588 (0. 5040) 0. 3725 (0. 4883) 0. 4800 (0. 5047) 1. 0600 (0. 4243) 0. 3030 (0. 4667) 0. 4688 (0. 5070) 0. 3939 (0. 4962) 0. 9063 (0. 2961) 0. 3438 (0. 4826) 0. 8182 (0. 3917) 0. 4000 (0. 4949) 0. 4694 (0. 5042)

52 (0. 3151)

51 (0. 4359)




Age

Age (squared)




Higher education





Autonomy



Married

Children

Number of cases (ratio)1

0.717 2.049** 2.091** 0.061 0.818 1.747* 1.575 1.699* 1.845* 0.665 0.834 0.378 1.376 0.890 1.204 0.493

Note: Data stem from the IfM Founder Panel. 1 Relating to the number of men and women which switched into self-employment between t-1 and t or still intend to switch in the near future. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent. Some variables have missing values.

33 Table 5: Ordered Probit Estimates for the willingness to become self-employed in t-1 Specification I

Specification II

Specification III

Coeff. (Std. Dev.)

Coeff. (Std. Dev.)

Coeff. (Std. Dev.)


Female

Age

Age (squared)




Higher education





Autonomy



Married

Children


Entrepreneurial self-perception x Female


-0. 0915 (0. 0695) 0. 0670*** (0. 0257) -0. 0008** (0. 0003) 0. 0334 (0. 1028) 0, 5462*** (0, 0746) 0. 1072 (0. 0778) -0. 0812 (0. 0740) 0. 1199* (0. 0695) 0. 1478** (0. 0686) 0. 0240 (0. 0694) 0. 1770** (0. 0715) 0. 2915*** (0. 1080) 0. 1156 (0. 0793) 0. 3385*** (0. 0767) -0. 0742 (0. 0798) -0. 0038 (0. 0842)

-1351. 07 (132. 22)***

-0. 0615 (0. 0719) 0. 0527** (0. 0268) -0. 0006* (0. 0003) -0. 0657 (0. 1074) 0, 4452*** (0, 0782) 0. 0193 (0. 0803) -0. 1071 (0. 0767) 0. 1304* (0. 0715) 0. 1998*** (0. 0711) -0. 0025 (0. 0719) 0. 1470** (0. 0738) 0. 1889* (0. 1112) 0. 0069 (0. 0821) 0. 3078*** (0. 0796) -0. 0751 (0. 0820) -0. 0597 (0. 0872) -0. 3709*** (0. 0392)

-1240. 85 (220. 72)***

Note: Data stem from the IfM Founder Panel. All regressions include regional dummies. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent.

-0. 1883 (0. 1913) 0. 0533** (0. 0268) -0. 0006* (0. 0003) -0. 0684 (0. 1075) 0. 4465*** (0. 0782) 0. 0205 (0. 0803) -0. 1092 (0. 0768) 0. 1320* (0. 0715) 0. 2022*** (0. 0712) -0. 0001 (0. 0720) 0. 1461** (0. 0738) 0. 1904* (0. 1113) 0. 0076 (0. 0821) 0. 3087*** (0. 0796) -0. 0783 (0. 0822) -0. 0588 (0. 0872) -0. 3907*** (0. 0481) 0. 0534 (0. 0747) -1240. 60 (221. 23)***

34 Table 6: Logit Estimates for the probability to switch into self-employment until t Specification I

Specification II

Specification III

Coeff. (Std. Dev.)

Coeff. (Std. Dev.)

Coeff. (Std. Dev.)


Female

Age

Age (squared)




Higher education





Autonomy



Married

Children

Second household income


Entrepreneurial self-perception x Female


0. 1067 (0. 2597) 0. 1460 (0. 1032) -0. 0021 (0. 0013) 0. 1986 (0. 3628) 1, 3116*** (0, 3760) -0. 0645 (0. 2934) 0. 0271 (0. 2732) -0. 1394 (0. 2514) 0. 5926** (0. 2653) -0. 1904 (0. 2537) 0. 0928 (0. 2661) 0. 4715 (0. 4038) 0. 2866 (0. 2991) -0. 1215 (0. 3267) -0. 1708 (0. 3001) 0. 1522 (0. 3198) 0. 9535*** (0. 2667)

-203. 16 (44. 75)***

-0. 1079 (0. 2727) 0. 1876* (0. 1137) -0. 0026* (0. 0014) 0. 0435 (0. 3750) 1. 1342*** (0. 3943) -0. 0516 (0. 3029) -0. 0337 (0. 2821) -0. 0446 (0. 2642) 0. 6703** (0. 2782) -0. 2434 (0. 2641) 0. 1236 (0. 2750) 0. 2913 (0. 4157) 0. 2575 (0. 3054) -0. 2507 (0. 3419) -0. 2255 (0. 3120) -0. 0784 (0. 3298) 0. 9535*** (0. 2794) -0. 4805*** (0. 1672)

-190. 94 (53. 43)***

Note: Data stem from the IfM Founder Panel. All regressions include regional dummies. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent.

0. 2601 (0. 7530) 0. 1847 (0. 1137) -0. 0026* (0. 0015) 0. 0412 (0. 3749) 1. 1343*** (0. 3948) -0. 0372 (0. 3040) -0. 0276 (0. 2828) 0. 0463 (0. 2651) 0. 6556** (0. 2795) -0. 2498 (0. 2645) 0. 1238 (0. 2749) 0. 2788 (0. 4158) 0. 2459 (0. 3061) -0. 2681 (0. 3432) -0. 2131 (0. 3131) 0. 0713 (0. 3300) 0. 9492*** (0. 2797) -0. 4248** (0. 1933) -0. 1861 (0. 3556) -190. 80 (53. 71)***

35 Table 7: Logit Estimates: Interaction terms for the impact of gender on entrepreneurial activity Willingness to become self-employed in t-1

Probability to switch into self-employment between t-1 and t

Coeff. (Std. Dev.)

Coeff. (Std. Dev.)

Variables (measured in t-1)

Entrepreneurial self-perception x female

Combining work with family responsibilities x female Need for achievement x female

Internal locus of control x female

Industry specific experience x female

Parent(s) self-employed x female

age x female

Self-employment experience x female


0. 0918 (0. 0797) 0. 2875** (0. 1461) -0. 1672 (0. 1715) -0. 0039 (0. 1679) -0. 2315 (0. 1567) 0. 3331** (0. 1678) 0. 0035 (0. 0081) -0. 0493 (0. 2426) -1235. 42 (231. 57)***

Note: Data stem from the IfM Founder Panel. All regressions include the complete set of control variables (main effects). *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent.

-0. 2872 (0. 3490) 0. 9162* (0. 5401) -0. 8813 (0. 6625) -0. 7692 (0. 6715) -0. 1740 (0. 7825) -0. 3278 (0. 6091) -0. 0157 (0. 0326) -0. 8647 (0. 7981) -198. 26 (49. 39)***

36 Table 8: Descriptive statistics (Part IV): Start-up characteristics by gender Self-employed in t (male)

Self-employed in t (female)

Mean (Std. Dev.)

Mean (Std. Dev.)

0. 2143 (0. 4124) 47. 8646 (19. 1770) 0. 1327 (0. 3409) 0. 6979 (0. 4616) 0. 2551 (0. 4382) 0. 1837 (0. 3892) 0. 1771 (0. 3837) 0. 3021 (0. 4616) 0. 3333 (0. 4739)

0. 1159 (0. 3225) 39. 7537 (18. 1033) 0. 1940 (0. 3984) 0. 6923 (0. 4651) 0. 1159 (0. 3225) 0. 0725 (0. 2612) 0. 0896 (0. 2870) 0. 1940 (0. 3984) 0. 6418 (0. 4831)

98

69


Start-up characteristics (measured in t)

Team start-up (1=”yes”; 0=”no”)

Weekly working hours (metric)

Second job (1=yes; 0=”no”)

Second household income (1=yes; 0=”no”)

Hired employees (1=yes; 0=”no”)

Construction (1=yes; 0=”no”)

Retail trade (1=yes; 0=”no”)

Business and professional services

Other services (1=yes; 0=”no”)

(1=yes; 0=”no”)

Number of cases

1. 727* 2. 746** 1. 029 0. 075 2. 364** 2. 209** 1. 663* 1. 554 4. 056***

Note: Data stem from the IfM Founder Panel. *, **, *** denote statistical significance at an error level of 10, 5, and 1 percent. Some variables have missing values.