Identity And Entrepreneurship - CiteSeerX

4 downloads 267 Views 182KB Size Report
Identity And Entrepreneurship: Do Peers At School Shape ..... and attractive success stories, and also often receive individual-level technical advice and assistance in starting up a business. ...... Akerlof, G. A. and R. E. Kranton (2005). “Identity ...
Program on Education Policy and Governance Working Papers Series

Identity And Entrepreneurship: Do Peers At School Shape Entrepreneurial Intentions?

Oliver Falck, Stephan Heblich, Elke Luedemann PEPG 09-05 November

Program on Education Policy and Governance Harvard Kennedy School 79 JFK Street, Taubman 304 Cambridge, MA 02138 Tel: 617-495-7976 Fax: 617-496-4428 www.ksg.harvard.edu/pepg/

DO NOT CITE WITHOUT PRIOR PERMISSION FROM AUTHORS

Identity And Entrepreneurship: Do Peers At School Shape Entrepreneurial Intentions?

Oliver Falck+, Stephan Heblich*, Elke Luedemann‡

November 2009

+

Ifo Institute for Economic Research, Poschingerstr. 5, D-81679 Munich (Germany), Phone: +49 89 9224 1370, Fax: +49 89 9224 1460, Email: [email protected], CESifo and Max Planck Institute of Economics.

*

Max Planck Institute of Economics, Entrepreneurship, Growth, and Public Policy Group, Kahlaischestr. 10, D-07745 Jena (Germany), Phone: +49 3641 686 733, Fax: +49 3641 686 710, Email: [email protected]. ‡ Ifo Institute for Economic Research, Poschingerstr. 5, D-81679 Munich (Germany), Phone: +49 89 9224 1369, Fax: +49 89 9224 1460, Email: [email protected] Acknowledgements: We are grateful to David Audretsch, Lee Flemming, Oliver Kirchkamp, Mirjam van Praag, Simon Parker, Olav Sorenson, and Ludger Wößmann for helpful comments on an earlier version of this paper. We also would like to thank the conference participants at the Spring 2009 Meeting Of Young Economists in Istanbul, the International Institute of Public Finance (IIPF) 2009 conference in Capetown, and the 2009 Conference of the German Economic Association (Verein fuer Socialpolitik) in Magdeburg for many helpful comments. Daniel Erdsiek provided capable research assistance. This research was completed during Oliver Falck’s research visit to the Program of Education Policy and Governance (PEPG) at the Taubman Center, Kennedy School of Governance, Harvard University. Oliver Falck is indebted to Paul E. Peterson and Edward Glaeser for their invitation. This research visit was partly funded by the Fritz Thyssen Foundation. Stephan Heblich received funding for the research leading to these results from the European Community’s Seventh Framework Programme (FP7/2007-2013) under grant no. 216813.

1

Identity And Entrepreneurship: Do Peers At School Shape Entrepreneurial Intentions?

Abstract We incorporate the concept of social identity into entrepreneurship and analyze whether an individual’s identity affects his or her decision to become entrepreneur. We argue that an entrepreneurial identity results from an individual’s socialization. This could be parental influence but, as argued in this paper, also peer influence. Based on PISA 2006 data in which students report their entrepreneurial intentions at the age of 15, we find that having an entrepreneurial peer group has a positive effect on an individual’s entrepreneurial intentions. We find that the strength of the peer effect in a country is moderated by prevailing values, namely individualism. Keywords: Social identity, entrepreneurship, peer effects JEL codes: J24, L26, Z13

2

1. Introduction There is a great deal of research into what motivates an individual to become an entrepreneur, taking into consideration extant conditions and institutions that either support or impede entrepreneurial intentions.1 One major finding in this research is that entrepreneurs are willing to accept a lower expected income than what standard economic models of occupational choice would imply (Evans and Leighton 1989; Hamilton 2000), suggesting that there is some type of nonpecuniary value to being an entrepreneur that, at first glance, makes little sense from a standard economic perspective. However, drawing on well-established insights from the fields of sociology and psychology, Akerlof and Kranton (2000) introduced the concept of identity, meaning a person’s self-image, into an economic utility function. They argue that individuals earn additional utility from an identity that matches their ideals. Eventually, this nonmonetary incentive can explain occupational choices that vary from what would be optimum for a rational actor in a standard economic model. Following Akerlof and Kranton’s concept of identity, this paper chooses entrepreneurship as setting and argues that identity plays an important role in explaining an individual’s intention to become an entrepreneur, with consequent effects on his or her economic future. To analyze where an entrepreneurial identity actually comes from, we start with the existing literature on the intergenerational transmission of entrepreneurship (cf. Aldrich et al. 1998; Hout and Rosen 2000; Johnson 2002). A large body of literature in this field shows that children who grow up in an entrepreneurial household are more likely to become entrepreneurs themselves. Of course, this association can only be partly ascribed to the intergenerational transmission of an entrepreneurial identity. For example, Björklund et al. (2007) further emphasize that both nature and nurture are at play in the intergenerational transmission of a socioeconomic status. Nicolaou et al. (2008) and Nicolaou and Shane (2009) analyze this in the context of entrepreneurship and their results indeed suggest that genetic factors provide an important explanation for entrepreneurial activities as well. However, in this study we do not further consider these innate abilities as our focus lies on

1

For instance, Kihlstrom and Laffont (1979) analyze occupational choice with regard to an individual’s risk aversion, Lucas (1978) considers innate abilities, and Lazear (2005) stresses the importance of an individual’s mix of skills. Yet others analyze the impact of external constraints (e.g., Holtz-Eakin et al. 1994; Michelacci and Silva 2007), social contacts (e.g., Bauernschuster et al. 2008; Stuart and Sorenson 2005). For an extensive overview, see Parker (2004).

3

developed characteristics from adolescent work experience in the family business (cf. Fairlie and Robb 2007), parental role modeling (cf. Sørenson 2007), or, as we argue, peer effects.2 The role of peers for becoming an entrepreneur is considerably less prominent in the literature. Existing studies on the role of peers for engaging in entrepreneurial activities have focused on older individuals, either in higher education or after labor market entry. For instance, based on a registerbased longitudinal Swedish data set, Nanda and Sorenson (2009) study peer effects in the workplace, and show that having co-workers that have had prior entrepreneurial experiences increases the likelihood of becoming an entrepreneur. Using data from Harvard Business School, Lerner and Malmendier (2007) find evidence of peer effects on MBA students’ individual entrepreneurial activity. It has also been shown that individuals who work at venture-backed firms are more likely to become entrepreneurs (Gompers et al. 2005). Again, this association can only be partly ascribed to the transmission of an entrepreneurial identity. For instance, it might also be the case that peers at work or at business school provide financial aid when it comes to the foundation of a new venture (cf. Sanders and Nee 1996). Yet, Giannetti and Simonov (2009) investigate peer effects at the neighborhood level on the decision to become an entrepreneur, and conclude that peer effects create non-pecuniary benefits from entrepreneurial activity, which is in line with our idea of peers forming an individual’s entrepreneurial identity. To concentrate on the transmission of an entrepreneurial identity, we look, in this paper, on the role of peers for developing an entrepreneurial intention at younger ages. We argue that at younger ages, say at the age of 15, adolescents have, for the most part, not collected any labor market experiences, and hence do not yet know whether they possess the necessary abilities to become a successful entrepreneur. In other words, we assume that their intentions to become an entrepreneur are not driven by ability considerations, but rather based on their identity which is influenced by parents’ and peers’ entrepreneurial identity. To our knowledge, we are the first to analyze the impact of parents and peers on entrepreneurial intentions instead of entrepreneurial activity. To identify the effect of peers empirically, we employ PISA 2006 data in which students report their entrepreneurial intentions at the age of 15 and find evidence that, after controlling for entrepreneurial parents, entrepreneurial peers still increase the likelihood that an individual will have entrepreneurial intentions. However, given our identification 2

With the exception of ruling out fields in which firm succession of children is more common (e.g., agriculture), we do not attempt to distinguish between the intergenerational transmission of identity either stemming from work experience

4

strategy, we can not distinguish between endogenous peer effects and contextual effects, but we can plausibly rule out correlated effects (Manski 1995). Moreover, we find some evidence that the strength of this influence varies across countries. When we complement our research by looking at data from the World Value Survey, we find that this variation across countries is associated to some extent with the role of individualism in the specific society. The rest of the paper is organized as follows. In Section 2, we first sketch our hypothesized connection between peers and the formation of an entrepreneurial identity. Focusing on how an entrepreneurial identity originates then lays the groundwork for our idea that a student’s social environment will play an important role in the formation of his or her entrepreneurial identity. In Section 3, we describe our data and then, in Section 4, we set out the empirical strategy we use to test our hypothesis of the impact of peers on the formation of an entrepreneurial identity. Section 5 presents the results of our empirical analysis and then, Section 6 concludes.

2. The Formation of an Entrepreneurial Identity Identity as a Contribution to the Entrepreneurship Literature Although a familiar and well-developed concept in the fields of psychology and sociology, identity, defined as a person’s sense of self, has not attracted too much interest among economists,3 until Akerlof and Kranton (2000) introduced the concept to this field.4 They argue that an individual’s utility might not only be determined exclusively by individual considerations but also influenced by social desirability considerations, i.e., by an individual’s view of who he or she is and what the individual and others should or should not do to live up to this ideal concept of the self. One area where the influence of identity on behavior and economic outcomes is likely to provide additional explanatory power is the field of entrepreneurship. Research in this field owes a large debt to the seminal contributions of Schumpeter (1912) and Knight (1921), which, when taken together, comprise an entrepreneur’s most essential attributes: innovativeness, opportunity recognition, and acceptance of a certain degree of risk (Baumol 1968). In a nutshell, Schumpeter sees the independent entrepreneur as the ultimate source of economic development by being the one

in the parental business or from a parental role modelling. We rather focus on the role of peers in the formation of an entrepreneurial identity. 3 “Because of its explanatory power, numerous scholars in psychology, sociology, political science, anthropology, and history have adopted identity as a central concept. This paper shows how identity can be brought into economic analysis, allowing a new view of many economic problems” (Akerlof and Kranton 2000: 716). 4 An exemption is the work by Sen (1977).

5

who recognizes the potential of an invention and introduces it to the market. By innovating, the entrepreneur initiates a process of creative destruction in which the new constantly replaces the old. With time, this ongoing crowding-out process guarantees that resources are shifted to the most productive sectors. As to what drives the entrepreneurial spirit, Schumpeter rather romantically describes it as “the will to conquer,” “the dream and the will to found a private kingdom,” and “the joy of creating, of getting things done” (1912: 93). From a standard economic perspective, the Schumpeterian motivation for entrepreneurial action— conquering, founding, and creating—that helps the entrepreneur to overcome Knightian (1921) uncertainty inherent in the endeavor seems rather lyrical than theory driven. Arrow (1962) provides some weightier economic reasoning when he argues that, under uncertainty, information becomes a commodity with economic value, implying that those individuals who find a way to overcome uncertainty can appropriate a pioneer rent that is in itself an incentive to engage in entrepreneurial action (Kanbur and Ravi 1990; Hamilton 2000). However true this may be, it still does not answer the crucial question of why some people manage to overcome uncertainty better and more successfully than others, that is: Why are some people more entrepreneurial than others? We argue that the concept of identity can help answer this question. The answer cannot be found in a purely economic environment, however, but will need to be looked for in an interdisciplinary arena. On the supra-individual level, the quest leads us to sociological network theory, which stresses the importance of social embeddedness (Coleman 1988; Granovetter 1985; Hayek 1937). According to this theory, social networks provide access to information that makes the future more predictable and thus decreases uncertainty (cf. Bauernschuster et al. 2008; Sanders and Nee 1996; Stuart and Sorenson 2005). On the individual level, we need to look to the emerging field of behavioral economics where psychology comes into play (Kahneman and Tversky, 1979).5 According to this field, many factors, including optimism, self-assessment, autonomy, and overall job satisfaction, influence the ideal that an individual will try to live up to (cf. Camerer and Lovallo 1999; Wu and Knott 2006). Thus, choosing an entrepreneurial identity means that an individual mentally frames certain situations as being entrepreneurial and then adjusts his or her behavior accordingly. The individual tries to live up to, or emulate, a real or imagined character, for example, Bill Gates or the Schumpeterian Entrepreneur, who is a personification of the

5

Note in this regard that there is a growing literature on entrepreneurial behavior in the field of strategic management research that focuses on entrepreneurial behavior from a psychological perspective. See, e.g., Baron (1998) and Mitchell and Shepherd (2008).

6

entrepreneur he or she would like to be. Consequently, the would-be entrepreneur suffers a loss in utility if his or her behavior strays from this ideal, thus determining the individual’s situationdependent utility (Kahneman and Tversky 1979). The Transmission of an Entrepreneurial Identiy It is nearly standard practice now for business schools to offer entrepreneurship courses, in which students learn how to write business plans, meet successful entrepreneurs who tell their powerful and attractive success stories, and also often receive individual-level technical advice and assistance in starting up a business.6 However, having an entrepreneurial identity, as we define it in this paper, is not something that can be taught in one term—even if the course is very practice-oriented:7 students can be taught specific practices and techniques, but they cannot be schooled in the famous “will to conquer.” This essential attribute of entrepreneurship is part of a persons’ identity that gradually developed out of a person’s background and experience.8 Economic research on what factors drive the formation of cognitive and non-cognitive skills usually adopts a life-cycle perspective, and stresses the important influence of experiences during early childhood years (cf. Heckman 2006; Cunha and Heckman 2007)9. The reasoning for why early childhood experiences have proven so important is that later investments in skills build on foundations that are laid down earlier. The fact that young children spend most of their time with their parents can help explain the strong impact of parental background on educational attainment and student performance we observe across countries all over the world (e.g. OECD 2007a, b). Considering that identity, along with various skills, results from an individual’s socialization at home and in school (cf. Akerlof and Kranton 2005: 12), it seems plausible to assume that an entrepreneurially-inclined identity is similarly developed (cf. Halaby 2003; Johnson 2002; Mortimer and Lorence 1979).

6

See Kuratko (2005) for further details about the emergence of entrepreneurship education. See also the European Commission’s (2006) report on “Entrepreneurship Education in Europe” and Oosterbeek et al. (2008) for an empirical attempt to evaluate the impact of entrepreneurship education. 7 In support of this statement, note that Oosterbeek et al. (2008) find that a leading entrepreneurship education program has no effect on college students’ intention to become an entrepreneur. Their empirical analysis is based on differencein-differences methodology. 8 “The power of example to activate and channel behavior has been abundantly documented. … One can get people to … converse on particular topics, to be inquisitive or passive, to think innovatively or conventionally, and to engage in almost any course or action by having such conduct exemplified” (Bandura 1986: 206). 9 However, it is important to note that independent research from the fields of developmental psychology and neuroscience emphasizes the role of early childhood experiences as well (cf. Heckman, 2006).

7

An individual’s parents are the initial role models and should thus have a seminal influence on the child’s self-image across the lifespan and if the parents are entrepreneurial it is quite possible that their child will choose that sort of identity also (cf. Bandura 1977).10 Or, as Marshall (1920) put it, “as years pass on, the child of the working man learns a great deal from what he sees and hears going on around him.” This transmission channel is well accepted and analyzed in the field of entrepreneurship (cf. Aldrich et al. 1998, Dunn and Holtz-Eakin 2000, Fairlie and Robb 2007; Hout and Rosen 2000; or Lentz and Laband 1990). Once entered, the informal school environment, with its various social categories and expectations (e.g., nerds, jocks, or burnouts as described in Coleman 1961), along with the formal school philosophy geared toward producing “certain types of human beings” (Bloom 1987: 26) are additional critical influences on the development of identity.11 Based on interviews, Eckert (1995) for example found that the jocks’ lives lie between the boundaries of the school and its extracurricular activities. This suggests that the influence of peers at school is more pronounced than neighborhood effects. It is largely how well one does at school, academically or socially, that determines one’s future occupation. So even if we assume that children of school age are rather not aware of which occupation would ideally complement her skills and thus earn the highest future returns, there are still identities formed that lead to ideas about the ideal job (cf. Bandura 1977). The underlying identities arise from assignment to social categories where children strive to live up to certain ideas. To illustrate this, let us assume that a certain fraction of a child’s peers think of themselves and others as future entrepreneurs, although perhaps not in that exact terminology at this point. These peers think it would be “cool” to be your own boss, run your own business, and not have to take orders from anyone else. Let us assume that these particular children are quite adventurous and generally fun to hang out with, characteristics that presumably make them “leaders of the pack.” And leadership, argues Baumol (1968), is one of the major ingredients for entrepreneurial success.12 And since these entrepreneurial peers are so attractive and fun to be around, it is plausible that they could have a great deal of influence on other children’s identity choice. This is the central 10

This assumption is also in line with findings by the Harvard Center for Entrepreneurial History. Miller (1952) and also Neu and Gregory (1952) both find that the most influencing business men during in the period of the great American Industrialization from 1870-1910 came from landowning or entrepreneurial families. 11 “The people with whom one regularly associates, either through preference or imposition, delimit the behavioral patterns that will be repeatedly observed, and hence, learned most thoroughly” (Bandura 1986: 55).

8

hypothesis which we put to an empirical test in an attempt to learn more about the role of peers in the formation of an entrepreneurial identity. We assume that children, by living up to their current intentions and aspirations in the school yard, playfully reinforce their identity which then drives long-term intentions and aspirations like the occupational choice. If this assumption turns out to be true and if the identity from belonging to a social category does indeed drive long-term intentions, it contains a meaningful implication because then the early developed identity patterns influence social welfare through intentions and aspirations. Of course, the importance of peers in the formation of an entrepreneurial identity is country-specific and plausibly depends on culture and social norms. Indeed, there is a large body of literature arguing that culture and social norms shape economic outcomes in a country. This strand of literature distinguishes between beliefs and preferences through which culture can affect economic outcomes while the standard economic assumption that each individual has one identity and maximizes the utility from this identity is maintained (cf. Guiso et al. 2006). Additionally considering the possibility of multiple identities (Akerlof and Kranton 2000) might then lead to an even more pronounced role of culture. This is because culture might influence the importance of different rolemodels which together shape an individual’s identity. Accordingly, a child’s decision of whether to follow in the parents’ footsteps or step down the same path as his or her peers might also be country specific and dependent on each a country’s social structures. In the following empirical analyses, we will also consider and test this possibility.

3. Data We use data from the 2006 cycle of the Programme for International Student Assessment (PISA) (OECD, 2007a, 2007b) to empirically test our model. The main objective of PISA is to assess the scientific, mathematical, and reading literacy of the student population in each of the participating countries. PISA is a representative sample of all 15-year-olds enrolled in school. Thus, in most of the countries assessed, the target population comprises young people near the end of their compulsory schooling. As for the PISA sampling procedure, most countries employ a two-stage sampling technique. The first stage draws a (usually stratified) random sample of schools in which 15-year-old students are enrolled. In the second stage, a random sample of usually 35 of the 15-year-

12

The entrepreneur’s job is “to locate new ideas and to put them into effect. He must lead, perhaps even inspire; he cannot allow things to get into a rut and for him today’s practice is never good enough for tomorrow. … He is the individual who exercises what in the business literature is called ‘leadership’” (Baumol 1968: 65).

9

old students in each of these schools is drawn, with each 15-year-old student in a school having an equal chance of being selected. Our main variable of interest is the students’ response to the question of what kind of job they intend to have when they are about 30 years old. Students are asked to write down the job title, which is then given a four-digit ISCO-88 code (International Standard Classification of Occupations; ILO 1990). One might argue that this variable is an unsuitable predictor for students’ future occupational choice because it only measures their intentions to become entrepreneurs. However, based on the longitudinal British Cohort Study (BCS), we show in the appendix of this paper that the occupational intentions at the age of 16 are indeed a good predictor for the future occupational choice (cf. Appendix A1). The students also provide information on their mother’s and father’s occupation, which again is given a four digit ISCO-88 code. For both the students’ intended occupations and the parents’ actual occupations we construct a dummy variable that takes the value 1 if the occupation is entrepreneurial and 0 otherwise. We employ two different definitions of an entrepreneurial occupation, as shown in Table 1: (i) a broad definition of entrepreneurial occupation containing all ISCO-88 codes starting with 13xx (Definition 1), and (ii) a more restrictive definition excluding agriculture, forestry, and fishing professions (Definition 2). All codes reflect occupations that are related to running small enterprises. Indeed, running a small business is commonly regarded as a good proxy for entrepreneurship (cf. Parker 2009). Our reason for making this distinction is that entrepreneurship in agriculture differs from that in other fields with respect to the share of individuals who run their own business. Moreover, the importance of agriculture differs across countries. Due to tradition and institutions, firm succession by children—often the son—is more common in agriculture than in other fields.

10

Table 1. Classification of entrepreneurial occupations according to definition 1 and 2 ISCO-88 Code 1300 1310 1311

1312 1313 1314

1315

1316 1317 1318 1319

[SMALL ENTERPRISE] GENERAL MANAGERS [SMALL ENTERPRISE] GENERAL MANAGERS [incl. Businessman, Trader, Manager nfs] [Small enterprise] General managers agriculture, forestry & fishing [incl. Farm Manager, Self-employed Farmer with personnel] [Small enterprise] General managers manufacturing [Small enterprise] General managers construction [incl. Building Contractor] [Small enterprise] General managers wholesale & retail trade [incl. Shop Owner/Manager, Retail Owner/Manager, Merchant] [Small enterprise] General managers restaurants & hotels [incl. Manager Camping Site, Bar Owner/Manager, Restaurateur] [Small enterprise] General managers transp., storage, & communications [incl. Owner Small Transport Company] [Small enterprise] General managers business services [incl. Manager Insurance Agency] [Small enterprise] General managers personal care, cleaning, etc. services [incl. Owner Laundry] [Small enterprise] General managers nec [incl. Manager Travel Agency, Manager Fitness Center, Garage Owner]

Definition 1 9 9

Definition 2 9 9

9 9 9

9 9

9

9

9

9

9

9

9

9

9

9

9

9

Table 2 contains descriptive statistics of the share of students intending to be an entrepreneur at age 30 and the share of students with parents in an entrepreneurial occupation.

11

Table 2. Descriptive Statistics: percentage of students intending to be in an entrepreneurial occupation, and percentage of students with parents in an entrepreneurial occupation, by country Percentage of students who report that...

AUS AUT BEL CAN CZE DEU DNK ESP FIN GBR GRC HUN IRL ISL ITA JPN KOR LUX MEX NLD NOR NZL POL PRT SVK SWE TUR USA All 28

Percentage of missing values

either parent is an entrepreneur (definition 1)

either parent is an entrepreneur (definition 2)

they intend to be in an entrepreneuria l occupation at age 30 (definition 1)

they intend to be in an entrepreneuri al occupation at age 30 (definition 2)

parents' occupation

students' intended occupation

10.92% 7.43% 9.69% 9.62% 21.16% 4.18% 2.30% 7.30% 4.20% 13.70% 19.36% 4.37% 13.72% 9.27% 17.80% 7.23% 29.24% 8.87% 3.30% 12.60% 11.40% 17.75% 9.60% 2.26% 6.02% 16.78% 16.18% 9.52% 10.96%

10.69% 7.40% 9.33% 6.69% 20.56% 4.16% 2.19% 7.28% 4.20% 13.46% 19.06% 4.15% 13.28% 9.23% 17.01% 7.23% 28.87% 8.70% 3.27% 10.58% 11.21% 14.43% 9.54% 2.26% 5.74% 16.04% 16.18% 9.27% 10.46%

3.21% 1.56% 3.47% 1.27% 8.09% 0.92% 0.73% 1.79% 1.61% 3.21% 1.88% 1.28% 3.25% 1.00% 4.29% 0.71% 3.54% 2.36% 1.01% 7.11% 2.03% 5.00% 1.46% 0.19% 5.78% 4.60% 2.83% 2.77% 2.71%

3.17% 1.56% 3.31% 0.87% 7.76% 0.90% 0.73% 1.78% 1.61% 3.15% 1.86% 1.28% 3.15% 1.00% 4.19% 0.71% 3.52% 2.36% 0.99% 6.50% 2.01% 4.34% 1.46% 0.19% 5.75% 4.53% 2.83% 2.75% 2.62%

1.94% 0.80% 1.09% 1.58% 1.60% 5.17% 1.57% 0.71% 0.81% 3.92% 1.01% 1.15% 2.79% 2.51% 0.79% 2.67% 0.53% 1.35% 1.37% 1.29% 2.71% 2.09% 0.81% 0.48% 0.77% 1.14% 0.77% 3.54%

13.45% 17.63% 9.57% 5.04% 20.86% 21.01% 6.33% 20.02% 9.55% 6.75% 20.52% 18.37% 10.42% 23.68% 7.96% 13.73% 3.44% 7.47% 16.92% 3.70% 19.07% 12.26% 12.72% 4.04% 12.77% 8.58% 20.44% 7.19%

Note: based on PISA2006, not imputed data, weighted by the inverse of students' sampling probabilities

For the purpose of the following analyses, observations with missing values for either students’ entrepreneurial intentions or their parents’ actual entrepreneurial status are dropped. Our dataset thus contains 204,074 students from 28 of the 30 OECD countries. Switzerland is excluded because no student from that country reported intending to be in any kind of an entrepreneurial occupation at age 30. As we control in most specifications for a large number of background variables to minimize 12

potential biases from omitted variables at the school level, we also dropped France from the sample because no school-level background information is provided for any of the schools sampled in this country. In addition, students provide detailed information on their personal characteristics and family backgrounds and school principals report details on their schools’ resource endowments and institutional settings. Thus, we are able to control for other influencing factors at the individual and school level respectively. In a first step, we entered a large set of control variables at the student and school level in our model. Then, using Wald tests, we tested which variables did not enter the regression equation jointly significantly, excluded them, and were thus left with a considerably smaller set of controls (cf. Appendix A2). Among the student and family background variables, there is information on the student’s gender, three indicators for the student’s immigrant status (namely, native, first and second generation immigrant students), and, finally, the scores from the student’s performance in science and mathematics. Regarding family background, we control for an indicator of family wealth as well as parents’ educational attainment. At the school level, we include four dummies as controls for the size of the community where the school is located, along with several aggregated measures of the schools’ socioeconomic composition. We also control for learning time in regular lessons provided to the student.13 Like any other survey data set, the PISA dataset contains missing values. Although the percentage of missing values is minor for almost any single control variable in our model, deletion of all student observations with a missing value on at least one variable would mean a severe reduction in sample size. We thus include missing dummies in all regressions and set the missing explanatory variables to zero if the respective variable is categorical, or replace the missing value by the weighted school (country) mean of the respective variable if it is continuous.

4. Econometric Model To assess whether peer groups have a causal effect on a student’s intention of being in an entrepreneurial occupation at age 30, we estimate cross-country regressions conditional on controls for different sets of background variables at the student and school levels. However, to be able to interpret the peer group effect causally, we need to go beyond a simple cross-sectional analysis and

13

The initial set of control variables was selected based on Fuchs and Wößmann (2007), who estimate an educational production function using PISA data.

13

need to show that the observed partial correlation between an individual’s entrepreneurial intention and her peers’ entrepreneurial intentions is in fact due to some form of social interaction. Manski (1995) identifies two broad forms of social interaction. The first refers to the case where youth behavior is influenced by the prevalence of a certain behavior in the group (endogenous effects). In our model of occupational choice, this would mean that an individual’s intention to become an entrepreneur is influenced by her peers’ intentions to become entrepreneurs. In the second form of social interaction, youth behavior is influenced by exogenous characteristics of the youth’s reference group. In our model, this would mean that an individual’s intention to become an entrepreneur is influenced by her peers’ background characteristics (so called exogenous or contextual effects). Such contextual effects might arise from students spending time at their peers’ homes and coming into contact with peers’ parents. In the presence of contextual effects, a student spending time at a peer’s home with entrepreneurial parents is going to be more exposed to entrepreneurship as a career option than a student who does not. On top of that, Manski (1995) raises the possibility of spurious estimates of peer-group effects that may be erroneously interpreted as true endogenous or contextual effects: the so-called correlated effects. These can arise when youths in the same reference group express the same occupational intentions because they share a common set of unobserved characteristics. The identification of a causal effect of having an entrepreneurial peer group on an individual’s entrepreneurial intentions as outcome is thus complicated by two potential types of endogeneity problems: first, a selection bias from non-random selection and/or sorting into peer groups and second unobserved correlated effects, i.e. Manski’s reflection problem. We address the problem of non-random selection of students into schools by using within-school variation in entrepreneurial peer groups, which we argue are formed randomly (following Ammermueller and Pischke, 2009). The reflection problem of students’ and their peer group’s entrepreneurial intentions being determined simultaneously can possibly be solved by estimating an instrumental variable approach or a reduced form equation. Gaviria and Raphael (2001), for instance, estimate a structural model, regressing individual behavior on the proportion of students in an individual’s school showing the same behavior, using peers’ parents background characteristics as an instrumental variable to address the simultaneity problem. However, for this to be a valid instrument, the authors have to assume that contextual effects are non-existent. They argue students who attend the same school are more likely to interact during school hours, and not at the students’ home. Therefore, they assume that background characteristics of peers’ parents do not have a direct impact on individual behavior. 14

Given this strong assumption, the authors can identify endogenous peer effects. Yet, in our context we believe that we cannot credibly make this assumption, since we rely on within school variation to solve the selection and sorting problem. Moreover, PISA data do not contain any information on how much time students spend in school per day which means that we cannot directly assess how plausible the assumption of the absence of contextual effects is. Therefore, we proceed on the assumption that peers in the same grade and school are relatively likely to also spend some time at each others’ home and use the more conservative strategy of estimating a reduced form equation. This means that we regress individual entrepreneurial intentions directly on peers’ parents’ entrepreneurial status. This approach relies on much weaker assumptions, but comes at the cost that we cannot separate exogenous and endogenous social interaction effects. However, we would like to point out that addressing the simultaneity problem by estimating a reduced form equation is common in the recent peer effects literature (cf. Ammermueller and Pischke, 2009), making no attempt at separating the two types of peer effects. To identify the causal effect of having an entrepreneurial peer group on an individual’s entrepreneurial intentions, we thus estimate the following linear probability model14: _________

OCC igspc = α ⋅ OCC

− igspc

parents

+ β ⋅ OCC igspc

parents

+ Bigspc γ + R spc δ + I spcθ + η c + μ pc + ν spc + λ gspc + ε igspc

(1)

where

OCCigspc conditional probability of student i in grade g, school s, school type p and country c intending to be in an entrepreneurial occupation at age 30 _________

OCC

−igspc

parents

: share of students at school/ grade with at least one parent who is an entrepreneur

(calculated after excluding student i)

OCCigspc

parents

: dummy for whether either one of student i’s parents is entrepreneur

and where, in some specifications, we include the following control variables

Bigspc : vector of family background variables Rspc : vector of school resource variables I spc : vector of institutional characteristics of school s

14

Because of the incidental parameters problem that arises when estimating fixed effects in non-linear models when group sizes are small (Neyman and Scott, 1948), we prefer to estimate linear probability models as opposed to a probit and logit models. However, we also estimated the country fixed effects specification using probit and logit models which yielded similar results in terms of direction and significance, but are not shown here. They are available from the authors upon request.

15

Our error term thus has four components at different levels of aggregation: a country level error component η c , a school-type-level component μ pc , a school level component ν spc , as well as an idiosyncratic error term ε ispc . In addition to country fixed effects, we use school type fixed effects and school fixed effects to address the problem of non-random selection and sorting of students to entrepreneurial peer groups at school and grade level. Many authors have argued that selection into schools is non-random, and have applied within-school estimators to identify causal effects of peer group characteristics on individual outcomes (e.g., Schneeweis and Winter-Ebmer 2007; Ammermueller and Pischke 2009). Although we additionally present the results from country fixed effects and school type fixed effects specifications, we consider the within-school estimator as our preferred specification. To be able to use the within-school estimator, we define the peer group as students attending the same grade level and school, following Schneeweis and Winter-Ebmer (2007). In contrast, in the country fixed effects and school type fixed effects specifications, we define the peer group as students attending the same school. To account for the two-stage survey sampling design, we use clustering robust linear regressions, where standard errors are clustered at the school level (cf. Moulton 1986; Deaton 1997: 74–78). Furthermore, we weight each student by the inverse of his or her sampling probability (DuMouchel and Duncan 1983; Wooldridge 2001). In all cross-country regressions, we also give equal weights to all countries. Given that we measure peer characteristics over a sub-sample of students in the grade or school, we consider β , our coefficient of the peer group, as a lower bound for the true causal effect of having an entrepreneurial peer group at school. In fact, Ammermueller and Pischke (2009) show that in this case, estimates of peer effects will be attenuated by a factor of

(N

sample

− 1)

(N actual − 1)

. The authors propose

two strategies to correct their estimate of the peer group effect. Unfortunately, both strategies are not feasible with our data for the following two reasons. First, we cannot adjust the coefficient estimate by the factor N actual because we do not observe it. Second, the authors propose an instrumental variables strategy, which, in our case, would mean that we instrument the students’ report of their parents’ actual occupation by their parents’ report of their actual occupation. Yet, the ‘Parent Questionnaire’ that was part of PISA 2006 was only administered in a small subsample of countries,

16

and there is a high proportion of missing values in parents’ responses15. Given these data limitations, we deem that the IV strategy would not solve but rather aggravate the problem of measurement error in our sample. As we cannot address the measurement error problem by using the suggested strategies, we try to reduce the attenuation bias indirectly by using only data from students that have a peer group of at least five students per grade (i.e. N sample ≥ 5 ) in the school fixed effects specification. Moreover, we can only use data from schools with at least two different grade levels in the PISA data, while observations from schools with only one grade level had to be discarded. To avoid problems due to grade repetition or skipping grades, we further restrict our observations to only those from the two most common grade levels in each country, i.e. grade nine and ten in most countries except Australia, where it is grade ten and eleven. For the school fixed effects specification, we are thus left with a sample of 52,783 students from 1937 schools in 14 countries16, which comes at the cost of a considerable reduction in sample size, but allows us to remove any bias from non-random selection of students into schools.

5. Results 5.1 The Impact of Entrepreneurial Peers on a Student’s Entrepreneurial Identity

Table 3 contains the estimation results. All specifications contain country fixed effects, and in all specifications, we find a significant positive effect of the entrepreneurial peer group on an individual’s entrepreneurial intentions. In particular, an increase of the share of peers having at least one parent who is entrepreneur by one standard deviation leads to an increase in an individual’s entrepreneurial intention by 0.4 percentage points. The effect is robust across various specifications. In particular, adding individual and school level covariates ( Bigspc , Rspc , and I spc in equation (1)) does not significantly alter the estimated coefficient. Given that, on average, in our sample only 2.7% of the students report having entrepreneurial intentions (cf. Table 2) our estimate for the peer group effect seems noteworthy. Moreover, the fact that including school type fixed effects (columns (3) and (4)) as well as school fixed effects (columns (5) and (6)) does not change the estimated coefficient indicates that nonrandom selection or sorting into schools is not an issue in our case. At first glance, this seems to 15

10 OECD countries administered the Parent Questionnaire, namely Germany, Denmark, Iceland, Italy, South Korea, Luxembourg, New Zealand, Poland, Portugal and Turkey. In this 10 countries subsample, 23.5 % contain missing values for parental occupation.

17

make little sense, given what we know from previous literature on the effects of peer groups on, e.g., academic achievement (e.g., Schneeweis and Winter-Ebmer 2007; Ammermueller and Pischke 2009). However, note that non-random sorting into school or grade level peer groups with respect to student ability or achievement would not necessarily pose a problem for our identification strategy. It would only bias our results if students with entrepreneurial intentions would choose to attend the same schools, or be sorted into the same schools. Moreover, the estimated effect of having an entrepreneurial peer group is very similar, irrespective of whether we use school or grade level peer groups. Lastly, it is interesting to note that the exclusion of three quarters of our observations in the school fixed effects specification, does not result in a different coefficient estimate.

Table 3. Impact of peer group and parents on individual entrepreneurial intentions

Share of peers having entrepreneurial parents (definition 1) Effect of an increase by 1 stdev in pp either parent is an entrepreneur (definition 1) Peer group defined at Country fixed effects School type fixed effects School fixed effects Background variables Observations Countries

(1) (2) 0.035*** 0.039*** (0.010) (0.010) 0.35 0.38 0.043*** 0.041*** (0.003) (0.002) School level 9 9

(3) (4) 0.039*** 0.038*** (0.009) (0.009) 0.39 0.38 0.043*** 0.041*** (0.003) (0.002) School level 9 9 9 9

No 204,074 28

no 204,074 28

yes 204,074 28

yes 204,074 28

(5) (6) 0.039*** 0.037*** (0.019) (0.019) 0.43 0.41 0.054*** 0.052*** (0.006) (0.005) Grade level 9 9 9 9 9 9 no yes 52,783 52,783 14 14

Notes: The Table shows regression coefficients from a linear probability model. Standard errors (clustered at school level) in parentheses; students weighted by inverse probability weights; each country weighted equally; for the background variable, see appendix A2. * significant at 10%; ** significant at 5%; *** significant at 1%

The reported results are based on Definition 1, the broader definition of an entrepreneurial occupation that includes the agricultural sector (cf. Table 1). In a specification based on Definition 2, we found that the results are insensitive to the definition of entrepreneurial occupation

used, i.e., we rule out the possibility that a correlation between parents’ actual entrepreneurial occupation and a child’s intended entrepreneurial occupation is simply driven by different countries having different shares of employment in the agricultural sector, where, due to a possibly more

16

The 14 countries are Australia, Austria, Belgium, Canada, the Czech Republic, Germany, Spain, Hungary, Italy, Luxembourg, the Netherlands, Portugal, Slovakia and Turkey.

18

traditional farm ownership succession, the impact of parents’ occupation on one’s own intended occupation may be stronger. We therefore restrict ourselves to Definition 1; the results based on Definition 2 are available from the authors upon request.

5.2 Heterogeneity between Countries

In this section, we investigate whether differences in the peer group influence on individual entrepreneurial intentions are related to differences in values between countries. To this end, we use data from the World Value Survey (2005) and hypothesize that the peer group’s impact is likely to be stronger in less individualistic countries. We measure individualism by the country averages of the following two items. •

I seek to be myself rather than follow others.



I decide my goals in life by myself.

The World Value Survey also contains two items measuring the importance of friends in life, and the extent to which individuals make an effort to live up to their friends’ expectations. But, as mentioned above, we estimate a reduced form equation and thus get an estimate of the total effect of having peers’ with entrepreneurial parents. The direct effect of peers’ entrepreneurial intentions on individuals’ entrepreneurial intentions is thus only imperfectly captured by our reduced form estimate, which is why we decided to use the above mentioned two measures instead. We thus estimate a slightly modified version of equation (1), where the share of peers’ having entrepreneurial parents is interacted with our two measures of individualism. As only 16 out of the 28 countries in our sample participated in the World Value Survey 2005, the sample size reduces accordingly.17

17

These countries are Australia, Canada, Germany, Spain, Finland, Great Britain, Italy, Japan, South Korea, Mexico, Netherlands, Norway, New Zealand, Poland, Sweden, Turkey, and USA.

19

Table 4. Individualism and the impact of the peer group on individual entrepreneurial intentions (1) (2) (3) (4) (5) (6) Share of peers having entrepreneurial 0.025** 0.025** 0.025** 0.025** 0.025** 0.024** parents (definition 1) (0.010) (0.010) (0.010) (0.010) (0.010) (0.010) Either parent is an entrepreneur (definition 0.039*** 0.037*** 0.039*** 0.037*** 0.039*** 0.037*** 1) (0.003) (0.003) (0.003) (0.003) (0.003) (0.003) Share of peers having entrepreneurial -0.058 -0.067 0.102 0.122 parents (definition 1) interacted with “I seek (0.055) (0.053) (0.101) (0.100) to be myself rather than follow others” (a) -0.117** -0.137** -0.201* -0.238** Share of peers having entrepreneurial (0.056) (0.0553) (0.105) (0.104) parents (definition 1) interacted with “I decide my goals in life by myself.” (a) 9 9 9 9 9 9 Country fixed effects 9 9 9 9 9 9 School type fixed effects Background variables no yes no yes no yes Observations 151,294 151,294 151,294 151,294 151,294 151,294 Countries 16 16 16 16 16 16 Notes: Standard errors (clustered at school level) in parentheses; students weighted using inverse probability weights; each country weighted equally. Peer group measured at the school level. For the background variable, see appendix A2. * significant at 10%; ** significant at 5%; *** significant at 1% (a) Responses are re-coded such that higher values indicate strong agreement. The variables from the World Value Survey were centered at the international mean, before computing the interaction term.

In line with our hypothesis, Table 4 shows that in those countries where, on average, more individuals report deciding about their goals in life by themselves the peer group influence seems weaker. However, we do not find a significant association with deciding about one’s goals in life by oneself and the peer group impact. Figures 1a and 1b depict these associations graphically: countryspecific estimates of the peer group influence on entrepreneurial intentions are plotted against the country average response to two items from the World Value Survey.

20

Figure 1a. Individualistic society and the peer effects on entrepreneurial intentions

Notes: The vertical axis shows the estimated peer group influence on entrepreneurial intentions. The regression model contains all control variables listed in the appendix A2 as well as school type fixed effects. The horizontal axis shows the country’s average response to the item “I seek to be myself rather than follow others.” from the World Value Survey Responses are re-coded such that higher values indicate strong agreement.

21

Figure 1b. Individualistic society and the peer effects on entrepreneurial intentions

Notes: The vertical axis shows the estimated peer group influence on entrepreneurial intentions. The regression model contains all control variables listed in the appendix A2 as well as school type fixed effects. The horizontal axis shows the country’s average response to the item “I decide my goals in life my myself.” from the World Value Survey Responses are re-coded such that higher values indicate strong agreement.

Of course, these associations between the country average responses from the World Value Survey and the estimates of the impact of the entrepreneurial peer group do not necessarily lend themselves to a causal interpretation. However, these results are in line with our idea that there is an identity component that may explain why some individuals develop entrepreneurial intentions while others do not. The impact of peers on an individual’s identity appears to be moderated by cultural values related to individualism. However, further research is necessary on this issue. Over time, cultural differences might manifest in different levels of entrepreneurial intentions across countries (cf. Audretsch 2007) which then influence student’s entrepreneurial intentions. Accordingly, one might ask if the peers’ influence on entrepreneurial intentions is stronger in countries with a higher self-employment rate. Figure 2 shows the relationship between the average impact of the entrepreneurial peer group and self-employment rates within countries. We do not find a significant association between these two variables. This suggests that encouraging youth entrepreneurial intentions might be important for both countries where entrepreneurship is less common and for those where there already is a high rate of self-employment.

22

50

Figure 2. Self-employment rates and the strength of the peer group effect

40

TUR

GRC

MEX

20

30

KOR

IRL

ITA

POL

PRT

NZL

BEL JPN

10

FIN SWE NOR

LUX

-.1

ESP CZE AUT

ISL GBR AUS SVK DEU

HUN NLD

CAN DNK USA

0 .1 peer group influence on entrepreneurial intentions Fitted values

.2

selfrates

Notes: The vertical axis shows the total self-employment rates as a percentage of total civilian employment, 2005 or latest available year (Source: OECD 2006). The horizontal axis shows the estimated peer group influence on entrepreneurial intentions. The regression model contains all control variables listed in the appendix A2 as well as school type fixed effects.

6. Conclusions In this paper, we argue that an individual’s entrepreneurial identity is shaped an individual’s peer group and parents. We empirically test our model using a representative sample of 15- year-olds from the 2006 cycle of the Programme for International Student Assessment (PISA), and find support for our theoretical prediction that having entrepreneurial peers at school has an impact for the development of entrepreneurial intentions in this age group. We further provide evidence that entrepreneurial intentions in adolescent ages are indeed a good predictor for the actual future choice to become an entrepreneur. We close by discussing some policy implication of our research. As the entrepreneur is regarded as a driving force behind economic development (Schumpeter 1912) or even as the ultimate source of economic development in the modern knowledge-based society (Audretsch 2007), there is a large interest among policymakers in how to foster entrepreneurship, especially in secondary and tertiary education. Programs in entrepreneurship education become increasingly recognized around the 23

world. However, at least for the Netherlands, Oosterbeek, van Praag, and IJsselstein (2008) find no effects of this type of program on entrepreneurial intentions of vocational college students, that are, on average about 4 to 5 years older than the students in our sample. Against this non-finding, our findings of the importance of the school environment for entrepreneurial intentions suggest that it might be promising to study in more detail the importance of the school environment and institutional features of the school system for developing entrepreneurial intentions. Moreover, our findings suggest that entrepreneurship education programs at younger ages might be more effective. We see our research as a contribution to push the discussion on educating entrepreneurship into this direction.

24

References Akerlof, G. A., and R. E. Kranton (2000). “Economics and Identity.” Quarterly Journal of Economics 105(3), 715–53. Akerlof, G. A. and R. E. Kranton (2005). “Identity and the Economics of Organizations.” Journal of Economic Perspectives, 19, 9–32. Aldrich, H., L. A. Renzulli, and N. Langton (1998). “Passing on Privilege: Resources Provided by Self-Employed Parents to Their Self-Employed Children.” Research in Social Stratification and Mobility, 16, 291–317. Ammermueller, A., and J.-S. Pischke. (2009). "Peer Effects in European Primary Schools: Evidence from the Progress in International Reading Literacy Study." Journal of Labor Economics, 27, 315-48. Arrow, K. J. (1962). “Economic Welfare and the Allocation of Resources for Innovation,” in: R. R. Nelson (ed.), The Rate and Direction of Inventive Activity, 609–26. Princeton, NJ: Princeton University Press. Audretsch, D. B. (2007). The Entrepreneurial Society, Oxford: Oxford University Press. Bandura, A. (1977). Social Learning Theory. Englewood Cliffs, NJ: Prentice Hall. Bandura, A. (1986). Social Foundations of Thought and Action. Englewood Cliffs, NJ: Prentice Hall. Baron, R. A. (1998). “Cognitive Mechanisms in Entrepreneurship: Why and When Entrepreneurs Think Differently Than Other People.” Journal of Business Venturing, 13, 275–94. Bauernschuster, S., O. Falck, and S. Heblich (2008). Occupational Choice and Social Contacts Across Regions. Jena Economic Research Papers 2008-079. Baumol, W. (1968). “Entrepreneurship in Economic Theory.” American Economic Review, 58, 64– 71. Björklund, A., M. Jäntti, and G. Solon (2007). Nature and Nurture in the Intergenerational Transmisson of Socioeconomic Status: Evidence from Swedish Children and Their Biological and Rearing Parents. Working Paper, Michigan State University. Bloom, A. (1987). The Closing of the American Mind. New York: Simon and Schuster. Camerer, C., and D. Lovallo (1999). “Overconfidence and Excess Entry: An Empirical Approach.” American Economic Review, 89, 306–18. Centre for Longitudinal Studies (2007). British Cohort Study Response Dataset, 1970-2005, www.cls.ioe.ac.uk. Coleman, J. S. (1961). The Adolescent Society: The Social Life of the Teenager and Its Impact on Education. New York: Free Press. 25

Coleman, J. S. (1988). “Social Capital in the Creation of Human Capital.” American Journal of Sociology, 9, 95–121. Cunha, F., and J. Heckman (2007). “The Technology of Skill Formation.” American Economic Review, 97, 31–47. Deaton, A. (1997). The Analysis of Household Surveys: A Microeconometric Approach to Development Policy. Baltimore: Johns Hopkins University Press. DuMouchel, W. H., and G. J. Duncan (1983). “Using Sample Survey Weights in Multiple Regression Analyses of Stratified Samples.” Journal of the American Statistical Association, 78, 535–43. Dunn, T., and D. Holtz-Eakin (2000). “Financial Capital, Human Capital, and the Transition to SelfEmployment: Evidence from Intergenerational Links.” Journal of Labor Economics, 18, 282– 305. Eckert, P. (1995). “Trajectory and forms of institutional participation. In: L. J. Crockett and A. C. Crouter (Eds.): Pathways Through Adolescence, 175-195, New Jersey: L. Erlbaum. European Commission (2006). “Entrepreneurship Education in Europe: Fostering Entrepreneurial Mindsets Through Education and Learning.” Final Proceedings of the Conference on Entrepreneurship Education in Oslo. Evans, D. S., and L. S. Leighton (1989). “Some Empirical Aspects of Entrepreneurship.” American Economic Review, 79, 519–35. Fairlie, R. W. and A. Robb. (2007). "Families, Human Capital, and Small Business: Evidence from the Characteristics of Business Owners Survey." Industrial & Labor Relations Review, 60, 225-45. Fuchs, T., and L. Wößmann (2007). “What Accounts for International Differences in Student Performance? A Re-Examination Using PISA Data.” Empirical Economics, 32, 433–64. Gaviria, A., and S. Raphael (2001). “School-based Peer Effects and Juvenile Behavior.” The Review of Economics and Statistics, 83, 257-268. Giannetti, M., and A. Simonov (2009). "Social Interactions and Entrepreneurial Activity." Journal of Economics & Management Strategy, 18, 665-709. Gompers, P., J. Lerner, and D. Scharfstein (2005), "Entrepreneurial Spawning: Public Corporations and the Genesis of New Ventures, 1986 to 1999," Journal of Finance, 60, 577-614. Granovetter, M. (1985). “Economic Action and Social Structures: The Problem of Embeddedness.” American Journal of Sociology, 91, 481–510. Guiso, L., P. Sapienza, and L. Zingales (2006). “Does Culture Affect Economic Outcomes?” Journal of Economic Perspectives, 20, 23-48. Halaby, C. N. (2003). “Where Job Values Come From: Family and Schooling Background, Cognitive Ability, and Gender”, American Sociological Review, 68, 251-278. Hamilton, B. H. (2000). “Does Entrepreneurship Pay? An Empirical Analysis of the Returns of SelfEmployment.” Journal of Political Economy, 108, 604–31. 26

Hayek, F. A. von (1937). “Economics and Knowledge.” Economica, New Series, 4, 33–54. Heckman (2006). Skill Formation and the Economics of Investing in Disadvantaged Children. Science, 312, 1900-1902. Holtz-Eakin, D., D. Joulfaian, and H. S. Rosen (1994). “Sticking It Out: Entrepreneurial Survival and Liquidity Constraints.” Journal of Political Economy, 102, 53–75. Hout, M., and H. Rosen (2000). “Self-Employment, Family Background, and Race.” Journal of Human Resources, 35, 670–92. International Labour Organisation (ILO) (1990). International Standard Classification of Occupations: ISCO-88. Geneva: International Labour Office. Johnson, M. K. (2002). “Social Origins, Adolescent Experiences, and Work Value Trajectories During the Transition to Adulthood.” Social Forces, 80, 1307-40. Kahneman, D., and A. Tversky (1979). “Prospect Theory: An Analysis of Decision Under Risk.” Econometrica, 47, 263–92. Kanbur, S., and M. Ravi (1990). “Entrepreneurial Risk Taking, Inequality, and Public Policy: An Application of Inequality Decomposition Analysis to the General Equilibrium Effects of Progressive Taxation.” Journal of Political Economy, 90, 1–21. Kihlstrom, R. E., and J.-J. Laffont (1979). “A General Equilibrium Entrepreneurial Theory of Firm Formation Based on Risk Aversion.” Journal of Political Economy, 87, 719–48. Knight, F. H. (1921). Risk, Uncertainty and Profit, New York: Houghton Mifflin. Kuratko, D. F. (2005). “The Emergence of Entrepreneurship Education: Development, Trends, and Challenges.” Entrepreneurship Theory and Practice, 29, 577–98. Lazear, E. P. (2005). “Entrepreneurship.” Journal of Labor Economics, 23, 649–80. Lentz , B. F. and S. Laband (1990). “Entrepreneurial Success and Occupational Inheritance Among Proprietors.” Canadian Journal of Economics, 23, 101–17. Lerner, J. and U. Malmendier (2007). “With a Little Help from My (Random) Friends: Success and Failure in Post-Business School Entrepreneurship.“ Working Paper. Lucas, R. E. Jr. (1978). “On the Size Distribution of Business Firms.” Bell Journal of Economics, 2, 508–23. Manski, C.F. (1995). Identification Problems in the Social Sciences. Cambridge: Harvard University Press. Marshall, A. (1920). Principles of Economics, 8th ed. London: MacMillan. Michelacci, C., and O. Silva (2007). “Why So Many Local Entrepreneurs?” Review of Economics and Statistics, 89, 615–33. Miller, W. (1952). The Business Elite in Business Bureaucracies. In: W. Miller (Ed.): Men in Business: Essays in the History of Entrepreneurship, 286-305. Cambridge: Harvard University Press. 27

Mitchell, R., and D. Shepherd (2008). “To Thine Own Self be True: Images of Self, Images of Opportunity, and Entrepreneurial Action.” Journal of Business Venturing, in press. Mortimer, J. T., and J. Lorence (1979). “Work Experience and Occupational Value Socialization: A Longitudinal Study.” American Journal of Sociology, 84, 1361-85. Moulton, B. R. (1986), “Random Group Effects and the Precision of Regression Estimates.” Journal of Econometrics, 32, 385–97. Nanda, R., and J.B.Sørensen (2008), “Peer Effects and Entrepreneurship”, Harvard Business School Working Paper 08-051. Neu, I.D., and F.W. Gregory (1952). The American Industrial Elite in the 1870s: Their Social Origins. In: W. Miller (Ed.): Men in Business: Essays in the History of Entrepreneurship, 193211. Cambridge: Harvard University Press. Neyman, J. and E. Scott (1948). “Consistent Estimates based on partially consistent observations.” Econometrica, 16, 1-32. Nicolaou, N., S. Shane, J. Hunkin, L. Cherkas, and T. Spector (2008). “Is the tendency to engage in entrepreneurship genetic?” Management Science, 54,167-179. Nicolaou, N., and S. Shane (2009). “Born entrepreneurs? The genetic foundations of entrepreneurship,” Journal of Business Venturing, Forthcoming. Oosterbeek H., M. van Praag, and A. IJsselstein (2008). The Impact of Entrepreneurship Education on Entrepreneurship Competencies and Intentions: An Evaluation of the Junior Achievement Student Mini-Company Program. IZA Discussion Paper No. 3641. Organisation for Economic Co-operation and Development (OECD) (2006). Labour Force Statistics. Paris: OECD. Organisation for Economic Co-operation and Development (OECD) (2007a). PISA 2006 Science Competencies for Tomorrow’s World. Volume 1: Analysis. Paris: OECD. Organisation for Economic Co-operation and Development (OECD) (2007b). PISA 2006 Science Competencies for Tomorrow’s World. Volume 2: Data. Paris: OECD. Parker, S. C. (2004). The Economics of Self-Employment and Entrepreneurship. Cambridge: Cambridge University Press. Parker, S. C. (2009). Small firms and innovation, forthcoming in: D. B. Audretsch et al. (eds.), Handbook of Research on Innovation and Entrepreneurship. Forthcoming, Cheltenham: Edward Elgar. Sanders, J. M., and V. Nee (1996). “Immigrant Self-Employment: The Family as Social Capital and the Value of Human Capital.” American Sociological Review, 61, 231–49. Schneeweis, N., Winter-Ebmer, R. (2007). “Peer Effects in Austrian schools.” Empirical Economics, 32, 387 – 409. Schumpeter, J. A. (1912). The Theory of Economic Development. New York: Oxford University Press. 28

Sen A. (1977). Rational Fools: A Critique of the Behavioral Foundations of Economic Theory, Philosophy and Public Affairs, 6, 317-344. Soerensen, J. B. 2007. "Closure and Exposure: Mechanisms in the Intergenerational Transmission of Self-Employment." Research in the Sociology of Organizations, 25, 83-124. Stuart, T. E., and O. Sorenson (2005). “Social Networks and Entrepreneurship,” in: S. Alvarez, R. Agarwal, and O. Sorenson (eds.), The Handbook of Entrepreneurship: Disciplinary Perspectives, 211–28. Berlin: Springer. Wooldridge, J. M. (2001). “Asymptotic Properties of Weighted M-Estimators for Standard Stratified Samples.” Econometric Theory, 17, 451–70. World Values Survey (2005). Official Data File v.20090901, 2009. World Values Survey Association (www.worldvaluessurvey.org) Wu, B., and A. M. Knott (2006). “Entrepreneurial Risk and Market Entry.” Management Science, 52, 1315–30.

29

Appendix Appendix A1. The Effect of Students' Entrepreneurial Intentions on Occupational Choice across the Lifecycle

In this appendix, we examine the effects of students’ job intentions on their occupational choice across the lifecycle. In particular, we show that students who stated entrepreneurial intentions at the age of 16 obtain a significant higher probability of being an entrepreneur until the age of 34 than students who did not state entrepreneurial intentions. We will show that this result is robust to the inclusion of several control variables for the individual’s educational and family background. Data are taken from the “1970 British Cohort Study (BCS70)” (cf. Centre for Longitudinal Studies 2007). BCS70 is a continuing, multi-disciplinary longitudinal study which follows individual living in England, Scotland and Wales who were born in one particular week in April 1970. The first wave was carried out in 1970 and it examined the social and biological characteristics of the mother in relation to neonatal morbidity. Since then, there have been six full data collection exercises in order to monitor the individuals' health, education, social and economic circumstances. These took place when respondents were aged 5, aged 10, aged 16, aged 26, aged 30, and aged 34. To analyze the persistence of students’ entrepreneurial intentions over time, we initially analyze the survey results at age 16, i.e. in 1986, when the individuals were asked if there is an actual job they wish to do. Subsequently they had the chance to state the name, nature and industry of their preferred job. Given these statements we construct a dummy variable for entrepreneurial intentions. This variable takes on a value of “1” if the student’s answer contains at least one of the following words: “own,” “self,” and “manage” or “chef” in combination with a well-defined business activity such as “restaurant,” “hotel,” “shop” or more general “firm” or “business.” Underlying our definition of entrepreneurial intentions, 164 out of 3,998 individuals at age 16 intended to become an entrepreneur. We further we distinguish three comparison groups: Reference group A contains all students with any other job intentions than entrepreneurial; reference group B consists of students who either have no or any other job intentions than being entrepreneurial; students with no job intentions at all constitute reference group C.18 Consequently, we construct three modifications of the dummy variable for entrepreneurial intentions.

18

This information is taken from a filter question, where respondents were asked whether there is an actual job they want to do. We assume that teenagers did not have any job intentions if they negated this question.

30

Descriptive statistics about entrepreneurial intentions subject to the underlying reference group are given in Table A1.1. Students with entrepreneurial intentions account for 4.10 percent of the students who stated any job intentions (A). This is a good match to the findings concerning entrepreneurial intentions of students based on PISA data (3.21 percent for Great Britain; cf. Table 2). Table A1.1. Descriptive statistics: entrepreneurial intentions subject to reference group Reference group: A B C Entrepreneurial intentions Freq. Percent Freq. Percent Freq. Percent at age 16 no 3,834 95.90 5,995 97.34 2,207 93.08 yes 164 4.10 164 2.66 164 6.92 total 3,998 100 6,159 100 2,371 100 Notes: Reference group A contains all students with any other job intentions than entrepreneurial; reference group B consists of students who either have no or any other job intentions than entrepreneurial; students with no job intentions at all constitute reference group C.

In a second step, we turn to the survey results and analyze data from the 30-year and 34-year surveys (sweep 2000 and 2004) to construct a dummy variable indicating if the individual ever has been an entrepreneur until the age of 34. In each questionnaire the individuals were able to report up to eleven occupations they have performed until the date of the survey. Consequently we can use each individual’s entire employment history up to age 34 for constructing this dummy variable. We then estimate the following equation by a linear probability model: Pr ob( Ent _ Occi ,