Risk, Balanced Skills and Entrepreneurship

This paper proposes that risk aversion encourages individuals to invest in balanced skill profiles, making them more likely to become entrepreneurs. By not having taken this possible linkage into account, previous research has underestimated the impacts both of risk aversion and balanced skills on the likelihood individuals choose entrepreneurship. Data on Dutch university graduates provides evidence which supports this contention. It thereby raises the possibility that even risk-averse people might be suited to entrepreneurship; and it may also help explain why prior research has generated mixed evidence about the effects of risk aversion on selection into entrepreneurship.


Introduction
Two of the most inuential theories of individual selection into entrepreneurship are based on the concepts of risk aversion, RA (Kihlstrom & Laont, 1979), and balanced skills, BS (Lazear, 2005). Specically, if entrepreneurship is a more risky occupation than paidemployment, and if individuals vary in their aversion to risk, then it follows that the least risk-averse people are most likely to become the entrepreneurs (Kihlstrom & Laont, 1979).
Moreover, because entrepreneurship requires expertise in a variety of roles while paidemployment rewards specialists, people with balanced skills are most likely to become entrepreneurs as well (Lazear, 2005).
Despite the prominence and continued inuence of the RA and BS theories, the evidence for them is decidedly mixed. For example, many psychology-based studies have failed to detect any dierence between entrepreneurs and non-entrepreneurs in terms of their risk attitudes (Brockhaus, 1980;Shaver & Scott, 1991). Meta-analyses of risk aversion and entrepreneurial selection have also generated conicting results (Stewart & Roth, 1991;Miner & Raju, 2004), with Miner & Raju (2004) concluding that the available evidence about the validity of the RA theory is inconclusive. Economics-based studies have also generated mixed ndings (Åstebro et al, 2012). While some research suggests that entrepreneurs are indeed typically less risk-averse than employees (Cramer et al, 2002;Brown et al, 2011), others have reported insignicant dierences between these groups (Barsky et al, 1997;Parker, 2008). And while several studies have measured balanced skills in terms of the number of prior job roles, and have generated evidence consistent with the BS theory (Lazear, 2005;Wagner, 2006;Hartog et al, 2010;Åstebro & Thompson, 2011), the robustness of these results has been called into question (Silva, 2007).
While RA and BS remain popular and inuential theories, not least because of their persuasive and attractive internal logics, their lack of clear empirical support raises several troubling questions. For example, does the inconclusive evidence about the role of risk aversion mean that any dierences of this sort do not aect occupational choice on net, perhaps because other factors dominate this choice (or because paid-employment is also risky: Parker, 1997)? Likewise, have the estimates of skill balance been weakened by using a awed proxy, namely the number of prior job roles or are they actually a mirage, masquerading as hard-to-measure personal abilities (Silva, 2007;Hartog et al, 2010), or preferences such as a`taste for variety' (Åstebro & Thompson, 2011)? Lacking answers to these questions, our knowledge about reasons why people become entrepreneurs is bound to remain limited. This paper proposes a dierent argument which may shed light on this issue. Specically, we propose that balanced skills and risk aversion are not the independent constructs which previous research has taken them for. Given evidence that risk-averse actors like to diversify their human capital (e.g. Amihud & Lev, 1981), one might expect highly specialized employees to be left with few competitive options if returns from specialism suddenly become less valuable in fast-changing, uncertain environments (Abernathy & Wayne, 1974).
Then risk-averse individuals who fear the loss of exibility associated with highly specialized human capital may respond by diversifying their human capital investments. As a result, risk-averse people could ironically end up acquiring the balanced skill sets which, it is argued, are especially conducive to entrepreneurship.
As well as being of interest in its own right, the possibility that risk aversion and balanced skills are positively related implies, as we go on to show, that empirical studies (which have ignored this interdependence hitherto) are prone to have underestimated both of their impacts on entrepreneurial selection. In principle, this point might help to explain the weak and mixed body of evidence pertaining to the RA and BS theories.
The paper makes the following contributions. First, it extends our theoretical understanding of entrepreneurship as an occupational choice by proposing a novel association between the two hitherto separate concepts of risk aversion and balanced skills. Our simple formulation extends the theory of BS from a certain environment (as in Lazear, 2005) to a risky one. Risk is present in both occupations; and the acquisition of balanced skills is treated as a choice variable, rather than being taken as given as in previous work. Second, our theorizing proposes a richer empirical specication, which is estimated using a sample of recent graduates from universities in the Netherlands. The dataset has two attractive properties.
One is that, in line with our theory, the survey respondents are homogeneous in terms of their education levels and labor market experience. The other is that, consistent with our theory, skills balance is measured prior to when occupational choices were observed, thereby avoiding problems of reverse causality. Furthermore, we depart from the conventional practice of proxying skills balance by the variety of prior labor market experience, which may be associated with unobserved abilities (Silva, 2007). Instead we propose a novel measure based on the observed multi-industry versatility of degree majors as well as on the spread of individual-level scholastic skills (whose levels we also control for). Third, the paper makes a further contribution by providing a platform for re-evaluating mixed prior evidence from tests of the RA and BS theories.

The model
Suppose there are two occupations, paid employment (P) and entrepreneurship (E), and also two skills which generate returns in both occupations, x 1 and x 2 . To abstract from issues of aggregate skill acquisition, which is not of interest here, assume that every agent obtains a unit endowment of total skill. This allows us to use the more compact notation In general, both skills x and 1 − x can generate output. In E, both skills are needed for any output to be produced [see (2) below]. In P, employers either oer workers a menu of jobs requiring dierent combinations of skill with given rates of return, or workers can freely mix jobs requiring only one skill or the other [see (1) below]. Either way, note that specialism in P is not predetermined by assumption. People specialize if they choose x * = 1 or x * = 0. If 0 < x * < 1 they choose some mixture of skills. We are interested in agents' optimal choices x * and how risk aects x * as well as occupational choice between P and E.
The production technology which maps x and 1 − x into returns diers in each occupation.
For tractability, we will use specications which dier from those proposed by Lazear (2005) in his analysis of riskless choice. We will rst show that our specications generate the same results in a riskless environment. Our specications of the returns in each occupation are: In the benchmark case of certainty considered by Lazear (2005), all parameters in the set Ω := {ω 1 , ω 2 , θ} are positive. It follows immediately that optimal choices of x in P are The optimal choice of x in E is x * = 1 2 . Hence employees specialize in one skill while entrepreneurs have balanced skills. Provided θ > 4 max{ω 1 , ω 2 }, individuals possessing balanced skills do best in E, whereas those possessing specialized skills do best in P. These predictions mirror Lazear's. Now we move into more novel territory by examining the roles of risk and risk preferences.
Consider the standard utility function where λ is the coecient of absolute risk aversion (ARA). To introduce risk, make Ω stochastic, with ω 1 ∼ N (µ 1 , φ), ω 2 ∼ N (µ 2 , φ) and θ ∼ N (m, ψ). Agents are assumed to know the parameters of all of these normal distributions, which all have positive means and variances. Restricting the variances of ω 1 and ω 2 to be equal results in no loss of generality for the analysis below.
The following assumption restricts admissible parameter values to ensure internal consistency of the model: Assumption 1(a) is needed to ensure that optimal choices of x * in P derived in (4) below are conned to the unit interval. Assumptions 1(b) and 1(c) ensure that positive mean eects dominate negative variance eects in terms of expected utility in both occupations.
As is well known, the combination of normally distributed payos with constant ARA utility (3) gives rise to simple mean-variance utility expressions (see e.g. Sargent, 1987, 15455).
So, for example, the problem max x EU (y P ) is equivalent to the problem The rst order condition for this problem yields This equation implies that the optimal skill prole for employees under risk can dier from their skill prole under certainty analyzed above. In general, risk gives employees some incentives to acquire more balanced skill sets, as can be seen in (4) as φ → ∞. The reason is that, when it is unknown a priori which skill will be most valuable, workers have incentives to choose a skill prole which diversies their labor market portfolio. Proposition 1 In occupation P, greater risk aversion is associated with a more balanced skill prole except for the special case where returns to the two skills are identical.
Proof. When µ 1 = µ 2 , (4) can be dierentiated to obtain ∂|x * − 1 2 |/∂λ < 0. Hence greater risk aversion is associated with a more balanced skill prole. When µ 1 = µ 2 , (4) implies The optimal skills balance for E is as follows. Write the optimization problem in E as The FOC for this problem is But h ∈ (0, 1 4 ] while ζ < 4, so ζh(x) < 1 and the above FOC requires h (x) = 0. This solves for x * = 1 2 in E. So introducing risk into E does not aect the incentives to obtain balanced skills in that occupation.
Analogous to the case of certainty above, we need a condition on m to ensure that E is a non-empty occupation in equilibrium. Using (4), the relevant condition is: We can now state the next proposition: Proposition 2 All else equal, an individual with a more balanced skill prole is more likely than an individual with a less balanced skill prole to choose occupation E over P.
Proof. Denote byx the values of x which solve the occupational choice equilibrium: By Assumptions 1(c) and 1(b), the LHS of (5) is monotonic in x while the RHS is a ∩shaped quadratic in x, with its maximum at one half. By Assumption 2 the LHS and RHS intersect. Hence there are two solutions to (5), denoted by ( Proof. (a) Let z * be the dierence in expected utility in E relative to P: The direct eects of risk aversion are given by This derivative is only certain to be negative if ψ is suciently large relative to φ, i.e. if In E, x * = 1 2 so ψ > 8φ is the necessary condition. In P, x * = 1 2 so the ψ/φ ratio must be greater still. Hence the condition ψ > 8φ is necessary (but not sucient) for an increase in λ to have a negative direct eect on incentives to choose E over P.
(b) Proposition 1 established that the indirect eect of greater λ on balanced skills in P is positive. Hence by Proposition 2, more employees prefer E to P. At the same time, the solution x * = 1 2 in E is obviously invariant to λ. (In the language of part (a) of this proof, a greater λ decreases the height of the quadratic return function in E without aecting its skew.) Since an increase in λ increases outows from P to E while leaving choices in E unchanged, the total number of entrepreneurs increases.
Proposition 3 shows that the existence of balanced skills considerations has subtle implications for the eects of risk aversion on occupational choice. On the one hand, when risk is present in both occupations the direct eects of risk aversion become ambiguous in principle (see also Parker, 1997). However, suciently pronounced income risk in entrepreneurship relative to paid employment predisposes risk-averse people to choose paid-employment over entrepreneurship. On the other hand, because greater risk aversion encourages people to acquire more balanced skill sets, and because balanced skills are more valuable in entrepreneurship, greater risk aversion also serves to make entrepreneurship more attractive relative to paid employment through the indirect balanced skills channel. An empirical analysis of risk aversion and balanced skills in entrepreneurship needs to take account of these distinct mechanisms. Consider the following equation to be estimated using a sample of individuals i: where z * i is a latent variable underlying a binary occupational choice variable [see (6) in the proof of Proposition 3] such that Here λ i and SB i are individual-level measures of risk aversion and skill balance, respectively; X i are a set of orthogonal control variables and u i is a disturbance term. According to Proposition 1, λ i and BS i are directly related; let γ > 0 denote the coecient of proportionality.
In terms of (7), Proposition 2 predicts β 2 > 0, while Proposition 3(a) predicts β 1 is ambiguous in principle though negative if entrepreneurship is much riskier than paid employment.
Hereafter, suppose β 1 < 0, in accordance with the RA theory of Kihlstrom and Laont (1979) (who ignored risk in P). Given these predictions, we can now deduce the bias that will occur if λ i or SB i are omitted from (7). First consider the case where SB i is omitted.
Then a standard result in econometrics (e.g. Greene, 2003) is that the bias from estimating Our empirical strategy is as follows. First, we examine whether SB i and λ i are positively related by using OLS to estimate γ in a regression of SB i on λ i . This tests Proposition 1.
Second, we estimate the eects of SB i and λ i by applying probit methods to (7) & (8). This tests Propositions 2 and 3(a). In each of these cases, we also take account of the possibility that skill balance and unobservables aecting occupational choices are more similar within degree elds than between them. We do so by additionally reporting clustered standard errors by degree eld (j = 40). And, we also provide estimates using robust estimation techniques to correct for heteroskedasticity.
Third, we statistically test the biases predicted above, which can be summarized as β 1 < 2 An alternative approach in principle is Instrumental Variable (IV) estimation of (7) and (8) where SBi is related to λi and some other variables. In practice, this approach requires valid identifying instruments, i.e., factors that aect the choice for investing in balanced skills but not the choice of entrepreneurship. Our dataset does not include variables that would qualify as identifying instruments. By not using IV estimation we implicitly assume that the investment in balanced skills is not aected by the prospect of a future occupational choice. This assumption is not implausible given that our measure of skills balance is based on choices of children between 12 and 18 years of age. nal sample comprises 3,002 respondents who graduated in 2002 with a Master's degree and who were working as paid employed or self-employed in January 2004.
An advantage of these data is that, consistent with the theory expounded in the previous section, the survey respondents are homogeneous in terms of education level and labor market experience. They dier however in terms of their investments in balanced skills.
Moreover, the data are rich enough to measure balanced skills in two distinct ways, as explained below. Crucially, the choices giving rise to both measures of BS i were made before any labor market participation decisions, thereby avoiding problems of reverse causality.

Variables
Occupational choice: self-employment versus wage employment. Consistent with the data, we operationalize entrepreneurship as self-employment, and use as the dependent  Table 1 around here > Risk attitude. Respondents were asked to value participation in a hypothetical lottery paying out 1, 000 euros with a 10 percent chance of success. The reservation price (p) for participating in such a hypothetical lottery has been shown to be a valid (inverse) indicator of risk aversion and behavior under risk (see Barsky  Generality. Some degree majors confer a skill set which is useful in a variety of dierent industries after graduation, whereas other majors have only a narrow, or specialized, range of applicability. We dene our Generality measure as the total number of distinct industry sectors employing graduates with a given major two years after graduation, scaled by the number of students graduating with that major. To minimize the impact of outliers, we only dene this variable for degree elds with more than thirty graduates in the sample. Data on both employees and the self-employed were used to construct this measure. Appendix Table   A1 lists all academic majors, the numbers of associated respondents, values of Generality, and self-employment rates. Majors such as sociology, applied computer science, languages and culture have high Generality scores, whereas medical sciences ranks lower. Appendix Table A2 lists the distinct industry sectors and the number of observations in each sector. Grade variance. This construct measures the variation in grades received by respondents while in secondary school. The smaller this variation, the more balanced is a person's foundation of learning skills. Grade variance equates to 1 − stdev(α, β, γ), where α = Grade Point Average (GPA) in humanities and languages, β = GPA in hard sciences, and γ = GPA in behavioral sciences.
Skill balance. We multiply`Generality' and`Grade variance' together to obtain a composite explanatory variable, SB. By combining a measure of skill balance which varies across degree elds with a measure which varies across individuals, SB provides a comprehensive overall measure of skill balance. We believe this is more informative than either of the underlying measures alone. For instance,`Generality' on its own says relatively little about skill balance at the individual level, while`Grade variance' on its own does not capture the industry context and applicability of diverse skills. 3 The main tables of results below will present results based on SB, although for completeness the Appendix will also present results obtained for each of the underlying measures.
Control variables Besides the key variables described above, we include a set of control variables including gender, age (varying from 22 to 29), parental education levels (measured on a 1-5 scale), and ability levels. The latter is measured as mean GPA scores both in secondary and in tertiary education, expressed on a scale from 110, where 6 is deemed a pass grade in the Netherlands. Table 1 presents descriptive statistics and correlations between the variables. There are no obvious problems of collinearity. Self-employment is correlated negatively with risk aversion and positively with`Generality' (though not with`Grade variance'), while risk aversion is associated positively with skill balance. Interestingly, the two main measures of skill balance are negatively correlated, suggesting that they are capturing distinct aspects of SB.

Estimation results
We rst test Proposition 1 by measuring the association between skill balance, SB, and risk aversion, λ, among employees. Column I of Table 2 presents the results for a`baseline' specication without control variables. It oers clear support for the proposition that 3 Previous measures of balanced skills have emphasized individual level variation, relying heavily on the number of previous job roles (though Lazear, 2005, also proposed the diversity of subjects studied at college). Unlike numbers of job roles, our SB variable is not time-varying, so panel data estimation could not be used to control for person-specic xed eects à lá Silva (2007), even if we had a panel.
people who are more risk averse acquire signicantly more balanced skill sets. These results continue to hold when control variables are included and alternative estimation methods, namely robust estimation and clustering, are used (columns IIIV). The results for the two underlying SB measures can be found in Appendix Table A3. Across the board, the results support Proposition 1. Tables 2 and 3 around here > Next, we test Proposition 2 by estimating a probit model of self-employment status. The results reported in Table 3 I and II with III and IV). The results continue to hold using the underlying measure`Generality', but not using the underlying measure`Grade variance' (see Appendix Table A4 for details). Table 3 Appendix Table A4).

< Insert
As noted in Section 2, Proposition 3(b) follows logically from Propositions 1 and 2, both of which received empirical support above. And as noted in Section 3, an implication of Proposition 3(b) is that excluding SB from (7) will increase the estimate of β 1 in this equation, while excluding λ from (7) will reduce the estimate of β 2 . Inspection of Table   3 indicates that the coecients change in the expected directions when these exclusion restrictions are imposed. But are these dierences statistically signicant? To answer this question, we adopt the testing approach outlined in the previous section, and report the χ 2 statistics in Table 4. These results clearly show that the expected biases are statistically signicant.
< Insert Table 4 around here > Finally, if risk aversion has a negative direct, and a positive indirect, eect on entrepreneurship, what is the overall (net) eect and how does it vary across sample cases? The estimated net eect of risk aversion on entrepreneurship is certainly negative at the sample mean; but it turns out to be positive for 12 per cent of the sample cases. For these cases, the impact of risk aversion on the acquisition of balanced skills is so powerful that it actually turns risk aversion into a force promoting entrepreneurship.

Conclusion
For the applied researcher, accurate estimation of the eects of balanced skills and risk aversion is obviously a desirable objective. This paper has proposed that accurate estimation needs to take into account the possible interdependence between these two constructs.
Such interdependence is also of interest in its own right. By making the acquisition of balanced skills more attractive, risk aversion can even end up as a positive force promoting entrepreneurship contrary to what might be expected from theories of RA which ignore BS arguments.
We believe that our arguments and empirical ndings may command interest beyond the community of entrepreneurship scholars, including among practitioners and entrepreneurs.
Our results reveal, perhaps surprisingly, that some risk-averse people, long deemed inherently ill-suited to entrepreneurship, might actually be well-suited to this occupation after all. This insight could have implications for entrepreneurship educators, who often stress the`negative' aspects of risk aversion for entrepreneurship without suggesting any positive aspects. It is also possible that young people under-estimate the future value of acquiring balanced skills, for instance by discounting the possibility of turning entrepreneur later in life. Our research suggests that the acquisition of balanced skills could be usefully encouraged at school and university since it builds a valuable future option for students.
It is also possible that some cultures or environments succeed, either deliberately or otherwise, in fostering balanced skills amongst their population, or in channeling risk aversion into the acquisition of balanced skills. For instance, formal education and corporate management training programs are known to dier in their emphasis on specialized relative to balanced skill acquisition. If governments genuinely wish to encourage entrepreneurship, a less specialized school curriculum might be one indirect, and long-term, way of doing so. Conversely, for rms concerned about losing employees to entrepreneurship (Hellmann, 2007) specialists might be favored over job candidates with balanced skills. Extending the logic in this paper, one is led to wonder whether there might be other unintuitive indirect relationships between balanced skills and individuals' preferences or personality traits. For example, people who have a`need for achievement' may spend a decade and longer in a single eld of study in order to attain the requisite expertise (Simon & Gilmartin, 1973).      Note: ***/**/* denotes signicance at the 1%/5%/10%-level.   Note: J = 40 clusters. Absolute t-values are given in parentheses. The sample excludes selfemployed entrepreneurs. They are based on robust estimates in specications 1 and 2, and based on clustered estimates in specications 3 and 4. ***/**/* denotes signicance at the 1%/5%/10%level.  observations) when using BS_tertiary as the measure of skill balance. Robust estimates are shown when using BS_secondary as the measure of skill balance. The results are similar when applying robust (clustered) estimation instead of clustered (robust) estimation. ***/**/* denotes signicance at the 1%/5%/10%-level. The controls included in specications (II) and (IV) are the same as in Table 2.