شواهد جدید در رابطه بین نگرش خطر و خود اشتغالی

This paper analyses the impact of risk attitudes on the decision to become self-employed among individuals who grew up under the communist regime in Ukraine, which banned self-employment so that individuals could not observe what it is like to be self-employed. Since the intra-family transmission of self-employment experiences was largely shut down, the observed correlation between risk preferences and self-employment after transition is unlikely to be driven by parents transmitting self-employment experience and risk preferences to their children. Robustness checks on a sample of East Germans confirm that such a third factor explanation is implausible, thus shedding light on the causal nature of the relation between risk preferences and the decision to become self-employed.

Self-employment is an important driver of economic activity. Not surprisingly, the determinants of self-employment have been on the economic research agenda for a long time, and several factors have been identified.1 Already in 1755, Cantillon recognised that entrepreneurial activities are risky. He argues that the employed face a more variable stream of earnings because of uncertain product demand, which results in revenue fluctuations and can lead to business failure.2 Given the risky nature of self-employment, heterogeneity in risk preferences across individuals should affect the choice of employment status (see also Kihlstrom and Laffont, 1979).3 A host of empirical studies have established a robust positive correlation between individual willingness to take risks and the probability of being self-employed (e.g. Cramer et al., 2002, Ekelund et al., 2005 and Fossen, 2011). Skeptics might raise the concern that self-employment experience makes individuals more willing to take risks. In order to preclude that such exposure effects drive the relation between risk preference and self-employment status, Caliendo et al. (2009) show that risk attitudes measured in the 2004 wave of the German Socio-Economic Panel Study (SOEP) predict the respondents' transition into self-employment in 2005. Similarly, Brown et al. (2011) document that risk attitudes measured in the 1996 wave of the Panel Study of Income Dynamics (PSID) predict the respondents' decision to become self-employed up to 9 years later, suggesting a causal effect of risk attitude on the probability of entering self-employment. But the authors caution the reader that an unobserved third factor that influences both the decision to become self-employed and risk attitude might drive their result. A plausible third factor is the influence of parents who shape their children's risk attitudes through intergenerational preference transmission (Dohmen et al., 2012) and potentially affect children's decision to become self-employed indirectly or directly. The most obvious direct channel is the transfer of businesses from parents to children.4 But there are also indirect channels such as the transmission of information and beliefs about self-employment through which children of self-employed parents might grow more likely to become self-employed themselves (Hout and Rosen, 2000, Falter, 2007a and Andersson and Hammarstedt, 2010). For example, growing up with self-employed parents can provide children with invaluable information on what it is like to be self-employed. In this paper, we analyse a setting in which the scope for this kind of third factor explanation is limited or even non-existent. We provide new evidence that supports a causal interpretation of the correlation between risk attitudes and the probability of becoming self-employed. We exploit the fact that self-employment was banned for at least 45 years in Ukraine, so that the cohorts, whose self-employment decision after transition we study, grew up without exposure to self-employment. A very substantial fraction of these individuals has parents who never had the right to become self-employed during their active labour market career.5 We describe our research strategy in the next section and describe the data that we use in Section 3. In Section 4, we report the findings from our empirical analysis, and end with a conclusion.