Entry into self-employment and individuals’ risk-taking propensities

See also Jaeger et al. (2010) for a similar result in the context of risk-taking propensities and migration.

Weber and Milliman (1997), for example, show that situational variables such as outcome framing or prior outcome history affect people’s risk perception. They find that the perceptions of the riskiness of choice alternatives (measured both in stock choices and in investor’s risk perception of stocks) systematically and substantially vary with the feedback on outcomes.

Our focus group consists of those individuals, who either report dependent employment or unemployment in the first wave eliciting risk-taking propensities (in 2004), but report self-employment in later waves of the panel survey. We call these individuals “future self-employed” and differentiate them from the “not self-employed” (who never indicate self-employment) and from the “2004 self-employed” who already indicated being self-employed in the first wave in 2004.

In addition to this evidence, Ahn (2010) addresses the issues of measurement error and time variation when using self-reported risk measures. He allows the measure of risk tolerance to vary with age and other covariates, but again basing his analysis partially on risk-taking propensities collected after the decision to become self-employed.

These inconclusive findings are also well in line with applied psychology literature. While Miner and Raju (2004) report that self-employed may be even more risk-averse than others, the results of Stewart and Roth (2001) provide evidence for a self-selection bias, i.e. findings suggest that risk-seeking individuals are more likely than others to choose self-employment.

Findings of Hetschko and Preuss (2015) suggest that job losses can affect an individual’s propensity to take risks. Given that a job loss is unexpected (e.g. a mass lay-off), their identification strategy provides an alternative setting for the analysis of causal effects. However, we only have an extremely limited number of persons that experienced a job loss and became self-employed. Therefore, we rely on other identification strategies.

For more detailed information on the SOEP, see Wagner et al. (1993) and Wagner et al. (2007) or http://​www.​diw.​de/​en/​diw_​02.​c.​221178.​en/​about_​soep.​html.

We use doi:https://​doi.​org/​10.​5684/​soep.​v29.

This means that we exclude non-employed individuals, individuals in vocational training, individuals on internship and individuals in military or civil service from the analysis. We also exclude individuals with missing information on any of the variables used to perform the analysis.

The appropriate measurement of risk-taking propensities is subject to a lively debate. We rely on self-reported risk-taking propensities as also used by numerous other authors (Caliendo et al. 2009, 2014; Jaeger et al. 2010; Dohmen et al. 2012). Some authors argue that risk preferences revealed in choice behaviour are more reliable that self-reported risk-taking propensities (e.g. Necker and Voskort 2014). Some authors use hypothetical choices to reveal risk preferences (e.g. Barsky et al. 1997; Cramer et al. 2002). Comparing self-reported risk-taking propensities with real lottery choices, Dohmen et al. (2011) confirm the behavioural relevance of self-reported risk-taking propensities.

We thank an anonymous reviewer for this suggestion.

The same holds true for unexpected changes in wealth and for changes in the employment status between 2004 and 2009. When we restrict our sample to situations where, first, the treatment group consists only of individuals who enter self-employment out of employment in 2004, second, all individuals never were unemployed between 2004 and 2009 and, third, the entire sample of individuals did not receive a gift, bequest or lottery win between 2002 and 2009, then we still find that individuals who enter self-employment experience a statistically significant increase in their willingness to take risks. This result remains robust when applying the alternative risk measure. And, we also find evidence for a common trend. These results are available upon request.

In a somewhat related literature, some authors argue that the propensity to take risks may be affected by peer comparisons in the individual’s interaction group (see, e.g. Knudsen 2008 or Ahern et al. 2014). These peer pressure theories are difficult to test with a broad general population panel, because the actual interaction groups usually cannot be identified. Furthermore, while peer group effects may be quite strong, it is unlikely that they only increase the risk taking propensities of the individuals switching to self-employment, but of no one else.

We thank an anonymous referee for pointing this out.

Empirical research demonstrates that the variance of earnings in self-employment is more than three times greater than that of wage workers (Rees and Shah 1986) and that self-employment entails risky decisions (Caliendo et al. 2009) in various domains, including acquiring clients, dealing with project financing and competing in the market. All these conditions are bound to define worst-case scenarios with lower reference point income in self-employment than in dependent employment.