تستوسترون مرتبط است با خود اشتغالی در میان مردان استرالیایی

Testosterone has pronounced effects on men's physiological development and smaller, more nuanced, impacts on their economic behavior. In this study of 1199 Australian adult males, we investigate the relationship between the self-employed and their serum testosterone levels. Because prior studies have identified that testosterone is a hormone that is responsive to external factors (e.g. competition, risk-taking), we explicitly control for omitted variable bias and reverse causality by using an instrumental variable approach. We use insulin as our primary instrument to account for endogeneity between testosterone and self-employment. This is because prior research has identified a relationship between insulin and testosterone but not between insulin and self-employment. Our results show that there is a positive association between total testosterone and self-employment. Robustness checks using bioavailable testosterone and another similar instrument (daily alcohol consumption) confirm this positive finding.

Testosterone, an androgen produced mainly in the testes, has been shown to have pronounced effects on men's physiological development and smaller, more nuanced, associations with economic behavior (Dabbs et al., 1990 and Dabbs, 1992). However, alongside cognate genetic research on monozygotic and dizygotic twins (Nicolaou and Shane, 2010 and Nicolaou et al., 2008), more recent research has begun to identify that individuals with higher levels of testosterone are associated with new venture creation activities (White et al., 2006 and White et al., 2007) and with behaviors and characteristics commonly associated with self-employment. Carney and Mason (2010) showed that those with high testosterone levels are more likely to make use of utilitarian decision making. Strong and Dabbs (2000) also found that high testosterone individuals are more likely to be independent, a motive that is characteristic of individuals that are self-employed (Birley and Westhead, 1994 and Cassar, 2007). Equally, Cashdan (1995) identified that high testosterone individuals are more self-centered which relates to findings that the self-employed are more likely to be achievement orientated and focused on their own personal goals (Shane, 2003 and Chell, 2008). We investigate if testosterone levels are associated with self-employment. We focus on self-employment because it differs in one key respect from other labor market states. At its root, indeed the sin non qua of self-employment is that the self-employed are uncertain about their future income ( Storey and Greene, 2010). While employees or the unemployed, at least in the short-term, can be fairly certain of their wages or benefits from the state, the self-employed are only certain of their costs. To achieve any income, therefore, the self-employed are reliant upon their own proactive endeavors. There are also other reasons for considering that there will be a positive association between testosterone and self-employment. Integral to self-employment is risk-taking behavior under uncertainty (Kirzner, 1973). Recent cognate research has investigated the link between both organizational and circulating testosterone and financial risk-taking proclivity.1 Using the ratio of second to fourth finger (2D:4D) as a marker of organizational testosterone, Brañas-Garza and Rustichini (2011) found in their study of 188 college students that lower digit ratio was associated with financial risk-taking in men. These results are similar to that found among 413 college students by Stenstrom et al. (2011) who found that rel2 (length of the second finger relative to the sum of the lengths of all four fingers) was associated with financial, social and recreational risk-taking. Using circulating testosterone, Sapienza et al. (2009) found that MBA students with higher testosterone levels were more likely to choose a career in finance rather than a less risky career after graduation. Coates and Herbert's (2008) small study of 17 male UK City of London traders also found that morning levels of testosterone were associated with their profitability for that day. Because self-employment involves proactive behaviors, other studies focused on testosterone's role in competition and status seeking/maintenance also provide clues as to why testosterone may be associated with self-employment. Competition studies (Archer, 2006) generally indicate that actual or prospective competitive activities lead to changes in testosterone levels (Fry et al., 2011 and van der Meij et al., 2012) while Saad and Vongas (2009) found that testosterone levels were linked to displays of social status: men that drove a Porsche had higher testosterone levels than when they drove a lower status vehicle such as a family sedan. Overall, given the direct evidence of a link between testosterone and self-employment (White et al., 2006 and White et al., 2007) and cognate evidence that links testosterone with risk-taking and agentic behaviors (Archer, 2006, Sellers et al., 2007 and Zyphur et al., 2009), we hypothesize that higher serum testosterone levels in men is associated with being self-employed – particularly compared with those in employment but also potentially those that are unemployed and inactive in the labor market. In doing so, we are conscious of two issues. First, we are unaware of any other large-scale community based study that specifically examines the effect of testosterone on self-employment. Second, the tendency in prior cognate studies has been to assume that because testosterone levels are, in part, heritable (Meikle et al., 1988) then the direction of causation flows from testosterone to the behavior. Testosterone, though, is a labile hormone. Studies show that it changes with age (Gray et al., 1991, Tennekoon and Karunanayake, 1993 and Harman et al., 2001); and is influenced by environmental conditions. Gray et al. (2006) found testosterone levels to be higher in urban rather than rural dwellers (Saad and Vongas, 2009). Mazur and Booth (1998), Booth et al. (2006), and van Anders and Goldey (2010) also found that higher testosterone levels are more likely among divorced and single males than in men who are married. Higher testosterone levels have also been found to be associated with increased alcohol consumption (Lenz et al., 2012). By contrast, further meta-analytic studies have shown that testosterone is inversely related to insulin levels with men with low testosterone levels being more prone to insulin resistance (Grossmann, 2011). This plasticity of testosterone means that even if testosterone levels are higher among the self-employed this may not reflect that role occupation has been shown to have a genetic basis (Zhang et al., 2009), particularly in relation to self-employment (Nicolaou et al., 2008). Instead, testosterone may be higher because the self-employed have to take risks and adopt agentic and competitive behaviors in order to survive in self-employment. Cognate research into honor subcultures in the US (Mazur and Booth, 1998) associates higher testosterone levels among young black men in communities in which there is a cultural expectation that individuals will adopt a defensive posture to protect their honor or reputation. In short, we cannot ignore, particularly as our data are cross-sectional, the potential for reverse causality. Our contribution is 2-fold. First, rather than using samples of particular groups (e.g. MBA students) or small groups of subjects, we use a large representative study of 1199 adult Australian males from the Florey Adelaide Male Aging Study. We also use blood samples (collected at similar times of the day) of the two types of activational testosterone: total testosterone levels and bioavailable testosterone levels. Second, we use an instrumental variable (IV) approach to control for endogeneity due to omitted variable bias and reverse causality in the relationship between testosterone and self-employment. The rationale for using an IV approach is explained in Section 2. In Section 3, we provide an overview of our data and detail the variables used in the study. We then present the results in Section 4 while conclusions are presented in Section 5.

To the best of our knowledge, there have been no prior large-scale community based studies of the association between testosterone and self-employment.7 This study hypothesized – based on earlier cognate studies of the association between testosterone and status seeking, competition and risk-taking – that testosterone is associated with self-employment. However, one other contribution of this study is to recognize that testosterone – either in terms of total or bioavailable testosterone – is labile and it is important to control for endogeneity. The study, therefore, used a two stage IV approach to control for endogeneity. Using this IV approach with insulin as an instrument, the study examined the association of total testosterone and self-employment using large-scale community based data. This approach is in contrast to earlier studies that have focused on small groups of subjects or particular groups and, typically, have not sought to control for endogeneity. Our IV probit results indicate that there is a positive relationship between total testosterone and self-employment. Using the alternative measure of testosterone (bioavailable testosterone) and another instrument (alcohol consumption per day) as robustness checks we found a similar positive relationship. Overall, this study found that testosterone levels are associated with the likelihood of being self-employed. An increase in total testosterone levels by a standard deviation increases the probability of self-employment by 10%. For bioavailable testosterone, a standard deviation increases the likelihood of self-employment among labor market active males by 12%. This finding provides succor to the view that testosterone is associated with self-employment and is in line with other cognate research that has identified that testosterone is associated with risk-taking proclivities and status seeking competition (Archer, 2006). Our results, therefore, contribute to the burgeoning literature that suggests that testosterone in men has a wide-ranging impact on behavior beyond its effects on sex and reproduction. Our data, though, do not allow us to distinguish if testosterone levels are higher because of risk-taking or status seeking competition. One reason for this is that we do not have a measure of risk-taking. Nor do we have information that considers the relationship between testosterone levels and performance (e.g. income) and particular types of self-employment (e.g. professional, trade). These are important factors because the status and risk profile of self-employment varies across the self-employment life cycle (Shane, 2003). For the nascent self-employed (activities associated with becoming self-employed), the typical risk is establishing economic viability of self-employment while the status challenge is ensuring the social credibility and legitimation of their self-employment (Stinchcombe, 1965). This differs from the actual self-employed (the focus of our study) whose the main challenge is status maintenance and ensuring that costs are lower than revenues (Henley, 2007), or those who have exited self-employment who may have adapt to status loss even if their income – if in employment – is likely to be much more stable (Shepherd, 2003). We, therefore, call on research to use longitudinal data to assess how testosterone changes over time in and out of self-employment; to consider different patterns of self-employment and the specific role that status seeking and risk play in the relationship between testosterone and self-employment. We also call for consideration of how testosterone levels are influenced by, inter alia, stress levels and hours worked (Smith et al., 2006). This may be particularly pertinent because the self-employed, despite having greater levels of job satisfaction than employees (Benz and Frey, 2008) and evidence that those with elevated testosterone have lower stress reactivity (Hermans et al., 2007), have been found to work relatively longer hours and have greater levels of stress. Overall, there is a need for much more research in this area, both to investigate if testosterone across the self-employment life cycle varies and how risk-taking, status seeking and other factors influence testosterone levels. Our hope is that further research can build upon our approach and findings. In conclusion, this study has sought to examine the association between testosterone and those that are self-employed. Using data on 1199 Australian males, our two stage IV approach indicate that testosterone is associated with agentic activities such as self-employment. While such results are novel, we call for further research to tease out how testosterone influences – and is itself influenced – by different forms of employment.