Testosterone and Financial Risk Taking

Testosterone levels have been directly associated with risk-taking behavior in multiple studies. For example, when trial lawyers who take courtroom risks are compared to non-trial lawyers, they have higher testosterone levels. So do criminals who commit violent crimes compared to those who don’t. This association of risk taking and testosterone levels extends to economic-decision making as well. I previously wrote about a study that showed that morning testosterone levels in Wall Street traders had a direct positive relationship with profits during the day.

Now, we have to rethink our concept about testosterone and risk taking during economic-decision making. It may not be a linear relationship as previously thought, i.e., the higher the testosterone level, the greater the propensity for risk taking.

Scott A. Huettel and colleagues from Duke University presented new evidence that endogenous testosterone levels have a U-shaped relationship to risk taking in both men and women. Individuals with testosterone within a standard range were risk averse. Those with testosterone levels more and less than 1.5 standard deviations from this mean were risk neutral -- meaning they were willing to take more risk. The study was published in the March issue of Psychological Science.

In the words of the authors, “In contrast to received wisdom regarding testosterone and risk, the present data provide the first robust evidence for a nonlinear association between economic preferences and levels of endogenous testosterone.”

The study involved 142 men and 156 women, in ages ranging from 18 to 48 years (the mean was 22.1 years). Economic-decision making tasks were performed that included risk and ambiguity preference, and loss aversion. Risk was determined through 120 trials in which the subjects chose between a secure outcome and an uncertain gamble. Saliva was taken to determine the subject’s testosterone level.

When the results were examined, women were more risk averse than men, as expected. The big surprise was that regression models showed that risk preference did not have a linear relationship with testosterone. Instead, it was U-shaped as I stated earlier. The groups on either end of the core were risk neutral -- meaning higher risk was accepted than the middle range, which was risk averse. This observation was true for each sex.

The authors offered several explanations for this unexpected result. One that I thought was particularly plausible was that risk seeking may become adaptive when resources are scarce. If so, the greater potential for variability in preferences among those with intermediate testosterone levels could reflect adaptive flexibility.

In addition, testosterone levels change all the time. Perhaps studying them as we have in relatively static ways is not the optimal method. A more dynamic approach may be needed, such as testing testosterone levels more frequently during economic decisions. If anything, this study shows us that we don’t know as much as we thought we did about testosterone and risk taking.