Strong Evidence for Gender Differences in Risk Taking
Highlights
► We consider the results of 15 different studies on risk-taking in investment, each of which gathered data by gender. ► Each study used the same very simple and easy-to-comprehend mechanism for eliciting risk preferences. ► We find the extremely robust result that women are more risk averse than men. ► Most of the data were not collected in order to study gender differences, but rather to study other hypotheses regarding investment behavior, ameliorating the issue of publication bias with respect to gender results.
Introduction
Many economic interactions involve some form of risk. Thus, it is not surprising that a substantial body of research in social science has tried to understand how decision makers incorporate risk in their choices. Expected utility, the dominant theory of decision under risk, makes some testable empirical predictions. However, under expected utility the actual level of risk-taking behavior by the agent is left as a free parameter, allowing for individual differences. This is also true for more behaviorally driven theories, such as prospect theory.
In this article we study one important systematic difference in risk taking between groups. In particular, we study the interaction of risk-taking with the gender of the decision maker. The common stereotype is that women are more risk averse than men; this stereotype is important since it can potentially explain important economic phenomena.1 Empirical investigation of gender differences in risk taking do point in the direction of less risk taking by women than by men (see the surveys in Eckel and Grossman, 2008 and Croson and Gneezy, 2009).
A major problem with the empirical investigation of individual differences in risk taking is the variation in the methods used to study the phenomenon. Considering only the experimental work (done mostly by psychologists), each experiment uses a different decision problem, which makes it hard to compare results. In addition, some of these articles found gender differences without looking for them and others were specifically designed to test for these differences. The problem with the former approach is that these articles report the results of one experiment, and we do not know how many experiments have looked for a difference and did not find one. We are left with a selection bias of articles with a positive finding. The problem with the latter approach is that when the goal is to find (or not to find) a gender difference, the design of the experiment may be based on a small set of incidents in which the researchers expect to find the results they are after; here again, finding no difference can lead to the researcher shelving the project. It is easier to publish an article that reports a gender difference in risk taking than a study that reports no difference. For example, when a study finds that women are more risk taking than men (“man bites dog”), this study has a much higher chance of being published than a study that finds no difference in risk taking. This bias in publication also creates incentives for researchers to design studies that will generate such a difference.
The novelty of our article is in using existing empirical results that were collected in a systematic way using thousands of observations by different researchers in a variety of setups, but based on one simple investment game. Most of the data were collected not in order to study gender differences but rather to study other hypotheses regarding investment behavior, and they vary with respect to the subject pools, country, age, different incentives and probabilities, repeated versus one-shot interaction, laboratory versus Internet, known probabilities versus uncertainty, and framing. Moreover, since the original data were not collected in order to facilitate comparisons, there was no effort made by the researchers to have a uniform design (e.g., use the same instructions). This variety allows us to test the robustness of the hypothesis.
The data are based on an investment decision that was introduced by Gneezy and Potters (1997). In this choice, the decision maker receives $X and is asked to choose how much of it, $x, she wishes to invest in a risky option and how much to keep. The amount invested yields a dividend of $kx (k > 1) with probability p and is lost with probability 1 − p. The money not invested $(X − x) is kept by the investor. The payoffs are then $(X − x + kx) with probability p, and $(X − x) with 1 − p. In all cases, p and k are chosen so that p × k > 1, making the expected value of investing higher that the expected value of not investing; thus, a risk-neutral (or risk-seeking) person should invest $X, while a risk-averse person may invest less. The choice of x is the only decision the participants make in the experiment. We report data from all studies (of which we are aware) using this method for testing risk aversion.
The striking and consistent result is that despite the large environmental differences among the sets of experiments, a consistent gender difference is reported: Men choose a higher x than women do.
Section snippets
Dreber et al. (2010)
This study was conducted in the field, with highly skilled people who consider probabilities and risk quite frequently: tournament bridge players; the financial stakes were also rather large. Participants at the Fall 2008 North American Bridge Championship in Boston were recruited for the study. “Almost all of the participants were serious tournament bridge players who play many dozens of session per year.” The primary focus of the study was with how risk-taking behavior correlates with
Conclusion
The results reported in this article are obtained by using data from previous studies based on one similar design in which the data of interest was recorded independently of the goal of the study. The field of experimental economics is growing quite rapidly, with experiments being relatively easy to run (compared with, e.g., analyzing real world data). There is a natural tendency to continue to collect new evidence without fully considering what we might learn from the data that are already
References (33)
- et al.
On the evolution of attitudes towards risk in winner-take-all games
Journal of Economic Theory
(1999) - et al.
Sex differences and statistical stereotyping in attitudes toward financial risk
Evolution and Human Behavior
(2002) - et al.
Men, women and risk aversion: experimental evidence
- et al.
Gender differences in risk behaviour in financial decision-making: an experimental analysis
Journal of Economic Psychology
(1997) - et al.
Risk aversion and retirement income adequacy
- et al.
Myopic loss aversion: information feedback vs. investment flexibility
Economics Letters
(2005) - et al.
Gender, risk and retirement
Journal of Economic Issues
(2001) - et al.
An experimental test of inequality and risk-sharing arrangements
Economic Journal
(2009) - et al.
Gender, Framing, and Investment
(2004) - et al.
Portfolio choice and risk attitudes: an experiment
Economic Inquiry
(2010)
Second-to-fourth digit ratio predicts success among high-frequency traders
Evolutionary psychology and conceptual integration
Gender difference in preferences
Journal of Economic Literature
2D:4D and Risk Aversion: Evidence that the Gender Gap in Preferences is Partly Biological
Dopamine and Risk Preferences in Different Domains
Cited by (1106)
Voluntary insurance vs. stabilization funds: An experimental analysis on bank runs
2024, Journal of Behavioral and Experimental FinanceDo ‘Lehman Sisters’ work in China? Women on boards and bank risk
2024, International Review of Financial AnalysisGender and choices in higher education
2024, Economics of Education ReviewBoard gender diversity and corporate litigation: evidence from China
2024, Gender in ManagementGender diversity of board of directors and shareholders: Machine learning exploration during COVID-19
2024, Gender in ManagementRural-urban migration, financial literacy, and entrepreneurship
2024, Journal of Business Research