Abstract :

Risk preferences define many aspects of human decision making, and assumptions about their stability drive many economic models. It nevertheless remains an open empirical question as to whether risk preferences are stable over time and space. To answer this, we develop a new risk measure extracted from approximately 10 million chess games. Games are taken from over 50 years of tournament praxis and 137 countries. This in turn, allows us to follow individual players across the lifecycle, and use within-player estimates. Doing so, we show that risk preferences decline most rapidly before the age of about 30, plateau somewhat between the ages of 40 and 60, before declining slowly for the remainder of the lifecycle. Similarly, we provide the first empirical evidence that risk preferences are rank order stable. Finally, we show that there are large gender differences in the lifecycle progression of risk preferences, which may explain the varied reports in the current literature on differences in risk preferences between men and women.