Small Stakes Risk Aversion in the Laboratory: A Reconsideration

Evidence of risk aversion in laboratory settings over small stakes leads to a priori implausible levels of risk aversion over large stakes under certain assumptions. One core assumption in standard statements of this calibration puzzle is that individuals define utility over terminal wealth, and that terminal wealth is defined as the sum of extra-lab wealth and any wealth accumulated in the lab. This assumption is often used in Expected Utility Theory, as well as popular alternatives such as RankDependent Utility theory. Another core assumption is that the small-stakes risk aversion is observed over all levels of wealth, or over a “sufficiently large” range of wealth. Although this second assumption if often viewed as self-evident from the vast experimental literature showing risk aversion over laboratory stakes, it actually requires that lab wealth be varied for a given subject as one takes the “risk aversion temperature” of the subject. We consider evidence from a simple design that tests this assumption. We find that the assumption is strikingly rejected for a large sample of subjects from a population of college students. However, we find that the assumption is not rejected for a large sample of subjects from the adult population of Denmark. We conclude that the implausible predictions that flow from these assumptions do not apply to one specialized population widely used to study economic behavior in experiments, but do apply to a broader population that is arguably of greater interest