Eliciting preferences for risk and altruism : experimental evidence

We apply the basic lessons and insights learned in the elicitation and estimation of risk and time preferences literature to the literature on social preferences. Following Andersen et al. (2008), we design a laboratory experiment to jointly elicit risk preferences and preferences for altruism. Consistent with theory, we find that the standard simplifying assumptions about risk preferences lead to significantly biased estimates of altruism. This is particularly problematic when comparing altruism across relevant sub-groups, such as gender and wealth, leading to possibly erroneous conclusions about which is the more generous sex and the self-regarding rich.