Previous research has documented a behavioral distinction between "social risk" and financial risk. For example, individuals tend to demand a premium on the objective probability of a favorable outcome when that outcome is determined by a human being instead of a randomizing device (Bohnet, Greig, Herrmann, and Zeckhauser 2008; Bohnet and Zeckhauser 2004). In this paper we ask whether social risk is always aversive, answering in the negative and identifying factors that can eliminate, or even change the sign of, the social risk premium. Motivated by the stereotype content model from the social psychology literature, which we argue has straightforward predictions for situations involving social risk (Fiske, Cuddy, and Glick 2007), we focus on two factors: "warmth," synonymous with intent, and "competence." We investigate these factors using a between-subjects experimental design that implements slight modifications of the binary trust game of Bohnet and Zeckhauser across treatments. Our results indicate that having risk generated by another human being does not, on its own, lead to a social risk premium. Instead, we find that a positive risk premium is demanded when a counter-party has interests confl icting with one's own (low warmth) and, additionally, is competent. We find a negative social risk premium - i.e., social risk seeking - when the counter-party has contrary interests but lacks competence.
