Calibration without Reduction for Non-Expected Utility

Calibration results in Rabin (2000) and Safra and Segal (2008; 2009) suggest that both expected and non-expected utility theories cannot produce non-negligible risk aversion over small stakes without producing implausible risk aversion over large stakes. This paper provides calibration results for recursive non-expected utility theories that relax the Reduction of Compound Lotteries axiom (as in Segal 1990). These calibration results imply that a broad class of non-expected utility theories can accommodate both small and large stakes risk aversion, even for a decision-maker who faces background risk.