Identifying Risk Attitude with Moment-Restricted Lotteries

Speaker: A/Prof Chen Zhao

Affiliation: University of Hong Kong

Location: Room 211, Chamberlain Building (#35), St Lucia Campus

Abstract: We analyze the identification of agents' risk attitude when the available lotteries adhere to a general moment condition. We find that an agent’s Bernoulli utility index is nonparametrically identified only up to a positive affine transformation and an additive multiple of the moment function. Assuming HARA utility functions restores generic identifiability, which demonstrates that parametric functional forms play a critical---yet potentially misleading---role in such settings: identified parameters may conflate risk attitudes with structural features of the moment constraints. To illustrate these challenges, we examine a canonical insurance problem where a menu of contracts is available in the market. We show that if full coverage offers the highest payout relative to the premium, the menu inherently satisfies a generalized moment condition dictated by its own structure. Consequently, insurance choices are intrinsically prone to misidentification when relying on parametric utility specifications. Finally, we characterize the data requirements for robust identification. We derive a dimensionality condition on lottery variation that ensures identifiability.