Betting under subjective uncertainty
I show that efficient markets equilibria may feature non-fundamental bets if some traders are uncertainty averse. This finding is robust to common priors and strictly convex preferences. Conditional on fundamentals, some candidate priors may disagree on the distribution of an otherwise irrelevant variable. This correlation cannot be exploited through contingent claims on fundamentals. I provide necessary and sufficient conditions which restore bet-free trade under maxmin, smooth, and variational preferences. While these conditions are typically stringent, I find that multiplier preferences (Hansen and Sargent, 2001; Strzalecki, 2011) allow for a natural generalization of the common priors condition from expected utility.
Matching to share risk without commitment (with Sarolta Laczó)
This paper reconsiders matching to share risk by requiring insurance transfers to be self-enforcing. We consider a marriage problem (i.e., two-sided, one-to-one matching without search frictions) with imperfectly transferable utility in a dynamic setting. With efficient risk sharing within the household, every stable matching is negative assortative with respect to spouses’ risk attitudes (Chiappori and Reny, 2006; Legros and Newman, 2007). We study the robustness of this result when risk sharing within the household is partial due to lack of commitment. We show that stable matchings might be positive assortative. More risk-averse agents can be more attractive risk-sharing partners because they are able to credibly promise larger insurance transfers.
Socially efficient discounting under ambiguity aversion (with Christian Gollier)
We provide sufficient conditions under which ambiguity aversion decreases the social discount rate when future consumption growth is uncertain. We identify two effects. The first is equivalent to a distortion of beliefs. This pessimism effect requires joint restrictions on the growth process and the agent’s preferences to be signed. The second effect is novel and similar to prudence under expected utility. We show that this decreases the rate if and only if preferences satisfy decreasing absolute ambiguity aversion. Calibrations suggest a net ambiguity effect between –2.5 and –4.5 percentage points for the rate at which cash flows occurring in 30 years should be discounted.
Older paper: Hedging Priors.