New paper on recursive stochastic choice

Drew Fudenberg and I just completed a new
. We provide axiomatic characterizations of two sorts of stochastic choice rules, where the agent makes his current decisions using a forward-looking value function that takes into account his future randomizations. One of the rules is the discounted logit model used in estimation, whereas the other one is novel. The rules differ in how the agent views future choice sets and how he views his future randomizations: one rule exhibits a “preference for flexibility” while the other is “error-averse.” Comments welcome.

Share on: