Date:
Location:
Tao Lin
Persuading a Behavioral Agent: Approximately Best Responding and Learning
Abstract: The classic Bayesian persuasion model assumes that the receiver is fully Bayesian and best responds to the sender’s signaling scheme. In this talk, I will talk about two alternative (and more realistic) behavioral models for the receiver: (1) approximately best responding, and (2) learning. We analyze how these two behavioral models affect the sender’s utility in the Bayesian persuasion problem. We first show that, under natural assumptions, approximately best-responding behaviors of the receiver do not affect the sender’s maximal achievable utility a lot. In particular, the sender can always find a signaling scheme that guarantees a utility similar to the sender’s optimal utility in the classic Bayesian persuasion model, no matter how the receiver approximately best responds. The sender cannot achieve a much higher utility than that, either. We then consider the case of a learning receiver in a repeated Bayesian persuasion setting. Interestingly, the previous conclusion no longer holds now: the sender can sometimes exploit the learning behavior of the receiver and achieve a much higher utility than its optimal utility in the classic Bayesian persuasion model.
Lily Xu
Optimizing ranger patrols to deter poaching: causal insights for improving protected area management
Abstract: Ranger patrols are a primary conservation tool for deterring poaching and conserving wildlife. However, patrol-induced deterrence has not been conclusively identified due to ecosystem–human complexities and detection difficulties surrounding poaching events. We present the first causal evidence of deterrence, using a field test in Uganda as a shock to control for unmeasured confounders and Bayesian inference to recover unobserved outcome labels. We found that 1 kilometer increased patrol effort reduced the probability of poaching in the next timestep by 47%. We show that improving resource allocation within protected areas can reduce the required quantity of resources: we estimated that deterrence is most effective near park boundaries and that improving patrol allocation can reduce the globally required number of rangers from 1.5 million to 978,000.
As usual, we will have a Zoom option to join the talk virtually, but please try to be there in-person this week if you can.
https://harvard.zoom.us/j/95184948637?pwd=bXBIc2U5MEZ0QmRUb01WQ0o0SXRCdz09
(password: econcs)