Brandon John Williams

Abstract:  Incentives are often deployed in eliciting beliefs, with the idea that in doing so we can help collect better beliefs. Implicitly, in wanting to provide incentives we are trying to counteract a cost of information acquisition or reporting, either in understanding the construct, formulating a clearer belief, or pushing against ego-relevant or partisan motives for distortions in the report. But if so, stark marginal incentives within our elicitations are paramount. In this paper we examine the effects of marginal incentives on the accuracy of belief formation. We find that starker marginal incentives (though a close-enough incentive) generate substantially more effort in forming a probabilistic belief than a BSR incentive. A series of designed experiments attempt to elicit the psychic costs of completing a Bayesian updating task.

Abstract: A recent vein in the behavioral economics literature has tried to evaluate how well theoretical mechanism design performs with real decision makers. Deviations from classical predictions due to non-standard preferences, bounded rationality, or misunderstanding of the mechanism can severely alter the mechanism’s efficacy. In this study, we examine a potential advantage of designed mechanisms: reducing the potential for behavioral deviations by relocating conflict within the mechanism. We focus on a recently proposed bargaining mechanism under uncertainty, from Kartik, Kleiner and Van Weelden (2021), where a seller and buyer negotiate over quantity, but where the buyer’s demand function is private information. Theory identifies an optimal solution—a constrained delegation mechanism in which the seller offers a menu with a minimum quantity requirement. The mechanism shifts the nature of the conflict between buyer and seller, from adversarial to one of revealing private information. In doing so, it optimally balances efficiency (how often agreements fail) against surplus extraction (the seller’s payoff) through the quantity constraint. In our experiment, we implement a two-by-two design varying the mechanism and the presence of pre-play communication. First, we compare this delegation mechanism to a simpler decision where the seller makes a single take-it-or-leave-it offer. In a secondary dimension, we examine the potential effects from revelation of information outside the mechanism, examining pre-play chat, which should generate considerable efficiency gains in the single-offer mechanism but offers no theoretical value to the delegation mechanism and may in fact worsen outcomes behaviorally as the chat becomes a venue for buyer-seller conflict.