Title:Revealed Preference Analysis Under Limited Attention

Abstract:An observer wants to understand a decision-maker's welfare from her choice. She believes that decisions are made under limited attention. We argue that the standard model of limited attention cannot help the observer greatly. To address this issue, we study a family of models of choice under limited attention by imposing an attention floor in the decision process. We construct an algorithm that recovers the revealed preference relation given an incomplete data set in these models. Next, we take these models to the experimental data. We first show that assuming that subjects make at least one comparison before finalizing decisions (that is, an attention floor of 2) is almost costless in terms of describing the behavior when compared to the standard model of limited attention. In terms of revealed preferences, on the other hand, the amended model does significantly better. We can not recover any preferences for 63% of the subjects in the standard model, while the amended model reveals some preferences for all subjects. In total, the amended model allows us to recover one-third of the preferences that would be recovered under full attention.