Revealed preference analysis under limited attention

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 generic model of limited attention (Masatlioglu et al., 2012) can only help the observer marginally. To address this issue, we study a family of models of choice under limited attention by imposing an attention floor in the decision process and test whether these models (1) are good at explaining the observed behavior, and (2) can be used to identify the underlying preferences. We then analyze our findings on these models using experimental data that has the novelty of making observations on the consideration sets (Dianat and Freer, 2025). Our findings are threefold. First, subjects consistently exhibit attention floors that are large relative to the size of the choice problem. This is evident both from direct observations on the size of consideration sets and from testing the model with the attention floor disregarding the information about consideration sets. Second, the generic model fails to identify any preferences. Third, although raising the attention floor improves the model’s identifiability, it also introduces false inferences about preferences: roughly one-third of the comparisons it generates differ from the “true” revealed preference relation recovered from direct data on consideration sets. Overall, while the attention floor model fits the observed data well, its usefulness for welfare analysis remains limited.