On Welfare Analysis under Limited Attention

An observer wants to understand a decision maker's welfare from her choices. She believes that decisions are made under limited attention -- i.e., not all alternatives are considered. Given the experimental evidence, we argue that existing models of limited attention are not helpful in achieving this observer's goal. We propose a \say{minimal} adjustment to the baseline model of choice under limited attention by assuming that the decision process involves \emph{at least one comparison}. Using experimental data, we illustrate that this adjustment significantly enriches the welfare relevance of the model with negligible costs in its explanatory power.