18 September 2024
12:50
CORE C.035
Hassan Nosratabadi (ULB. ECARES)
On the Welfare (Ir)Relevance of Two-Stage Models
This paper investigates the welfare relevance of two-stage models of choice. That is, we ask to what extent these models may be used to identify preferences. We show that a large class of the two-stage models can only be identified through direct choice reversals. This class includes the models of limited attention, choice overload, rational shortlist method, among others. Moreover, we provide characterization results for some of these models to find the exact nature of direct reversals needed to learn about preferences. Finally, applying our result on experimental data, we show that almost no preference are identified via these models. On the other hand, the data seems to suggest that welfare relevance, almost entirety, responds to ad-hoc assumptions about the mechanics of the first stage (e.g., the size of the filter) rather than its nature, where the focus of the literature has been so far.