Essays on Stochastic Choice and Welfare

Abstract

Motivated by the literature on preference elicitation and welfare analysis, Chapter I studies the properties of aggregators of choice datasets into preferences. Novel normative principles and their theoretical implications are provided. I analyse numerous approaches proposed by the literature in view of the introduced principles. I also propose and characterize two counting procedures that are foundational for the analysis. Motivated by the theoretical framework of the first chapter, in Chapter II, I propose a novel experimental design to test two normative principles: (1) Informational Responsiveness guarantees that no choice data is ignored; (2) Revealed Preference constrains the preference elicitation process to a particular reorganization of data. These principles are summarized by a method denoted as Counting Reveal Preference procedure. I show that approaches founded on this procedure provide more reliable results in terms of preference relation. Motivated by the literature on stochastic choice, Chapter III studies the relation between imperfect discrimination and the transitivity of preferences. I show that the degree of transitivity depends on the degree of discrimination between pairs of alternatives. I characterize the notions of Weak, Moderate and Strong stochastic transitivity. The results allow us to organize a wide range of stochastic models in accordance with Fechnerian models and imperfect discrimination.