Essays on Stochastic Choice

Abstract

This thesis delves into the realm of stochastic choice theory, offering novel perspectives across three distinct but related chapters. In Chapter One, we introduce a Salience Biased Nested Logit model, constructively incorporating salience bias into choice probabilities, thus exploring variation from conventional nested logit approaches. In this first chapter, we approach the theoretical construction of the model from two different perspectives, the decision theory framework and the random utility estimation framework, presenting a rich set of price elasticities in the applied context. Chapter two shifts focus to the Eliminationby- Aspects framework of Tversky, proposing the entitled Elimination-by-Salience model, which simplifies traditional EBA by dropping sharply the number of parameter while capturing the nuanced impacts of salience on choice. Moreover, we propose a competing model where agents are fully attentive and no salience influences transition probabilities. Chapter three, “Replicas Without Full Support,” extends the Luce model with replicas, integrating non-positive probabilities and offering a broader understanding of choice under uncertainty. Collectively, these chapters contribute to the stochastic choice literature by introducing innovative models that enhance our understanding of decision-making processes, highlighting the significance of salience and the rationality of choice within stochastic decision frameworks.