An Experimental and Theoretical Investigation of Decision-Making Under Risk

Citation

Lucia, Aldo (2024) An Experimental and Theoretical Investigation of Decision-Making Under Risk. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/2sk6-j508. https://resolver.caltech.edu/CaltechTHESIS:05042024-221411474

Abstract

This dissertation comprises three chapters related to the fields of decision theory, game theory, and experimental economics. Chapters 1 and 2 use experimental and structural methods to study individual decision-making in the domain of risk, while Chapter 3 examines decision-making under risk in settings of strategic interaction.

In Chapter 1, co-authored with Shunto Kobayashi, we conduct the first experiment that studies two classical behaviors under risk inconsistent with Expected Utility together: the common ratio effect and preferences for randomization. We show that these two behaviors are strongly positively correlated in a manner inconsistent with the predictions of leading economic models and machine learning algorithms. Motivated by this observation, we develop a novel empirical approach which, unlike machine learning algorithms, imposes some basic assumptions on preferences but does not rely on specific decision models. We further demonstrate that this approach provides more accurate predictions---both inside and outside laboratory settings---compared to leading economic models and machine learning algorithms.

In Chapter 2, I design an experiment testing Expected Utility's central independence axiom and contemporaneously eliciting measures of decision confidence. Recent theoretical work implicates decision confidence as a central component of decision-making under risk, attributing failures of Expected Utility to a lack of confidence. I find that choices characterized by high self-reported levels of decision confidence and low response times are more likely to comply with the independence axiom. Contrary to the common certainty effect rationale for independence violations, I show that subjects predominantly violate Expected Utility by choosing risky lotteries over certain amounts when they are unconfident in their choices.

In Chapter 3, co-authored with Marco Loseto, we study static games in which players have convex preferences. Under convexity, players' preferences admit a conservative multi-utility representation: each utility generates an evaluation for each action, and actions are ranked according to the lowest evaluation. We characterize the set of optimal actions for players with convex preferences and propose an efficiency criterion to rank them. Next, we derive a new class of mixed Nash equilibria that we call ``strict'' because players strictly prefer randomization. In general, convexity may lead to a multiplicity of mixed Nash equilibria. However, we show that when they exist, only strict equilibria ensure that all mixed actions are efficient.

Thesis Files