Date of Completion


Embargo Period


Major Advisor

Robert Day

Associate Advisor

Manuel Nunez

Associate Advisor

Mikhael Shor

Field of Study

Business Administration


Doctor of Philosophy

Open Access

Campus Access


In four essays, this dissertation introduces models of decision making under risk. Standard models of rational choice have assumed decisions should not be influenced by changes in context, the way equivalent choice alternatives are presented, or the mode of preference elicitation. As a result, these models fail to predict certain consistently observed behavioral effects, including many that are widely viewed as “rational,” if not reasonable and predictable. Chapters 1, 2, and 3 provide background material. The first essay (Chapters 4-5) investigates a new model of reference-dependent preferences known as Target-Adjusted Utility (TAU), in which the target payoff used for evaluating a set of risky choices is described as varying systematically across choice sets. Though several “target functions” are plausible, the “maximin” target (among the most simple and natural possible implementations) is shown to resolve many of the classical paradoxes for decisions under risk. The resulting model also makes novel predictions supported by the experimental literature, and can be viewed as a mild relaxation of standard expected utility theory. The second essay (Chapter 6) embeds expected utility theory in a new formal model, in which risk perception is based on the framing of alternatives as a set of direct comparisons of the component probabilities and payoffs. This descriptive model explains classic framing effects documented in the literature, as well as other features of commonly observed risk preferences. For the third essay (Chapter 7), two behavioral experiments were conducted to test for the prevalence of a classic decision bias – the common ratio paradox – under three qualitatively different modes of preference elicitation: choice, pricing, and happiness rating tasks. The experimental results reveal that both the consistency and distribution of responses vary significantly across modes of elicitation. The final essay (Chapter 8) of the dissertation considers decision making under risk in an applied setting: core-selecting combinatorial auctions with risk-averse or loss-averse bidders. Using a stylized combinatorial auction setting from the literature, results include Bayes-Nash equilibrium bidding strategies for loss-aversion, and bounds on the equilibrium strategies for risk-aversion. Implications for revenue and efficiency, and extensions to more complex auction environments are also investigated.