Universität zu Köln

Armouti-Hansen, Jesper (2022). Essays on the Behavioral Foundations of Choice. PhD thesis, Universität zu Köln.

Preview

PDF dissertation_JAH_v2.pdf - Accepted Version Download (781kB) | Preview

Abstract

The research presented in this dissertation contributes to the literature on choice and decision theory. Chapter 2 presents generalizations of two well-known boundedly rational choice proce- dures. Our generalization consists of defining these procedures as choice correspondences, instead of choice functions. In turn, this imposes less of a restriction on the contained rationales and allows for the decision maker to be indecisive. We provide the axiomatic characterizations of our generalizations by extending the axioms used to characterize the original procedures. Furthermore, we discuss ways in which an indecisive decision maker may arrive at a unique choice and show that our proposed generalizations can explain choice anomalies that cannot be accommodated in the original setup. Chapter 3 introduces a two-period model of individual decision-making, in which the decision maker derives utility from both future consumption and present anticipation of such consumption. Specifically, we consider a setting in which the decision maker may choose her anticipation and where this choice of anticipation, in turn, determines her reference point. In this setting, we formulate equilibrium concepts that dictate feasible choices of anticipation and consumption lotteries based on when the decision maker com- mits to her decision. In addition, we show that our model on the domain of choice is equivalent to a two-stage choice procedure based on the concept of consideration sets. We provide the axiomatic characterization of the choice procedure, and hence by extension, a characterization of our model of anticipation-based reference-dependent preferences. Fi- nally, we show the extent to which the decision maker’s preferences and consideration set can be identified from choice data. Chapter 4 assesses the predictive capability of simple linear social preference models by using flexible machine learning models as a benchmark. Specifically, based on exper- imental observations from the lab on binary dictator games and reciprocity games, we apply the recent introduced concept of a model’s completeness by comparing its predic- tive performance to that of (i) a naive baseline model that is stripped of other-regarding preferences, and (ii) a non-parametric machine learning model capable of capturing the predictive variation in the data. In turn, this provides us with information on (i) how large a fraction of the predictable variation in the data a given social preference model captures, and (ii) how large a gain in performance the model brings by introducing other-regarding motives compared to a naive baseline model. To address the potential remaining patterns in the data that are not captured on the level of the representative agent, we also conduct the analysis in a mixture model framework allowing for heterogeneity in the estimated parameters.

Item Type:	Thesis (PhD thesis)
Creators:	Creators Email ORCID ORCID Put Code Armouti-Hansen, Jesper jeshan49@gmail.com UNSPECIFIED UNSPECIFIED
URN:	urn:nbn:de:hbz:38-550429
Date:	2022
Language:	English
Faculty:	Faculty of Management, Economy and Social Sciences
Divisions:	Faculty of Management, Economics and Social Sciences > Business Administration > Corporate Development > Professorship for Business Administration and Human Resources Management
Subjects:	Economics
Uncontrolled Keywords:	Keywords Language sequential rationalization UNSPECIFIED anticipation UNSPECIFIED reference-dependence UNSPECIFIED social preferences UNSPECIFIED choice theory UNSPECIFIED decision theory UNSPECIFIED machine learning UNSPECIFIED
Date of oral exam:	21 December 2021
Referee:	Name Academic Title Sliwka, Dirk Prof. Dr. Mariotti, Marco Prof. Ph.D. Puppe, Clemens Prof. Dr.
Refereed:	Yes
URI:	http://kups.ub.uni-koeln.de/id/eprint/55042