Moritz LOEWENFELD will defend his thesis on Friday 7 June at 09:30 (Auditorium 5 and zoom)
« Essays on correlation-sensitivity in economic decision-making - a theoretical and experimental exploration »
Supervisor: Professor Astrid HOPPFENSITZ
To attend the conference, please contact the secretariat Christelle Fotso Tatchum
Memberships are:
- Astrid HOPFENSITZ : Professor in Economics, Université Lumière Lyon 2 Supervisor
- Sébastien POUGET : Professor in Economics, TSM-TSE, Université Toulouse Capitole Examinateur
- Aurélien BAILLON : Professor in Economics, EM Lyon Business School Rapporteur
- Ferdinand VIEIDER : Professor in Economics, Ghent University Rapporteur
Abstract :
This thesis is dedicated to studying correlation sensitivity in decision-making under risk. Allowing risk preferences to be sensitive to the correlation between lottery outcomes can explain classical deviations from expected utility theory as well as phenomena in various real-world settings. However, experimental evidence on correlation sensitivity is limited and mixed. Moreover, the concept of correlation sensitivity has thus far been studied almost exclusively in the context of individual decision-making. Chapters 1 and 2 seek to contribute to our understanding of correlation-sensitive risk preferences, whereas Chapter 3 explores how choices can become correlation-sensitive in a setting of delegated decision-making, even when none of the involved parties has correlation-sensitive preferences.
In the first chapter, together with Jiakun Zheng, we study correlation-sensitive preferences in individual decision-making. We show that correlation-sensitive preferences in the general framework of (Lanzani, 2022) can be classified into three categories. We propose a choice task to classify experimental subjects accordingly. In multiple experiments, we find that aggregate choices display correlation sensitivity but in the opposite direction as assumed in regret and salience theory. Clustering analysis identifies a consistently correlation-sensitive minority driving aggregate patterns, with the majority showing no correlation sensitivity. Crucially, the analysis does not produce a regret/salience theory type. We disentangle correlation sensitivity arising from deliberate within-state comparisons from incidental payoff comparisons due to the framing of decision problems. Both produce correlation sensitivity, with deliberate comparisons exerting a somewhat greater influence.
In the second chapter, also together with Jiakun Zheng, we reconsider recent experiments that seem to show evidence for correlation sensitivity as assumed in salience theory. However, these studies fail to control for event-splitting effects (ESE). We seek to disentangle the role of correlation and event-splitting in two settings: 1) the common consequence Allais paradox as studied by Bordalo et al. (2012), Frydman and Mormann (2018), Bruhin et al. (2022) 2) choices between Mao pairs as studied by Dertwinkel-Kalt and Köster (2019). In both settings, we find evidence suggesting that recent findings supporting correlation effects are largely driven by ESE. Once controlling for ESE, we find no consistent evidence of correlation effects. We conclude that our results thus shed doubt on the validity of salience theory in describing risky behavior.
In the third chapter, I leave the realm of individual decision-making and consider a setting of delegated decision-making. In a theory-guided experiment, I study how outcome bias, a tendency of principals to reward and punish economic agents as if they could have anticipated a random state of the world, shapes the incentives and choices of agents. Agents choose between two lotteries on behalf of their principal. One lottery is first-order stochastically dominant, but the dominated lottery is more likely to yield a higher payoff state-by-state. Despite perfectly observing the agent’s choice, principals tend to reward agents if they choose the lottery, which realizes a higher payoff. As a result, they incentivize agents to choose the dominated lottery. Although most agents anticipate these incentives, only strategically sophisticated agents tend to choose the dominated lottery when they believe they have an incentive to do so. Structural estimation suggests that principals are either fully outcome-biased or fully unbiased, with less cognitively sophisticated principals displaying more outcome bias. The results imply that outcome bias might be most relevant in settings where sophisticated agents meet relatively unsophisticated principals.
