Hungry and Rational: Preference Shifts and Rationality under Food Deprivation (Working Paper)
(with Ulrich Schmidt and Lukas Baumann)
Previous field studies suggested that scarcity could increase rationality in decision-making. We systematically investigate this hypothesis in a controlled lab experiment that manipulates short-term food scarcity. Our experiment examines multiple facets of rationality across the core domains of decision-making, i.e., risk, time, and social preferences. We find that food deprivation increases both goal rationality — by sharpening material outcome orientation — and normative rationality — by reducing probability weighting, loss aversion, and decision noise, thereby enhancing focus and decision consistency while leaving underlying attitudes, such as self-interest, largely unaffected.
Status: Submitted
Hyperopic Loss Aversion (Working Paper)
(with Ulrich Schmidt)
Time and risk are integral components of many real-world decisions. Despite the importance of their interaction, both domains are primarily studied in isolation. A concept common to models of risk and time preferences is reference dependence. In Prospect Theory, it underlies the property of loss aversion, and in discounting the sign effect, which describes that gains are discounted more than losses. This paper analyzes the interaction of loss aversion and the sign effect. If gains are discounted more than losses, then loss aversion should increase for more distant risks, i.e., preferences should exhibit Hyperopic Loss Aversion. We validate this hypothesis experimentally and show empirically that Hyperopic Loss Aversion offers a novel perspective on the equity premium puzzle. In a first experiment, we propose a new mechanism to incentivize losses in the lab and show that the sign effect, so far only observed for hypothetical decisions, survives incentivization. In a second experiment, we show that mixed lotteries become less attractive when delayed, supporting the hypothesis of Hyperopic Loss Aversion.
Status: R&R at Management Science
Intelligence and Measurement Errors in Risky Decision-Making: Disentangling Risk Attitudes, Rationality, and Noise (Working Paper)
(with Lukas Baumann)
Although it is often assumed that more intelligent individuals are less risk-averse, empirical evidence on the relationship between intelligence and risk preferences remains mixed. We argue that this inconsistency reflects a narrow focus on risk attitudes and propose a broader conceptualization based on three distinct mechanisms: risk attitudes, normative rationality (i.e., deviations from Expected Utility), and decision noise. To examine how intelligence relates to each component, we conduct two incentivized experiments (N = 369). We combine a validated measure of fluid intelligence with structurally estimated risk preferences under Cumulative Prospect Theory, thereby addressing common sources of measurement error in both constructs. The experimental design covers the gain, loss, and mixed domains, allowing us to test whether effects differ systematically across contexts. Our results show no association between intelligence and either normative rationality or risk attitudes. However, we find a consistent negative relationship between intelligence and decision noise. These findings suggest that intelligence improves the internal consistency of choices under risk but does not affect individuals’ underlying preferences. Our findings highlight how construct validity, an often overlooked source of measurement error, can obscure the relationship between intelligence and economic decision-making.
Status: Submitted
Sign effects in Quasi-Hyperbolic Discounting: Evidence from an incentivized Experiment (Working Paper)
Quasi-hyperbolic preferences are actively studied in the gain domain, with two recent meta-analyses providing evidence of a sizeable present bias. However, little is known about a possible sign effect for present biases. Even though the sign effect for discount rates is an established and often replicated phenomenon, evidence from incentivized experiments is rare. This paper reports evidence from an incentivized experiment where subjects completed a comprehensive risk task that allows estimating all parameters of a Prospect Theory model and an additional, comprehensive time preference task designed to elicit present biases and discount factors in the gain and loss domain. I introduce a novel approach to measure time preferences on the population and individual level by equating the instantaneous utility function and Prospect Theory value function. The results show a sign effect for both discount rates and present biases. Specifically, I find a present bias for losses but not gains at the population level. On the individual level, many subjects are present-biased in one domain and future-biased in the other domain.
Status: Revising Draft
Quasi-hyperbolic Prospect Theory: Axiomatizations of Intertemporal Risk Preferences (Working Paper)
Intertemporal extensions of Prospect Theory are slowly emerging as alternatives to Discounted Expected Utility. But, while these models assume sign-dependence with respect to probability weighting, they do not extend it to discounting behavior, i.e., potential gains and losses are discounted with the same factor. This is surprising, as sign-dependence is a central aspect of Prospect Theory and as there is a sizable literature on sign-dependent discounting, i.e., on the sign effect which describes that gains are discounted more than losses. In this paper, I introduce and present an axiomatic analysis of Quasi-Hyperbolic Prospect Theory (QHPT), a model of intertemporal risk preferences that combines Quasi-hyperbolic Discounted Utility and Prospect Theory. In line with Prospect Theory, QHPT extends sign-dependence to the discounting parameters β and δ. The axiomatization approach can be adapted to extend a large number of risk preference representations to the temporal domain and provides important insights for the discussion of whether intertemporal risks are evaluated risk-first or time-first.
Status: Revising Draft
Perceived Similarity in Risk Taking for Others (Working Paper)
(with Christoph Schütt)
Recent studies on risk-taking for others report either a “risky” or a “cautious shift”, that is, more or less risk-taking for a partner compared to oneself. In addition to contradicting results, there exists no consensus about the channels involved. This study provides an experimental test of Construal Level Theory as an explanation for differences in risk-taking for oneself and a partner. By varying the perceived similarity to a partner, we manipulate the construal level while keeping information about the partner’s risk preferences constant. Using this manipulation, we find that subjects take significantly more risk for a dissimilar partner, for whom the construal level is high, than for a similar partner, for whom the construal level is low. In the gain domain, the target decision gap, that is, the difference in risk-taking for oneself and a partner, is significantly positive for a dissimilar partner and negligible for a similar partner. In the mixed domain, we find positive target decision gaps for similar and dissimilar partners.
Status: Revising Draft
Noisy or Not? A Hierarchical Bayesian Test of Impulsivity and Risk Preferences. (Draft available on Request)
(with Lukas Baumann)
The relationship between impulsivity and risk preferences has yielded mixed findings in the literature. A comprehensive review reveals that most studies have relied on general measures of risk preferences, which may overlook established empirical patterns or confound risk preferences with other behavioral factors. To gain a more accurate understanding of the relationship between impulsivity and risk preferences, this paper employs a refined measure of risk preferences within a Hierarchical Bayesian structural model grounded in Prospect Theory. This approach enables a detailed analysis of risk preferences across gains, losses, and mixed outcomes. Using data from an incentivized experiment, Prospect Theory parameters at both individual and aggregate levels are estimated, and the influence of impulsivity on these parameters is examined. Our results reveal no significant effect of impulsivity on risk preferences, underscoring the importance of combining precise risk measures with advanced modeling techniques.
Status: Revising Draft