Economic Experiments on Behavior, Beliefs and Preferences

Sammanfattning: This dissertation consists of four independent papers, which share the use of experimental methods. Using different settings – the field, the lab, an online platform, and telephone interviews – I study economic behavior and its underpinning beliefs and preferences.The first paper investigates how an incentive for a hedonic but risky activity affects the risk assessment of young adults. I conduct a field experiment and show that when offered a voucher for a café visit during the COVID­19 pandemic, young adults not only visit that café more often, but also start to convince themselves that the risk of a café visit is low. They downplay the risk in anticipation of the visit and in the absence of information that would justify an assessment update. This finding contradicts Bayesian updating theory but is consistent with the notion of motivated reasoning.In the second paper, we conduct two online experiments to study the demand for gender­related affirmative action (AA). We document evidence that around one third of our subjects who act as employers choose a to hire workers using a female quota, which guarantees that at least one of two vacant positions is filled with a woman. Surprisingly, this result is stable across three treatment conditions despite varying costs to employ the quota and varying beliefs about the productivity of workers. We interpret this finding as evidence for an inelastic demand.My third paper seeks to solve an empirical puzzle: How does time affect prosocial behavior? I begin by conducting a meta­analysis collecting 14 experiments that addressed this question but report strikingly mixed evidence. This allows me to compare design features that were kept constant within the experiments but may have moderated their treatment effects. I identify repeated interaction as a potential moderator. In a second step, I test this feature in a controlled donation experiment that is set up through telephone interviews to mimic a fundraiser. The results of that experiment do not support the intuition of my meta study. I do not find that repeated interaction reinforces or alleviates the effect of time on donations.In the fourth paper, we analyze the drivers of time discounting. We conduct a lab experiment and show that people discount lotteries less than their certainty equivalents. Next, we exploit this variation to structurally estimate parameters of two important drivers: Future uncertainty and impatience. We find that more than 70%of the observed time discounting can be explained by an aversion against future uncertainty. This contradicts the common interpretation of the discount factor as a measure of impatience.