Work in progress

Abstract: La crise de la reproductibilité, révélée par Ioannidis (2005) et confirmée par de nombreuses études, soulève des inquiétudes sur la fiabilité des résultats scientifiques. Bien que l’économie affiche un taux de réplication relativement élevé, elle reste vulnérable aux biais de publication, au p-hacking et à l’effet-tiroir. La science ouverte propose des solutions pour renforcer la crédibilité des recherches. Trois leviers sont essentiels : (1) l’accès libre aux publications, données et codes, (2) le pré-enregistrement et les registered reports pour limiter les biais analytiques, et (3) la réplication systématique des études. En examinant leurs mécanismes et défis, nous soulignons leur impact sur la fiabilité scientifique. L’adoption de ces pratiques marque un tournant vers une science plus robuste et cumulative, mais nécessite une évolution des incitations académiques, un soutien institutionnel et un engagement collectif pour assurer la reproductibilité des recherches.

Abstract: We study experimentally two variants of the beauty contest game (BCG+ and BCG-, thereafter) with an interior equilibrium. Subjects choose an integer number between 1 and 100. The winner is the one whose chosen number is closest to 2/3×(mean+30) in BCG+ or 100-2/3×mean in BCG-. For both games, the unique equilibrium, assuming IEWDS, is 60. The key difference between the two elimination processes is alternation. Under positive feedback elimination occurs only on one side of the equilibrium point, either the upper side or the lower side, depending at which extremity of the interval the reasoning process is initiated. In contrast, under negative feedback, elimination alternates on both sides of the equilibrium point, whatever the extremity that is chosen as the starting point. We use eye-tracking to identify whether subjects rely on one-sided elimination under BCG+ and two-sided elimination with alternation under BCG-.

Abstract: We assess the impact of signed environmental externalities on individual portfolio decisions in a lab-in-the-field experiment on finance professionals and students. Participants are prone to accept lower returns for positive environmental impact but will not bear increased risk. They show asymmetric pro-environmental preferences depending on the sign of the externality. Finance professionals are more pro-environment than students, particularly regarding positive externalities, and less influenced by a ranking signal about environmental performance. Control tasks show that experimental measures of pro-social and environmental preferences are significant predictors for students, while actual market practices for professionals explain a significant part of professionals portfolio composition.

This paper investigates the effect of an additive unfair background risk on the optimal donation behavior of a purely altruistic decision-maker in

the dictator game. Assuming both the dictator and the recipient face this background risk, and that the dictator is risk-vulnerable, we find that the

presence of background risk reduces the deviation of the optimal donation from an equal split. This outcome holds whether the dictator’s fairness

is evaluated ex-ante or ex-post, offering new insights into the implications of risk vulnerability.

Abstract: We conducted a study to assess the impact of a delegation option on moral decision-making using an online questionnaire based on the "moral machine" paradigm1. Interestingly, the inclusion of a delegation option did not significantly alter individuals' moral tradeoffs. Nevertheless, when presented with the option, most participants opted for delegation as a means to avoid the moral burden of challenging decisions, regardless of the delegate's profile. Factors influencing this choice included gender (favoring females), occupation (doctors), education level (lower), a strong sense of fairness, less frequent driving, and greater risk aversion. Additionally, participants displayed a preference for saving more lives, with particular emphasis on babies, pregnant women, doctors, and law-abiding victims, indicating a general aversion to death.

Abstract: We study theoretically and experimentally the approval mechanism (Masuda et al., 2013, 2014) in the case of a three-player voluntary contribution game with the minimum disapproval benchmark. The three-player game raises the issue of aggregating approval votes. We compare the unanimity and the majority approval rules. Backward elimination of weakly dominated strategies predicts null contributions under the majority rule and any contribution vector under the unanimity rule. The experimental results show that the approval mechanism is less efficient in three-player than in two-player voluntary contribution games. Furthermore, the unanimity rule is more effective at increasing contributions than the majority rule. Overall, our results suggest that the efficiency of the AM is mitigated in the case of three-player public goods games.

Abstract: We study experimentally a three player sequential and symmetric coalition formation game with empty core. In each round a randomly chosen proposer must choose between a two players coalition or a three players coalition and decide about the payoff division among the coalition members. Players who receive a proposition can accept or reject it. In case of acceptance the game ends. If it is rejected, a new proposer is randomly selected. The game was played repeatedly, with randomly rematched groups. We observe that over 86% of the realized coalitions are two-players coalitions. Three players coalitions are often observed in early rounds but are frequently rejected. Equal splits are the most frequently observed divisions among coalition members, and their frequency increases sharply over time. We propose an extension of von Neumann and Morgenstern (1944)’s notion of stable set to account for our results.

Abstract: Experimenting with dynamic games raises issues about implementing discounting in experiments. Theoretical rational decision-makers evaluate payoff streams by converting them to a reference period, often “time zero.” In experiments, subjects can adapt their strategy continuously. We explore individual behavior in a dynamic resource extraction experiment with two treatments: “z-discounting” (evaluating gains at time zero) and “p-discounting” (evaluating gains in present-time equivalents). Contrary to theoretical predictions, our data shows a significant positive treatment effect, indicating more substantial extraction under p-discounting. This challenges the theoretical model and prompts discussion on methodological considerations for discounting in laboratory settings.

Abstract: Conventional finance models assume that stock price magnitude should not influence portfolio choices or future returns. This view is contradicted, however, by empirical evidence. In this paper, we show that trading prices, in experimental markets, are processed differently by participants depending on their magnitude. Our experiment has two consecutive treatments. One where the fundamental value is a small number (small price market) and one where the fundamental value is a large number (large price market). Small price markets exhibit greater mispricing than large price markets. We obtain this result both between-participants and within-participants. Our findings indicate that price magnitude has a direct impact on how individuals perceive the distribution of future returns. This result is at odds with standard finance theory but is consistent with: (1) evidence in neuropsychology on the use of different mental scales for small and large numbers, and (2) empirical results in the finance and accounting literature.