Abstract: The paper analyzes the theoretical effect of risk aversion on the demand for vaccines that reduce the severity of a disease but are also perceived as presenting a risk of side effects. Risk aversion is first shown to have ambiguous effect on the propensity to vaccinate. However, the model highlights how this relationship is influenced by the characteristics of the vaccine’s side effects. Namely, risk averse individuals are more prone to get vaccinated than risk neutral ones when the side effects potentially induced by the vaccine are relatively likely but not severe. The opposite result holds – i.e., the propensity of getting vaccinated is higher among risk-neutral individuals than among risk-averse ones – when side effects are more severe but less likely to occur. Finally, we show that the combination of the severity and probability of side effects that reverses the relationship between risk aversion and vaccination depends on the individual’s degree of absolute prudence, suggesting that this attitude toward risk is another driver of the demand for vaccination.
Co-auteur: Sarat Chandra Akella (PSE)
Abstract: After a decision under ambiguity or compound risk, the probabilities associated with the resolution of the final outcome may remain unknown forever or be revealed either before (ex-ante) or after (ex-post) the resolution. Using an integrated discrete-choice experiment, this paper investigates preferences for ex-ante and ex-post information about probabilities under ambiguity and compound risk, in addition to attitudes toward uncertainty in both domains. Our results reveal model-free behavioral patterns and also support an econometric implementation of source theory, capturing preferences along two key dimensions: elevation and likelihood sensitivity. We find demand for both ex-ante and ex-post information that is likelihood-independent and highly correlated both within and across domains, suggesting the presence of a stable behavioral trait. Moreover, the information premium is positively correlated with ambiguity attitudes, but not with compound risk attitudes. These findings offer new insights into the relationship between ambiguity aversion, compound risk aversion, and the demand for information.
Co-author: Pauline Madies (Sciences Po)
Abstract: Gender homophily, the tendency to associate with people of the same gender, affects hiring, mentoring, and labor market outcomes. We propose a new experiment to study gender homophily with endogenous team formation and performance feedback. Each participant chooses a male or female partner before performing an individual real-effort task with a shared payoff. We vary performance feedback across treatments to test whether it affects the gender of the chosen partner. We find no gender differences in performance or beliefs, but clear differences in homophily. Men show no homophily, whereas women more often choose female partners. This pattern is entirely driven by top-performing women. A simple conceptual framework shows how partner-specific altruism lowers the effort threshold, making female teams more attractive to high-performing women.
Abstract :We conducted a contextualized laboratory experiment to determine whether competition between wood buyers affects pro-environmental behaviour, that is, the adoption of new sus-
tainable biomass harvesting practices. We also tested the effect of a non-monetary incentive,that is signing a declaration that commits wood buyers who voluntarily sign it to act in a sustainable manner, on pro-environmental behaviour. Our results provide evidence that competition between wood buyers has an adverse effect on pro-environmental behaviour. Nevertheless, presenting a declaration to sign is effective in inducing wood buyers to act in a sustainable manner as it tends to correct unsustainable behaviour that emerges from competition.
Keywords : Forest; Timber harvest; Laboratory experiment; Non-monetary incentives; Commitment; Competition
Abstract : Reducing greenhouse gas emissions and lowering the risk of non-communicable diseases requires large-scale lifestyle changes, particularly in transport and food consumption. Yet behaviours such as car use or meat consumption are not merely individual choices: they are shaped by daily routines, structural constraints, and social context. This talk examines how high-impact lifestyle behaviours can be effectively shifted through three complementary pillars.
Auteurs : Pavlo Blavatskyy (Montpellier Business School), Elias Bouacida (Universite Paris 8), Jianying Qiu (IMR, Department of Economics, Radboud University) et Liu Shi (IMR, Department of Economics, Radboud University)
Abstract : This paper studies individuals' perceived stochasticity in their own future choices. We elicit subjects' beliefs about how they will decide in 26 risky lottery choice problems and subsequently observe their realized choices. To further characterize individual-level choice stochasticity, we repeat 15 of these problems six times for each subject.
We find that 89% of subjects report non-degenerate beliefs about their future choices, which arise in 69% of all decision problems. These beliefs are predictive of realized behavior and match the stochastic choice distributions revealed through repeated decisions, suggesting that individuals have meaningful insight into the probabilistic nature of their own future choices. Beliefs elicited as intended choices display systematic behavioral biases: they exhibit a pronounced common ratio effect and, to a lesser extent, a reverse common consequence effect. We provide evidence that these patterns largely reflect underlying preferences for such effects, although stochasticity remains a contributing factor (Loomes, 2005; McGranaghan et al., 2024). Finally, we show that the prevalence of non-degenerate beliefs cannot be accounted for by choice noise, indifference, or random preferences. While quasi-concavity in probabilities explains behavior for a subset of subjects, it does not account for the majority, highlighting the role of other mechanisms - most notably preference uncertainty - in shaping stochastic choice and beliefs about one's own decisions.
Abstract : Nudges like Home Energy Reports have been shown to reduce energy use, but their impact is highly heterogeneous. We provide one of the first experimental explanation of this heterogeneity through a year-long field experiment with 150 students living in 82 apartments, combining detailed heat consumption data with survey and experimental measures of pro-environmental attitudes and economic preferences.
We find that HERs reduce average heat consumption by around 7%, consistent with the upper range of existing estimates. Yet, households with stronger pro-environmental attitudes—already consuming less than their peers—reduced usage by nearly 25% more, while the highest-consuming households showed no reduction or even an increase. These findings suggest that social-comparison nudges may act as a form of “moral tax,” affecting individuals unevenly across the consumption distribution and raising important questions about fairness in the design of energy efficiency policies.
Co-authors: Ilke Aydogan and Loïc Berger
Abstract: Measuring ambiguity attitudes with monetary incentives is often logistically complex, costly, and time-consuming. We propose and validate a simplified, introspection-based method that retains the core strengths of state-of-the-art approaches while being suitable for survey implementation. The method involves a hypothetical binary choice between a risky and an ambiguous option, followed by an introspective assessment of the strength of preference. We assess its validity by comparing it to incentivized matching probability measures in two experiments: a laboratory study with university students and an online study with a representative sample of the French population. Our findings show that the introspection-based method reliably captures ambiguity aversion but falls short in measuring ambiguity-generated likelihood insensitivity. These results offer new evidence on the strengths and limitations of non-incentivized approaches to eliciting ambiguity preferences and highlight their practical value in empirical research beyond the lab.
co-écrit avec Emmanuelle Lavaine et Marc Willinger
Abstract: Social interactions frequently involve situations in which rational, self-interested individualsare unwilling to sacrifice their private benefits for the sake of collective welfare. In the absence of adequate incentives, such behavior often leads to inefficient outcomes, where the pursuit of individual goals undermines the common good, a situation widely known as a social dilemma.
Two classic examples illustrate this problem: the public goods dilemma and the common-pool resources dilemma. The former captures the difficulty of financing goods or servicesthat benefit everyone, such as public safety, infrastructure, or healthcare. Withoutcompulsion or incentives to contribute, individuals are naturally tempted to free ride on thecontributions of others, thereby jeopardizing the quality and sustainability of these collectivegoods. The latter reflects the tension surrounding the exploitation of shared resources, suchas fisheries, forests, or water basins, where open access and rivalry in consumption often lead to overexploitation, the so-called “tragedy of the commons,” threatening long-termsustainability.
To address these imbalances, economists have proposed a range of institutional andbehavioral mechanisms. Some rely on top-down regulation, imposed by external authorities,including quotas, property rights, or corrective taxes. Others take a bottom-up approach, based on voluntary cooperation and community governance, where users monitor and enforce rules themselves through arrangements such as communication, collective voting, or binding agreements. While these approaches have sometimes succeeded in restoring cooperation, their effectiveness varies depending on the institutional context and the specific nature of the social dilemma under consideration. Our research explores new incentive-based frameworks designed to enhance coordination and align individual actions with socially desirable outcomes. These mechanisms typically operate through a two-stage process: in the first stage, individuals make their initial decisions or proposals; in the second, they are invited to evaluate or respond to the choices of others before implementation. This structure promotes transparency, mutual accountability, and convergence toward more efficient collective solutions.
The overarching goal of this research is to assess the ability of such mechanisms to mitigate social dilemmas, particularly in the contexts of public goods and common-pool resources, and to identify the institutional and behavioral conditions that underpin their success. The analysis pays particular attention to decision-making parameters, such as group size and collective approval rules, to better understand how different coordination structures influence cooperative behavior.
Co-auteurs : Julien Jacob, Eve-Angéline Lambert, Emmanuel Peterle.
Résumé : "Artificial intelligence (AI) technologies are increasingly being integrated into healthcare practice, particularly through decision support systems that aim to improve physicians' diagnostic and therapeutic judgments. While these tools have the potential to enhance clinical accuracy and efficiency, they also introduce risks of moral hazard, particularly when AI systems provide an initial medical diagnosis that may unduly influence the physician's independent assessment. This dynamic highlights the need to establish a consistent legal framework to ensure that healthcare professionals remain liable for verifying and interpreting AI-generated information. In the medical field, several liability regimes, such as strict liability and loss of chance, govern professional liability. Strict liability is a legal regime in which a person can be held liable for damage, regardless of any proven fault, as long as there is a causal link between the activity performed and the damage suffered. The doctrine of loss of chance, meanwhile, allows patients or their relatives to hold a physician liable when his or her negligent inaction deprives the patient of a significant opportunity to avoid harm. This article examines how different rules of legal liability influence physicians' behavior when an AI-generated preliminary diagnosis is presented, and how the existence, absence, or nature of these rules influences the degree to which physicians rely on AI recommendations. In addition, the study examines the potential substitutability between AI assistance and physician effort in formulating a diagnosis. Using an experimental methodology, theoretical and empirical analyses are conducted to assess the impact of different liability regimes on healthcare professionals' decision-making processes and their perception of AI-assisted diagnosis. The objective is to elucidate how different legal contexts (namely strict liability, loss of chance, and no liability) influence physicians' behavior in the presence or absence of AI-based diagnostic tools."
Abstract : The presentation combines two experimental studies on intergenerational cooperation in common-pool resources. The first examines how intragenerational conflict —when several individuals must coordinate within a generation— undermines sustainability, even when future generations are clearly identified. The second tests how framing the future —as near or distant generations, close kin, or through self-projection— shapes extraction decisions. Together, these experiments provide new evidence on how social conflict and psychological distance influence people’s willingness to preserve resources for future generations.
Avec Francis Bloch (PSE) et Sudipta Sarangi (Virginia Tech)
Abstract: We investigate rumors through a lab experiment where subjects who prefer sharing truthful information, are exposed to biased agents attempting to promote a specific opinion, regardless of its veracity. In our experiment, message content is minimal (Heads or Tails), network structure and agent types are known, but the source of the message is not known. Participants must assess the veracity of the message before deciding to share or block it. We find that network structure impacts the diffusion of rumors. While the majority of our subjects make few errors, they are much better at using the network as a filter than at computing the credibility of a message on their own. We also uncover two fundamental features of human behavior with regard to transmitting messages: subjects tend to overshare but they also care about the identity of the receiver. Thus, subjects want to be heard but care about who is listening to them.
Abstract:
Perceptions of inequality are shaped by how different dimensions of inequality are portrayed and communicated across traditional media, digital platforms, and online conversations. Leveraging large language models (LLMs), we study the global trends in the salience and pupillarity of inequality and its dimensions as well as the narratives driving its portrayal, and the conversation from YouTube videos and its discussions across English-speaking countries. This approach allows us to trace how public discourse evolves over time, identify the key drivers of change, and examine how these dynamics respond to turning points triggered by political and business cycles. We examine the main drivers of such trends and study how such trends are affected by turning points resulting from political and business cycles. Finally, we show that when political and business recessions coincide, these conversations tend to amplify polarization, accelerating the adoption of more extreme preferences and underscoring the powerful role of discourse and dialogue in shaping public attitudes toward inequality.
Co-écrit avec François Cochard, Xavier Hollandts, Eli Spiegelman
Abstract: Despite the explicit integration of social goals into their governance, members’ adherence to the cooperative is not guaranteed. Some of them are reluctant: they participate, but lack intrinsic commitment; opportunistic behaviors then can trouble the collective project. External observers may not trust that cooperative principles are really endorsed by members. This paper studies these issues with laboratory and lab-in-the-field experiments. Based on interviews and pilots with cooperative members, we adapt a standard linear public goods game to isolate the basic principles of cooperatives in a menu of institutions. To see whether the effects are specific to cooperatives, we implement choice frames of a cooperative, a general business enterprise, and an unframed “group”. Because cooperatives are fundamentally social structures, and therefore likely rely on norms for their effectiveness, we also apply a Keynesian “beauty contest” procedure to test the extent to which expectations of others’ behavior influences cooperation. Our main results show that, when compared to other groups, individuals in a cooperative frame contribute more. Nevertheless, in the lab, cooperative members have no strict preferences for institutions meant to induce more cooperation. In the field, wine producers engage in effective redistribution public goods, while their business partners fail to anticipate it.
Abstract: "The experimental literature on bargaining has largely focused on identifying bargaining power and the intriguing role of equal-split solutions as reference points in diverse bargaining settings. In contrast, the theoretical literature has devoted comparatively less attention to reference points and the related concept of loss aversion. Typically, most theoretical papers concentrate on solutions that Pareto-dominate the disagreement point and are Pareto efficient; hence, the reference point naturally inherits those properties. In this paper, we seek to bridge the gap between the theoretical and empirical literatures by examining whether bargaining agreements are influenced by an experimentally induced reference point. We conducted a laboratory experiment with 166 participants involved in bilateral bargaining, where participants were first endowed with points earned through a real-effort task. The results show that the reference point is not always chosen, even when it is Pareto-efficient. Notably, in scenarios with asymmetric fallback positions, some degree of disagreement remains, and a minority of pairs continues to favor egalitarian outcomes. Our results also show the emergence of an Equal-change solution as an alternative to the standard Egalitarian and Nash solutions."
Co-auteurs: David Craininch, Wanda Mimra
Abstract: Industry gifts to experts tasked with making decisions on behalf of their clients are a common phenomenon. A notable example is the pharmaceutical industry's provision of gifts to physicians, such as conference invitations and complimentary meals. In this paper, we analyze the impact of such gifts on experts' diagnostic effort and product recommendation behavior. Using a simple model, we demonstrate that a sufficient condition for these gifts to reduce an expert’s propensity to seek diagnostic information is that the expert's valuation of their client’s welfare remains constant or decreases with reciprocity for gifts. Additionally, we propose an experimental framework to empirically test these theoretical predictions.
Arash Talebi (EDHEC), Sonja Prokopec (ESSEC), Ayse Onculer (ESSEC)
Abstract: This research explores the paradoxical effects of uncertain price promotions, focusing on how low-probability gains can influence evaluations and future choices. Our findings reveal that receiving a low-value reward in an uncertain promotion is more disappointing than receiving no promotion at all—a result that contradicts conventional utility-based predictions. This effect arises from the interplay of upward and downward counterfactual comparisons: individuals compare their realized outcome both to the unattained high-value prize and the baseline of no promotion. This dual reference-point mechanism challenge the assumption that any positive gain enhances consumer satisfaction and offer new insights into the emotional and cognitive biases that shape decision-making under uncertainty. Implications for designing more effective and psychologically attuned promotional strategies are discussed.
Co-auteurs : Noémi Berlin, Tarek Jaber-Lopez et Anne-Laure Levet
Abstract :
In a context of underconsumption of fruit and vegetables, particularly among disadvantaged populations, as well as rising poverty, food insecurity, and inflation, we are conducting an experimental study in France on the impact of financial incentives on the consumption of fruit and vegetables among low-income households. The aim is to guarantee universal access to healthy and sustainable food for all and thereby mitigate long-term health issues associated with poor diets.
This longitudinal study analyzes the effects of two distinct incentives (immediate or deferred) over a four-month period, divided into three phases: pre-intervention, intervention, and post-intervention. In addition to weekly purchase records, we conducted a qualitative phase designed to identify the drivers and barriers associated with each type of intervention.
This research enhances the understanding of financial incentive mechanisms regarding fruit and vegetable consumption and helps identify the optimal incentive measure for low-income households in France, considering individual preferences, socio-economic barriers, and perceptions related to autonomy and dignity.
Abstract: Tuition fees of business schools typically represent an important financial burden for students and their families. Could this financial strain impair the cognitive functions of students from more disadvantaged backgrounds, and in turn their academic performance? We address this question by analysing the impact of the introduction of means-tested tuition fees for first-year students at a French business school. Using cognitive measures, we find that the difference in the performance at a Raven’s matrices test between economically disadvantaged first-year students, who obtained a fees reduction, and their counterpart in later-years, who paid the full fees, is significantly improved (more positive) in favour of the former, compared to the difference between first and later-year students overall. We find no such effect on the other cognitive measures (i.e. Stroop task and Cognitive Reflection Test) nor on a synthetic measure based on the three cognitive tasks. At the intensive margin, the increase with poverty levels of the fees’ spread between beneficiaries and non-beneficiaries does not accentuate the positive effect on the performance to the Raven’s matrices test, but it does so on the synthetic measure. This suggests a more broad-based positive cognitive effect of the program for the most economically disadvantaged. At the academic level, using administrative data, we found no total effect of the reduced tuition fees on the average grade of students, but mediation analyses demonstrate a positive indirect effect through improved cognitive abilities, at both the extensive and the intensive margins.
Abstract : A persuader influences a decision-maker by providing a model for interpreting some upcoming news. The decision-maker adopts the model if it does not distort the marginal
distribution of news. Both parties can benefit if the persuader can provide news contingent, overall incoherent, models, privately learn the truth, or design the process of news arrival.
Coécrit avec Vassili Vergopoulos (Paris 2, Lemma).
Abstract: In this paper, we propose an axiomatic characterization of new allocation rules that take as input the marginal value of each agent in various possible worlds and output a potential division of the value produced in each world. We axiomatize a specific family of rules that integrates both marginalist and egalitarian principles. Finally, we show that this framework extends the cooperative game theory and we establish a connection with the Shapley value.
coécrit avec Yannick Gabuthy, Eve-Angéline Lambert, et Sébastien Massoni
Please wait ...