I am a behavioral experimental economist, meaning that I use laboratory and field experiments to determine the empirical validity of relatively specific assumptions concerning the determinants, and to some extent the limits, of rational action. A list of my more recent and ongoing work can be found below, categorized by current state of advancement. The specific hypotheses vary with the project; however, several themes emerge.

Several of the papers, for instance, deal with the effects of real or experimentally induced group identification. In [1], experimentally induced “minimal groups” are used as lever to encourage sustainable exploitation of an intergenerational public good, while in [2] and [8] real social groups – as indicated in practice by participation in a complementary currency scheme in the French Basque country – are the focus, while increasing prosocial behavior (increasing trustworthiness and reducing negative reciprocity respectively) is the behavioral outcome. [3], by contrast, looks at somehow antisocial effects of group integration with real groups, measuring the effect of studying at a business school on the adoption of personal norms in favor of “selfish” behavior. Finally, [10] studies the effect of induced group association on the so-called democracy premium. It has previously been found that people cooperate with regulations more easily when they are involved in their formulation. What happens, this paper asks, when one arbitrary group is systematically excluded from the (otherwise) democratic process?

Another theme emerges in the effect of contextual effects that standard theory would not predict to influence behavior. For instance, [12] looks at the effect of the outcome of one uncertain event on later – similar but transparently independent – choices under ambiguity. [15] investigates parallel effects in a strategic setting, finding that a “presumption of truth” in a sender-receiver game with payoff uncertainty is strikingly influenced by private information in a previous encounter. Also related to contextual effects, with Raúl López-Pérez I have worked on a number of projects related to the classic preference-reversal phenomenon, in which the preference elicitation procedure appears to substantially influence revealed risk aversion. This venerable phenomenon tends to find that risk aversion is lower when subjects choose between pairs of binary gambles by giving reservation prices for each than when they make a direct choice between them. [6] and [7] carefully manipulate particular components of the procedure to disentangle what about the classic design generates these changes. We were surprised by the direction of the result in the former case, and the complete absence of a result in the latter.

A potentially interesting avenue to think about contextual effects involves attention and mental effort. Assumptions about such variables are different from much of the other work I’ve done recently, in that they do not exactly specify the nature of rational choice but rather its extent. The basic idea is that a variable is incorporated into the utility function only if one is aware of it. Context can, therefore, by changing awareness, also change the effective utility function. Study [6] and particularly [4] investigate attention mechanisms in preference reversals, the latter using eye-tracking technology. The behavioral effects of differential attention can also be seen in [9] and [3].

Finally, I have a long-standing interest in problems of ethical and unethical behavior. The choice to lie or not is a standard experimental lever for this kind of study, because a lie’s normative negative weight is relatively universally recognized, while remaining both minor enough not to generate significant ethical risk to participants, and easy to implement in the lab. Studies [3] and [14] study the determinants and consequences of dishonesty in different settings. It is relatively clear from the literature that some elements of ethical conformity belong in many people’s descriptive utility functions, but that there is significant variation across situations in the extent to which they bind. A particular interest of mine in this field concerns internal defense strategies against the negative utility from committing unethical acts, notably the idea of dissonance reduction/wishful thinking. [3] addresses this directly in the context of ethical choice, while [16] and [5] investigate related forms of motivated inference to develop more general intuitions.

The overall vision that can be discerned through this work is that economic behavior is not uniquely motivated by material incentives. Rather, people give some interpretation or meaning to the choices they have, and the importance of different actions’ meaning will weigh for or against payoff differentials in determining choice. The effects of irrelevant context and group identification found above exemplify this. This meaning-generation, or interpretive stage of the decision process may well require attention by the decision maker to the relevant dimensions of the choice problem, and may be susceptible to defensive mechanisms such as dissonance reduction strategies. Therefore, investigating all these factors represents an important movement towards a descriptively accurate and normatively satisfactory account of economic behavior.

List of projects

Published or accepted

1. Selfish now or altruistic (for) tomorrow? Inter- versus intra-generational strategies for the preservation of a common resource

With Ivan Ajdukovic and Angela Sutan

Environmental protection requires costly actions taken for other people’s benefit. Those actions are irrational for selfish individuals, but may not be for those who care for the well-being of people close to them, for instance in next generations. We compare two situations of exploitation of a common resource: a situation in which individuals are members of successive generations, without dynastic link (1), and a situation in which individuals of various generations are members of dynasties (2). Our results show that individuals who know that their extraction decision will impact members of their dynasty belonging to future generations make a significantly lower extraction than others.

2. Convertible local currency and trust: ‘It’s Not You, It’s Me’ – A field experiment in the French Basque Country

With Hayyan Alia

We present a field experiment investigating the mechanism by which community currencies enhance trust. Our question is the following: do I trust more when using a community currency because I am a trusting-type person or because I think that you are trustworthy? We call the former preference-based trust; while the latter is belief-based trust. We apply a modification of the standard trust game from the experimental economics literature to disentangle these mechanisms. Player A has to choose whether or not to trust player B, and player B can either reciprocate that trust or not. Our innovation is in experimentally separating the currency in which the game is played (effective currency), from the currency preferred by the participant (preferred currency). If the mechanism is preference-based, then preferred currency will determine trust more than effective; if it is belief-based, then the effective currency will be determinant. We find strong evidence of the preference-based mechanism of community currencies on trust, and only weak evidence of the belief-based mechanism.

3. Embracing the dark side? Testing the socialization of a maximizing mindset

Previous literature suggests an “economist effect”: business and economics students behave more like Homo economicus than do those from other disciplines. Does this represent selection of selfish maximizers into the disciplines, or a causal effect of study? We argue that common findings that behavior gets no “worse” over time represent overly simplistic tests. Our experiment investigates changes not just in behavior, but also in how participants feel about this behavior. Although we replicate the previous behavioral result, we find evidence that students learn to (1) attach less-negative normative weight to maximizing behavior, and (2) employ greater sophistication in its implementation.

Submitted/circulating

4. Using eye-tracking techniques to understand the role of attention on choice and reversals

In classical preference reversal (PR) experiments, participants often choose the safer of two gambles but express a higher monetary evaluation for the riskier. We use eye-tracking techniques to explore three manners in which attention may influence such behavior. First, we find that consistency is predicted by spending more time on Evaluation than Choice: the difficult task seems to require more attention to be performed without ‘mistakes’. Second, choice (but not relative evaluation) tracks relative attention across bets (but not across attributes of the bets): a higher focus on the safer bet during Choice predicts PRs, but the share of fixations on probabilities versus prizes has little influence. Third, people compare bets differently in the two tasks, making substantially more transitions between bets in Choice, but variation in this trait does not correlate with PRs. These results indicate behavioral impacts of attention and present intuitive explanations for PRs.

5. Flavors of regret: sour grapes and counterfactual information in a binary lottery choice experiment

This paper compares the implications of two mechanisms by which restricting counterfactual information about foregone outcomes may limit regret. One suggests simply that people do not pay attention to uncertain outcomes whose resolution is not revealed: out of sight, out of mind (OSOM). The other holds that, like the fox in the Aesop fable, they “sour” unresolved uncertainty in their minds. Our experiment allows an estimation of both the diminution of regret posited by OSOM, and the distortion in beliefs inherent in “sour grapes”. The results show a modest but significant effect for the latter, but no evidence of the former.

6. Nil evidence of timing and order effects in classical preference reversal experiments

In preference reversals, subjects express different orderings of two options depending on how preferences are elicited. In classical reversal tasks, the so-called “standard reversal” sees subjects select a safe bet over a risky one with a similar expected value in a straight choice, but give a higher monetary evaluation to the risky one. One interpretation of the reversals is that they are asymmetric “mistakes”, either due to imprecise preferences, insufficient adjustment from an anchor, contingent weighting, or unstable reference points. In this last account, reversals occur because deciders fail to think about how some bet B compares with the alternative when they evaluate B in isolation, which influences the way in which B is perceived. If we further assume that memory decay affects the likelihood that the alternative bet comes to mind, the account predicts order and timing effects on the reversal rate. We run several experiments where we vary the chances that the subjects’ considered reference points change between decisions, or more specifically, that they have in mind the alternative bet in a pair while evaluating bet B. To our surprise, the reversal rate shows no variation across treatments and we find no order or timing effects. It seems therefore that, at least in our design, reversals are not reduced if subjects are more likely to recall during the evaluations how bet B and its alternative compare.

7. An experiment where non-standard preference reversals become the norm

With Raúl López-Pérez

In preference reversals (PRs), subjects express different orderings of two options depending on how preferences are elicited. We report data from a PR experiment where the frequency of standard reversals, i.e., those most common in the literature, is almost nil, but non-standard ones become extremely frequent.

8. Complementary currencies and social capital: the case of negative reciprocity in the French Basque country

With Hayyan Alia

We present a lab-in-the-field experiment using a standard ultimatum game to investigate negative reciprocity between the members of a community that uses a local currency. Half of the participants played in euro while the others played in a local currency. We find a strong evidence of the impact of local currency on reducing the negative reciprocity of the subjects who are local currency users.

9. How to avoid errors in work environments: an exploratory investigation of anxiety

With Guillermo Mateu and Evelyn Agüero

The purpose of this study is to analyze the effect of cognitive pressure on performance of a simple task. Specifically, 284 participants complete a series of basic arithmetic operations in one of four conditions: baseline (limited time); time awareness (identical but with a salient clock indicating the time remaining); punishment (identical to baseline but with a penalty for errors); and time plus punishment (combining both previous treatments). Results are clear; treatment variables (time awareness and punishment) significantly increase the individual number of failures in a mathematical task. This study could help teachers and managers to better design work environments to avoid errors.

10. Preferences after pandemics: time and risk in the shadow of Covid-19

This paper investigates whether risk, time, and ambiguity preferences are affected by the covid-19 health crisis by analyzing individuals’ choices through an incentivized questionnaire before, during, and after the lockdown in France. Our results show that subjects’ time, risk, and amiguity preferences were affected during the lockdown. After the lock down, preferences tend to gradually return to thier initial level without reaching it 4 months later. Specifically, subjects showed a greater level of impatience, risk aversion, and ambiguity aversion from the lockdown. Thse results have implications for emergency policies aimed at managing the spread of the virus, as well as for economic policies aimed at supporting the population during the crisis.

In progress

11. Discrimination in the large and the small: how social exclusion affects sensitivity to notional fines

With Claude Fluet

This study investigates how exclusion from a decision-making process influences people’s respect for the decisions made. Subjects played a 10-round linear public goods game (LPG), in which mild law, in the sense of a non-deterrent penalty, was sometimes applied to uncontributed tokens. Fines were applied based on votes by the individuals in a round-specific matching group. A first dimension of exclusion, exclusion “in the large”, was implemented through the induction of near-minimal identities: half the participants in any session were assigned the red identity, while the rest were blue. Only red votes counted in the determination of whether a fine would be applied. In the ID condition, this was made salient, while it was not indicated in the No-ID condition. We define and measure effective exclusion as the difference between color groups in the No-ID condition. Symbolic exclusion, by contrast, is studied as the difference between the ID and No-ID conditions for each color group. We find no evidence that effective exclusion influences behavior in our experiment. However, symbolic exclusion – that is, the added information that exclusion was based on arbitrary group assignment – affects both excluded and “privileged” groups in different ways. The second dimension of exclusion, exclusion “in the small”, varied the number of participants in the matching group whose votes were counted. Based on the literature on so-called democracy premiums, we expected that when more people’s votes counted, the fines would be more effective. However, we do not see this pattern emerge in our data. Thus, while we do find interesting results on questions of discrimination, the democracy premium as such is not replicated in our study. We conclude with some suggestions as to why this might be the case.

12. Once bitten: memory of recent disappointment increases ambiguity aversion but not risk aversion

Since the financial crisis of 2008, an active literature has developed linking ambiguity to macroeconomic-level market performance. In part, it associates Keynes’ famous animal spirits (Keynes 1936) with behavioral biases such as ambiguity aversion. However, the nature of the mechanism is not clear. One proposal posits broadly that recessions or market downturns are periods of higher inherent market ambiguity. This drives ambiguity-averse traders out of the market, which can lead to a “freeze” where no trades occur. Several theoretical papers have developed models in which economic fluctuations were due to such ambiguity shocks. This paper experimentally investigates a potential corroborative factor, by which the result of one uncertain outcome may influence the propensity to take ambiguous risks later on. Such swings in ambiguity aversion due to previous experience would result in even ambiguity-neutral negative shocks possibly generating the kind of freeze described above, tipping participants into a more ambiguity averse behavioral mode. In the experiment, participants observe a series of binary bets drawn from an unknown distribution. Each period, they must either play the bet proposed, or pass it up for another draw from the same distribution. The choice is clearly related to choosing between options of first- and second-order uncertainty, and therefore participants more averse to second-order uncertainty – the ambiguity-averse – will be more likely to take the offered bet, conditional on its characteristics. The key finding is that losing a bet in one period results in a 7.8% increase in the probability of choosing a risky, versus ambiguous, prospect in the next period. This amounts to experimental evidence for the conjecture that bad news can increase ambiguity aversion later on. On the other hand, (1) a loss more than one period previous has no predictive power; and (2) previous losses do not seem to affect risk aversion, calculated as the impact of the probability of winning a bet on the probability of taking it.

13. Different measures: testing gender discrimination in wine service in France

With Jeremy Celse, Brit Grosskopf, Alex Karakostas, and Graeme Pearce

Gender discrimination is a widespread and economically harmful phenomenon, in addition to its inherent injustice. This study investigates a facet of systemic, institutionalized discrimination, in the form of the quantity of wine served in an ordered glass. Following a fixed script, pairs of testers (MM, MF, FF) went to wine bars in Dijon and Bordeaux, France, ordering two glasses of the same wine. The quantity of wine in each glass was measured by weighing the glasses before and after drinking the wine. Pre-registered analysis turns mainly on the interaction of the tester gender and quantity variables. Control variables included fixed effects for the bar and server, the gender of the server, the gender of the person ordering the wine, the price of the wine, and observable characteristics of servers and testers. Data collection currently delayed by Covid-19 related closure of wine bars.

14. Would you believe that!? Irrelevant information and equilibrium selection in a cheap talk game

This paper reports a cheap talk deception experiment in which the underlying game is random, either of common or conflicting interest. A first hypothesis is that when the uncertainty about the underlying game is more vague (specifically, when the message receiver does not know the probabilities of the game types), then the message will be more focal. While this hypothesis receives little empirical support, the realization of the game type itself has a strong effect on the Sender's strategy and beliefs. Common-interest realizations generate not just more truthful messages, but also greater expectation that the messages will be followed. This is puzzling, as it implies that the Sender expects Receiver behavior to depend on information the Receiver clearly does not know. More puzzling still, this effect is persistent: when the Sender plays the game again as a Receiver without receiving feedback, her beliefs about and reaction to the message received are strongly influenced by the (transparently irrelevant) realization of the previous interaction. The results suggest that the strategic requirements for a strategy to be focal may sometimes be quite weak. It also has implications in line with a literature on spillovers and behavioral contamination across repeated interactions.

15. Evolution of trust: testing direct against indirect evolutionary models in the lab

With Sigrid Suetens

In many models, evolution serves to destabilize inefficient equilibria through a “secret handshake" mechanism. The essential intuition is that a fully rational player cannot be relied upon to behave in the “right" way. A behavioral type that is "hard-wired" by evolution constrains behavior, so that the “carrier" cannot be tempted to engage in opportunistic actions. Together with the idea that the evolutionary forces are shaped by the payoff that "matters", cooperative types can gain an advantage over uncooperative. This is formalized in Guth, Dekel as models of indirect evolution of nonselfish preferences. We design an experiment to investigate the role of observability of preferences. In our context, players face three simultaneous games, so we can assess consistency of preferences across them. They also have a certain probability of seeing the move (strategy) of their partner beforehand. Classic evolutionary game theory (direct evolution) has different predictions in this context from indirect evolution. For instance, direct evolution predicts mixed strategy equilibria in these games that imply a particular rate of inconsistency. Indirect evolutionary models predict that rate to be near zero. Our results support the conclusion that individuals in the lab are doing something closer to application of a rule (“local preference”) to the problem than to the mixed strategy equilibrium of direct evolution.

16. Aesop’s conjecture: theory and experiments on motivated inference

With Raúl López-Pérez

This paper presents a behavioral model of motivated inference. The model posits that the rule used to form expectations about future events is unbiased, but may be applied to a biased subsample of past experience. We derive predictions for the extent of an agent’s motivated bias, compare our theory with previous ones, and provide new experimental data in line with our predictions.

More peripheral work

17. Impact of the WISE in France: a critical overview (submitted)

With Arvind Ashta, Guillaume Anais Camus, Mathieu-Claude Chaboud, and Roberto Hernan Gonzalez

Social Entrepreneurs solve social problems. One such social problem in Europe is persistent unemployment despite enormous State investments in education. Therefore, social entrepreneurs have introduced a solution called Work Integration Social Enterprises (WISE) that provide on-the-job training. Our research objective is to provide a critical overview of this sector in France. In a study of WISE cases in France over a nine-year period 2010-18, we look at data of over 1000 enterprises based on the annual reports of the industry federation. Overall, we find that the sector is growing slowly. We find that subsidies per person have increased, much of care is present in terms of beneficiaries per permanent staff, and more people are taking up short-term precarious jobs, rather than go back to unemployment. We find that the beneficiaries are predominantly and increasingly male. While they are also logically predominantly low-educated and long-term unemployed, the latter is decreasing in recent years. Impact data shows that over the nine-year period we considered, there is some reduction in those falling back into unemployment. But overall, France mirrors EU data as one third of program participants fall back into unemployment.

18. The Economic Psychology of Gabriel Tarde: something new for behavioral economics? (published)

With Ivan Ajdukovic and Sylvain Max

In the last years of the 19^th Century, Gabriel Tarde developed a theory of economic psychology based on interpersonal transmission of subjective values and beliefs. While his work has several aspects that resonate with behavioral economics as practiced today, its differences are profound. His interest in dynamics of change, and in particular in the role of innovation and invention in driving this process, is hard to fit into a behavioral economics framework. However, new empirical techniques leveraging social media and big data seem well suited to addressing his themes.