Search

We provide experimental evidence that core intertemporal choice anomalies – including extreme short-run impatience, structural estimates of present bias, hyperbolicity and transitivity violations – are driven by complexity rather than time or risk preferences. First, all anomalies also arise in structurally similar atemporal decision problems involving valuation of iteratively discounted (but immediately paid) rewards. These computational errors are strongly predictive of intertemporal decisions. Second, intertemporal choice anomalies are highly correlated with indices of complexity responses including cognitive uncertainty and choice inconsistency. We show that model misspecification resulting from ignoring behavioral responses to complexity severely inflates structural estimates of present bias.

The paper studies whether a drought in 1788 affected the outbreak of peasant revolts during the French Revolution. I construct a community-level data set with information on local drought severity and peasant uprisings in 1789. Results indicate that those more affected by the drought more often participated in peasant revolts against the feudal system. Then, I investigate a mechanism through which drought may have affected peasant revolts. I find that those more affected by the drought had higher demand for institutional change as expressed in the lists of grievances. The results provide evidence on specific ways in which the drought of 1788 impacted the French Revolution, a milestone in the democratization of Western Europe.

We consider the psychological and social foundations of human contributions and punishments in a voluntary contributions mechanism with punishment (VCMP). We eliminate 'dynamic economic linkages' between the two stages of our 'modified' VCMP to rule out other potential explanations. We use a beliefs-based model, rooted in psychological game theory, to derive rigorous theoretical predictions that are then tested with pre-registered experiments in China and the UK. Social norms, culture, and endogenous emotions are the key determinants of contributions and punishments. The emotions of shame, frustration, and anger, play a key role in our theoretical and empirical analysis through 'dynamic psychological linkages'. We provide potential microfoundations for the inherent human tendency to follow social norms and punish norm violators, while respecting boundedly rational strategic decision making.

This paper studies how litigation and settlement behavior is affected by agents motivated by spiteful preferences under the American and the English fee-shifting rule. We conduct an experiment and find that litigation expenditures and settlement requests are higher for more spiteful participants. The relative increase in litigation expenditures due to spite is more pronounced under the American fee-shifting rule. We further find that the expected payoff for more spiteful societies is lower than for less spiteful societies. This effect is particularly pronounced for low-merit cases under the English rule compared to a constant cost under the American rule.

We consider income-source-dependent tax evasion and show that this is a generalization of the well-known endowment effect. We show that loss aversion, moral costs, mental accounting, and risk preferences play a key role in explaining key features of source-dependent tax evasion. We provide evidence of the first direct link between subject-specific loss aversion and tax evasion, which is central to most successful modern theoretical accounts of tax evasion. We provide some evidence that risk aversion strengthens the cautionary effect of loss aversion and risk loving behavior attenuates, or reverses, it. However, the underlying effect is also influenced by the source of income. Evasion is increasing in the tax rate and decreasing in the audit penalty, as predicted. Our paper provides novel theoretical insights; proposes new methods in the estimation of the underlying behavioral parameters; and confirms the central predictions of the theory, while pointing out challenges for further developments that existing theory is unable to account for.

Descriptive norms – the behavior of other individuals in one's reference group – play a key role in shaping individual decisions. When characterizing the behavior of others, a standard approach in the literature is to focus on average behavior. In this paper, we argue both theoretically and empirically that not only averages but also the shape of the whole distribution of behavior can play a crucial role in how people react to descriptive norms. Using a representative sample of the U.S. population, we experimentally investigate how individuals react to strategic environments that are characterized by different distributions of behavior, focusing on the distinction between tight (i.e., characterized by low behavioral variance), loose (i.e., characterized by high behavioral variance), and polarized (i.e., characterized by u-shaped behavior) environments. We find that individuals indeed strongly respond to differences in the variance and shape of the descriptive norm they are facing: loose norms generate greater behavioral variance and polarization generates polarized responses. In polarized environments, most individuals prefer extreme actions – which expose them to considerable strategic risk – to intermediate actions that minimize such risk. Importantly, we also find that relative to tight environments, in polarized and loose environments, personal traits and values play a larger role in determining actual behavior. This provides important insights into how individuals navigate environments that contain strategic uncertainty.

We use a ten-round online mind game to determine whether the effect of honesty nudges depends on timing and content. Reminding individuals about the right thing to do increases honesty. Including information that it is possible to assess an individual's dishonesty strengthens the effect of the intervention. Both types of intervention are similarly effective when they take place before an individual has made any decision or after individuals have played five rounds of the mind game. Nudging an individual after they have made five decisions allows us to add personalized information based on the individual's previous response; however, this does not increase honesty. Examining the reaction to nudges based on previous behavior shows that (presumably) honest and dishonest individuals respond by reducing overreporting. The effect of the different nudge content is driven by those previously dishonest.

Economists and management scholars have argued that the scope of incentives to increase cooperation in organizations is limited as their use signals the prevalence of free-riding among employees. This paper tests this hypothesis experimentally, using a sample of managers and employees from a large company. We exogenously vary whether managers are informed about prevailing cooperation levels among employees before they can set incentives to promote cooperation. In addition, employees matched to informed managers learn that the manager could base their incentive choice on cooperation levels. We find no evidence for the hypothesized signaling effect. Having an informed manager set the incentive does not change employees' be-liefs about the cooperativeness of others. Incentives hence have strong positive effects on cooperative beliefs, irrespective of information. The absence of the signaling effect seems related to the perception of managers' intentions, a mitigating but understudied factor.

We examine how people redistribute income when there is uncertainty about the role luck plays in determining opportunities and outcomes. We elicit redistribution decisions from a U.S.-representative sample who observe worker outcomes and whether luck magnified workers' effort (``lucky opportunities'') or determined workers' income directly (``lucky outcomes''). We find that participants redistribute less and are less reactive to changes in the importance of luck in environments with lucky opportunities. Our findings have implications for models that seek to understand and predict redistribution attitudes and help to explain the gap between lab evidence on support for redistribution and U.S. inequality trends., Comment: JEL codes: C91, D63

Personal autonomy has been argued to be fundamental to well-being and is often discussed as an important driver of economic and political behavior. Yet, preferences for autonomy are not well understood, because their identification requires the separation of instrumental value attached to autonomous choice. We propose a novel elicitation method that solves this identification challenge. We establish the existence of intrinsic preferences for choice autonomy and show substantial heterogeneity in a large online sample. We further study their antecedents by relating them to existing personality scales and socioeconomic characteristics. Finally, we test their association with other preferences, attitudes and beliefs.

In an experiment that elicits subjects' willingness to pay (WTP) for the outcome of a lottery, we confirm the fourfold pattern of risk attitudes described by Kahneman and Tversky. In addition, we document a systematic effect of stake sizes on the magnitude and sign of the relative risk premium, holding fixed both the probability that a lottery pays off and the sign of its payoff (gain vs. loss). We further show that in our data, there is a log-linear relationship between the monetary payoff of the lottery and WTP, conditional on the probability of the payoff and its sign. We account quantitatively for this relationship, and the way in which it varies with both the probability and sign of the lottery payoff, in a model in which all departures from risk-neutral bidding are attributed to an optimal adaptation of bidding behaviour to the presence of cognitive noise. Moreover, the cognitive noise required by our hypothesis is consistent with patterns of bias and variability in judgments about numerical magnitudes and probabilities that have been observed in other contexts.

Many occupations and industries are highly segregated with respect to gender. This segregation could be due to perceived job-specific productivity differences between men and women. It could also result from the belief that single-gender teams perform better. We investigate the two explanations in a lab experiment with students and in an online experiment with personnel managers. The subjects bet on the productivity of teams of different gender compositions in tasks that differ with respect to gender stereotypes. We obtain similar results in both samples. Women are picked more often for the stereotypically female task and men more often for the stereotypically male task. Subjects do not believe that homogeneous teams perform better but bet more on diverse teams, especially in the task with complementarities. Elicited expectations about the bets of others reveal that subjects expect the effect of the gender stereotypes of tasks but underestimate others’ bets on diversity.

Do groups and individuals behave differently in dictator games with varying deservingness of the recipient? Does the involvement in group-decision making affect the decisions of group members in subsequent individual decisions? We address these questions using a controlled dictator-game experiment and find the following main results. First, groups and individuals are not different w.r.t. their dictator-game decisions and recipient deservingness does not have a different effect on groups than on individuals. Second, participants who were previously part of a group decision process are more generous in a subsequent individual-level decision than participants who previously made individual decisions. We exploit the chat protocols of group discussions to shed light on the mechanism behind this result. Consistent with moral balancing, we show that the effect of group-decision making on subsequent individual decisions is driven by subjects who intent to make good for the initial group decision.

Debates about affirmative action often revolve around fairness. Accordingly, we document substantial heterogeneity in the fairness perception of various affirmative action policies. But do these differences translate into different consequences? In a laboratory experiment, we study three different quota rules in tournaments that favor individuals whose performance is low, either due to discrimination, low productivity, or choice of a short working time. Affirmative action favoring discriminated individuals is perceived as fairest, followed by that targeting individuals with a short working time, while favoring low productivity individuals is not perceived as fairer than an absence of affirmative action. Higher fairness perceptions coincide with a higher willingness to compete and less retaliation against winners, underlining that fairness perceptions matter for the consequences of affirmative action. No policy harms overall productivity or post-competition teamwork, but affirmative action may reduce the average output of tournament winners.

Algorithm-based decision support systems play an increasingly important role in decisions involving exploration tasks, such as product searches, portfolio choices, and human resource procurement. These tasks often involve a trade-off between exploration and exploitation, which can be highly dependent on individual preferences. In an online experiment, we study whether the willingness of participants to follow the advice of a reinforcement learning algorithm depends on the fit between their own exploration preferences and the algorithm's advice. We vary the weight that the algorithm places on exploration rather than exploitation, and model the participants' decision-making processes using a learning model comparable to the algorithm's. This allows us to measure the degree to which one's willingness to accept the algorithm's advice depends on the weight it places on exploration and on the similarity between the exploration tendencies of the algorithm and the participant. We find that the algorithm's advice affects and improves participants' choices in all treatments. However, the degree to which participants are willing to follow the advice depends heavily on the algorithm's exploration tendency. Participants are more likely to follow an algorithm that is more exploitative than they are, possibly interpreting the algorithm's relative consistency over time as a signal of expertise. Similarity between human choices and the algorithm's recommendations does not increase humans' willingness to follow the recommendations. Hence, our results suggest that the consistency of an algorithm's recommendations over time is key to inducing people to follow algorithmic advice in exploration tasks.

We use the trust and the dictator games to explore the effects of religious identity on trust, trustworthiness, prosociality, and conditional reciprocity within a beliefs-based model. We provide a novel and rigorous theoretical model to derive the relevant predictions, which are then tested in pre-registered lab-in-the-field experiments from villages in the Indian states of Bihar and Uttar Pradesh. We find strong evidence of religious identity effects in the beliefs, and the chosen actions, for Hindu and Muslim subjects. Priming has little effect on Hindu subjects but it enhances religious polarization in beliefs and actions among Muslim subjects. There is taste-based discrimination but no statistical discrimination. All our underlying assumptions on beliefs, and their dependence on priming and identity are confirmed by the data, identifying a precise beliefs-based mechanism for the effects of religious identity. More religious subjects expect greater prosociality/reciprocity and often are more prosocial/reciprocal.

Competitive rewards are often assigned on a regular basis, e.g., in annual salary negotiations or employee-of-the-month schemes. The repetition of competitions can imply that opponents are matched based on earlier outcomes. Using a real-effort experiment, we examine how cheating and effort evolve in two rounds of competitions in which subjects compete with different types of opponents in the second round (random/based on first-round outcome). We find that (i) losing causes competitors to increase cheating in the second round while winning implies a tendency to reduce cheating. A similar effect is found with regard to effort, which losers increase to a larger extent than winners. (ii) Competitor matching does not significantly affect behavior.

Does being lucky (or unlucky) affect honest decision-making? We examine (1) whether luck-based income strengthens or erodes the moral value of honesty; (2) whether the perceived level of agency over an uncertain event affects the relationship between luck and honesty; and (3) whether accumulated luck affects honesty. To this end, we conducted a lab experiment where participants self-report a dice roll outcome, which is associated with effort-based income, after having received luck-based income. We manipulated the participants' perceptions regarding their influence on luck-based income. In the exogenous luck treatment, computerized coin tosses determines the luck-based income, whereas in the endogenous luck treatment, the participants choose the coin's winning side before the computerized coin toss. Our results are as follows: (1) lying behaviour increases when contemporaneous luck-based income is high, (2) lying is not affected by the perceived level of agency, and (3) lying is not affected by the previous outcomes of the luck-based income. Our observations challenge the relative importance of context that may render moral justification. Therefore, our findings indicate that differences in dishonest behavior may be largely due to individual-specific heterogeneity.

This study investigates the influence of gender composition on allocation decisions involving a rank–inequality tradeoff. In a laboratory experiment, participants chose to either alleviate inequality by relinquishing their current relative rank or exacerbate inequality by maintaining their current rank. Two essential features of the experiment are: 1) participants' relative rank is the outcome of their real-effort performance and luck; 2) participants' genders are naturally revealed by gender-specific nicknames. We found that female participants are more reluctant to relinquish their current relative rank when the persons ranked below and above them are of the opposite gender. This tendency was less pronounced in the male participants.

An extensive literature documents that people are willing to sacrifice personal material gain to adhere to a moral motive. Yet, less is known about what happens when moral motives are in conflict. We hypothesize that individuals engage in what we term "motive selection," namely adhering to the moral motive that aligns with their self-interest. We test this hypothesis using a laboratory experiment that induces a conflict between two of the most-studied moral motives: fairness and truth-telling. In line with our hypothesis, our results show that individuals prefer to adhere to the moral motive that is more aligned with their self-interest.

We consider infinite-horizon bargaining in which an uninformed seller sequentially makes a price offer to a privately informed buyer who decides whether to accept or reject it in every bargaining round. Existing theories suggest that the presence (absence) of an arbitrarily small outside option available to the buyer in this dynamic screening problem leads the seller (buyer) to enjoy a substantial surplus, and we examine the validity of the differences in the share of the surplus theoretically and experimentally. We first show that the theoretical differences collapse if an arbitrarily small fraction of optimistic buyers would believe that the sellers sometimes ask for a lower price in the subsequent rounds. We then present experimental evidence that the earnings of both the buyers and the sellers when the buyer has an outside option are not significantly different from those without the outside option. We find supporting evidence that some buyers reject the current-round offers, optimistically believing that the next offer would be more favorable to them.

This paper reports results of two controlled experiments on the behavioural effects of relative performance information (RPI) in different organizational structures. Our baseline study 1 focuses on a centralized organizational structure where employees are exogenously assigned to either a high-performing or a low-performing peer group. We find that RPI boosts performances when employees are assigned to the low-performing group. In contrast, when assigned to the high-performing group, our results point to a discouragement effect of RPI that can be attributed to low-performers. In study 2, we show that this or similarly undesired effects do not play a crucial role under a decentralized organizational structure where employees can self-select. In fact, we demonstrate that RPI especially induces employees with a relatively low performance to voluntarily choose the high-performing group. Analyzing subsequent performances suggests that providing self-selection options allows employees to use the high-performing group as a self-set target to spur motivation.

We investigate in a laboratory setting whether revealing information on intelligence affects behavior in games with repeated interactions. In our experimental design we communicate information on the cognitive ability of both players. We use three stage games: Prisoners’ Dilemma (PD) and two versions of Battle of Sexes (BoS), with high and low payoff inequality. We find that the information affects strategic behavior significantly in two distinct ways. In PD, disclosure markedly hampers cooperation, as higher intelligence players are less cooperative once they are made aware that they play against someone of lower ability than themselves in the disclosure treatment. Similarly, in BoS with low payoff inequality, disclosure disrupts coordination on outcomes with positive payoffs, as higher intelligence players try to force their most preferred outcome onto the less intelligent. However, in BoS with high payoff inequality, this pattern of behavior dramatically changes. Disclosure does not significantly affect coordination rates. Differently from the low payoff inequality game, coordination is achieved more often on outcomes that favour less intelligent players. We conjecture that when coordination becomes more difficult, because of the high inequality between payoffs, intelligence and inequality together form a coordination device.

We analyze linear, weakest-link and best-shot public goods games in which a distinguished team member, the team allocator, has property rights over the benefits from the public good and can distribute them among team members. These team allocator games are intended to capture natural asymmetries in hierarchical teams facing social dilemmas, such as those that exist in work teams. Our results show that the introduction of a team allocator leads to pronounced cooperation in both linear and best-shot public-good games, while it has no effect in the weakest-link public good. The team allocator uses her allocation power to distribute benefits from the public good in a way that motivates people to contribute. Re-allocating team payoffs allows the team allocator to reward cooperating team members and to sanction non-cooperating members at no efficiency losses from explicit sanctioning costs. As a result, team profits are higher in the linear team allocator game but not in the best-shot case, where the lack of coordination leads to a welfare decrease for the remaining team members.

There is abundant evidence on individual preferences for policies that reduce national inequality, but only little evidence on preferences for policies addressing global inequality. To investigate the latter, we conduct a two-year, face-to-face survey experiment on a representative sample of Germans. We measure how individuals form perceptions of their ranks in the national and global income distributions, and how those perceptions relate to their national and global policy preferences. We find that Germans systematically underestimate their true place in the world���s income distribution, but that correcting those misperceptions does not affect their support for policies related to global inequality.

We report experimental evidence on the voluntary provision of public goods under threshold uncertainty. By explicitly comparing two prominent technologies, summation and weakest link, we show that uncertainty is particularly detrimental to threshold attainment under weakest link, where low contributions by one subject cannot be compensated by others. In contrast, threshold uncertainty does not affect contributions under summation. We demonstrate non-binding pledges as one mechanism to improve chances of threshold attainment under both technologies, yet in particular under weakest link.

The impact of culture on non-kin cooperation has been singled out as critical for economic activity. However, causal evidence of culture’s influence on cooperation remains scant. In this paper we provide such evidence, focusing on two key components of culture: preferences and beliefs. Adopting the view that culture is one aspect of an individual’s multi-faceted self-concept (identity) we conduct an experiment with foreign- and US-born Chinese immigrants at a large US public university. In a two-by-two design, we exogenously vary: i) the salience of participants’ American or Chinese cultural identities; and ii) the capacity for culture to affect beliefs by randomly providing previous-session cooperation-rate information. Comparing behavior across cultures and information conditions, our results suggest a prominent role for both preferences and beliefs. In particular, we find that culture’s effects through beliefs are as important as its effects through preferences.

Overbidding in sealed-bid second-price auctions (SPAs) has been shown to be persistent and associated with cognitive ability. We study experimentally to what extent cross-game learning can reduce overbidding in SPAs, taking into account cognitive skills. Employing an order-balanced design, we use first-price auctions (FPAs) to expose participants to an auction format in which losses from high bids are more salient than in SPAs. Experience in FPAs causes substantial cross-game learning for cognitively less able participants but does not affect overbidding for the cognitively more able. Vice versa, experiencing SPAs before bidding in an FPA does not substantially affect bidding behavior by the cognitively less able but, somewhat surprisingly, reduces bid shading by cognitively more able participants, resulting in lower profits in FPAs. Thus, 'cross-game learning' may rather be understood as 'cross-game transfer', as it has the potential to benefit bidders with lower cognitive ability whereas it has little or even adverse effects for higher-ability bidders.

Research examining the effect of weak punishment on conformity indicates that punishment can backfire and lead to suboptimal social outcomes. We examine whether this effect is due to a lack of perceived legitimacy of rule enforcement, which would enable agents to justify selfish behavior. We address the question of legitimacy by shedding light upon the importance of social norms and their interplay with weak punishment in the context of a trust game. Across six conditions, we systematically vary the combination of existence of weak punishment and norm information. Norm information may refer either to what most others do (empirical) or to what most others deem appropriate (normative). We show that in isolation, neither weak punishment nor empirical/normative information increase prosocial, reciprocal behavior. We instead find that reciprocity significantly increases when normative information and weak punishment are combined, but only when compliance is relatively cheap. When compliance is more costly, we find that the combination of punishment and generic empirical information about others’ conformity can have detrimental effects. In additional experiments, we show that this negative effect can be attributed to the punishment being perceived as unjustified, at least in some individuals. Our results have important implications for researchers and practitioners alike.

Under weak contract enforcement the trading parties’ trust, defined as their belief in other’s trustworthiness, appears important for realizing gains from trade. In contrast, under strong contract enforcement beliefs about other’s trustworthiness appear less important, suggesting that trust and contract enforcement are substitutes. Here we show, however, that trust and contract enforcement are complements. We demonstrate that under weak contract enforcement trust has no effect on gains from trade, but when we successively improve contract enforcement, larger effects of trust emerge. Likewise, improvements in contract enforcement generate no increases in gains from trade under low initial trust, but cause high increases when initial trust is high. Thus, the effect of improvements in contract enforcement is trust-dependent, and the effect of increases in trust is dependent on the strength of contract enforcement. We identify three key ingredients underlying this complementarity: (1) heterogeneity in trustworthiness; (2) strength of contract enforcement affecting the ability to elicit reciprocal behavior from trustworthy types, and screen out untrustworthy types; (3) trust beliefs determining willingness to try such strategies.

Our results from a laboratory experiment offer new evidence for the detrimental effects that cheating behaviour in the workplace may have on the degree of reciprocity between firms and workers. First, we replicate existing findings showing that in the absence of monitoring (cheating is possible) workers cheat on their actual performance in a real-effort task. The extent of cheating influences how firms (managers) decide to set their wages in a subsequent gift-exchange game. Specifically, firms offer higher wages to workers who cheat and interestingly, workers expect such behaviour by firms. These higher wages are not matched by workers’ performance in the gift-exchange game, where cheating is not possible, resulting in a flat wage-effort relationship. In contrast, in the presence of monitoring (cheating is not possible), we find a positive relationship between wages offered and effort provided by the workers. Our findings have implications for adopting measures at the workplace that eliminate workers’ opportunities to cheat on their performance.

Many committees—juries, political task forces, etc.—spend time gathering costly information before reaching a decision. We report results from lab experiments focused on such information-collection processes. We consider decisions governed by individuals and groups and compare how voting rules affect outcomes. We also contrast static information collection, as in classical hypothesis testing, with dynamic collection, as in sequential hypothesis testing. Several insights emerge. Static information collection is excessive, and sequential information collection is non-stationary, producing declining decision accuracies over time. Furthermore, groups using majority rule yield especially hasty and inaccurate decisions. Nonetheless, sequential information collection is welfare enhancing relative to static collection, particularly when unanimous rules are used.

Perceived urgency and regret are common in many sequential search processes; for example, sellers often pressure buyers in search of the best offer, both time-wise and in terms of potential regret of forgoing unique purchasing opportunities. theoretically, these strategies result in anticipated and experienced regret, which systematically affect search behavior and thereby distort optimal search. In addition, urgency may alter decision-making processes and thereby the salience of regret. To understand the empirical relevance of these aspects, we study the causal effects of regret, urgency, and their interaction on search behavior in a pre-registered, theory-based, and well-powered experiment. Empirically, we and that anticipated regret does not affect search behavior either with or without time pressure, while experienced regret leads to systematic adjustments in search length. Urgency reduces decision times and perceived decision quality, but does not generally alter search length. Only very inexperienced decision-makers buy earlier when pressured. Thus, consumer protection measures against pressure selling tactics can help inexperienced consumers in particular.

In cases of conflict of interest, people can lie directly about payoff relevant private information, or they can evade the truth without lying directly. We analyse this situation theoretically and test the key predictions in an experimental sender-receiver setting. We find senders prefer to deceive through evasion rather than direct lying. This is because they do nοt want to deceive others, and they do nοt want to be seen as deceptive. The specific language of evasion does not matter. The results suggest deception should be tested in more naturalistic contexts with richer language.

Expert advice is often biased in ways that benefit the advisor. We demonstrate how self-deception helps advisors be biased while preserving their self-image as ethical and identify limits to advisors' ability to self-deceive. In experiments where advisors recommend one of two investments to a client and receive a commission that depends on their recommendation, we vary the timing at which advisors learn about their own incentives. When advisors learn about their incentives before evaluating the available investments, they are more likely to be biased than when they learn about their incentives only after privately evaluating the investments. Consistent with self-deception, learning about the incentive before evaluating the options affects advisors' beliefs and preferences over the investments. Biased advice persists with minimal justifications but is eliminated when all justifications are removed. These findings show how self-deception can be constrained to improve advice provision.

The evidence shows source-dependent entitlement to income sources and individuals are reluctant to part with income they feel more entitled to, e.g., earned labor income. Taxpayers may also be more reluctant to part with tax payments (evade more) from income sources they feel more entitled to- a form of mental accounting. We embed two main hypotheses within a rigorous theoretical model based on prospect theory. From incomes sources they feel more entitled to, taxpayers experience (i) greater loss aversion from paying taxes, and (ii) lower moral costs of evasion. We confirm the predictions of our model through MTurk experiments. Evasion is increasing in the tax rate and decreasing in the audit penalty. Moral costs influence taxpayers' decisions. Loss aversion, measured "directly" for the first time for each individual in an evasion experiment, reduces evasion, as predicted by our theory. Loss aversion, risk aversion, and their interaction, are critical determinants of evasion.

This paper uses a novel experimental design to study the contagion of pro- and antisocial behavior and the role of social proximity among peers. Across systematic variations thereof, we find that anti-social behavior is generally more contagious than pro-social behavior. Surprisingly, we also find that social proximity amplifies the contagion of anti-social behavior more strongly than the contagion of pro-social behavior, and that anti-social individuals are most susceptible to behavioral contagion of other anti-social peers. These findings paired with the methodological contribution are informative for the design of effective norm-based policy interventions directed at facilitating (reducing) pro- (anti-)social behavior in social and economic environments.

Promises are prevalent in many competitive environments, but promise keeping is often difficult to observe. Do promises still offer an opportunity to honor future obligations, if promise keeping is unobservable? Focusing on campaign promises, we study the value of transparency. We show how preferences for truth-telling shape promise competition when promise keeping can(not) be observed. We identify the causal effects of transparency in a laboratory experiment. Transparency leads to less generous promises, but also to less promise breaking. Nonetheless, officials appropriate similar rents as in opaque institutions. Preferences for truth-telling and (instrumental) reputational concerns explain these results.

A large literature in behavioral economics has emphasized in the last decades the role of individual differences in social preferences (such as trust and altruism) and in influencing behavior in strategic environments. Here we emphasize the role of attention and working memory, and show that social interactions among heterogeneous groups are likely to be mediated by differences in cognitive skills. Our design uses a Repeated Prisoner's Dilemma, and we compare rates of cooperation in groups of subjects grouped according to their IQ, with those in combined groups. While in combined groups we observe higher cooperation rates and profits than in separated groups (with consistent gains among lower IQ subjects and relatively smaller losses for higher IQ subjects), higher IQ subjects become less lenient when they are matched with lower IQ subjects than when they play separately. We argue that this is an instance of a general phenomenon, which we demonstrate in an evolutionary game theory model, where higher IQ among subjects determines – through better working memory – a lower frequency of errors in strategy implementation. In our data, we show that players indeed choose less lenient strategies in environments where subjects have higher error rates. The estimations of errors and strategies from the experimental data are consistent with the hypothesis and the predictions of the model.

Do labor market concerns affect support for immigration? Using a large, representative sample of the US population, we first elicit beliefs about the labor market impact of immigration. To generate exogenous variation in beliefs, we then provide respondents in the treatment group with research evidence showing no adverse labor market impacts of immigration. Treated respondents update their beliefs and become more supportive of immigration, as measured by self-reported policy views and petition signatures. Treatment effects also persist in an obfuscated follow-up study. Our results demonstrate that information about the labor market impact of immigration causally affects support for immigration.

Survey measures of the reservation wage reflect both the consumption-leisure trade-off and job search concerns (the arrival rate of job offers and the wage distribution). We examine what a survey measure of the reservation wage reveals about labor supply when search concerns are absent. To this end, we combine the reservation wage measure from a large labor market survey with the reservation wage for a one-hour job that we elicit in an online experiment. The two measures show a strong positive association. For unemployed individuals, the experimental reservation wage increases on average by around one Euro for every Euro increase in the survey measure. For employed individuals, the association between the two measures is weaker, but still positive and statistically significant. We show that these results are robust to selection into the experiment, and that demographic variables have a similar influence on both reservation wage measures.

Both theory and recent empirical evidence on nudging suggest that observability of behavior acts as an instrument for promoting (discouraging) pro-social (anti-social) behavior. We connect three streams of literature (nudging, social preferences, and social norms) to investigate the universality of these claims. By employing a series of high-powered laboratory and online studies, we report here on an investigation of the questions of when and in what form backfiring occurs, the mechanism behind the backfiring, and how to mitigate it. We find that inequality aversion moderates the effectiveness of such nudges and that increasing the focus on social norms can counteract the backfiring effects of such behavioral interventions. Our results are informative for those who work on nudging and behavioral change, including scholars, company officials, and policy-makers.

We examine whether belief-based preferences - caring about what transgressors believe - play a crucial role in punishment decisions: Do punishers want to make sure that transgressors understand why they are being punished, and is this desire to affect beliefs often prioritized over distributive and retributive preferences? We test whether punishers derive utility from three distinct sources: material outcomes (their own and the transgressor's payoff), affective states (the transgressor's suffering), and cognitive states (the transgressor's beliefs about the cause of that suffering). In a novel, preregistered experiment (N = 1,959) we demonstrate that consideration for transgressors' beliefs affects punishment decisions on its own, regardless of the considerations for material outcomes (distributional preferences) and affective states (retributive preferences). By contrast, we find very little evidence for pure retributive preferences (i.e., to merely inflict suffering on transgressors). We also show that people who would otherwise enact harsh punishments, are willing to punish less severely, if by doing so they can tell the transgressor why they are punishing them. Finally, we demonstrate that the preference for affecting transgressors' beliefs cannot be explained by deterrence motives (i.e., to make transgressors behave better in the future).

We experimentally investigate whether individuals strategically distort their beliefs about dominant norms. Embedded in the context of lying, we systematically vary both the nature of elicited beliefs (descriptive about what others do, or normative about what others approve of) and whether subjects are aware of the forthcoming lying opportunity at the belief-formation stage. We build a dual-self model of belief distortion applied to the context of social norms and derive a number of precise predictions. Our findings provide a perspective on why, when and which norm-relevant beliefs are strategically distorted and show that not all belief distortions are created equal.

Do people anticipate the conditions that enable them to manipulate their beliefs when confronted with unpleasant information? We investigate whether individuals seek out the "cognitive flexibility" needed to distort beliefs in self-serving ways, or instead attempt to constrain it, committing to unbiased judgment. Experiments with 6500 participants, including financial and legal professionals, show that preferences are heterogeneous: over 40% of advisors prefer flexibility, even if costly. Actively seeking flexibility does not preclude belief distortion. Individuals anticipate the effects of cognitive flexibility and their choice to pursue it responds to incentives, suggesting some sophistication about the cognitive constraints to belief distortion.

We generalize the Rubinstein (1982) bargaining model by disentangling payoff delay from bargaining delay. We show that our extension is isomorphic to generalized discounting with dynamic consistency and characterize the unique equilibrium. Using a novel experimental design to control for various confounds, we then test comparative statics predictions with respect to time discounting. All bargaining takes place within a single experimental session, so bargaining delay is negligible and dynamic consistency holds by design, while payoff delay per disagreement round is significant and randomized transparently at the individual level (week/month, with/without front-end delay). In contrast to prior experiments, we obtain strong behavioral support for the basic predictions that hold regardless of the details of discounting. Testing differential predictions of different forms of discounting, we strongly reject exponential discounting in favor of present-biased discounting.

We consider an expanded notion of social norms that renders them belief-dependent and partial, formulate a series of related testable predictions, and design an experiment based on a variant of the dictator game that tests for empirical relevance. Main results: Normative beliefs influence generosity, as predicted. Degree of partiality leads to more dispersion in giving behavior, as predicted.

In this paper we use an experiment to compare a theory of risk aversion and a theory of spite as an explanation for overbidding in auctions. As a workhorse we use the second-price all-pay and the first-price winner-pay auction. Both risk and spite can be used to rationalize deviations from risk neutral equilibrium bids in auctions. We exploit that equilibrium predictions in the second-price all-pay auctions for spiteful preferences are different than those for risk averse preferences. Indeed, we find that spite is a more convincing explanation for bidding behavior for the second-price all-pay auction. Not only can spite rationalize observed bids, also our measure for spite is consistent with observed bids.

We study the role of face-to-face interaction for gender differences in deceptive behavior and perceived honesty. In the first part, we compare women's to men's deceptive behavior using data from an incentivized income reporting experiment in which lies can be detected in the course of an audit. Between the three treatments of that experiment, (i) the degree, and (ii) the impact of the face-to-face interaction vary from none (computerized baseline treatment) to a little (treatment in which face-to-face communication triggers psychological effects such as greater lying aversion) to much (treatment in which the perception by others also enters as a strategic effect determining the probability of detecting a lie). In the computerized baseline treatment men and women lie alike. Women's truthfulness increases when psychological effects of face-to-face interaction come into play. In contrast, male deceptive behavior does not change until the strategic effect of perceived honesty matters and men's truthfulness rises way beyond the level of women. To elaborate on these gender differences, in the second part, participants are asked to assess the honesty of videotaped statements from an experimental setting identical to the third treatment. We find that more men are assessed as rather dishonest. Men's dishonest perception is independent of whether they are actually truthful or not and whether they are assessed by men or women. We conclude that men anticipate their low perceived honesty in a face-to-face setting and, therefore, deceive less compared to women.