Due to the COVID19 global pandemic, the Economic Science Association organized a seminar series featuring junior job-market candidates in economics who use experimental methods. The virtual seminar series ran from October to December 2020 with sessions occuring at three different times to accomodate audiences around the world. Each sessions consisted of two to three speakers, each speaking for 20 minutes, followed by a discussant.
The scheduled speakers and discussants are listed below.
Women are under-represented and more likely to "leak" out of male-dominated fields. This paper tests whether women's status as the minority gender causes their dropout from male-dominated fields. I conduct a field experiment in an introductory economics course where I randomly assign students to study groups with different gender compositions. Results show that women assigned to male-dominated groups become 10 percentage points more likely to drop out from the course. In contrast, minority status in study groups does not significantly affect men's dropout behavior. I present suggestive evidence on educational expectations and peer-to-peer interaction as underlying mechanisms through which minority status increases female attrition. Women assigned as the minority gender form more pessimistic expectations about future academic achievement and interact less frequently with their peers. This paper demonstrates the harm of minority status on educational persistence and suggests that small groups with non-minority women can potentially help increase female representation in male-dominated fields.
Recent studies have shown that prosocial incentives do not increase motivation, as measured by workers’ performance, as rewards increase. To overcome this shortcoming, we implement a self-chosen goal scheme along with the incentives. We design an online experiment in which participants set goals for themselves engaging in real effort tasks. Participants obtain the prosocial rewards only when they reach their goals. Our results show that workers who receive prosocial incentives improve their performance by setting higher goals and achieving them. Moreover, when provided with the opportunity to receive large rewards, workers who are matched with the charity’s mission will set higher goals to further motivate themselves to make additional efforts. Our findings suggest prosocial incentives are comparable to monetary incentives in motivating workers within a self-chosen goal scheme. The preferred type of incentives depends on the firms’ targets.
Corporations, non-profits, and public sector bureaucracies use organizational mission to motivate workers. In a field experiment, with the Department of Health in Pakistan, we show that mission matters to workers. The organizational mission helps improve the performance of workers by 16.2%. This effect exists for multiple tasks, persists even after the experiment has ended, and leads to better health outcomes for the client population. In comparison, financial incentive improves performance by 27.5% but this effect is visible only for the task linked to the incentive, disappears as soon as the project ends, and results in a smaller improvement in health outcomes compared to the mission treatment. When the two treatments are combined, workers exhibit crowding out behavior that is driven by the multi-tasking nature of the job. We find suggestive evidence that the mission treatment effects are not driven by social norms but they activate mission-dependent intrinsic preferences of workers. These results highlight that organizational mission is a powerful motivator to improve performance when contracts are incomplete.
I derive a theoretical model of choice bracketing from two behavioral axioms on top of expected utility. The first behavioral axiom establishes a direct link between narrow bracketing and correlation neglect. The second behavioral axiom identifies the reference point as the place where broad and narrow preferences are connected. In my model the narrow bracketer is characterized by an inability to process changes from the reference point in different dimensions simultaneously. As a result, her trade-offs between dimensions are distorted. While she disregards interactions between actual outcomes, she appreciates these interactions mistakenly with respect to the reference point. In addition to the theoretical contribution, I present the results of an experiment to test the predictions and underlying assumptions of my model.
I study morally controversial transactions and focus on those which are only objected to when they are coupled with high incentives. To this end, I run a preregistered framed field experiment with a large representative sample. Participants decide whether they approve offers to a third person that concern monetary compensation in exchange for registering as a stem cell donor. I provide the following insights. First, some people want to prevent a controversial transaction only when incentives are high and their preference can be revealed using incentivized choice data. Second, most of the people who only object to monetary incentives when incentives are high do so, because offering high incentives draws in individuals with high reservation prices. Third, the objection to high incentives is most prevalent for people who consider the offer to be coercive. My study advances our understanding of whether and why high incentives for certain transactions are considered to be ethically inappropriate and frequently objected to.
In this paper we introduce the “endowment effect for information”: that the value of information is higher when information is expected than when information is not expected. We demonstrate such an endowment effect for information in an experiment in which we manipulate expectations. Furthermore, we show that this result is consistent with predictions from two (standard and expanded) leading theories of expectations-based reference-dependent preferences. Our endowment effect for information implies that net benefits from information policies may vary with people’s expectations (e.g., consumers who regularly see calorie information on menus might value the information more because they expect it).
We study how people perform risky experimentation to acquire information when they can also learn from each other. We develop and experimentally test a two-armed bandit model adapted from that of Keller et al. (2005). Our model predicts that the information generated by a group of players is no more than that generated by a single player. To implement this model in the lab, we design a novel dynamic information structure that can trivialize the posterior calculation for any sequence of signal realizations. We find that 1) when experimenting alone, the median subject generates almost exactly the same amount as the theoretical prediction, that 2) when experimenting with others, the subjects tend to generate more information than they would do alone, which is against the theoretical prediction, and that 3) the subjects only react to the posterior belief and do not condition their actions on what other players' past actions, thus excluding the folk theorem type argument as the explanation for more information generated in groups.
We investigated the physiological and information processing mechanisms underlying the effect of incentives on risk aversion. We recorded reaction time, gaze fixation, pupil dilation, pulse rate, and skin conductance while participants (N=39) chose between lotteries involving low real (1x), high hypothetical (50x), and high real (50x) stakes. Each block involved 10 choices repeated twice, with the location and order of the attributes randomly assigned, for a total of 80 choices. Increased risk aversion under high real stakes was associated with changes in tonic arousal levels and in attention (as measured by gaze fixation duration). We find that the high and low attributes for available options were attended differently during both non-decision time (NDT: passive exposure phase) and decision-time (DT). Also, increased dwell time advantage on high attributes, and not low ones, was strongly associated with increased risk aversion. We develop and estimate an arousal-modulated Attentional Drift Diffusion model (aADDM) of the interaction of arousal and attention during evidence accumulation. We estimate the attribute-based aADDM using our behavioral, eye-tracking, and physiological data. NDT and DT gaze amplified the value difference for high attributes only (multiplicative gaze bias). Trial-to-trial variation in arousal modulated this bias for NDT gaze. In other words, the high attributes’ value difference is discounted steeply when attending to lower attributes, with heightened arousal further amplifying this process. No such effect was found for low attributes. Additionally, trial-to-trial increases in arousal strengthened DT additive (simple) gaze bias for low attributes and weakened it for high attributes. We also find that increased trial-to-trial arousal levels widen the decision threshold signifying higher response caution. Physiological measures and attention differ significantly when the payoffs are large and real compared to small or hypothetical stakes. Our sequential sampling model that incorporates both arousal and attention shows that the effect of arousal-gaze interaction differs between gaze for non-decision time and that of decision-time, with the former influencing multiplicative gaze bias and the latter influencing additive one. Our results demonstrate the value of integrating physiological and attention measures for understanding the decision-making process.
This paper offers experimental evidence on the significance of role-models on fostering hope, increasing effort and academic performance of primary school students in India. Students from private schools were individually randomised to a treatment or a placebo group. Treated students watch a short film produced as a part of the experiment in Jaipur, Rajasthan - the study location. The placebo group students watch a television show for kids, 'Malgudi Days'. I find a 0.17 standard deviation (s.d.) increase in student hope and 0.25 s.d increase in their effort, immediately after the intervention. The one-off treatment leads to a 0.16 s.d. increase on standardised test scores in English, six-weeks after the intervention. Along with hope, I find significant improvements in students' self-efficacy or optimism and happiness. A cost-effectiveness analysis highlights role-models as a promising treatment intervention tool that can have an effect on student motivation and their learning outcomes.
Committee quotas have been introduced during the last years for combatting the underrepresentation of women in male-stereotyped environments. This paper provides experimental evidence on the effect of the committees’ gender composition on female candidates’ probabilities of success in recruitment processes. I designed a laboratory experiment in which groups of three subjects jointly select two candidates in a pool of six to perform a task. I found that male-majority committees were the most beneficial for female candidates while female-majority committees were the most detrimental. I also propose a mechanism based on the gender differences in group dynamics that could explain this counter-intuitive result. In terms of voice and influence, I found that men and women behave similarly in male-majority committees. In female-majority committees, however, men are more influential than women and disproportionately proposed to recruit only male candidates. The paper suggests that more women in the committee do not necessarily benefit female candidates and examines some reasons that question the effectiveness of this policy.
We examine how professional traders behave in two classic experiments about financial markets. In the first experiment, traders trade an asset over multiple periods, after receiving private information about its value. In the second experiment, they play the Guessing Game. We complement these experiments with a novel, individual decision making version of the Guessing Game and collect data on cognitive abilities, risk preferences and confidence. We run the same tests with a control group of undergraduate students. We find that professional traders behave very differently from students. In the first experiment, professional traders do not produce the classic bubble pattern in the price of the asset observed in previous studies; on the contrary, the median price tracks the fundamental value rather closely. Moreover, they aggregate private information better than students. In the Guessing Game, they choose significantly lower numbers (are strategically more sophisticated, in the level-k interpretation) than students. These results are not driven by different individual characteristics (e.g., cognitive abilities).
In this project, we propose that holding debt causes poorer financial decisions using a new experimental design where we randomly assign debt. Our design isolates the consequences of having debt while controlling for potential confounding factors such as initial wealth levels, risk, and time preferences. Our findings show that debt causes more financial mistakes, detrimental to participants' payoffs. Subjects exhibit a debt-biased behavior in which they first pay their debt and then maximize their payoffs, consistent with a model where debt balances are over-weighted. Furthermore, subjects miss profitable opportunities due to their reluctancy to borrow when it involves incurring debt, providing a rationale for debt aversion.
Market design has provided many managerial insights into why certain market institutions fail while others succeed in allocating scarce resources in both the for- and non-profit sectors. In this paper, we analyze a new form of dynamic matching mechanism enabled by innovations in information technology. We provide a theoretical and experimental examination of this mechanism in the context of college admissions in Inner Mongolia, China, where students are given real-time allocation feedback and are allowed to revise their choices. Theoretically, we show that efficient and stable outcomes arise in every rationalizable strategy profile if there is a sufficient number of revision opportunities. Experimentally, we find that when the coordination cost of reaching equilibrium is high, the Inner Mongolia Dynamic mechanism performs better than theoretical predictions compared with the Deferred Acceptance and the Boston mechanisms. These results suggest that the Inner Mongolia Dynamic mechanism can be a good substitute for static mechanisms when the coordination cost of reaching equilibrium is high. The Inner Mongolia Dynamic mechanism may also be useful in matching potential employers and employees in the labor market.
This paper centers on whether the timing of the regulator's commitment to an environmental policy affects the firms' behavior concerning the adoption of an advanced abatement technology. Our experimental design mimics an industry with small asymmetric polluting firms regulated by either emission taxes or permits. We compare two different timing regimes concerning the regulator's commitment to a given emission control policy: In the ex-ante regime, the regulator announces the instrument level (tax rate or amount of permit) before the firms' technology adoption decisions, whereas in the ex-post regime the regulator sets the optimal instrument level after the firms' technology adoption decision. We find that in the two regimes, the adoption of advanced technology is close to the social optimum. Our findings support market-based environmental policies because an ex-post regime may be more politically feasible than an ex-ante regime.
For some questions, such as what the best policy to address a problem is, it is uncertain if the answer will ever be known. Asking experts yields two practical problems: how can their truth-telling be incentivized if the correct answer is unknowable? And if experts disagree, who should be trusted? This paper solves both problems simultaneously. Experts decide whether to endorse a statement and trade an asset whose value depends on the endorsement rate. The respective payoffs of buyers and sellers indicate whom to trust. We demonstrate theoretically and illustrate empirically that "following the money" outperforms selecting the majority opinion.
This paper uses a laboratory experiment to determine how charity religious affiliation drives donor preference and giving. Subjects choose from one of eight charities, with each charity varying in religious affiliation. Masked and unmasked sessions differ in the knowledge of the religious affiliation of half the charities, with masked sessions omitting the religious affiliation of the aforementioned charities. My results show that adding additional religious language decreases donation frequency for Christian charities competing exclusively against other religious charities. Additionally, adding religious language increases the average donation size for secular charities competing against Christian charities, but decreases average donations for Christian charities competing exclusively against other religious charities. Subjects prefer charity religious affiliation to match their own religious identity; however, subject strength of religiosity is more predictive in charity choice than religious affiliation of the donor. My results indicate that religiously affiliated charities have financial incentive to selectively display their affiliation.
Does the local nature of information in a networked environment impede coordination and cooperation? How does the communication structure affect the actions group's coordinate on? This paper explores the interplay between two communication structures and decisions in a best-shot public good game across three five-person networks. The network determines which group members share a local public good. Across the networks, the number of neighbors each person has is varied. In the Line network, an individual can have either one or two neighbors. In the Asymmetric network, an individual can have one, two, or three neighbors. Lastly, the Circle network is symmetric, and everyone has two neighbors. I then introduce two communication structures. The first is global, where all group members can communicate. Second is local, where only neighbors can communicate. Unlike global, the local structure overlays the communication structure on the underlying network of externalities, which reinforces the strategic advantage of individuals with more neighbors in the Line and Asymmetric networks. Compared to no communication, both global and local structures successfully raise efficiency in the Line and Asymmetric networks. However, in the Circle network, only local communication is successful in raising efficiency. Global communication does not improve efficiency or the likelihood of coordinating on an equilibrium. With communication, individuals learn to take turns to invest, which promotes equity and efficiency.
Choice overload is a phenomenon whereby decision makers are overwhelmed by the choices they face, which can lead to poor decisions and reductions in welfare. I conduct an experiment in which subjects face three donation lists of differing lengths and are asked whether or not they would like to donate to the charities. On the extensive margin, I find a U-shape exists for giving, the donations least frequent with an intermediate number of options. On the intensive margin, there is no significant difference between the donated amounts with the list size treatments. I also find some strong evidence that some measured personality traits influence giving as well. Using an ex-post model of choice overload in altruism, when the list size treatment is very small, any costs of search will be negligible. As the list size treatments continue to get larger, the cost of search increases. But, depending on the individual’s strength for their preference for finding the right charity to donate to, the expected match quality deviation is larger than the cognitive cost of searching. This shows a U-shape on the choice to donate and the list size treatments.
We investigate the tendency to avoid information on externalities associated with collective and individual decisions in different institutional contexts with a laboratory experiment. Externalities are implemented as contributions to carbon offsets. Subjects need to choose between two allocations which differ in own payoffs and offset contributions. Depending on the treatment, the size of the contribution is either directly observable or needs to be actively revealed by the decision maker. In a between-subjects design, we vary the institutional setting: market, dictatorship and majority voting in small and large groups. We find self-serving information avoidance on markets and in large democracies. In small groups and dictatorships, selfish choices are frequent even if all information is provided. Interestingly, differences in the shares of selfish choices between institutional contexts disappear if information is to be revealed. Our results suggest that information avoidance effectively substitutes expressive voting as an instrument to manage self-image on the part of the voter, highlighting the importance of avoided information in democratic decisions.
Choice Architects have the power to design the choice environment in which Decision Makers later operate. Architects may use nudges, changes to the decision environment that do not alter monetary payoffs or limit options, to predictably influence the behaviour of the Decision Makers. This project investigates how Decision Makers' awareness about participating in such a game, what I call transparency, affects how the nudges influence behaviour and the Choice Architects use these nudges. The model distinguishes between System 1 nudges that provide shortcuts (for example, default options) and System 2 nudges that encourage reflective thinking (for example, through the promotion of cost-benefit analysis). The model predicts that a) System 1 nudges have less effect with transparency while System 2 nudges keep their effectiveness and that b) the use of System 1 nudges is reduced by transparency while the use of System 2 nudges is increased. These predictions are tested with an online experiment. I find that while transparency does not change choice architects' behaviour it does however have the effects predicted for System 1 and System 2 nudges, particularly among women.
Access to morally relevant information could lead to behavioral change, but only if individuals attend to such information. We investigate this issue in the context of food consumption, where the decision to eat meat from intensive farming entails animal suffering. Based on a pre-registered experiment, we find that about 30 percent of subjects avoid information on animals' living conditions in intensive farming. When receiving information, subjects significantly reduce their propensity to consume meat on average by about 12 percentage points in the laboratory and 6 to 8 percentage points in university canteens. Our analysis of heterogeneous effects suggests that responsive individuals select out of information. This selection pattern impedes the effectiveness of information provision, even when information is provided for free.
I investigate whether remaining partially ignorant of the consequences of one’s decision leads to a decrease in prosocial behavior using a laboratory experiment in the style of Dana et al. (2007), where the dictator can choose whether or not to know the recipient’s payoffs. I modify this framework by introducing a noisy signal about the recipient’s payoffs in one state of the world to investigate how much information dictators will acquire. I use a simple model to motivate the following predictions: information acquisition will be decreasing in costs of information acquisition, increasing in beliefs that the agent’s actions are prosocial, and increasing in the expected prosociality of the available choices. I find that subjects acquire more free information than costly information and will acquire more signals when they believe that the unknown distribution is prosocial. Subjects also act more prosocially when the prosocial distribution is the known distribution. I also find evidence that subjects look for excuses to be selfish. These findings, taken together, provide support for the intuition that agents often look for opportunities to self-deceive, by choosing to know just enough about a particular situation that they experience utility from the outcome, even if their actions are selfish.
This paper explores how an individual can deceive herself in order to acquire a more favorable self-view. I present a theoretical model and a laboratory experiment in which each subject’s donation to a charity gets transformed into an ambiguous lottery, i.e., a lottery where the winning probability is unknown. A subject who holds a pessimistic belief about this probability has a legitimate reason to not donate. However, a subject who is not pessimistic can deceive herself and falsely adopt a pessimistic belief, giving herself the same excuse that is available to a truly pessimistic subject. She can then behave selfishly without damaging her self-view of an altruistic person. In an experiment, I find that only selfish subjects self-deceptively adopt pessimistic beliefs. Altruistic subjects, who obtain an altruistic self-view by donating money, do not deceive themselves in this way.
Recent studies of communication and information have embraced more flexible models allowing for omission or vagueness in messages. Applications include a public-good provision game, a sender-receiver disclosure game, and many more. Yet little is known about the relationship between the strategic usage of vague messages and individuals attitudes toward lying. Research on lying behavior in economics such as Fischbacher and Follmi-Heusi 2013 and its successors have mainly focused on subjects making outright lies and excluded a broader set of possibly misleading communication. This paper attempts to fill the gap by incorporating vagueness into the model of costly lying and provide empirical knowledge through an online experiment. The theoretical predictions of the paper derive from a variation of the Fischbacher and Follmi-Heusi 2013 model in which an agent may send a vague, set-valued message, or a precise, single-valued message. We focus on two motivations for honesty: the agent cares not only about her monetary payoff but also the reputation/external concern for being seen as honest and the internal motivation for being honest. The main hypothesis is that a vague yet truthful message affects only the agent's external but not her internal motivation. As a result, the agents who make small outright lies under the constraint of precise messages would choose to be truthful and vague if allowed. This suggests that the restricted message space could be another source of the observed aversion for monetary-payoff-maximization in previous experiments. We propose a 2x2 experimental design that varies the anonymity of subjects and the availability of vague messages to test the hypotheses and isolate the effect of vagueness in the subjects internal motivation from their external motivation.
During the COVID-19 pandemic, the introduction of mandatory face mask usage was accompanied by a heated debate. It was argued that community use of masks creates a false sense of security that could decrease social distancing, thus making matters worse. We conducted a randomized field experiment in Berlin, Germany, to investigate whether masks lead to decreases in distancing and whether this mask effect interacts with the introduction of a mask mandate in Berlin. Joining lines in front of stores, we measured the distance kept from the experimenter in two treatment conditions – the experimenter wore a mask in one and no face covering in the other – both before and after the introduction of mandatory mask use in stores. We find no evidence that mandatory masking has a negative effect on distance keeping. To the contrary, in our study, masks significantly increase distancing and the effect does not differ between the two periods. Further, we find no evidence that the mask mandate affected distancing. However, our results suggest that the relaxation of shop opening restrictions had a negative effect on distancing.
I document that people may choose to avoid information about the consequences of a decision when it can signal an undesired characteristics like being selfish or betraying someone's trust. Even when there is no other observer than themselves, people may have self-image concerns and worry about what does a decision show about their characteristics. There is a clear pattern: people who make selfish decisions in a distributive game are more likely to avoid information about the consequences of their decisions. This behaviour is observed even when the distributive decision is already made and the information cannot change the outcome, particularly when information can show that the decision caused a loss for someone else. This paper contributes to the experimental research on information preferences and moral decisions by documenting that distributive decisions may not only be motivated by the desire for certain outcomes but also desire to self identify as a good person by oneself.
This paper provides evidence that biases attributed to the perception of probabilities affect cooperation levels in repeated games. In an experiment, subjects completed a prisoner's dilemma game that continued to the next round with a fixed probability. Under the standard assumption of (constant) discounted expected utility, such a probability can be interpreted as time discounting. The presence of probability biases leads to deviations from constant discounting, as shown in Halevy (2008, AER), which then affects the evaluation of the outcomes in the game and, the hypothesis is that this affects the cooperate-defect decision of subjects. Using an incentive-compatible mechanism based on scoring rules, I quantify the direction and magnitude of subjects probability bias. I find that 53% of subjects are expected utility (EU) subjects; 21% of subjects reveal biases that accord with prospect theory: small probabilities are overweighted and medium to large ones are underweighted (inverse-S); 26% of subjects underestimate small probabilities and overestimate large probabilities (S-shape). The main finding is that the cooperation level is correlated with the type of biases. Specifically, for all continuation probabilities, inverse-S subjects cooperate more than EU subjects, and S-shape subjects cooperate less than EU subjects. I explain this behaviour in the repeated games by adopting Halevy's impatience index.
Using a multiple-price list dictator game, this paper provides experimental evidence that the level of generosity is affected when we vary the probability that the dictator’s decision will be implemented. We also show that framing matters for generosity in that subjects are less generous when their choices under role uncertainty are such that subjects perceive that they are in the role of dictators and know that their choices will be implemented with a certain probability, compared with a setting in which subjects are told that they are in the role of recipients and know that their choices will not be implemented with certain probability.
In developing countries where formal institutions are often weak, the community is responsible to enforce local agreements. In such settings, peer monitoring represents a natural mechanism for the enforcement of agreements. This paper studies the demand for monitoring and its effectiveness in sustaining cooperation across social groups. We map social networks of 19 villages in rural Nepal and conduct an experiment to explore the role of the endogenous choice of monitors. Individuals play in groups of three, both with their close friends and with people socially distant in their network. They receive the opportunity to anonymously choose their preferred institution. We combine a theoretical model and a unique lab-in-the-field experiment to show that closely knit groups are significantly more likely not to choose a monitor, while sparse groups tend to prefer a monitor who is highly central in the network. The endogenous selection of monitoring ensures higher cooperation compared to an exogenous assignment, but only in sparse groups. Further, we observe that the outcome of the vote acts as a signal of intra-group trust.
An option is especially tempting when it is both immediate and certain. I conduct a framed field experiment on longitudinal real-effort provision to test this hypothesis and quantify the effect of risk on present bias. The observation of dynamic inconsistency requires decisions made in advance of real-effort exertion, as well as when exertion is imminent. My novel experiment methodologically eliminates two types of risk created by the random incentive scheme, while still identifying present bias. I find that the extent of present bias is no different given certainty, but its average intensity increases dramatically under certainty.
This paper explores how ambiguous signals and ambiguity aversion influence individuals' expectations and the pricing of asset in experimental financial markets. In line with the theory of Epstein and Schneider (2008), we find that subjects' degree of ambiguity aversion is positively correlated with their expectations about the variance of ambiguous signals. These signals matter for the determination of asset prices. We find that the distribution of the excess return of the asset exhibits negative skewness, and that price volatility is significantly larger under ambiguous signals. Our findings provide evidence in support of the idea that ambiguous information and ambiguity aversion may be a source of negative skewness and excess volatility in financial markets.
We study a mechanism for crime deterrence that utilizes insider information contained in social networks. Regulators may have limited information about a crime but happen to identify a suspect. We propose a mechanism where this suspect can redirect the penalty to another person from the network who is deemed to be more responsible. The regulator examines the criminal activities of both and obtains two noisy signals about their actions. The one with the higher signal is punished and the other goes free. We show theoretically that, for a given probability and magnitude of the penalty, crime levels are lower with this mechanism than in the case where the first suspect is automatically punished. In equilibrium, crime levels depend on the given criminal’s position in the network and on the network structure. Our experiment confirms that this mechanism effectively deters crime, but the magnitude is above the Nash equilibrium predictions and is less sensitive to changes in the network structure than the theory predicts. Level-k reasoning helps to explain these patterns.
Economists model self-control problems through time-inconsistent preferences. Empirical tests of these preferences largely rely on experimental elicitation methods using monetary rewards, with several recent studies failing to find present bias for money. In this paper, we compare estimates of present bias for money with estimates for healthy and unhealthy foods. In a within-subjects longitudinal experiment with 697 low-income Chinese high school students we find strong present bias for both money and food, and that individual measures of present bias are moderately correlated across reward types. Our experimental measures of time preferences over money predict field behaviours better than preferences elicited over foods.
If preferences over job characteristics are believed to vary between genders, gender segregation in the labor market may be interpreted as signals of unobserved job characteristics. Labor market choices of one generation of workers can then affect the next generation’s interpretation of job characteristics - and thus their choices. I present a formalisation of this idea, before exploring its predictions empirically by means of an online experiment. I find that when there is extant uncertainty about job characteristics, while the distribution of men and women in various jobs is observable, segregation reproduces itself. Individuals’ beliefs about the relative distance between their own and men and women’ preferences predict whether they choose the male or female dominated job.
Labor market opportunities and wages may be unfair for various reasons, and how workers respond to different types of unfairness can have major economic consequences. Using an online labor platform, where workers engage in an individual task for a piece-rate wage, we investigate the causal effect of neutral and gender-discriminatory unfair chances on labor supply. We randomize workers into treatments where we control relative pay and chances to receive a low or a high wage. Chances can be fair, unfair based on an unspecified source, or unfair based on gender discrimination. Unequal pay reduces labor supply of low-wage workers, irrespective of whether the low wage is the result of fair or unfair chances. Importantly, the source of unfair chances matters. When a low wage is the result of gender-discriminatory chances, workers matched with a high-wage worker substantially reduce their labor supply compared to the case of equal low wages (-22%). This decrease is twice as large as those induced by low wages due to fair chances or unfair chances coming from an unspecified source. An additional experiment confirms the deleterious effect of gender discrimination on labor supply in a work environment devoid of chances, and highlights that workers' beliefs about facing discrimination matter for their responses. Our results concerning gender discrimination indicate a new reason for the lower labor supply of women, which is a prominent explanation for the gender gap in earnings.
There exists extensive evidence documenting the economic consequences of discrimination patterns between individuals belonging to the same or different group identities. However, many group identities rely on convictions and beliefs that are non-observable, and therefore, might be uncertain. This paper investigates the effects of group identity, and its uncertainty, on individuals’ interaction preferences and their willingness to coordinate. Results from a laboratory experiment using a repeated weakest-link game with endogenous group formation, show that unknown-group individuals are more negatively discriminated than out-group individuals in the long term. Nevertheless, all discrimination patterns vanish when interactions entail high and mutual potential economic incentives. Our findings offer several managerial implications for deterring discrimination among individuals to interact, and increasing their coordination efficiency when working in teams.
This paper presents a puzzle in the behavior of experimental subjects in what we call common-probability auctions. In common-value auctions, uncertainty is defined over values while, in common-probability auctions, uncertainty is defined over probabilities. We find that in contrast to the substantial overbidding found in common-value auctions, bidding in strategically equivalent common-probability auctions is consistent with Nash-equilibrium. Additional treatments reveal that subjects valued the auctioned items equally, implying that differences in bidding behavior originate in the strategic uncertainty of the auction.
Many committees, juries, political task forces, etc.. spend time gathering costly information in order to reach a decision. We report results from lab experiments on such information-collection processes. We consider decisions governed by individuals and groups and compare how different voting rules affect outcomes. We also contrast static information collection, as in classical hypothesis testing, with dynamic collection, as in sequential hypothesis testing. Generally, outcomes approximate the theoretical benchmark, and sequential information collection is welfare enhancing relative to static collection. Nonetheless, several important departures emerge. Static information collection is excessive, and sequential information collection is non-stationary, producing declining decision accuracies over time. Furthermore, groups using majority rule often reach especially hasty and inaccurate decisions.
This paper analyzes how individuals resolve an exploration versus exploitation trade-off in a laboratory experiment. The experiment implements the single-agent exponential bandit model. We find that subjects respond in the predicted direction to changes in the prior belief, safe action, and discount factor. However, we find that subjects typically explore less than predicted. We estimate a structural model that incorporates risk preferences, base rate neglect/conservatism, and non-linear probability weighting. We find support for risk aversion, conservatism, and non-linear probability weighting. Our estimated structural model suggests that risk aversion is the predominant factor driving subjects' under-exploration in the experiment.
We provide a direct, experimental test of the buffer stock model of savings behavior. We use a three-period intertemporal model of consumption/savings decisions where liquidity in the second period is constrained (and, thus, borrowing is not possible). We contrast behavior in this constrained version of the model with an unconstrained version where there is no liquidity constraint. A second treatment variable is the variance of the stochastic income process, resulting in a 2x2 experimental design. We test the comparative statics predictions of the model and find that, in contrast to these predictions, the liquidity constraint does not increase savings in the first period of the constrained model relative to the first period of the unconstrained model. However, we find strong support for all the other comparative statics predictions of the model, e.g., the impact of a higher variance of income on savings behavior and differences between period 1 and period 2 savings. In further analyses, we find that we can rationalize the departures we observe from model predictions by some combination of debt aversion, heterogeneity in cognitive abilities and/or learning.
This paper presents, to our knowledge, the first empirical investigation of the direct relation between the opportunity to evade and overall tax revenues. Resuscitating the old work by Weiss (1976), we test the hypothesis how an optimal income tax might include incentives to evade to increase overall tax revenues: Such incentives partially offset the undesirable distortions of taxes on labor supply by lowering effective tax rates. We implemented an original real effort experiment in an online labor market (MTurk) with over 1,000 participants. Our findings not only show significant positive labor supply responses to the opportunity to evade (increased labor supply by on average 37%). Also the expected tax revenue significantly and substantially increased by more than 50%. Strikingly, this effect persists when comparing effective tax rates: Lowering effective tax rates through the opportunity to evade is more efficient than simply lowering statutory tax rates. Since this effect is strongest for low productive individuals, our work also contributes to the literature on optimal tax administration: Facing limited resources, tax enforcement should focus on high productive individuals.
We experimentally test whether the gap between expected and actual income impacts subsequent altruism. Participants first perform a real-effort task for a fixed wage and then play a dictator game. Between conditions, we vary the level and the timing of the revelation of the wage. In some conditions, participants know the wage before the real effort task and are not informed of the other potential levels. In some other conditions, they are informed of the distribution of wages before the real effort task, but the actual wage is only revealed afterward. Participants in the latter conditions can form expectations that may be higher or lower than their actual wage. Our model predicts that the gap between the expected and the actual wage impacts transfers in the subsequent dictator game. The results support this hypothesis: participants who get the low wage transfer less and are less likely to transfer when they are informed of the other potential levels than when they are not. Conversely, participants who get the high wage are more likely to transfer positive amounts when they are informed of the other potential levels. We use physiological (skin conductance response) and declarative data to discuss the role of emotions in our treatment effects.
Procrastination behaviour has been extensively studied as an individual problem. However, the effect of such behaviour on the receiving end is rarely discussed. Many large governmental programs - such as census, tax returns, grant application, subsidy application - have to deal with large volumes of submissions that take place right before the deadline. The last-moment submitting behaviour can substantially challenge the handling system in terms of quality of service and maximum capacity. To study how to alleviate last-moment overload, we conducted two waves of field experiments in a national census that includes submissions from all farming enterprises in the Netherlands. Using randomized-controlled trials, we compared four types of zero-stake interventions: 1) raising awareness with a peak-day calendar, 2) adding a short list of information to help prepare for submission, 3) providing a soft incentive of early submission, early evaluating, 4) setting an earlier non-binding target date for submission. Our results show that setting an earlier target date is a robust and effective method of nudging towards earlier submission, hence could reduce the overload of the peak submission period. This result suggests that anchoring a date, even when it does not have the power of a deadline, could still effectively influence procrastination.
This paper studies the sources of suboptimal allocations observed in credit card repayments using a diagnostic laboratory experiment. We find that optimization ability and limited attention are jointly insufficient to explain the puzzle. Moving beyond existing results, we find that the inherent negative frame of the debt payment problem interferes with subjects’ ability to optimize and hinders learning. We show that subjects predominantly rely on the irrelevant balance information while forming their decisions, regardless of how vividly the balance information is displayed. Using additional treatments, we find that the debt frame increases subjects’ focus on the irrelevant balance information.
Commitment devices such as coalitions can increase outcome efficiency in public goods provision. This paper studies the role of social preference in a two stage public good game where, in the first stage, heterogeneous agents first choose whether or not to join a coalition then, in the next stage, the coalition votes on whether its members will contribute. We find that individuals with stronger social preferences are more likely to join the coalition and vote for the coalition to contribute to the public good. We further show that higher marginal benefits of contribution leads to more people joining the coalition and contributing to the public good. These results hold whether the coalitions decision is determined by a majority voting or a unanimous voting rule. The results are also robust to different model specifications. We also test our predictions in an experimental setup.
This paper investigates the effects of complexity in various voting systems on individual behavior in small group electoral competitions. Using a laboratory experiment, I observe individual behavior within one of three voting systems -- plurality, instant runoff voting (IRV), and score then automatic runoff (STAR). I then estimate subjects' behavior in three different models of bounded rationality. The estimated models are a model of Level-K thinking (Nagel, 1995), the Cognitive Hierarchy (CH) model (Camerer, et al. 2004), and a Quantal Response Equilibrium (QRE) (McKelvey and Palfrey 1995). I consistently find that more complex voting systems induce lower levels of strategic thinking. Models of strategic thinking (Level-K/CH) fit the data best and are consistent with the high frequency of sincere strategies observed in the data.
We design a new experimental framework to study policy interventions in secular stagnations and liquidity traps using an overlapping-generations environment where participants form expectations and make real economic decisions. We explore the ability of unconventional monetary policy to lead economies out of deflationary traps and away from a binding zero lower bound. We observe that participants are able to coordinate on high inflation full-employment equilibria. Permanent exogenous deleveraging shocks induce pessimistic expectations that precipitate persistently deflationary episodes. We explore policies aimed at re-anchoring expectations and the economy on the high inflation steady state. Permanently increasing the central bank's inflation target is insufficient to generate inflationary expectations. Eliminating the zero lower bound, on the other hand, is consistently effective at stimulating spending and generating the necessary inflation for the economies to escape the zero lower bound. Our findings suggest that inflation expectations and demand is more effectively stimulated through realized wealth effects than coordination on rational expectations equilibria.
Building on rich psychological literature regarding the negative effect of stereotypes (stereotype threat) on the stereotyped group (Spencer and Steele, 1999), I study the channels of how statistical role models, i.e., examples of other successful participants, affect both self-selection into the stereotyped domain, and performance in it. First, I provide a theoretical model of stereotype threat, and show the importance of belief updating in combating an existing stereotype. I identify situations in which a stereotype may be self-correcting, i.e., stimulating the negatively stereotyped group to outperform the positively stereotyped group. Second, I conduct an online experiment in order to test the model’s predictions, making use of a common gender stereotype regarding mathematical ability. In my experiment, I use a real-effort task with two properties: 1) men and women perform equally well on this task, and 2) both genders expect men to perform better (Reuben et al., 2014). I use these facts to either strengthen or diminish the stereotype threat, and subsequently study the impact of introducing role models. My experimental design allows me to quantify to what extent do people update their priors about their own ability, as well as the decisions and performance of others in response both to stereotypes and to role models. I show how context affects the importance of this belief change for self-selection as well as task performance. Moreover, I test whether the provision of positive role models (i.e., successful people) can backfire, i.e., prove to be demotivating for those who see success as unattainable. My paper complements the existing literature using identifiable people as role models, since my design abstracts away from storytelling, charisma, and other social aspects, and instead focuses on the role of information and belief change.
In recent years, a number of online labor markets have emerged that allow workers from around the world to sell their labor to an equally global pool of buyers. The creators of these markets play the role of labor market intermediary by providing institutional support and remedying informational asymmetries. In this paper, we explore the difficulty and complexity of the online tasks, and discuss the determinants of the workers’ productivity, efficiency, effort and quality levels. The paper concludes with a discussion on the personality and incentive effects on workers’ job behavior.
The use of hypothetical instead of real decision-making incentives remains under debate after decades of economic experiments. Standard incentivized experiments involve substantial monetary costs due to participants' earnings and often logistic costs as well. In time preferences experiments, which involve future payments, real payments are particularly problematic. Since immediate rewards frequently have lower transaction costs than delayed rewards in experimental tasks, among other issues, (quasi)hyperbolic functional forms cannot be accurately estimated. What if hypothetical payments provide accurate data which, moreover, avoid transaction cost problems? In this paper, we test whether the use of hypothetical - versus real - payments affects the elicitation of short-term and long-term discounting in a standard multiple price list task. One-out-of-ten participants probabilistic payment schemes are also considered. We analyze data from three studies: a lab experiment in Spain, a well-powered field experiment in Nigeria, and an online extension focused on probabilistic payments. Our results indicate that paid and hypothetical time preferences are mostly the same and, therefore, that hypothetical rewards are a good alternative to real rewards. However, our data suggest that probabilistic payments are not.
This paper explores dismissal threats as an incentive device in an experiment. Subjects perform real effort tasks for a fixed wage per period and may be dismissed at the end of each period for low performance. A novel design mechanism allows decomposing the productivity gains from stronger dismissal threats into the selection effect and the incentive effect. Extensive trait measurements of subjects allows analysis of multidimensional sorting into high-paying, high-dismissal-threat jobs.
This paper investigates how workers' job application decisions are affected by their beliefs about hiring managers' beliefs regarding the relative productivity of women and men. To this end, I combine a natural field experiment with a lab experiment. In the field experiment, I partner with a firm to solicit approximately 5,000 job applications using ads that randomize over the gender of the hiring manager and the gender associations of the product sector. I then recruit the same job-seekers to a structured online lab experiment to elicit their beliefs about hiring managers' beliefs, based on the manager's gender and product sector. Truth-telling is incentivized with the Binarized Scoring Rule, using a procedure I adapt from Dustan, Koutout, and Leo (2020). I find that men are more likely to apply for a job with a manager whom they believe has beliefs that favor men more. A one standard deviation increase in beliefs about the manager's beliefs increases the probability a man applies by 70%. On the other hand, women are unresponsive to their beliefs about managers' beliefs. These results have important implications for the sorting by gender behavior driving a large part of the gender wage gap.
This paper studies the impact of continuous time environments on how lab subjects interact in battle of the sexes games in the laboratory environments. Compared to the Nash equilibrium in other continuous time games, alternating dynamics in battle of the sexes games require subjects to switch between two pure Nash equilibria and cause extra difficulty for subjects to coordinate in continuous time environments, as they need to coordinate both the order and the duration of alternations. The experimental results show that the coordination rate is consistently lower and subjects are less likely to play alternating dynamics in continuous time than in discrete time environments, even though subjects converge to their first Nash equilibrium faster. Compared to the learning and strong pattern of alternating dynamics in discrete time environments, subjects learn much slower and their dynamics are noisy and unstable in continuous time environments. The experimental evidence sheds light on the dynamics in continuous time battle of the sexes. There is no evidence that traditional theories that explain alternating dynamics work in continuous time.
In a series of two lab experiments, we study the willingness to trust in- versus outgroup members and its determinants, with a focus on betrayal aversion. Participants are students who are assigned quasi-randomly to groups outside the lab. The first experiment collects data soon after the groups have been formed, at the beginning of an academic year (at T1). The second experiment is a slightly modified version of the first one, and it is run seven months later, at T2. The two experiments allow us to study how the variables of interest change as the social groups mature. At T1, there is virtually no difference in the willingness to trust, nor in betrayal aversion towards in- versus outgroup members. Betrayal aversion is positive and indistinguishable towards members of both groups. At T2, betrayal aversion to members of both groups vanishes. Aggregate results also show no difference in the willingness to trust members of either group at T2. We zoom into the first of the two decisions participants make at T2, as there is evidence of order effects at this stage. In this first decision, 60% of participants, who select an ingroup member as a recipient in a hypothetical allocation task at T2, show a higher willingness to trust an ingroup member than an outgroup member. The remaining 40%, who select as recipient in the allocation task any other participant, are equally willing to trust members of both groups. Neither those who prefer the ingroup, nor those who do not show betrayal aversion to either group. We conclude that betrayal aversion at the aggregate level may disappear in time. Some participants might be relatively less willing to trust outgroup members at T2 because they are more ambiguity averse towards outgroup members. Other possible explanations are the competence hypothesis (Heath and Tversky, 1991), guilt aversion, or that these participants value intergroup inequality more by T2.
Welfare analysis in policy evaluation assumes the knowledge of individuals' behavioural models. Preferences are elicited consequently. We study preference elicitation when individuals behave according to heterogeneous and unobserved models. Since choices may be inconsistent with utility maximization, this problem is central for welfare analysis. Our solution is based only on two normative guidelines. First, a principle called Informational Responsiveness: choices should not be ignored even when inconsistencies are displayed. Second, these choices should be organized according to direct Revealed Preference. We test the joint importance of Informational Responsiveness and Revealed Preference using experimental data. We conduct a novel experiment in which participants firstly face a sequence of questions regarding time and risk outcomes and secondly report their preference relation over some of the alternatives. We find that when our generalized welfare analysis satisfies Informational Responsiveness and is based on direct Revealed Preference, it provides a significantly better match between the elicited and the reported welfare relation.
Individuals often behave impatiently when making financial decisions, which can have long-term economic consequences. This paper proposes and experimentally confirms an important factor for promoting more patient financial decision-making: the timing of financial decisions relative to payday. In a large-scale online experiment, participants choose between an early payment or a late payment. When participants are asked to make this decision eight days before their payday, rather than one day after their payday, they are 38% more likely to choose the late payment. By documenting the importance of the timing of financial decisions relative to payday, this paper is informative for policymakers and deepens our understanding of why impatient decisions persist.
Mental wellness apps are cost-effective, scalable, and effective in treating some common mental health conditions; however, a lack of clear information about data privacy may distort the market. I propose and test a simple five-star privacy rating system as a possible solution. I use a field experiment to evaluate changes in patterns of app selection, uptake, and persistence between participants with and without access to the ratings. I also measure willingness-to-pay for access to the ratings and estimate the social-welfare value of improving transparency regarding privacy policies for mental wellness apps in the United States. I find that a one-star increase in an app’s privacy rating leads to a six percent greater chance of app selection, without affecting app uptake or persistence relative to the control group. By participating in the experiment, both treatment and control participants increase usage of mental wellness apps, from 25% usage before the experiment to 41% usage after the experiment. I find participants are willing to pay about $2 for access to the privacy rating information, suggesting potential social welfare improvements on the scale of $80 million from improving transparency regarding the privacy policies of mental wellness apps in the United States.
Tax enforcing agencies have limited government budget for audits. As a result, it is important to develop cost-efficient audit strategies that deter tax evasion. One of the methods the IRS uses for audits is to pick individuals who are engaged in transactions with other taxpayers whose tax returns were selected for audit. This joint liability in the audit mechanism introduces an externality where one’s action affects the likelihood of others being audited. Therefore, it can induce different tax compliance behavior compared to an individual liability audit mechanism. We study the effectiveness of this group accountability (GA) audit mechanism against a purely random selection individual accountability (IA) audit mechanism using a lab experiment. In the GA audit mechanism, taxpayers are exogenously combined into several groups. Cheating by an audited person in the group triggers audits to all members of the group. Our data show that tax compliance is higher in the GA audit mechanism compared to the IA audit mechanism. The effect is robust with or without anonymity among group members and across various group sizes.