Research

Selected publications


Book chapter


Working Papers


Work in Process

Most literature on social preferences has documented altruistic and other-regarding behavior. A possible explanation suggests that this behaviour reveals underlying preferences for fairness. A competing hypothesis claims that the concern for social image motivates unselfish behavior which disappears when a psychological justification comes at hand (moral wiggle room). Ultimately the extent of both phenomena is an empirical question that we aim to address in this paper.  We propose a within-subject experimental design to provide a measure of the effect of the moral wiggle room at the individual level. Our design also allows us to correlate this measure with positive and negative reciprocity in an attempt to shed light on whether and how other traits of other-regarding behavior are related.    We implement variations of the modified Dictator Game by Blanco et al. (2011) to elicit equity-efficiency tradeoff, inequality aversion, moral wiggle room, and positive and negative reciprocity. Differently from most of the literature, all measurements are model free and not dichotomous.  Experiments were run at REBEL, with a total of 288 students. Preliminar results show that the distribution of altruistic concerns when the action is visible first order stochastically dominates the distribution when the action is hidden. Additionally, in order to be willing to reward other’s altruistic behavior, subjects demand more altruism than what they themselves show in the “standard” dictator game.  We also contribute to the literature that identifies a restricted number of types, within a model based measurement.


In this paper we report results from an online experiment conducted with over $1,000$ participants from Colombia's general population. The experiment is designed to examine the impact of exposition to a COVID priming and a negative economic shock on trusting behavior. Overall, we find that participants under the neutral prime who are exposed to a negative economic shock become less trusting. In addition, we find that trustors who receive the shock become more trusting, increasing the proportion of the endowment they transfer. This result is not an artifact of the modification of the trsutor's action set due to the negative shock received, and is consistent with beliefs of higher returned amount and stronger normative expectations of reciprocity, as well as general pro-sociality.


We implement a 2x2 between subjects design to identify the potencial benefits of a gradual reform, and how different  timings of the announcement of the reform can affect the reaction of tax payers to the reform execution. The main setup of the experiment is that subjects perform the encryption real effort task (RET) for 5 minutes. Each word encrypted correctly earns one token. Subjects learn their performance and earnings and then have to declare income (which they can under-report). A tax is collected from the declared income. There is 20\% probability of being audited. If caught under-reporting, participants have to pay a penalty. The tax collected goes to an NGO of the participant’s choice in the post-experimental survey. The experiment lasts for 40 periods, 4 blocks of 10 periods. Each block starts with 5 minutes of the RET, followed by 10 periods of tax-declaration (where the participant has to declare income to then learn whether he would be audited or not. Every treatment starts with a 10% tax rate. In the baseline treatment, the tax reform is abrupt and there is no information. This means that after the second block ends, subjects are informed that the new tax rate will be 40%. In Treatment 1 (Abrupt with information) the increment of the tax rate in announced before. In the gradual treatments, sessions starts with one block with a 10% tax rate, in the second block the tax rate is 20%, in the third it is 30% to end with a block of 40% tax rate. In Treatment 2 the tax reform is gradual and surprising (no information treatment), while in Treatment 3 the tax reform is gradual, but the tax increase is announced before the first RETperiod. The experiment was ran at REBEL with 410 university students. Preliminar analysis of results suggests a positive correlation between tax rate and tax compliance, regardless of whether the reform is implemented gradually or abrupt. At the highest tax rate, there is a treatment effect of information: under the no information treatment, tax compliance is lower than in the information treatments.


Our decisions are often influenced by what those around us do, but we are also inclined to seek out similar others. Research on social influence has focused more on choice behavior and less on process data such as eye movement patterns produced during the deliberation process. However, process data can reveal hidden information about a person’s preferences in situations where choices alone might not be informative. Using eye-tracking in the lab, we plan to study the effect of social influence on altruism. We record participants’ eye movements while inspecting others’ previous choices before making their own giving decisions. We plan to correlate the types of behavioral responses to social information with specific attention patterns and to test whether attention patterns alone can be used to infer giving behavior.


A large political economy literature identifies the source of people’s tolerance of inequality and preferences for redistribution in their beliefs on whether income and wealth are mainly the result of effort and ability rather than luck and connections. In the aggregate, such preferences for redistribution determine the tax rate and the redistributive policies. However, in principle they can also shape people’s willingness to engage in predatory redistribution via appropriation of others' wealth. This papers studies this implication in the lab. In the experiment, income inequality can be the result of differences in ability across individuals with equal opportunities, it may be determined by (negative) income shocks, or may result from a combination of both ability and bad luck. After their place in the income distribution is determined, subjects are allowed to “take" any share of the other subjects’ income. We find that anti-social behavior is not explained by subjects’ relative income per se, but rather by the process that determined their position in the income distribution. Subjects at the bottom of the income distribution engage in differential taking behavior if their endowment was determined by a negative shock in a context of an initially meritocratic income distribution. This is not the case for income distributions determined by either full merit of pure luck. In a second experiment that allows subjects to choose whether they want to be antisocial (take income from others) or prosocial (give to others), we find that most subjects self-select into taking, and that this selection stage makes richer subjects more prone to stealing relative to the poor.