My research is motivated by the practical problems faced by organizations delivering human /social services. My published and ongoing work can be divided into three areas: prosocial behavior, information aggregation, and policy experiments in the social services sector.
Abstract: We unpack the psychological influence of a Muslim cleric’s power over the poor in an experiment in Afghanistan. The same cleric requests contributions for a hospital from day laborers when dressed as a civilian and as a cleric. In Civilian condition, 50% contribute and 17% make large contributions; in Cleric condition, 83% contribute but large contributions fall. Through counterfactual simulations, we find that the clerical garb compels unmotivated subjects to contribute (selection), but causes those who initially were generous to reduce their contribution (crowding out). The backlash is present only among those with formal education but is counteracted when the cleric adds a recitation of Qur’anic verses. Overall, this suggests that education mediates whether people automatically associate religious authorities with the omnipresent.
Abstract: An extensive literature on reputation signaling has focused on the desire for positive reputation. In our paper we provide field and lab evidence that some individuals are averse to any form of reputation; this aversion correlates with gender in a prosocial setting. We formalize our hypotheses of these “wallflower” types in a theoretical model. The model predicts that wallflowers will deflect unwanted attention by choosing actions that signal that they are an “average altruism type” relative to their audience. Our laboratory experiment supports these predictions. Online Appendix B, C, D.
Abstract: We study the effect of the social environment on the quantity and quality of voluntary labor contributions. By extending Benabou and Tirole’s (2006) image signaling framework, we derive theoretical predictions on time volunteered given (1) the availability of excuses to stop volunteering and (2) the presence of an authority ﬁgure. We test these predictions in an experiment where laboratory subjects are directly involved in a local nonproﬁt operation.We ﬁnd that in the absence of excuses to stop volunteering, subjects volunteer longer without working less productively. This increase is partially driven by subjects’ reluctance to be the ﬁrst to stop volunteering. The presence of an authority ﬁgure has little impact, but the presence of peers has a positive and signiﬁcant impact.
Abstract: This paper tests the robustness of the “two-tiered labor market” in which efficient bilateral contracts emerge between firms and workers (Brown, Falk and Fehr, 2004). Our experiment introduces stochastic interruptions in firm’s ability to offer contracts. Involuntarily laid off workers are eager to be reemployed; they are unselective about job offers and do not shirk. Firm’s preference for these “temp workers” induces all workers to compete harder to enter relational contracts. Wages in low-tier markets rise dramatically, suggesting that the stigma of unemployment is removed. The results show that interruptions may shorten relational contracts without harming market efficiency.
Abstract: Can globalization change our willingness to redistribute to the poor? We propose the hypothesis that in developing countries, the ‘glitter’ of foreign direct investment (FDI) reduces public support for redistribution by creating perceptions of better employment opportunities for the poor. Initial evidence is derived from World Value Survey responses from developing economies. Delving deeper, a framed field experiment in India reveals foreign ownership of low-skilled firms reduces redistribution to the poor. We further find that rich conservatives drive this reduction. This analysis provides the first experimental evidence of the causal impact of globalization on redistribution, mediated by ideology and income.
Abstract: Does willingness to aid “others” change when in their physical presence? We argue that studies cueing non-coethnics through names and photos may underestimate discrimination resulting from actual interethnic interaction. In an experiment in Kabul, Afghanistan, Dari-speaking day-laborers contribute their earnings to a hospital under one of three randomly-assigned experimental conditions. When Pashtuns are absent, the findings accord with other experiments that find little to no out-group discrimination. However, the physical presence of Pashtuns in the waiting area decreases contributions by 25%. Consistent with the threat hypothesis, contributions decrease the longer Dari-speakers wait with Pashtuns and the Dari-speakers bilingualism in Pashto mediate this negative effect.
Abstract: Double auction prediction markets have proven successful in large-scale applications such as elections and sporting events. Consequently, several large corporations have adopted these markets for smaller-scale internal applications where information may be complex and the number of traders is small. Using laboratory experiments, we test the performance of the double auction in complex environments with few traders and compare it to three alternative mechanisms. When information is complex we ﬁnd that an iterated poll (or Delphi method) outperforms the double auction mechanism. We present ﬁve behavioral observations that may explain why the poll performs better in these settings.
Abstract: This paper shows that the basic unit of information aggregation described by the Geanakoplos and Polemarchakis (1982) posterior revision process does not always produce public statistics that are closer to the full information posterior than the common prior. I study this process of back and forth communication between two individuals with private signals by introducing white noise into payoff computations, defining the evolution of common knowledge, and providing conjectures on the resulting public statistics. I then develop a computational method to ex-ante rank information structures on likelihood of beating the prior. Through a laboratory experiment I find that subjects behave in ways consistent with the noise model: even though initial revisions bring reports closer to each other and to the full information posterior, subsequent revisions have little effect. A comparison of report errors across four information structures provides evidence that the ex-ante ranking matters; more than half of subjects’ reports are further from the full information posterior than the common prior in the two lowest ranked structures.
Abstract: Small-scale political violence is not well-understood, despite its insidious effect on inter-group relations and potential escalation into national-level politics and economics. Some regard it as unanticipated, but others argue that this is due to lack of information by outside observers. This paper provides the first quantitative evidence that low-level violent incidents are preceded by widespread changes in local behavior. By combining network traffic from Orange Telecom and peacekeeping records from the United Nations Operations in Côte d’Ivoire, we find an increase in local mobile phone usage driven by a higher than normal number of callers, conversations between phones that share the same provider, and activity between geographically closer antennas. This unique communication pattern varies with the intensity of violence and is distinct from the anticipation of positive events, such as football matches and festivals. Our findings are consistent with psychological research on changes in communication networks in stressful situations and suggest that some of the mechanisms necessary for rumor emergence are in place.
POVERTY AND SOCIAL SERVICE UTILIZATION
Abstract: This paper describes a randomized field experiment testing the impact of a savings competition on the behavior of homeless individuals staying at a transitional shelter. When monetary prizes were offered for achieving the highest saving rates within a particular month, average savings increased by $80 (a 30% increase) while income and attendance at case management meetings remained unchanged. However, repeating the competition in the following month had no effect. This is because individuals who increased their savings rate in response to the competition left the shelter after the first month. In summary, while a savings competition can indeed increase average savings in the short run, its effect may be limited to the intensive margin and may diminish with repetition.
Abstract: Self-reported survey measures of self-control continue to appeal to practitioners for ease of implementation, necessitating further investigation of how well they predict outcomes. We show theoretically and empirically how variation in self control survey responses in some populations may be driven by awareness of self-control problems. Previous empirical evidence, derived from highly successful populations, has shown that larger self-reported self control problems is correlated with worse outcomes. We complement this with evidence from the homeless population: here larger self-reported self control problems is positively correlated to larger savings in shelter lockboxes, a commitment savings device.