We analyze how Costa Rican coffee farmer's behavior in an experimental public good game depends on the institutional structure of the farmers buying point (cooperativevs.privately owned mills), and on the background of their game partners (partners selling to the same type of mill or not). We find that cooperative farmers do not display more public good orientation than private market farmers when playing with partners from the same type of mill. However, though farmers selling to private mills make no difference with respect to the background of partners, farmers selling to cooperatives significantly decrease contributions when paired with non-cooperative members. Finally, we study how self-selection into a mechanism that punishes the lowest contributors effects contributions both inside the group and with partners of the opposite background, and we show that it increases contributions by cooperative farmers interacting with non-cooperative farmers by more than 100%.

No abstract

How do informal transfers affect work incentives? The question matters in developing countries, where labor markets are intertwined with transfer networks. The tax-and-subsidy component of transfers would dilute work incentives, but their pro-social element could encourage people to work harder. Such crosscurrents are hard to disentangle because participation in informal networks is likely endogenous. We tackle this problem with a lab-in-the-field experiment that uses a real-effort task. Our main finding is that participants do not reduce their effort in the presence of transfers. This suggests that the impact of informal transfers may extend beyond just the sharing of risk.

We conducted an experimental study in Haiti testing for the relationship between religious belief and individual risk taking behavior. 774 subjects played lotteries in a standard neutral protocol and subsequently with reduced endowments but in the presence of religious images of Catholic, Protestant and Voodoo tradition. Subjects chose between paying to play a lottery with an image of their choice, and saving their money to play with no image. Those who chose the former are defined as image buyers and those who chose the latter as non-buyers. Image buyers, who tend to be less educated, more rural, and to exhibit greater religiosity, bet more than non-buyers in all games. In addition, in the presence of religious images all participants took more risk, and buyers took more risk when playing in the presence of their chosen images than when playing with other images. We develop a theoretical model calibrated with our experimental data to explore the channels through which religious images might affect risk-taking. Our results suggest that the presence of images tends to increase individuals' subjective probability of winning the lottery, and that subjects therefore believe in a god who intervenes actively in the world in response to their requests.

This paper challenges the conventional wisdom that giving voters more power-both formally through the use of more "open" electoral systems and informally through easier access to information on politicians' wrongdoingswill necessarily result in them voting corrupt politicians out of office. Focusing on a comparison between closed-list and open-list proportional representation systems, we theoretically show that opening the lists is likely to generate a large shift of vote shares in favor of the traditional, most corrupt parties. We design a survey experiment to test these predictions in Paraguay and find strong supporting evidence. We do not find in our context that the lack of information is a major obstacle preventing voters from voting out corrupt politicians; if anything, under the more open system, supporters of the incumbent party tend to cast more votes for politicians with a recent history of corruption.