اعلام رفتار اجتماعی و اقدامات آشکار: شواهد از شش کشور آمریکای لاتین

Do attitudinal surveys and incentivized experiments predict actual behavior? We answer this question using data on trust and pro-sociality from experiments and surveys conducted on six Latin American cities. Individuals in agreement with a set of pro-social statements who also either are willing to trust others more or are interested in risk-pooling, end up investing more in maintaining their social capital in the form of social organizations such as charities, religion, politics, sports and culture. Both, experiments and surveys carry useful information to understand motivations and intentions in pro-social behavior and social capital formation.

Whereas several of the findings derived from laboratory economic experiments have become widely accepted, it is unclear the extent to which they are linked to the responses that individuals provide in surveys, especially in regard to their predictive power about real life situations. In this paper we study the link between people's stated preferences regarding pro-sociality, their actions and participation in pro-social activities, and their corresponding actions when exposed to laboratory experiments on the same issues with the aim to test whether surveys and experiments carry useful complementary information to understand what people actually do. To our knowledge, the complementarity of surveys and experiments with representative samples has not yet been broached in the literature. Typically, the experimental literature has placed great emphasis on design but less so on sampling issues. By contrast, household and individual surveys measuring attitudes and preferences have placed considerable focus on sample representativeness, but the credibility of the responses is frequently put in doubt due to the hypothetical nature of the questions and the potential sensitivity to different biases. In this paper we combine the virtues of both tools to explore the potential of their complementarities for economic analysis.1 In this paper, we collect data for a sample of 3100 individuals from different backgrounds, socio-economic levels, age cohorts, and both sexes, from six Latin American capital cities (Bogotá, Buenos Aires, Caracas, Lima, Montevideo and San Jose) and have the individuals participate in three well known and commonly applied incentivized economic experiments, namely (i) a trust game, to measure the basis on which social capital is built, (ii) a voluntary contribution game, in order to capture willingness to cooperate; and (iii) a risk sharing game, to measure propensity to form an income pooling group when facing uncertain outcomes.2 On the other hand, we capture information on participants' attitudes towards pro-social behavior using a survey about stated preferences on pro-sociality. Furthermore, actual behavior is measured by asking participants whether they are involved in social organizations, as well as their degree of attendance, dedication, and involvement in the decision making process. The latter allows us to go beyond the mere counting of membership, providing a more complete picture of what people do when building their social capital. We use experimental behavior as a predictor of actions taken for building and maintaining social capital, including stated preferences regarding pro-sociality. Such empirical strategy, as discussed in Carter and Castillo (2011) and Cárdenas and Carpenter (2005), may provide insights about the complementarities of experimental methods with surveys as they can increase the power of explaining variation in the data collected in surveys, and help solve some of the endogeneity problems that remain in the social capital literature when trying to link economic outcomes and social capital survey questions. Our findings suggest that trusting in our incentivized experiments predicts individuals' participation in building and sustaining social capital through membership, attendance and volunteering in social organizations, along with stated preferences regarding pro-sociality. On the other hand, we find that trustworthiness (the responder's behavior in the trust game), risk sharing and voluntary contributions do not explain variation regarding building and sustaining social capital, although stated preferences keep having predictor power. All in all, the results presented here are telling on the complementarities of experiments and surveys to enhance the value of surveys in development studies and, most importantly, how they help to predict actual behavior. The paper is organized as follows. The next section describes the sample and the experimental design. Section 3 describes the methodological approach to measure the link between what people say and what people do. Section 4 presents our main findings, including robustness checks for our variable of interest. Finally, Section 5 concludes.

Household and individual surveys measuring attitudes and preferences have placed great emphasis on sample representation but not so much on the actual content of the responses as they typically involve only hypothetical situations. On the other hand, the experimental literature has placed great emphasis on design, in order to obtain more credible answers to the questions on preferences, but less so on sampling issues. In this paper we provide an attempt to combine the positive treats of both approaches. We apply well-known experimental activities and survey questions on pro-social preferences and attitudes to large samples for six Latin American cities. We test whether both sources of information act as complementary or substitutes on the prediction of actual pro-social behavior. Experiments provide additional information to what people state on surveys, in some cases having greater explanatory power that what people say they do which reinforces the idea that using multiple experiments might be a productive strategy to capture behavior and preferences. In fact, we find that statements and actions act as complementary predictors of social capital formation, with some variation depending on the experimental measure and pro-social indicator. In particular, we find that trusting in the experiments and being willing to share risk with others within a group are associated with individuals' attitudes towards social capital. Stated pro-sociality also shows statistically significant correlation with social capital in most of the specifications. Finally, risk-pooling also serves as a good predictor of these pro-social attitudes and actions.18