Essays on public goods and social learning

This dissertation consists of three independent chapters. The first chapter is co-authored with Professor Doctor Mehmet Yiğit Gürdal and Professor Doctor Özgür Gürerk. The second chapter is co-authored with Professor Doctor Mehmet Yiğit Gürdal and Associate Professor Doctor Orhan Torul. The third chapter is solo-authored. The first chapter investigates the impact of culture on the acceptance and prevalence of sanctioning institutions, and report the results of controlled lab experiments in two countries with substantial cultural differences: Germany and Turkey. We find, if a sanctioning institution is one of two alternatives that individuals can freely choose in an endogenous setting, then it is the clear winner against a non-sanctioning institution, both in Germany and Turkey. Though there are some differences in initial institutional preferences and contributions in both countries, the dynamics of institution choice, the evolution of contributions and sanctioning behavior are remarkably similar. Our results extend the findings of a previous study by Herrmann et al. [2008] that finds less cooperative behaviors in some countries (among them Turkey) if the sanctioning institutions are exogenous. We show that in one of those countries, Turkey, in an endogenous setting, sanctioning institutions beneficial to societies are adapted and prevail. The second chapter examines the cooperation and punishment in a public goods game in Istanbul. Unlike prior within-subject designs, we implement a between-subject design with separate no-punishment and punishment conditions. This approach reveals that punishment significantly increases contributions, demonstrating the detrimental effect of prior experience without sanctions. We highlight two critical factors - heterogeneous initial contributions across groups and estimated contribution updating rules based on received sanctions. An agent-based model verifies that their interaction produces strong persistence. Analyzing related data shows similar patterns emerge in comparable cities, suggesting our findings likely generalize if employing a between-subject design. We conclude that overlooking within-group heterogeneity biases cross-society comparisons and subsequent policy implications. The third chapter analyzes how the structure of the network and the relevancy of the object of interest affects the decisions of the subjects in a social learning experiment. In their most classical form, social learning experiments include a series of subjects lined up in a linear network. In such a design, each subject receives a private signal to predict the state of the world sequentially. The novelty of the chapter stems from its circular line-up design such that the first mover has to make two decisions. Initially, the first mover makes a decision based on her own private signal and another party subsequently makes a decision. Next, the first mover observes the decision made by the other party in its group and updates its prediction. We ask subjects to predict a randomly chosen target number and find that when the predicted item holds no self relevance for any players in the network, we lack sufficient evidence to suggest that the structure of the line-up influences outcomes. We also ask subjects to predict their own and another party's rank in a Raven Progressive Matrices test [Raven et al., 2003]. When the predicted item holds self-relevance for at least one party in the network, we observe a bias in predictions. In particular, the initial assessment of the subjects reveals that there is an overconfidence with respect to their performance in the test. Even though the bias of the predictions decreases compared to their initial beliefs after receiving an information about their rank, they still overestimate their rank. While subjects overestimate their rank more compared to other party's rank with respect to the earlier predictions, the difference between predicting one's own rank and another party's rank vanishes with later predictions.