Abstract:Considerable experimental evidence has been collected on how to solve the public-good dilemma. In a 'first generation' of experiments, this was done by presenting subjects with a pre-specified game out of a huge variety of rules. A 'second generation' of experiments introduced subjects to two different environments and had subjects choose between those. The present study is part of a 'third generation', asking subjects not only to choose between two environments but to design their own rule sets for the public-good problem. Whereas preceding 'third-generation' experiments had subjects design and improve their strategies for a specified game, this study is the first to make an attempt at answering the question of how people would shape their environment to solve the publicgood dilemma were they given full discretion over the rules of the game. We explore this question of endogenous institution design in an iterated design-and-play procedure. We observe a strong usage of punishment and redistribution components, which diminishes over time. Instead, subjects successfully contextualize the situation. Interestingly, feedback on fellow-players' individual behavior tends to be rendered opaque. On average, rules do improve with respect to the welfare they elicit, albeit only to a limited degree.