Implications of Game Triads for Observations of Trust and Reciprocity

This paper develops a triadic design for conducting trust and reciprocity experiments. A large literature on single-game trust and reciprocity experiments is based on the assumption that subjects’ utility payoffs are the same as their own monetary payoffs in the experiments. Such designs test compound hypotheses that include the hypothesis that other-regarding preferences do not affect behavior. In contrast, experiments with the triadic design do discriminate between transfers resulting from trust or reciprocity and transfers resulting from other-regarding preferences. Decomposing trust from altruism and reciprocity from altruism is critical to obtaining empirical information that can guide the process of constructing models that can increase the empirical validity of game theory.