Do Multiple Contacts Matter?

Abstract

We experimentally investigate how simultaneous interaction in multiple strategic settings influences the collusive behavior of agents. We utilize theory developed by Bernheim and Whinston (1990), which studies multi-market contact between price-setting firms, to derive the conditions under which agents can support collusion in symmetric and asymmetric environments (i.e., in the symmetric environment, every individual receives the same payoffs, in the asymmetric environment, one individual receives higher payoffs than the other). Theoretically, under both symmetry and asymmetry of payoffs, multiple contacts can facilitate collusion. We test our theoretical prediction experimentally. We find no significant increase in collusion due to multiple interactions when analyzing the average actions individuals chose. While we do not see a significant difference in average actions, theory and our analysis of the experimental data suggest that the effects of multiple contacts can be uncovered by studying the strategies that generated the observed actions as different strategies can lead to similar levels of collusion. Consequently, we estimate the underlying strategies that subjects use. To this end, we extend popular strategies (e.g., Grim Trigger, Tit-for-Tat, etc.) to condition on the history observed in multiple strategic settings. Our strategy estimation results show that only for asymmetric payoffs subjects use these new strategies