ONLINE APPENDIX Collusion through Communication in Auctions

In this Online Appendix, we conduct additional analysis of the data and robustness checks as well as provide some theoretical analysis. We proceed as follows. At the end of Section 1, we describe the type of statistical analysis we perform to compare treatments and behaviors in different parts of the experiment. This section also contains definitions we use in the Online Appendix. In Section 2, we present additional analysis of the Pure Communication treatment. In Section 3, we present additional analysis of the Communication and Transfers treatment. In Section 4, we consider the choice of “small perturbation” that was fixed at two experimental units for the results reported in the paper. We show the robustness of our results to alternative levels of perturbation. In Section 5, we illustrate that the risk measure elicited in some sessions does not seem to explain subjects’ behavior either during communication or while bidding. In Section 6, we discuss learning effects and compare outcomes and behavior in the first and second halves of the experiment. In Section 7, we present analysis of additional experimental sessions that we conducted with complete strangers matching protocol. Finally, in Section 8, we provide some theoretical analysis of one-shot independent private value sealed-bid auctions corresponding to our different treatments.