Cournot Competition with an Uncertain Number of Players : Experimental Evidence

We examine repeated Cournot oligopolies when there is uncertainty about the number of players. Already in the static game, such uncertainty can imply outputs above Nash. For the repeated game, we argue that the uncertainty may lead to a novel strategy, based on a notion of limited depth of reasoning. A sophisticated player produces the output of multiple players, attempting to fool rivals into believing there are more players than is actually the case. Total output increases, but quantities are distributed asymmetrically. An experiment largely confirms the existence of such sophisticated play. Whereas a first supergame is predominantly characterized by standard symmetric Nash play, a second supergame shows marked signs of increased and more asymmetric outputs. Using cluster analyses, we identify groups particularly characterized by sophisticated play and separate them from groups that play Nash or collude tacitly