Collusion among many firms: The disciplinary power of targeted punishment

Abstract

a b s t r a c tWe explore targeted punishment as an explanation for collusion among many firms. We run a series of Cournot oligopoly experiments with and without the possibility of targeting punishment at specific market participants. In markets with two, four, six, and eight firms, we analyze to what extent targeted punishment helps firms to restrict output. We find that targeted punishment leads to more collusion across all markets. Furthermore, beyond two firms, this collusive effect turns out to be even stronger in markets with more competitors, suggesting a reversal of the conventional wisdom that collusion is easier with fewer firms. for valuable comments and discussions. We would also like to thank participants at various conferences and seminars. Financial support from the Richard Büchner-Stiftung is gratefully acknowledged. The usual disclaimer applies.

Number Effects and Tacit Collusion in Experimental Oligopolies

“…In other words, a deviation from a collusive agreement inevitably punishes all competitors in markets with more than two firms. Yet, Roux and Thöni [2015] find that the possibility of punishing individual rivals facilitates tacit collusion in larger oligopolies. Finally, as in Huck et al [2004] and Bosch-Domènech and Vriend [2003] we normalized equilibrium payoffs across treatments in order to be able to isolate the impact of number effects on tacit collusion.…”

Information-sensitive Leviathans

“…It is, however, unclear whether this means that punishment is more severe in larger groups, because larger groups might be faced with more deviators, or subjects might be more likely to act as a bystander on the punishment stage. For a discussion of the adaptation of the punishment mechanism to various group sizes see also Roux and Thöni (2015). punishment point assigned to a citizen leads to a deduction of three units from the punished citizen's payoff and costs one unit.…”

Do control questions influence behavior in experiments?

“…Cournot game with punishment. The results are discussed in Roux and Thöni (2013). Insignificant results must, of course, be interpreted with great caution.…”