Renegotiation in Repeated Oligopoly Interaction

Standard repeated-interaction theories of oligopoly make collusion seem much easier than a \structural consensus" suggests that it is. I show that more intuitive results can emerge if colluders could renegotiate after a deviation. In repeated Bertrand oligopoly, if agreements are subject to a certain kind of frictionless renegotiation, then full collusion is impossible with more than three rms, however high the discount factor. With more rms, partial collusion is possible, but its deadweight losses are small compared to monopoly e ects. I also analyze results for repeated Cournot interaction. Finally, I try to relate the game-theoretic literature, the structural consensus, and the concern over collusion. Forthcoming in the Festschrift in honour of James A. Mirrlees, edited by G. Myles and P. Hammond: Oxford University Press, 2000. y Professor of Economics and Director of the Competition Policy Center, University of California, Berkeley, CA 94720-3880. email: [email protected]. I thank Luis Cabral, Dennis Carlton, Andrew Dick, Richard Gilbert, Matthew Rabin, and Carl Shapiro for helpful comments, Rene Kamita for research assistance, and the National Science Foundation for early nancial support (Grant SES{9111095).