Introduction

This chapter uses a unique data set to test for the degree and to discriminate among models of tacit collusion. The market investigated is the Vancouver, British Columbia retail gasoline market. The players are service stations who meet daily to compete. Station operators choose price, and the quantity of gasoline sold at each station each day depends on the prices chosen by all.

Oligopolistic interaction is modelled as a repeated game. In an earlier paper (Slade, 1989) I derive a method of producing price wars in price-setting supergames. The method is illustrated with two examples: one where firms use discontinuous strategies in the spirit of Friedman (1971) and the other where strategies are continuous as in Kalai and Stanford (1985).

The present chapter attempts to distinguish between these two supergame price-war models. The object of the chapter is to examine the behaviour of firms during wars. Because many punishment strategies can support the same collusive outcome, it is of interest to determine what strategies firms actually use in a particular situation. To anticipate results, weak support is given to the continuous-reply model, where small deviations lead to small punishments, over its discontinuous alternative.

The exercise is very different from the Porter (1983b) empirical work, where tests are made for the occurrence and timing of price wars.