The paper considers the application of the non-parametric test proposed by StateBerry and Briggs (1988) for the model of collusion with imperfect monitoring by Abreu et al (1986) in the context of regional cement markets in Brazil during the 1992-2003 period. The test focuses on the implication of that model according to which the prices would follow a first-order Markov process. The evidence does not indicate the prevalence of optimal collusion in those markets.
Game Theory; Cement Market; PriceWar