Testing static oligopoly models: Conduct and cost in the sugar industry, 1890-1914

We explore the widespread methodology of using demand information to infer market conduct and unobserved cost components under the hypothesis of static oligopoly behavior. Direct measures of marginal cost and conduct, indicating small market power, serve as benchmarks. The more competitive models yield better cost estimates. The best cost estimates occur when conduct is estimated as a free parameter, which in turn only slightly underestimates our direct measure. It also tracks the decline in market power accompanying the industry's structural changes. The methodology is largely validated, although partial cost information can improve its predictive power. Conclusions are robust to the demand function.