Further tests of static oligopoly models: Whiskey, 1882-1898

Using data from the turn-of-the-century whiskey industry, we conduct tests of the NEIO methodology similar to those conducted by Genesove and Mullin [1998]. Like Genesove and Mullin, we find that the NEIO methodology appears to perform reasonably well for low levels of market power. Conduct is somewhat overestimated, with estimates ranging from 0.17 to 0.35 as compared to direct estimates of 0.09. Cost parameters are generally underestimated. Estimates of conduct and remaining cost parameters improve significantly, however, with additional information on cost. Estimates improve further if conduct is allowed to have two regimes, which are identified based on historical evidence.