Validating the conjectural variation method: the sugar industry, 1890-1914

The Conjectural Variations (CV) methodology uses the responsiveness of price to cost determinants under differing demand conditions to infer market power and cost. It thus substitutes demand information for complete cost information. In this paper we use the American sugar refining industry at the turn of the century to assess the efficacy of the CV approach. We do so by comparing direct measures of marginal cost and price-cost markups with the indirect estimates obtained from the CV method. We find that the CV method performs reasonably well. It yields estimates of industry conduct that are close to the direct measure we derive from full cost information, and robust to the choice of the functional form of demand. The conduct parameter is underestimated, but the deviation is minimal in our context. The CV methodology does a better job of detecting differences in conduct arising from different structural regimes (corresponding to the aftermath of entry), but only when the researcher imposes the (a priori known) restriction of cost stability.