Oligopolistic "Agreement" and/or "Superiority"?: New Findings from New Methodologies and Data

The influential Scherer and Ross text (1990, p. 411) states that the "main question" in empirical industrial organization in the latter part of the twentieth century is Bain's (1951) "collusion" or "agreement" hypothesis versus Demsetz's (1973) "superior firm" hypothesis. Prior to the Federal Trade Commission Line-of-Business (LOB) studies the "contending schools were deadlocked," but these studies led to a win being declared for the superiority hypothesis by Scherer writing with seven other LOB researchers (1987). These studies found that the effect of concentration on profits disappeared when controlling for firm shares. As many economists agreed, merger policy shifted away from a focus on agreement to applying a "unilateral effects" (non-cooperative Nash) approach. We develop a nine year panel LOB data set for Korea. We perform three types of tests, all of which support both hypotheses, but which show that the agreement effect overwhelmingly dominates the superiority effect in pricing. First we examine a secondary implication of the superiority model: profit aggregation should imply that if share is negatively related to firm profits, so should concentration be negatively related to industry profits. Instead, we find that for those industries with a negative share relationship, the concentration profits relationship is positive and virtually identical to the relationship for the full sample in both within and between panel tests. Next we introduce a commonly cited model in the empirical literature. This model is cited to motivate the proposition that both share and concentration should have an effect on firm profits. However, authors who cite this model then typically use an ad hoc specification rather than estimating this as a structural model. We develop our structural model and define latent variables to distinguish between domestic and export price cost margins (PCMs) and to identify firm "conjectures" as they impact the domestic PCM. Demand elasticities are captured in non-linear industry fixed effects. We show that concentration plays an overwhelming role in determining firm PCMs, with firm share playing a far smaller role. We additionally exploit the structural characteristics of the model to deal with the possibility that deviations between marginal costs and average costs might be driving the results. For supporting evidence we construct a new latent variable identifying the domestic/export price ratio. We find a strong "within" relationship between concentration and the domestic/export price ratio, again firm shares play a weaker role. Finally, we discuss why our results differ from the FTC-LOB studies and provide evidence that would suggest that the FTC studies' conclusions are biased due to the 1973 removal of price controls and energy crisis, the "stagflation" of the 1970s, and the use of national firm shares along with geographically weighted averages of concentration ratios.