Collusion and efficiency hypotheses have been usual explanations given for the positive correlation between profits and concentration at the industry level. In this study, and following Schmalensee (1987), we develop testable predictions of these hypotheses. The paper focuses on the implications of relaxing the assumption of constant returns to scale, given that the observable profitability measure is the margin between price and average variable cost. The results obtained for a sample of manufacturing firms over the period 1990-1993 provide support for an interpretation where both hypotheses are consistent with the data. This evidence illustrates the difficulty of distinguishing between monopoly and Ricardian rents under the traditional "structure-conduct-performance" paradigm.

Keywords: collusion, efficiency, price-cost margin.