Sample-selection procedures for estimating the relationship between concentration and profitability from cross-industry data

Abstract

Aggregation of homogeneous product markets into industry classifications or disequilibrium in terms of differences between expected and actual market shares can lead to bias in the estimated relationship between concentration and profitability from regressions using cross-industry data. The relationship between concentration and profitability over the period from 1965 to 1980 is estimated for various samples of Canadian manufacturing industries. Regressions for the full sample of industries lead to rejection of the hypotheses from either structure-conduct-performance (structuralist) analysis or a model of oligopolistic equilibrium. However, regressions for subsamples of industries selected to remove industries subject to aggregation provide results consistent with the structuralist hypothesis, while regressions for industries selected to remove industries subject to both aggregation and disequilibrium provide results consistent with the oligopolistic equilibrium hypothesis.