An Econometric Workbench for Comparing the Substantive and Dominance Tests in Horizontal Merger Analysis

We investigate for the individual and relative accuracy of two major market power tests: the Herfindahl Hirschman Index (HHI), and the merger simulation model of unilateral effects. We propose a methodology to compare the predictive and screening power of both tests by implementing them in a hypothetical economy. We find that when a structural econometric approach is missed (substantive test) the concentration approach (dominance test) is biased upwards. We also find that the substantive test is influenced by the market size or more precisely the market size of the outside good. In particular, when it is large, a decision based on the HHI test would have nothing to do with a decision based on the unilateral effects test. Still, in the overall, the substantive test performs better in capturing the true situation of the market compared to the dominance test of concentration.