Merger Analysis with Sufficient Statistics

Organised by

• Department of Economics

We develop a sufficient-statistics framework for merger analysis allowing for multiproduct firms and arbitrary firm and product heterogeneity. The framework satisfies three desiderata: The multiproduct-firm pricing game is aggregative; the aggregator is also a sufficient statistic for consumer surplus; and each firm's product portfolio can be summarised into a uni-dimensional sufficient statistic, the firm's type. We show that those desiderata are satisfied if and only if the demand system is of the generalised CES (GCES) or generalized multinomial logit (GMNL) type. Relative to standard CES and MNL demand, our generalisation implies that prices, locally, can be strategic complements or substitutes, depending on the local behaviour of the curvature of indirect utility. In turn, this distinction is shown to determine whether competition authorities should be more or less strict on mergers in more competitive industries. We obtain further results on the static and dynamic consumer surplus effects of mergers, the aggregate surplus and external effects of mergers, and on the relationship between the market power effect of a merger and the merger-induced change in the Herfindahl index.