Simulating media platform mergers

Résumé

The empirical analysis of media platforms economics has often neglected the multi-homing behaviour of advertisers. Assuming away the cross-substitutability and/or complementarity between the advertising slots of dierent platforms could damage the quality and the robustness of counterfactual analysis. To evaluate the consequence of such an abstraction, we compare the simulation results of hypothetical platform mergers when the demand on the advertising side is derived from a Translog cost model which allows for multi-homing, and when it is approximated by using a simple log-linear inverse demand model that ignores the dierentiation among media platforms' advertising slots. Ignoring the existence of substitutes or complements on the advertising side would result in overpredicting the losses of the viewers' surplus and in underpredicting the gains in platforms' revenues.