One approach to merger simulations used in antitrust cases is to calibrate demand from market shares and a few additional parameters. When the products involved in the merger case are differentiated along several dimensions, actual diversion ratios may be very different from those calculated from market shares. This again may affect the predicted post-merger price effects. This article shows how merger simulation can be performed using observed diversion ratios. To illustrate the potential effects of this approach we use diversion ratios from a local grocery market in Norway. In this case diversions from the acquired to the acquiring stores were considerably smaller than suggested by market shares, and the predicted average price increase from the acquisition was 40% lower using this model rather than a model based upon market shares. This analysis also suggests that even a subset of observed diversion ratios may significantly change the prediction from a merger simulation based upon market shares.
