Sonia Jaffe, Glen Weyl, Mar 14, 2011
The innovations of the new U.K. and U.S. merger guidelines released last year have excited many economists. On the one hand, they apply in nearly as broad a range of merger contexts as traditional market definition and HHI-based approaches do. On the other hand, they incorporate the explicit economic grounding in the logic of differentiated products competition enjoyed by merger simulation. Furthermore, the core intuition behind the guidelines, as clearly exposited by Farrell and Shapiro, is simple and transparent. It can be explained as follows:
If Crest merges with Colgate, Crest must consider that every time it sells a tube of toothpaste there is a partial tube of Colgate that will go unsold as a result of an additional sale of Crest. Thus, post-merger the mark-up that would have been earned on the unsold partial tube is a new opportunity cost of selling Crest. This will encourage Crest to raise its price(s).
This logic indicates that in reviewing a hypothetical merger both the diversion ratios between the products (e.g., the fraction of a tube of Colgate lost when an extra tube of Crest is sold) and the firms’ mark-ups over marginal cost are important. The product of the diversion ratio and the mark-up is referred to as the “Upward Pricing Pressure” (UPP).
Despite the clear intuition behind UPP, a number of objections have been raised against its use in merger analysis. First, Coate and Simons argued that, unlike market definition, traditional UPP-based approaches rely on the assumptions that firms have constant marginal costs and take other firms’ prices as given in a static Nash-in-prices (differentiated Bertrand) equilibrium. Second, Schmalensee, Hausman et al., and Carlton have argued that the UPP approach can predict only the direction, rather than the magnitude, of price changes, and that even directional predictions require assumptions about “default efficiencies” to avoid flagging every horizontal merger as anticompetitive. Finally, Carlton pointed out that even if quantitative predictions could be obtained, the aggregation of these predictions into a single welfare number may be difficult, especially when not all effects on consumer welfare come directly from changes in prices. While some of these concerns apply to nearly all alternative approaches, they still offer opportunities for improvement.
