Horizontal Merger Guidelines Review Project: Proposed New Horizontal Merger Guidelines #548050-00004

Submission Number:
Robert Steiner
Initiative Name:
Horizontal Merger Guidelines Review Project: Proposed New Horizontal Merger Guidelines
REPLACE SINGLE-STAGE PROCEDURES WITH A MULTI-STAGE ANALYSIS In consumer goods industries changes in costs and elasticities at one stage profoundly affect results at adjacent stages and are in turn affected by them. Imperfect competition typically prevails at all stages, and firms compete both horizontally and vertically. Unfortunately, the new HMGs, like their predecessors, frequently analyze this complex horizontal/vertical system in a single-stage fashion – a procedure that can block welfare enhancing mergers and, less frequently, approve anticompetitive ones. In the standard UPP analysis of a merger between 2 large consumer goods manufacturers whose brands are the closest substitutes for each other, the investigation proceeds at the manufacturer level. Pre-merger if one producer raised price materially, it lost substantial sales to its competitor. Post merger, these diversions are internalized, so the merged manufacturing firm gains market power. After allowing for internal efficiencies, the Agencies may estimate that the merged firm will raise price by 10% and reduce output. But this is not an equilibrium. Looking upstream, the merged firm should be able to reduce its invoice costs and increase its margin. It can place larger orders with its suppliers who can no longer play off the pre-merger firms against each other. The larger margin will be used to raise factory prices and to strengthen its brand franchises through larger outlays on advertising and R&D following the Dorfman-Steiner theorem which shows how to jointly optimize for price and advertising. By consolidation and increased advertising, the merged firm’s brands will become more popular, and their factory prices will be raised. The classic single-stage error is to assume that their retail prices have increased by the same amount, e.g. that the retailing stage is perfectly competitive. However, retailers face downward sloping demand curves and bargain with manufacturers over price and terms. A KEY REGULARITY in consumer goods industries, of which the Agencies appear unaware, is the inverse association between the margins of manufacturers and retailers on both highly demanded manufacturers’ brands and on weak manufacturers’ brands and store brands. (See Steiner, Lynch attachments for empirical evidence and analysis). Makers of the most popular brands (such as Tide) enjoy very wide margins, while their retailers have thin ones. With weak brands the margin relationships are reversed. Famous brands have high manufacturers’ margins because they have successfully differentiated themselves from rival brands. Yet they have the tiniest retail gross margins because they are the least differentiated and the most homogeneous in the downstream intrabrand market. Consumers quickly recognize that the Tide sold at different stores is the same item. Retailers fear to be caught with a higher price on what consumers recognize is the same thing. They also believe that consumers judge whether the store is a high or low price vendor by its relative prices on the few popular brands whose prices across stores are familiar to them. Therefore, the retail prices of the merged firm’s brands will rise by less than their factory prices. Indeed, where the brands’ popularity has greatly increased, there are instances in which the fall in their $ retail gross margin exceeded the $ rise in their factory prices, so their retail prices fell despite their increased factory prices. Interbrand competition then forced down the retail prices of rival brands, raising output and consumer welfare. See Steiner attached. However, should the merger simply change the firm’s costs without materially changing its market power, we get the opposite result. Retail prices will rise or fall further than factory prices because % retail gross margins will not significantly change. Again, outcomes can’t be predicted by single-stage models. An informed multi-stage analysis is required.