FTC vs. Amazon: If there is no solution (remedy), there is no problem (liability)

The FTC asked a Washington court to split its monopolization case against Amazon (earlier blog post) in two: 1. to determine whether Amazon has a monopoly (liability), and then 2. what to do about it (remedy). 

Their argument (not yet posted) likely would be that bifurcation is more efficient because if FTC loses on liability, there would be no need for a remedy trial.  But if the FTC has to litigate both at once, it would change the trial.  The FTC would have to answer the questions "what should they have done differently?"  and "why is the choice they made bad for competition?"  The burden of answering it would fall on the FTC, and its economic expert. 

For example, suppose that the FTC thinks that the source of Amazon's market power is the integration of its fulfillment network with its electronic marketplace, and proposes a remedy to force divestiture of the two.  An economic witness would be forced to admit that divested companies may not perform as well as the integrated one (See Chapters 22, 23), e.g., by creating a double markup problem, or by foregoing some other economy of integration. 

The Dept of Justice brought (I was Chief Economist then), and then lost, the ATT/TimeWarner vertical merger challenge, in part because it could not overcome this burden of proof.

SOME RELATED ACADEMIC WORK:

• Willem H. Boshoff, Luke M. Froeb, Wihan Marais, Roan J. Minnie, Steven Tschantz. Bargaining Competition and Vertical Mergers: The Problem of Model Selection, Review of Industrial Organization (SSRN). 
• Cooper, James, Luke Froeb, Daniel O'Brien, and Michael Vita, Vertical Antitrust Policy as a Problem of Inference, International Journal of Industrial Organization, 23 (2005) 639–664. (SSRN)
• Comment by Frederick Scherer (2005) 
• Comment by Ralph Winter (2005) 
• Reply by authors (2006) 
• Rejoinder by Ralph Winter (2006)

TRUTH IN BLOGGING: I have done consulting work for Amazon.

Post will be updated as information becomes available.