Federal Trade Commission and Justice Department
February 18, 2004
Gibson, Dunn & Crutcher LLP
The Insight of the 1992 Guidelines
- New insight of 1992 guidelines is that unilateral effects can arise outside monopoly context
- Daimler-Chrysler not the same as Daimler-BMW
- Return to sub-markets?
- Most important change in merger law since 1982 Merger Guidelines
- Synthesis of economic theory and practical judgment about merger investigations
- Profound impact on merger enforcement
What the Guidelines Say
- Focus is on localized competition where “individual sellers compete more directly with those rivals selling closer substitutes”
- Requires “a significant share of sales” to “consumers who regard the products of the merging firms as their first and second choices”
- 35% screen for merged entity’s market share
- Pragmatic tests reflecting quality of analytical tools available in merger investigations
The Lawyer’s Approach
- The lawyer’s approach focuses on defining markets and identifying next best substitutes
- Why? That’s what the Guidelines say
- Analysis driven by interviews and documents
- The economist’s critique: Lawyers use models but don’t articulate them sufficiently
- 35% share screen and next-best substitutes analysis embody assumptions about the effects of cannibalization on merged firm’s incentives
The Economist’s Alternative
- Economists are less tethered to the Guidelines
- Institutional bias against the market definition-driven approach of the Merger Guidelines and case law
- Attempt to get closer to the “real answer” through econometrics or merger simulations
- Models with well-articulated specifications
- Data is to economists what documents are to lawyers
- The jargon of dead Frenchmen
The Role of Simulations
- Modeling and simulations can be useful if sufficiently attuned to market realities and based on defensible assumptions
- Small differences in assumptions can often make big differences in results
- What are we trying to show?
- Magnitude of price increase?
- Whether to get to next step of analysis?
- The risk of false empiricism
Some Issues With Simulations
- Merger simulations are likely to find more problematic mergers than the lawyer’s approach
- Models are designed to predict a price increase
- How does the plaintiff’s expert defend a model that predicts a price increase for safe harbor mergers?
- Economist’s retort: “But this is before entry, repositioning, and efficiencies are considered”
- That still means shifting the burden of proof to the merging parties
More Issues With Simulations
- What happens to next-best substitutes in a world of logit models?
- IIA assumption of identical cross-elasticities of all products with respect to a given product
- More complex models impose great costs and have insatiable appetite for data
- Issues with retail-level data as proxy for wholesale competition
- Assumptions regarding elasticities over relevant range
- Do models follow the Guidelines?
- Source of anticompetitive unilateral effects
- Burden of proof
Another Burden of Merger Review
- Even proponents concede that modeling can be very expensive but may yield little of value
- Costly for parties to get data, clean it up for economists, and analyze it
- Scheffman critique raises valid implementation and theoretical issues
- Reassuring to read Froeb critique of $100,000 rebuttal report
- Merger proponents forced to perform defensive modeling