Herbert Hovenkamp on Revising the Merger Guidelines

Herbert Hovencamp —  27 October 2009
  1. Yes, the Merger Guidelines should be revised; in particular:

a.  The discussion of concentration thresholds for collusion facilitating mergers must be aligned more closely with both recent case law and actual enforcement practices; otherwise they fail to provide guidance.  The current Guidelines indicate that concentrations greater than 1800 HHI and a post-merger increase exceeding 100HHI presumptively indicates a challenge. In fact, mergers with post-merger HHIs in excess of both these numbers are routinely permitted. While the standards in the current Guidelines are too aggressive, the George W. Bush administration policy was too lenient.  More fundamentally, the HHI creates an illusion of precision in coordinated effects analysis that is simply not warranted, particularly not when market definitions are ambiguous or when the merger market is subject to product differentiation.  Further, the “other factors” portion of the Guidelines tends to dominate the analysis.  A better approach is reduced reliance on the HHI and more on simpler observations about who the 3 or 4 largest firms in the market are, the effects of eliminating the acquired firm as an independent market entity, and the likely responses of rivals to an output reduction by the post-merger firm.

b.  The unilateral effects analysis requires elaboration that brings it into line with actual agency analysis.  Some of the economists would like to jettison a market definition approach and attempt to measure price increase potential more directly.  While I am sympathetic with that, there is too much water over the bridge; it could lead to courts’ simply rejecting the Guidelines.  A better approach is to write market definition criteria that permit narrower market definitions in situations where a unilateral price increase is likely.  This requires that some thought be given to situations where the merger is between one firm and its second closest rather than its closest rival.  Right now it is simply too easy for defendants to argue that the closer rival will steal sales and competition will not be affected.  So the government is currently losing cases that it should not be losing.  Finally,the Guidelines should quell discontent in lawyers’ perception about unilateral effects analysis that it is more speculative or less rigorous than traditional coordinated effects analysis.  The fact is that both types of analysis rely on many assumptions, some of which are no more than conjectures about future firm behavior.  If anything, the robustness edge belongs to the unilateral effects approach.

2 responses to Herbert Hovenkamp on Revising the Merger Guidelines

  1. 

    Mandelbrot explains better than I could why the alternative of concentration ratios is not a step forward.

  2. 

    More fundamentally, the HHI creates an illusion of precision in coordinated effects analysis that is simply not warranted, particularly not when market definitions are ambiguous or when the merger market is subject to product differentiation.

    Right on! But your suggested alternative (“simpler observations about who the 3 or 4 largest firms in the market are, the effects of eliminating the acquired firm as an independent market entity, and the likely responses of rivals to an output reduction by the post-merger firm”) fails to provide a workable quantitative alternative to HHI. Those are the right qualitative observations, but without any quantitative metrics, we can argue endlessly about whether the relative magnitude of competing qualitative factors cancels one way or the other.

    A quantitative metric is needed for characterizing competition. HHI is bad, but better than nothing.

    On my view, the core problem with HHI is a problem HHI shares with most modern institutional ruels of accounting — namely, the assumption that distributions of outcomes for particular firms or markets will be stationary. In practice, nearly any actual distribution of outcomes will not be stationary over different time-windows or scales.