Firms That Participate Through a Supply Response: Uncommitted Entry

FTC/DOJ Workshop on the
Horizontal Merger Guidelines

Firms That Participate Through a Supply Response:
Uncommitted Entry

Outline of Remarks of
Mark D. Whitener
Antitrust Counsel
General Electric Company

February 18, 2004

1. An important principle in writing or revising antitrust enforcement should be "do no harm"
   1. This is especially true for guidelines that are generally considered to be successful — theoretically sound, and good approximations of actual enforcement policy — which is the case with the Merger Guidelines
   2. Revisions are sometimes justified — to better reflect actual enforcement policies (arguably in order now with respect to the HHI-based presumptions); to expand on existing themes (efficiencies revisions in late 1990s); to address additional issues (vertical mergers?); or to refine/improve the analytics (uncommitted entry?)
   3. Revisions designed to fix one perceived problem can create other problems — if for no other reason than that new verbiage is created, requiring the creation of a "common law" of application and interpretation before there is a consensus about what it means
   4. If the worst thing that can be said of a portion of the Merger Guidelines is that it is seldom used in practice, that does not necessarily mean it needs to be changed — as long as it is fundamentally sound in those instances when it applies
   5. The uncommitted entry section does add value, and there could be some potentially adverse consequences if it were eliminated or folded into a single entry analysis
2. Is the Guidelines' analysis of uncommitted entry ("firms that participate through supply response," Section 1.32) economically unsound?
   1. No one seems to challenge the fundamental economic analysis underlying this section, and in my view it is more economically solid than some other aspects of the Guidelines. The introduction of a serious consideration of sunk costs in the analysis of supply elasticity was a significant contribution to the Guidelines in the early 1990s
   2. Critiques of the uncommitted entry analysis seem to focus on questions of usefulness, given the infrequency in which the necessary conditions for its applicability are met; on the arguably awkward placement of the analysis in the Guidelines' sequential logic; and on questions about the relevance of distinctions between types of supply responses (committed vs. uncommitted; market participants vs. new entrants)
   3. There are some legitimate questions about the analysis in the uncommitted entry section — e.g., whether it is analytically useful to focus on "uncommitted" firms that are likely to enter without the expenditure of significant sunk costs of both entry and exit. It has been pointed out that firms may typically be more likely to affect prices in the relevant market if incumbents view them as not only likely to enter readily in response to a SSNIP but also, having entered, as likely to remain in the market. These are fair points and could warrant a look at some refinements to the Guidelines
3. Is that analysis ever useful as a practical matter?
   1. Not often, for several reasons:
      1. This section's analytical approach is primarily aimed at cases in which it is important to examine the number of firms that are viewed as market participants and the shares assigned to those firms — typically, factors that are important to a coordinated interaction theory. But in such cases, market concentration and shares are (both as articulated in the Guidelines, and in practice) truly only a starting point for the analysis. So the fairly complex factual analysis required to identify and assign shares to uncommitted entrants is not only difficult to do, it usually will not be decisive anyway
      2. As a practical matter, significant sunk costs of entry (at least, and probably of exit as well) are usually present in markets in which antitrust concerns are plausible. This is often due less to impediments to "swing production" (the classic uncommitted entry case) than to other post-production factors noted in the Guidelines: the need to develop reputation and achieve customer acceptance, distribution, etc. So in the types of markets in which a rigorous competitive analysis may be required, uncommitted entry analysis often will have little impact
      3. The uncommitted entry analysis relies on line drawing that is inherently somewhat arbitrary — supply responses must take place within one year, and "significant sunk costs" are those that would not be recouped within a year assuming a SSNIP. Yet line-drawing occurs throughout the Guidelines (two years for new entry; the HHI thresholds; the market share indicators for unilateral effects), and while in their absence the Guidelines might be a more analytically correct statement of economic principles, they would arguably be less effective in accomplishing their intended goal of providing guidance
      4. The line-drawing exercise between committed and uncommitted entry may be of little value in most cases — the entry analysis in Section 3 of the Guidelines may suffice in most instances when supply responses are relevant to the analysis
   2. But the uncommitted entry analysis can be useful
      1. As noted above, if the analysis is applicable in only a limited number of instances, that alone may not be sufficient reason to fundamentally alter or abandon it assuming it is analytically sound
      2. There have been a number of cases in which an assessment of uncommitted entry/supply responses was an important part of the analysis — case involving price discrimination markets; unilateral effects cases where competitor repositioning is relevant; cases in industries such as petrochemicals where "swing" capacity and production is a central feature of the industry; and cases involving the role of imports
      3. Much of the discussion of uncommitted entry has focused on supply responses in a product sense, but the Guidelines make clear that it also to new sources of supply in the geographic market. Cases where the significance attributable to imports is an important factor may lend themselves well to the uncommitted entry analysis
         1. The concept of "hit and run" supply that can be done quickly and with relatively low sunk costs can be particularly useful in examining the role of imports
         2. A fairly typical scenario would involve a relevant antitrust market that is defined regionally from a demand standpoint (e.g. a U.S. market in which U.S. customer requirements are differentiated vs. customers in other parts of the world), and where the regional supply base is concentrated, but where imports can and do "swing" into the region in response to market conditions. In such cases imports will not always register as significant on a current-sales-share basis because recent market conditions have not so warranted
         3. While there are always alternative modes of analysis in which to view the same facts — for example, one could choose to view market shares over a longer time period in order to pick up the existence of imports in the past, or could simply look at importers under the analysis of new entry in Section 3 of the Guidelines — the existing Guidelines analysis of uncommitted entry can be instructive in situations like this
   3. The uncommitted entry analysis is potentially important in unilateral effects cases
      1. The Guidelines' analysis of unilateral effects focuses (Section 2.21) on whether the merging firms are each other's closest competitors for a significant share of sales, and on whether "repositioning of the non-parties' product lines to replace the localized competition lost through the merger [is] unlikely." The timing and likelihood of repositioning responses is to be analyzed using the same methodology as used in analyzing uncommitted entry or committed entry, depending on the significance of the sunk costs involved (fn. 23)
      2. As more transactional data have become available in many industries, and as the agencies have focused more on econometric analysis of those data to discern competitive "closeness" between various market participants' products, a number of unilateral effects cases have been premised on the competitive proximity of the merging firms as revealed by quantitative analysis
      3. But such a threshold finding, which typically is based on relatively static transactional data, requires that we proceed to the next step and examine supply responses — so we need to ask, as the Guidelines do, whether rival sellers are likely to "replace any localized competition lost through the merger by repositioning their product lines"
      4. In these cases, the type of relatively quick, relatively low cost repositioning that the Guidelines envision may be quite apt. By definition, unilateral effects/differentiated products cases turn on differences between certain products/suppliers that do not generalize to all suppliers in the broader market. Repositioning by other suppliers whose products have somewhat different feature sets or other characteristics — making them relatively distant in product space from the merging firms' products — can in some cases occur relatively quickly and without high sunk costs (though the costs of such repositioning, even if not high, will often be sunk)
      5. It would certainly be possible to apply one seamless "entry" analysis in such a case, without distinguishing between committed and uncommitted entry. But the new entry analysis does not appear to fit such a case as well as the uncommitted entry analysis
4. Are there reasons to keep the existing Guidelines analysis of uncommitted entry more or less intact?
   1. The key principle of enforcement guidelines noted at the outset — do no harm — suggests that we should avoid unnecessary changes to existing guidance that is analytically sound and that is useful in at least some cases
   2. As noted above, I believe there are some types of cases in which the uncommitted entry analysis adds value, although there may be aspects of that analysis that could be improved or refined