Joshua Gans has replied to my comments on the ACCC draft merger guidelines on his blog site. While my comments touched on several aspects of the draft guidelines, Gans only deals with one, which is the guidelines’ emphasis on vertical mergers. In my comments, I suggested that this emphasis was difficult to understand, as vertical mergers rarely raise competition issues (and more often promote efficiency).

Gans makes two points in this respect. The first is to say that the study I cite – a study which surveys an extensive literature on vertical mergers and finds strong evidence that these mergers generally increase efficiency and benefit consumers – is unreliable, as it is likely that harmful vertical mergers are prevented under the competition laws. The second is to then cite a case which he believes shows that vertical mergers can be harmful, namely AGL’s acquisition of a partial interest in the Loy Yang power station.

There are obvious problems with this chain of reasoning, in that the second assertion seems to contradict the first. In effect, his reasoning in respect of the AGL case is plainly that the Federal Court, which allowed that merger to proceed, erred: rather than allowing the merger to proceed, it should have prevented it. But his first claim rests on the assertion that anti-competitive vertical mergers are not observed in the data set because the authorities would have blocked them.

But if Australian courts can err, so too can courts and competition authorities elsewhere. As a result, were it indeed the case that vertical mergers are a substantial cause for concern, surely some harmful mergers would have found their way into the data set. This is all the more the case, as in the US, it was general policy from the Reagan years up to the Clinton administration for purely vertical mergers to be approved, and vertical mergers, including some that arguably had significant horizontal aspects, were almost never challenged. Going on Gans’ claim, one would therefore expect several (if not many) instances of such harmful mergers to have occurred. In the event, they do not appear in the most comprehensive data set yet assembled. Rather, as the authors of the data set say, the evidence is “overwhelming” that vertical mergers generally benefit consumers and efficiency. Indeed, the authors were surprised at just how strong the evidence was.
 
I am consequently confident of the conclusion I drew (and which was also that of the ACCC’s earlier merger guidelines): that while it is certainly not impossible for a vertical merger to be harmful, that is rarely the case in practice – so the focus of policy, and of the guidelines, should be squarely on horizontal mergers.

As I have noted, it is not impossible for a vertical merger to be harmful. Gans suggests that AGL’s acquisition of a minority stake in the Loy Yang power station was such a case, and should have been blocked. I found this claim completely implausible at the time of the merger, and my view has not changed. The ACCC argued that the merger (which involved an aggregation of less than 2 per cent of the market) would result in a 20 to 25 per cent increase in prices; that claim was based on modelling which assumed AGL would have perfect foresight and would be able to use that perfect foresight in its bidding. All the evidence told against that assumption. The Federal Court found that while the ACCC’s modelling was technically impressive, it was of no obvious relevance to the real world. So it was then, and so it remains.

Going back to the draft guidelines, for those readers who have not had a chance to work through them but did read my earlier note, they use the SSNIP test both for market definition and for determining whether a substantial lessening of competition (an “SLC”) is likely to occur. I have no problems with their use for both of these purposes. However, if you use a price elevation test to determine whether a merger is harmful (in other words, if the test is whether the merger will increase prices materially relative to a counterfactual), then you must take account of any merger-specific efficiencies. The draft guidelines try to have their cake and eat it too, by using a price-elevation test to determine whether competition is being harmed but excluding from the test full consideration of the efficiencies which the merger will cause. This makes no sense and, in any event, is a test that would be impossible to implement.

As for market definition, the ACCC’s use of the SSNIP test merely continues the approach adopted in the earlier guidelines. However, the ACCC now says (at 3.3) that despite the fact of having defined a relevant market in this way, there can be:

“no presumption that other firms within a relevant market necessarily provide an effective constraint on the merged firm.”

This is, of course, wrong: the whole point of using the SSNIP test to define the relevant market is that it does create the presumption that the market so defined comprises the firms that impose a competitive constraint on the merged firm. But that is a presumption (i.e. a starting point or supposition based on reasonable evidence), rather than a settled conclusion: and like other presumptions, it can be rebutted, and rejected, where the facts warrant so doing. For example, even in a homogenous goods market, a firm within the relevant market may be capacity constrained, and hence unable to expand output should the merged entity seek to increase prices. 

This is the sort of error a bit of peer review prior to issuing the draft might well have picked up; and in the great scale of things, it is small change. But it highlights and is symptomatic of an underlying issue that is of importance. That is the issue of what the guidelines should do. There are, in my view, two inter-related goals they should pursue.

First, they should help clarify the analytical approach the Commission will adopt in assessing mergers. This is what the draft seeks to do, and though one can argue about whether the Commission’s proposed approach is fully thought out, the draft guidelines do a useful job of stating it.

Second, they should seek to provide guidance to firms involved in the merger process in a way that helps avoid unnecessary regulatory burdens. The fact of the matter is that regulatory processes, including merger review, are costly and imperfect. As they cannot achieve first best solutions, society is better off if cases where it is unlikely, albeit not impossible, that harm will be caused are excluded from the regulatory net: on balance, the social savings from so doing will outweigh the likely costs. Moreover, even when cases are brought within the review process, it is desirable for there to be presumptions which structure the process, allowing participants to prepare their case in a timely and effective manner.

It is for this reason that the earlier guidelines had significant safe-harbours and presumptions; the proposed new guidelines entirely remove them. Should that remain the case in the guidelines’ final version, we will be the only major jurisdiction in the OECD area which does not rely on safe-harbours and presumptions in the administration of its merger laws. Now, one can of course argue about how safe-harbours and presumptions should be set and about what impact they should have; but to abolish them altogether would require some explaining. The ACCC has given no such explanation. It would be desirable for it to do so.

Henry Ergas
26 February 2008