Is the Intel/AMD Settlement Illegal?

Cite this Article
Geoffrey A. Manne, Is the Intel/AMD Settlement Illegal?, Truth on the Market (November 13, 2009),

So, AMD and Intel settled.  Its a case we’ve covered here in significant detail.  Terms haven’t been announced publicly.  AAI has predictably argued that the settlement shouldn’t preclude further enforcement action from NY and the FTC.   The NY Times suggests the same.  They may be right, although Herb Hovenkamp, among others, has suggested that the settlement “has taken a lot of the wind out of the sails” of the FTC case.  Here’s the joint press release:

Under terms of the agreement, AMD and Intel obtain patent rights from a new 5-year cross license agreement, Intel and AMD will give up any claims of breach from the previous license agreement, and Intel will pay AMD $1.25 billion. Intel has also agreed to abide by a set of business practice provisions. As a result, AMD will drop all pending litigation including the case in U.S. District Court in Delaware and two cases pending in Japan. AMD will also withdraw all of its regulatory complaints worldwide. The agreement will be made public in filings with the Securities and Exchange Commission.

What I find most interesting about the settlement are the reported conduct terms.  The settlement reportedly includes provisions whereupon AMD and Intel agree that Intel will not engage in the business practices that gave rise to the Section 2 complaint, i.e. loyalty or all units discounts.  Here is a reported summary of the business practice provisions:

  • Offering inducements to customers in exchange for their agreement to buy all of their microprocessor needs from Intel, whether on a geographic, market segment, or any other basis (Section 2.1.1.a)
  • Offering inducements to customers in exchange for their agreement to limit or delay their purchase of microprocessors from AMD, whether on a geographic, market segment, or any other basis (Section 2.1.1.b)
  • Offering inducements to customers in exchange for their agreement to limit their engagement with AMD or their promotion or distribution of products containing AMD microprocessors, whether on a geographic, channel, market segment, or any other basis (Section 2.1.2a-b)
  • Offering inducements to customers in exchange for their agreement to abstain from or delay their participation in AMD product launches, announcements, advertising, or other promotional activities (Section 2.1.2.b)
  • Offering inducements to customers or others to delay or forebear in the development or release of computer systems or platforms containing AMD microprocessors, whether on a geographic, market segment, or any other basis (Section 2.2.2 and 2.1.2)
  • Offering inducements to retailers or distributors to limit or delay their purchase or distribution of computer systems or platforms containing AMD microprocessors, whether on a geographic, market segment, or any other basis (Section 2.2.1)
  • Withholding any benefit or threatening retaliation against anyone for their refusal to enter into a prohibited arrangement such as the ones listed above.

As I read the summary of Sections 2.1.1.a and 2.1.1.b, assuming this summary is accurate and complete, they would prevent loyalty discounts or other inducements paired with exclusivity or market share commitments.  Exclusive dealing arrangements would also seem to be prohibited so long as Intel had to give something up to get the exclusivity.  For example, if Intel offers a rebate conditioned on Dell purchasing 80 percent of its microprocessor purchases from Intel, this necessarily “limits” purchases from AMD.  It’s unclear if the language would apply to discounting practices such as pure volume discounts that clearly provide incentives for purchasers to buy more of their requirements from Intel and less from AMD, but apparently Intel maintains that the agreement will not preclude discounts per se.  And arguably these practices would not be prohibited since, as summarized, the provisions would be triggered only if there was an agreement with a retailer to limit AMD commitments.   Nonetheless, doesn’t it seem fairly straightforward that the settlement would amount to an agreement between competitors for one of them not to engage in some forms of legitimate competition for distribution that have been broadly recognized as such under the antitrust laws?

So does the Intel/AMD settlement raise issues under Section 1?  We know that the FTC’s position on private settlements that reduce competition between rivals, such as those observed in the branded/generic pharmaceutical context, is that such settlements not only can but are likely to violate Section 1 of the Sherman Act despite any of the pro-competitive efficiencies associated with settling underlying legal disputes.   The potential distinction that settling antitrust cases somehow counts “more” than the settlement of patent infringement cases seems weak (and anyway this agreement also settles a patent infringement case).

Settlements only violate Section 1 of the Sherman Act if they satisfy its prerequisites.  So, does a horizontal agreement reached between two rivals that includes a provision aimed at preventing loyalty discounts violate Section 1 as a per se violation?  At a minimum, isn’t such a settlement at least “inherently suspect” or presumptively unlawful under Sherman Act jurisprudence?  Judge Ginsburg described the standard as follows in Polygram:

Although the Commission uses the term “inherently suspect” to describe those restraints that judicial experience and economic learning have shown to be likely to harm consumers, see FTC Op. at 29, we note that, under the Commission’s own framework, the rebuttable presumption of illegality arises not necessarily from anything “inherent” in a business practice but from the close family resemblance between the suspect practice and another practice that already stands convicted in the court of consumer welfare.

Are agreements to prohibit loyalty discounts close in “family resemblance” to other agreements that have been convicted in the court of consumer welfare and “almost or always almost” reduce output?  The tentative answer has to be “yes” doesn’t it?  These practices are, at their core, discounts, even if they are conditioned on market share commitments from buyers.  The counter-argument is that the underlying theory of the AMD and NY and EU cases is that there is something special about loyalty discounts of this particular form, at least as applied in these markets, that suggests that they are anticompetitive and so the per se logic is inapplicable in this context.  While the Section 2 analysis turns on whether Intel’s inducements to customers were likely to harm competition and consumers, the Section 1 question is whether an agreement to stop such inducements is likely, from judicial and economic learning, to always or almost always reduce output.  The questions are obviously related, but they are different inquiries.  The fact that one has a strong belief on the Section 2 question, for example, that the inducements will harm competition and consumers and violate Section 2 does not mean that an agreement between competitors to prohibit the conduct is necessarily immune (although it does complicate the argument as a rhetorical matter for those making it).

Yes, there are theoretical arguments in the literature for why loyalty discounts might be anticompetitive.  But best I can tell from the literature, there is no strong empirical evidence supporting these theories.  Nor is there consensus in the theoretical literature on the relative likelihood that loyalty discounts will have anticompetitive effects (remember that the theoretical models assume away potential efficiencies–a large assumption in the context of a discounting practice).  Further, consider a loyalty discount with a volume threshold that requires the retailer to grant 100% of its purchases to a particular manufacturer.  That would be an exclusive dealing contract, and we know that courts have accepted a plethora of competitive justifications for exclusive dealing arrangements.  The empirical literature also contains evidence that such agreements are generally pro-competitive.

In short, the argument that the discounts are alleged to violate Section 2 doesn’t seem to save the settlement from a Section 1 analysis.  Of course, if the enforcement agencies’ prior is that the contracts are anticompetitive (which is why it will bring the Section 2 case), it is unlikely to believe that the settlement violates Section 1 because it will believe that the settlement is good for consumers.  But that belief is not the law.  The law is that an agreement to stop a practice that generally results in consumer benefits is subject to antitrust scrutiny under Section 1 and might even be per se illegal. For example, the FTC challenges reverse payment settlements despite the fact that there are arguments that they do not “always or almost always” harm competition because they argue that those agreements amount to agreements between competitors to divide the market or reduce output.

I’m not making an argument by analogy to reverse payment settlements.  What I am saying is that a settlement between two competitors that prohibits discounting conduct could raise important issues about the legality of the settlement under Section 1.  (And, for what it’s worth, this WSJ story notes that both stocks had a favorable reaction to the announcement of the settlement (see also here)).   I don’t believe that the fact that the underlying conduct has been challenged under Section 2 changes the analysis.  Nor does the fact that other courts in Europe, Japan and elsewhere have condemned the conduct–especially since those jurisdictions have done so applying a very different welfare standard than that embedded in the U.S. antitrust law.  Nor does the fact that there are economic theories that would predict that loyalty discounts can under some conditions produce consumer harm.  None of these bears upon the appropriate legal standard.  In Polygram, for example, the existence of pro-competitive  theories certainly did not stop the Commission or the D.C. Circuit from concluding that the conduct was inherently suspect.

If the settlement goes beyond prohibiting these particular loyalty discounts and includes other forms of conventional volume discounts, I think the argument that such a settlement violates Section 1 would be even stronger.  But even if the agreement only applies to loyalty discounts, and not other forms of discounting, there may still be problems.  Isn’t there a legitimate reading of the Section 2 case law as sufficiently demonstrating the proposition in antitrust law that single firm discounting conduct is highly likely to be pro-competitive to support an argument that an agreement not to engage in such conduct would be presumptively unlawful whether in the context of a settlement or otherwise?

Obviously, this is hypothetical thus far because we haven’t actually seen the settlement, and what we have seen reported doesn’t seem to cover conventional volume discounts.  And even if I’m right, there are other things that could be in the settlement that might prevent it from violating Section 1 or at least influence the analysis.  It seems highly unlikely that the FTC, who is investigating the underlying conduct, would challenge the settlement since it would then be forced to simultaneously take the position that the discounts were likely to benefit and harm consumers (and one can also imagine the FTC looking for a broader settlement, applying he same conduct provisions with respect to all competitors and not just AMD, for example, or finding some of the precise language too loose).

But there are, presumably, other parties with standing.  And presumably court authority to review the settlement.  This could get interesting.

Any thoughts from others on this?  I confess to being uncomfortable with this analysis, in that I am delighted for Intel and for consumers if the AMD litigation goes away and the wind is taken out of the FTC’s sails as these are cases that I think are fundamentally misguided.  But that same logic compels some concern over the legality of this agreement.  So what’s the best defense of the settlement? Where is the flaw in this argument?