Article & Essay

The Fundamental Similarity of the Surcharge Theories in Qualcomm (2019), United Shoe Machinery (1922) and Microsoft (1995)

  • DATE WRITTEN : 2020-11-02
  • WRITER : Sang Seong
  • VIEW : 886
FILE1 DOWNLOAD
During the oral arguments before the 9th Circuit in February 2020, both the FTC¡¯s and Qualcomm¡¯s counsel made extensive references to the 1995 consent decree between the Department of Justice and Microsoft banning the so-called ¡°per-processor licensing contracts¡±. Qualcomm claimed that its ¡°No license, no chips¡± policy is ¡°totally different¡±.
According to Qualcomm, in the per-processor case, while a computer maker paid two royalties (one to Microsoft and one to Digital Research) when it installed DR-DOS, it only paid one royalty to Microsoft. In contrast, a smartphone maker paid the same royalties to Qualcomm whether it bought a modem from a rival or from Qualcomm. The non-discriminatory nature of its surcharges distinguished Qualcomm¡¯s conduct from Microsoft.
In this note, I show that Qualcomm¡¯s argument is wrong. Both Microsoft and Qualcomm levy a tax when the customer buys from a rival, like the United Shoe Machinery did in the 1922 case. It is ironic that the DOJ Antitrust Division which developed the anti-competitive theory of harm in the 1995 case did not understand this fundamental similarity between Qualcomm¡¯s and Microsoft¡¯s conduct. The Ninth Circuit simply repeated DOJ¡¯s rendition of the superficial difference between the 1995 per-processor case and Qualcomm¡¯s practices.

Microsoft¡¯s Per-Processor License Contract

Under the ¡°per-processor licenses,¡± Microsoft required computer makers to pay Microsoft a royalty for each computer the computer sells, whether or not the OEM has included a Microsoft operating system with that computer.

The But-For World

Begin with the hypothetical world in which the practice is prohibited. Let Microsoft¡¯s monopoly price (license fee) of MS-DOS be p^MS. There is a small rival, Digital Research, which charges p^DR for its OS called DR-DOS. The simple but important point is that in this but-for world, Microsoft charges p^MS for MS-DOS only on PCs on which an OEM decides to install MS-DOS. If the OEM decides to install DR-DOS instead, it pays p^DR to Digital Research and 0 to Microsoft.

In this but-for world, competition on the merits takes place: While Microsoft charges a monopoly price and earns a monopoly profit&\#8212\;which is encouraged as long as the monopolist acquires and maintains its monopoly power legitimately&\#8212\;an OEM decides which OS to install based on the prices and features of the two operating systems and its strategy on how to meet consumer demand and compete with other OEMs. Indeed, the monopoly price invites entry, disciplining the monopolist not to be complacent.

The World With Per-Processor Contract

It looks like that Microsoft continues to sell the same product to the OEMs under the per-processor contract as in the but-for world. However, careful analysis shows that the scheme allows Microsoft (the monopolist) to split MS-DOS (the monopoly product) into two components: the right to access MS-DOS and actual installation of MS-DOS itself. Leveraging its monopoly power, Microsoft requires the PC makers to pay two separate prices to Microsoft: the OS access fee (collected whether or not MS-DOS is installed on a computer\; denoted by p_access^MS) and the OS usage fee (collected only on machines on which MS-DOS is installed\; denoted by p_usage^MS).
Under the Chicago School¡¯s Single Monopoly Profit Theorem , the sum of the access and the usage fees cannot exceed the monopoly price of MS-DOS in the but-for world:

A strict inequality will hold if the monopolist (Microsoft) needs to compensate a buyer (an OEM) for charging OS access fees for units that the OEM does not want to install MS-DOS:


Under some circumstances, the Post-Chicago School shows that the monopolist can play one buyer against the others so that it can foreclose rivals at (almost) no cost:

Whether (2) holds or (3) holds, in general, if Microsoft tries to increase the access fee, it will need to reduce p_usage accordingly.

In the actual antitrust case, it happened that MS set p_usage^MS=0 so that it looks like MS collected only one price whether or not OEMs installed MS-DOS on a particular machine. However, this fact does not alter the four key points that emerge from the above analysis: First, Microsoft used its monopoly power to impose an OS access fee to OEMs.
Second, the OS access fee is non-discriminatory. That is, OEMs pay p_access^MS to Microsoft whether or not they buy from Microsoft or from the rival. It is secondary that in the actual antitrust case, Microsoft set p_usage^MS=0.
Third, the DOJ and (ultimately the D.C. Circuit) did not base its theory of harm on predatory pricing that is, whether or not p_usage^MS was below Microsoft¡¯s cost. Rather the DOJ¡¯s theory was rasing rivals¡¯ costs: the monopolist used its monopoly power to collect a tax on transactions between its customers and rivals with no or little off-setting procompetitive justifications.
Fourth, the DOJ¡¯s liability theory also did not rest on the precise quantification of how the burden of OS access fee was split between an OEM and the rival.

Qualcomm¡¯s No-License, No Chips Policy

A Simplified Example: Invalid Patents

It is instructive to imagine a situation in which Qualcomm¡¯s patents are invalid so that the reasonable royalty for Qualcomm¡¯s patent portfolio is 0.

No-License, No Chip Policy Prohibited

Let p^QC be Qualcomm¡¯s monopoly price of its modem chips in the but-for world in which the no-license, no-chips policy is banned. Let Intel be its rival and denote its chip price by p^Intel. As in Section 1.1, competition on the merits ensues in the but-for world: while Qualcomm is allowed (encouraged) to charge monopoly price for its chips (again, assuming that it acquired and has maintained it legitimately), OEMs can make a free choice between Qualcomm and Intel chips based on the merits.

No-License, No Chip Policy Allowed

However, in the acutal world with the no-license, no-chips policy, Qualcomm requires OEMs to first sign a patent license agreement which obligates them to pay Qualcomm a royalty, denoted by p_access^QC, on all handsets regardless of whether they install a Qualcomm chip or an Intel chip. It is clear that since Qualcomm¡¯s patents are invalid in the current expositional example, the royalty is just a chip access fee which OEMs have to pay Qualcomm in order to gain access to Qualcomm¡¯s monopoly chips. Let p_chip^QC be Qualcomm¡¯s chip price under the NLNC policy. As discussed in Section 1.1, when the Single Monopoly Profit Theorem holds,


It should be clear that in this example, Qualcomm¡¯s conduct is exactly analogous to Microsoft¡¯s: just like Microsoft levied an OS access fee on PC makers, Qualcomm imposes a chip access fee on smartphone makers. Patent royalty is in name only: it is an access fee to the monopoly product.
Moreover, both Microsoft and Qualcomm impose non-discriminatory access fees. Qualcomm emphasizes that it ¡°does not discriminate\; it does not charge higher royalties if the OEM uses a rival¡¯s chip¡±. But neither did Microsoft discriminate: it did not charge higher OS access fees (which Microsoft styled as licensee fees) if the PC maker used a rival¡¯s OS. Crucially, it is immaterial if Qualcomm¡¯s chip price is below cost or how the burden of the chip access fee is split between a smartphone maker and the rival chip maker in order to find Qualcomm¡¯s conduct anti-competitive. Once one accepts that Microsoft¡¯s per-processor contract which imposes a (non-discriminatory) tax on a rival OS is anti-competitive, so is Qualcomm¡¯s NLNC which levies a (non-discriminatory) surcharge on a rival chip.

A More Realistic Scenario: Valid Patents
In reality, Qualcomm holds a valid SEP portfolio. However, the fundamental point does not change: Qualcomm imposes a naked tax on transactions between rival chip makers and OEMs through the NLNC. The only change in the analysis is that we need to allow for the reasonable royalty that Qualcomm is entitled to collect: let p_FRAND^QC be the FRAND royalty for Qualcomm¡¯s patent portfolio.
No-License, No Chip Policy Prohibited

When NLNC is banned, below shows that the only change from
is that Qualcoms simply collects a FRAND royalty on all handsets\; competition in the chip market continues to take place based on chip prices.

o-License, No Chip Policy Prohibited

Under the NLNC scheme, Qualcomm again splits its monopoly price for chips (p^QC) into two components: a chip access fee which Qualcomm labels a royalty on its patents (p_access^QC\; so that Qualcom collects p_access^QC+p_FRAND^QC as the total royalty on each smartphone) and a chip price (p_chip^QC).

But as long as p_access^QC is positive&\#8212\;that is, as long as Qualcomm¡¯s royalty exceeds the FRAND level&\#8212\;the fact that this surcharge is non-discriminatory does not alter the fact that it is anticompetitive, just like Microsoft¡¯s (and United Shoe Machinery¡¯s) non-discriminatory surcharge was determined to be anticompetitive by the DOJ.
      
Prev THE FOURTH INDUSTRIAL REVOLUTION AND COMPETITION LAW IN KOREA - PART 1
Next