73 F.3d 756
1996-1 Trade Cases P 71,260
DIGITAL EQUIPMENT CORPORATION, Plaintiff-Appellee,
UNIQ DIGITAL TECHNOLOGIES, INC., Defendant-Appellant.
United States Court of Appeals,
Argued Sept. 6, 1995.
Decided Jan. 16, 1996.
Paul F. Donahue (argued), Seyfarth, Shaw, Fairweather & Geraldson, Chicago, IL, for Plaintiff-Appellee.
Richard P. Campbell, Anthony S. Divincenzo (argued), Campbell & Divencenzo, Chicago, IL, for Defendant-Appellant.
Before CUDAHY, FLAUM, and EASTERBROOK, Circuit Judges.
EASTERBROOK, Circuit Judge.
Between 1977 and 1986 Uniq Digital Technologies sold Digital Equipment Corporation (DEC) computers equipped with Unix operating systems. The operating system mediates between application programs and the computer's hardware. Unix, an operating system developed by Bell Laboratories at AT & T, was and is widely used for scientific and networking functions. Each central processing unit (the calculating engine of a computer) needs an operating system customized to its requirements, and Unix is adaptable. AT & T has licensed other firms to develop versions of Unix suited to different CPUs. Uniq "ported" the Unix-V system to DEC's computers. This enabled DEC machines to run many of the same application programs that customers had used on other manufacturers' computers. DEC gave Uniq a discount available only to distributors ("value added resellers" or "original equipment manufacturers," abbreviated VAR and OEM) that added value to the equipment, and DEC treated the Unix operating system as added value. In 1985 DEC, which by then had developed its own version of Unix, told Uniq that the OEM discount would no longer be available to resellers whose only contribution was Unix. Uniq proposed, and DEC accepted, a marketing plan that omitted Unix as an element of added value; Uniq sells and services suites of applications and treated these as the added value. But in 1987 Uniq did not sell a single DEC computer, leading to the cancellation of the distributorship agreement. DEC filed this diversity action in 1988 seeking to collect a note for $67,000. Lengthy delay ensued while the district court addressed Uniq's antitrust counterclaim, which it eventually dismissed (along with all of Uniq's contractual defenses) in advance of trial. Entry of judgment on DEC's claim followed ineluctably. 1995 U.S.Dist.
The contract governing the parties' dealings was a "standard volume agreement" (the "OEM agreement"). This contract ran from year to year and could be terminated by either party. Several addenda to this contract were negotiated over the years; one of these, the "authorized digital computer distributor addendum" (the "ADCD addendum"), entitled Uniq to use Digital's trademarks in promotion and to participate in a cooperative advertising program. The ADCD addendum was to last indefinitely, but not longer than the OEM agreement; Sec. 8.3 of the ADCD addendum provides that, "[s]hould the [OEM] Agreement be terminated for any reason, this appointment shall also terminate concurrently." The ADCD addendum also ended automatically if sales fell below a threshold, and Uniq's failure to sell a single DEC computer for one year is what terminated the ADCD addendum on October 1987. In light of this, Uniq's argument that DEC broke its promise by failing to extend the ADCD addendum is untenable. DEC ended the OEM agreement, and the ADCD addendum would have ended with it, had it not ended earlier because of Uniq's decline in sales. Unless the termination of Uniq's status as an OEM was wrongful--Judge Castillo held not, see 1995 WL 12297, 1995 U.S.Dist. LEXIS 190--Uniq's contractual defenses fail.
Uniq believes that its termination as an OEM was wrongful because DEC "was obliged to renew Uniq's contract on the same terms" every year; and by "same terms" Uniq means "with the same things treated as 'added value.' " For several years DEC treated the Unix operating system as added value, and that course of dealings "transcended the written agreements", Uniq tells us. Judge Castillo thought that Illinois law (which the parties agree governs) offered no support for that argument--and he was right. Illinois enforces written contracts, and Uniq agreed, in writing, to a contract that allowed DEC to terminate at the end of any year. Under the OEM agreement, Uniq had to propose a marketing plan every year. This plan had to be satisfactory to DEC; and what was satisfactory in one year might prove unsatisfactory in another, as market conditions and technology changed; that was the whole point of making renewal annual, rather than giving Uniq a right to OEM status for a longer term. Contributions valuable to both DEC and customers in one year might be old hat the next. A firm anxious to press forward technologically insists that its vendors come up with new products and services if they want to get an extra discount. For that discount is what an OEM agreement is about. It sets price, and nothing else. Uniq was (and is) eligible for an ordinary wholesale price. The lower OEM price is a subvention for extra services, and DEC, as the "buyer" of those services, is entitled to determine whether it is getting what it is paying for. In arguing that DEC "was obliged to renew Uniq's contract on the same terms" every year, Uniq really is arguing that DEC must pay the same price for Uniq's inputs in perpetuity, without regard to changes in their market value.
Trying to get past this obstacle, Uniq repackages its contention as an invocation of the "duty of good faith." Although Uniq proposed (and DEC accepted) a marketing plan in which Unix did not count as added value, Uniq tells us that this agreement should be ignored because DEC did not act in good faith when demanding that concession. "Good faith" is a ringing term, but there is no abstract "duty of good faith" even in the law of fiduciary obligations--and Uniq does not argue that DEC was its fiduciary. Some parts of the law create discrete duties to do things in good faith; labor law, for example, creates a duty of good faith bargaining, but even that branch of law does not require either side to make concessions. 29 U.S.C. Sec. 158(d). Employers may take a hard line and may insist on give-backs; unions may demand large wage increases and comfortable working conditions; and if one side's bargaining position does not suit the other, each retains its economic options. In the law of contract between merchants, there is no comparable duty; the Uniform Commercial Code defines "good faith" as "honesty in fact in the conduct or transaction concerned." UCC Sec. 1-201(19); see also PSI Energy, Inc. v. Exxon Coal USA, Inc., 17 F.3d 969 (7th Cir.1994); Kham & Nate's Shoes No. 2 v. First Bank of Whiting, 908 F.2d 1351 (7th Cir.1990). There is no duty to be kind, or considerate, or make concessions in bargaining. So we held, applying Illinois law, in First National Bank of Chicago v. Atlantic Tele-Network Co., 946 F.2d 516, 520 (7th Cir.1991). No state case decided since then has expressed any doubt about that holding.
Limiting the parties' ability to change the definition of "added value" would be the functional equivalent of forbidding vertical integration by any firm that initially buys inputs in the market. From one perspective an OEM or VAR is a manufacturer, buying inputs, reworking them (the "added value"), and selling the results. But for many purposes it is more helpful to think of a VAR as a vendor to the manufacturer. For example, IBM can equip its computers with disk drives in one of three ways: (i) it can buy disk drives from vendors such as Conner and put those third-party drives in its machines; (ii) it can make disk drives in its own plants, and package those with its machines; or (iii) it can sell disk-drive-less computers to third parties, who add them and resell the resulting bundle. In each case IBM pays for the drive-in cash to Conner if it buys before manufacture, in cash to its employees if it makes the drives, and in discounted prices if it has a VAR add the drive later. Most manufacturers use a combination of these methods, according to the costs of production and the demands of consumers. (There is a fourth option: sell a machine without a disk drive and let the consumer add one or not. Most producers use that option, too, for some fraction of their output.)
Features that few customers want--or that everyone wants a little differently--are added by VARs. That way the manufacturer can take advantage of economies of scale in production, while allowing specialists to add distinctive features. If every customer wants a feature, or if large numbers want a similar configuration, the manufacturer adds it. Thus General Motors builds the engine, chassis, and body for all of its cars; but for trucks and locomotives, GM builds only the engine and chassis, sending the product to VARs who add bodies customized to customers' desires. For limousines, GM does a little of each. When few drivers wanted compact disc audio systems, Ford and GM left these to the aftermarket; as demand rose, the auto companies started adding this feature on the assembly line. Similarly in the computer business. When few customers of DEC systems wanted Unix, DEC hired firms like Uniq to add Unix later and sell the system. As demand rose, DEC developed its own Unix operating system. Data General, Prime, Pyramid, Unisys, Wang, IBM, and DEC's many other competitors followed a similar path. After developing its own Unix operating system--that is, after vertical integration into the production of Unix operating systems--DEC had no need to pay Uniq for the same input.
This would be clear enough if the DEC-Uniq arrangement had begun as the sale of Unix by Uniq to DEC, followed by DEC's sale of Unix-equipped computers to consumers. Uniq could not expect to continue selling Unix to DEC indefinitely on the ground that, once the sales began, the "duty of good faith" meant that DEC had to renew on the same terms indefinitely, even though the contract provided for termination at the end of any year. Uniq stresses that it invested $1 million in "porting" Unix to DEC computers. Such investments are common; Goodyear invests many millions in designing tires for use on GM's cars, but it cannot complain if at the end of the contract term GM starts buying from Firestone--or starts making its own tires. Uniq has no greater complaint when DEC decides to stop buying Unix through the OEM discount. Firms protect their sunk costs through contracts long enough to recover them, or through buyout clauses. As it happens, a year was enough to recover the investment (in 1984 alone, Uniq had margins exceeding $2 million on DEC computers). Whether a year was long enough or not, however, the length of dealings is for the market rather than the courts to establish. See Industrial Representatives, Inc. v. CP Clare Corp., 74 F.3d 128 (7th Cir.1996) (Illinois law).
Illinois does not leave things entirely to the market in franchise transactions. If Uniq was a "franchisee," as 815 ILCS 705/3 defines that term, then it has some additional entitlements, which DEC may not have honored. Judge Holderman concluded that Uniq was not a franchisee. 1993 WL 338985, 1993 U.S.Dist. LEXIS 12136, 1993-2 Trade Cas. p 70,378. (Judge Holderman also resolved the antitrust counterclaims, before the case was transferred to Judge Castillo. All further discussion in this opinion concerns Judge Holderman's portion of the case.) One ingredient of a franchise relation in Illinois is a franchise fee. The district court held that DEC did not charge Uniq a franchise fee. Because that decision is well supported, we need not canvass the district court's other grounds.
Uniq paid for computers; there was no fee for being a dealer. So much is agreed. Relying on cases under Indiana's franchise act, Uniq argues that it bore an "indirect" franchise fee: the cost of inventory. We explained in Wright-Moore Corp. v. Ricoh Corp., 908 F.2d 128, 136 (7th Cir.1990), that an obligation to carry a large inventory can be the economic equivalent of a franchise fee. An excessively large inventory transfers cash to the seller without producing benefits for the buyer; and the interest the seller earns by making the sales earlier is a kind of fee. Like a cash payment, it transfers wealth from buyer to seller (one can speak of a "transfer" because, by hypothesis, the buyer gets no benefit from the inventory). We may assume that Illinois would take this approach to the definition of a franchise fee under 815 ILCS 705/3(c), (14). This does not assist Uniq, because the OEM agreement does not require it to carry any inventory of DEC's products. It does establish a 12-month lag between order and delivery. Delay meant that Uniq had to order on the hope of selling, and what it could not sell went into inventory. DEC observes that the long lead time reflects the fact that many of its computers are built to order. Perhaps; the record does not show whether 12 months is a good estimate of manufacture time. Whatever factual doubt there may be does not require a trial.
The lag between orders and deliveries is much greater in other businesses (airframe manufacture, for example) without leading to the accumulation of inventory. What of this industry? Did Uniq accumulate a stockpile of DEC computers? That question has an answer in the record: No. Uniq's inventory as a percentage of annual sales was approximately 4.5%. Uniq sold computers from other manufacturers but does not argue that DEC's goods languished for an unusually long time. Incoming computers and other parts therefore left the inventory an average of 16 days after they arrived. Although 16 days may be a tad long for practitioners of just-in-time inventory control systems, it is an unusually short time for a dealer that needs to have stock on hand to fill unexpected orders. (Just-in-time systems depend on a predictable production schedule; retailers and wholesalers accordingly hold more days worth of finished goods than manufacturers hold of parts.) The time value of advancing payments to DEC by 16 days is small, too small to be called a franchise fee. An average inventory of 16 days' supply is a considerable distance from what we had in mind in Wright-Moore. It does not distinguish Uniq from a normal OEM or VAR in the computer business. Similarly unimpressive is Uniq's argument that DEC charged it a hidden franchise fee by requiring it to pay for DEC's own operating system. DEC replies that this is no different from the requirement that the customer pay for the chips, circuit boards, power supplies, and other ordinary constituents of computers. DEC was selling computers, after all, and not parts for hobbyists to assemble. This just propels us toward the antitrust claim, to which we turn.
Uniq charged DEC with attempting to monopolize the market for operating systems for DEC's own computers, in violation of 15 U.S.C. Sec. 2. According to Uniq, DEC not only attempted to do this but also succeeded, by shipping an operating system with each computer, whether the customer wanted one or not. This tie of operating systems to computers is unlawful per se, Uniq submits. Now we doubt that it is useful to call DEC's practice a tying arrangement. An operating system is essential to make a bunch of silicon chips a "computer." No OS, no computation. One might as well complain that General Motors includes an electrical system with every car. Some firms sell central processing units, others sell memory chips or logic boards, others sell disk drives, and still others sell operating systems and applications software; it is possible to buy parts and software to assemble a computer. But competition is more vital, and consumers are better off, when it is possible to sell entire functioning units in boxes. Every manufacturer in the business includes an operating system with the physical parts that make up a computer. Consumers are free to buy and install additional operating systems; Apple's Macintosh, for example, comes with the Mac OS, but users can install and run Unix (or Microsoft Windows!) if they want. Sun's computers come with its proprietary OS; IBM's come with one of its several operating systems. This is not a reduction of competition; the process is competition. Computer manufacturers are vigorous rivals; prices drop daily; this is one of our economy's most competitive sectors. Calling the selection of components for one's product a "tie-in" does not help to uncover practices that restrict output, drive up prices, and transfer wealth from consumers to producers.
Let us ask, then, whether the inclusion of an OS with one's equipment--"tie-in" or not--is a means to reduce output and create monopoly profits. Unless the seller has market power, the answer is no, and the practice is of no antitrust concern. So Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2, 104 S.Ct. 1551, 80 L.Ed.2d 2 (1984), holds, even for practices properly called tie-ins. Uniq argues that it is entitled to press a claim under the Rule of Reason even if the absence of market power prevents it, under Hyde, from making a per se claim. The law of this circuit is otherwise; substantial market power is an indispensable ingredient of every claim under the Rule of Reason. See, e.g., Sanjuan v. American Board of Psychiatry & Neurology, Inc., 40 F.3d 247, 251 (7th Cir.1994); Hardy v. City Optical, Inc., 39 F.3d 765, 767 (7th Cir.1994); Chicago Professional Sports Limited Partnership v. National Basketball Association, 961 F.2d 667, 673 (7th Cir.1992); Will v. Comprehensive Accounting Corp., 776 F.2d 665, 670-74 (7th Cir.1985); Carl Sandburg Village Condominium Ass'n No. 1 v. First Condominium Development Co., 758 F.2d 203, 210 (7th Cir.1985).
The smallest plausible market--other than one in DEC's own computers--is that of mid-range computer systems, those between mainframes and desktop personal computers in size, price, and computational power. Most persons interested in performing computational tasks could substitute time on a mainframe (which are designed for time sharing), or multiple smaller computers, for a mid-range DEC, so we think it likely that the market is larger. Unix is available for all of these systems. People who want to perform computational tasks using software (or human capital) specialized to Unix can migrate freely from one brand of computer to another. If the price of DEC computers goes up, customers can substitute hardware from Pyramid or Omnibyte (both of which Uniq sold, with Unix operating systems it had ported to those models), or Data General, or any of dozens of other vendors of "Unix boxes." Any market power belongs to AT & T, the proprietor of Unix; but then other operating systems compete with Unix. We need not pursue the subject. DEC's share of the mid-range business is substantially less than 30%, and Hyde holds that 30% is not enough to confer substantial market power unless there are high barriers to competition. Rapid and continuing entry into the computer business, and the ease with which existing firms increase output, dispel any concern about barriers to entry.
DEC operates in a fiercely competitive business. Which means that it is entitled to choose which items to include in the bundles it sells. Suppose DEC includes extensive manuals, or "free" support services, or operating systems, that consumers do not want. Buyers may have a license to use Unix; or they may have their own support staff and not want a fancy set of manuals plus a tech support group at the other end of a phone line. "Free" tech support is not free; customers pay for it in the price of the product, and a marginal cost of zero for calling means that it will be overused, and hence can add quite a bit to the price of a computer or a software package. Instead of paying DEC for something they do not want, customers will turn elsewhere--for other firms can increase their profit by satisfying users' demands. For example, many have cut the "free" tech support, reducing the price of the product while charging for assistance, an appealing combination for customers that have their own tech support groups. IBM would like to include its OS/2 operating system with all of its small computers, but it will substitute Microsoft Windows (or Unix) if customers want. In a competitive market a pig-headed refusal to satisfy customers' preferences, or an attempt to charge for unwanted items, does not lead to monopoly prices; instead it leads to ruin as rivals step in to take the business. Preserving that market process, rather than regulating through the judicial system, is the objective of the antitrust laws. Although one court of appeals has held that firms lose the right to design their product packages when one element is a copyrighted operating system, see Digidyne Corp. v. Data General Corp., 734 F.2d 1336 (9th Cir.1984), we have rejected that position, Will, 776 F.2d at 673 n. 4, and see no reason to revisit the subject. Accord, A.I. Root Co. v. Computer/Dynamics, Inc., 806 F.2d 673, 675 (6th Cir.1986) (holding that OS software for a particular computer is not a relevant market).
According to Uniq, everything changed when the Supreme Court issued its opinion in Eastman Kodak Co. v. Image Technical Services, Inc., 504 U.S. 451, 112 S.Ct. 2072, 119 L.Ed.2d 265 (1992). Kodak sold plain paper copiers in a market with three substantial rivals. At the time of the sale, the Court assumed, Kodak sold replacement parts, enabling users to repair their copiers or hire independent service organizations (ISOs) to do so. Later Kodak changed its policy and refused to sell parts to ISOs, who alleged that this enabled Kodak to claim the repair business for itself, at supra-competitive prices. The Court held that evidence in the record that prices had increased prevented a grant of summary judgment. It conceded that customers who had anticipated the change of policy could not be exploited; they would have shopped around and purchased copiers based on full-life-cycle costs (purchase price, plus costs of repair, plus costs of consumables). Competition among manufacturers fully protects buyers who accurately calculate life-cycle costs. But not all customers do this, if only because they do not anticipate all changes of policy. Thus Kodak had some ability to extract additional money by raising prices. It could not so do again; once the new policy was known, consumers could shop with full information; but there was a material dispute about the effect of its change of policy. Kodak assumed that fooling consumers in this way is an antitrust rather than a deception problem, because the parties had not debated this question; we need not inquire whether it would convert this assumption into a holding. See Philip E. Areeda & Herbert Hovenkamp, Antitrust Law p 1709 at 1162-63, 1182-83 (1992 Supp.) (doubting that the assumption is justified or that Kodak supports a tie-in theory for parts when the principal market is highly competitive); cf. Comment, Parts and Service Included: An Information-Centered Approach to Kodak and the Problem of Aftermarket Monopolies, 62 U.Chi.L.Rev. 1521 (1995).
Uniq sees in Kodak the message that there is a market in a firm's own products, even if it sells in vigorous competition. If that is so, then Kodak overruled Hyde, which held the opposite. A hospital signed a contract giving one group of physicians the exclusive right to practice anesthesiology within its walls. The fifth circuit held this a tie-in, unlawful per se, because it precluded competition throughout the hospital. Yet the Supreme Court reversed, holding among other things that this is the wrong way to look at competition. A patient can elect where to go for an operation; from the patient's perspective, the market includes other hospitals. Physicians at East Jefferson Hospital did only 30% of the surgery in Jefferson Parish, and many patients could search outside the parish for alternatives. That sufficed to protect patients, the Court held. By Uniq's lights, however, the decision of the fifth circuit should have been affirmed on the ground that there is a market in anesthesiology at East Jefferson Hospital, just as (by Uniq's lights) there is a market in spare parts for Kodak copiers, and a market in operating systems for DEC computers.
Kodak did not undercut Hyde. The Court did not doubt in Kodak that if spare parts had been bundled with Kodak's copiers from the outset, or Kodak had informed customers about its policies before they bought its machines, purchasers could have shopped around for competitive life-cycle prices. The material dispute that called for a trial was whether the change in policy enabled Kodak to extract supra-competitive prices from customers who had already purchased its machines. Compare Mr. Sprout, Inc. v. United States, 8 F.3d 118 (2d Cir.1993), with Allen-Myland, Inc. v. IBM Corp., 33 F.3d 194 (3d Cir.1994). Concrete evidence that it had done so entitled the plaintiffs to a trial, the Court held. Uniq has not supplied such evidence; nothing in this record suggests that DEC was able to raise prices, or exploit any customer, by deciding to include an OS with every machine. And there is a reason the record lacks such evidence: DEC's policy does not enable it to exploit any customer locked-in to its equipment. Existing customers are unaffected by DEC's decision to include an OS with every new machine; potential customers can shop around. Facilitating such shopping is one benefit of Unix. Customers who use Unix are not committed to any manufacturer's hardware; they can play the field. One could hardly imagine a weaker case for the claim that DEC's computers are a market unto themselves. DEC is selling a fungible commodity (CPU cycles), to customers who can substitute brands without changing operating systems, in a market with rapid increase in production. Monopoly this is not. A dangerous probability of monopoly is not on the horizon. See Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 113 S.Ct. 884, 122 L.Ed.2d 247 (1993). This is a mundane commercial case, in which a buyer has used the antitrust laws to postpone paying its debts. Time for payment is at hand.