Barbarians at Bill Gates

It was a glorious fall day on the East Coast when Assistant Attorney General for Antitrust Joel Klein stepped up to the microphone at a news conference called hard on the heels of the release of Judge Thomas Penfield Jackson’s “findings of fact” in the government’s case against the Microsoft Corporation. An obviously pleased Mr. Klein declared that Judge Jackson’s findings, which accepted the Justice Department’s charges nearly in total, proved once again that no company—and no man—is above the law. Adding unintended irony to the scene, Bill Clinton’s guardian angel, Attorney General Janet Reno, stood at Mr. Klein’s shoulder.

Thus ended the first act in a continuing drama that began in the summer of 1993, when, after the Federal Trade Commission twice failed to vote out a complaint against Microsoft, the Justice Department launched its own investigation of the company’s possible antitrust violations. The Department of Justice’s intervention in its sister agency’s three-year-long inquiry was both unprecedented—in order to avoid unseemly competition the FTC and Justice have since 1948 operated under a liaison agreement that divides the antitrust territory between them—and politically inspired. Anne Bingaman, President Clinton’s first assistant attorney general for antitrust, requested the files accumulated during the commission’s investigation in response to pressure from Senators Howard Metzenbaum (since retired) and Orrin Hatch, both of whom urged her to take over the Microsoft case.

The substance of the charges of unlawful behavior against Microsoft has been in metamorphosis over the course of the company’s protracted encounter with the federal antitrust authorities—and with the 19 states and the District of Columbia that have filed their own antitrust lawsuits against it. At the outset the FTC’s attorneys maintained that Microsoft had illegally foreclosed the market for computer operating systems by entering into contracts with manufacturers of personal computers, granting them substantial discounts in return for agreeing to preload Microsoft’s own operating system, MS-DOS (later upgraded and marketed as Windows), on all their PCs. Indeed, some of the contracts obligated the manufacturers to pay Microsoft a royalty on every computer processor they shipped even if some other operating system had been installed. The FTC’s investigation had centered particularly on the lack of success experienced by DR-DOS, an operating system marketed by Novell, Inc., headquartered in Utah and represented in the U.S. Senate by Orrin Hatch.

When the Justice Department took over the investigation, the focus shifted to Web-browsing software. Reacting to the spectacular growth of the Internet and to consumers’ desires for a product that would allow them to navigate easily around the information superhighway it spawned, Microsoft developed its Internet Explorer Web browser and formed plans ultimately to integrate it with Windows. Those plans bore fruit with the release of Windows 95, raising the competitive stakes between Microsoft and Netscape, Inc., which marketed its own Web browser, Navigator, as a stand-alone software product.

Tying that Binds?

Following meetings between James Barksdale, then CEO of Netscape, and Assistant Attorney General Klein, the Justice Department filed suit in federal district court seeking to force Microsoft to offer a version of Windows without Internet Explorer. The Justice Department charged that Microsoft’s practice of bundling Internet Explorer with its operating system violated the Clayton Act’s proscription on “tying,” whereby the sale of one product is conditioned on the purchase of another. By requiring purchasers of Windows to also buy its Web browser, Microsoft, according to the government, had leveraged its market power in operating systems to foreclose Netscape from distributing Navigator. Not only did Microsoft’s strategy represent an unlawful restriction on consumer choice, but, given that Internet Explorer was included in Windows at no additional charge, the strategy also may have been intended predatorily to drive Netscape from the browser market.

The district court agreed and ordered Microsoft to offer two versions of Windows, with and without Internet Explorer. But only a month after the Justice Department filed more far-reaching charges of anti-competitive behavior against Microsoft on May 18, 1998, the D.C. Circuit Court of Appeals overturned the lower court’s ruling. Wading through mind-numbing details on the question of whether Internet Explorer was truly “integrated” with Windows 95, the appeals court held that because Intemet Explorer 3.0 contained features that went beyond browsing capabilities, it did not constitute a separate product within the meaning of the law’s anti-tying language; hence, Microsoft’s decision to include Internet Explorer in Windows did not violate the terms of a 1995 consent decree in which the company agreed not to bundle computer software applications with its operating system.

And so, by the time the Department of Justice submitted its proposed findings of fact to Judge Jackson, the basis of the government’s case against Microsoft had shifted once again. The Justice Department’s new theory, ultimately embraced by the judge, is that Microsoft has engaged in a variety of anti-competitive practices—including its decision to incorporate Web-browsing capabilities within Windows—calculated to protect and extend its monopoly power in the market for “Intel-compatible PC operating systems” (about which more below) by erecting and maintaining a so-called applications barrier to entry.

That entry barrier is based on the positive feedback loop between a computer’s operating system and the software applications that run on top of it. The greater the number of copies of a particular operating system in use, the more potentially profitable it is for software vendors to write compatible applications (such as word processors, spreadsheets, and tax-preparation and e-mail programs), which after all are why consumers want to buy personal computers in the first place. The more applications written to run on a particular operating system, the greater the demand for that operating system will in turn be. Hence, applications and operating systems are subject to a proverbial “chicken-and-egg” problem. With its large installed base of users—something on the order of 90 percent of new PCs are shipped with Microsoft’s operating system—thousands of Windows-compatible software applications are available on the market. Few applications are written for alter native operating systems because few copies of those operating systems are in use, and few alternative operating systems are in use because few software vendors write applications compatible with them.

The threat posed by Netscape’s Navigator (and by Sun Microsystems’ Java programming language) was not that they competed with Microsoft’s Web browser per se, but that they exposed interfaces that might serve as an alternative platform for running network-centric application programs—applications capable of running on Web servers rather than on a computer’s hard drive. Navigator and other such “middleware” thus endangered Microsoft’s dominance of the operating system market, and according to the Justice Department, the company moved aggressively and unlawfully to quash the competition.

Microsoft was particularly worried that Navigator, which had captured about 80 percent of the browser market (and had “been ported to more than fifteen different operating systems”¹), would be seen as such an attractive platform for applications by software developers that, in the words of Netscape’s co-founder, Marc Andreessen, Windows eventually would be relegated to an unimportant collection of “slightly buggy” device drivers.²

The Relevant Antitrust Market

As in any antitrust case, the definition of the market that is relevant for analyzing the competitive effects of Microsoft’s business practices is the key issue in Judge Jackson’s findings of fact. Drawn narrowly enough, the boundaries of relevant antitrust markets can limit the number of products that ostensibly compete with one another so severely that almost any firm can be determined to possess “market power”—the ability to raise prices above competitive levels. In the case at hand, Judge Jackson chose to define the relevant market as “Intel-compatible PC operating systems,” the narrow definition advocated in the government’s pleadings.

The curious conclusion produced by the judge’s finding on the relevant market definition is that Microsoft does not compete with the very firms it is accused of injuring. Excluded from the market so defined are Netscape and Sun, which, as indicated, market software products that interface with a computer’s operating system (and which might also serve as alternative platforms for running application programs); Apple, which markets an operating system (Mac OS) that is incompatible with Intel-based PCs; and the many other companies striving to develop the next ubiquitous applications platform. The only other firms in the “market” are those offering either alternative Intel-compatible operating systems, such as IBM (OS/2) and Be (BeOS), or “cross-platform” operating systems, such as Linux, capable of running on both Intel-based and non-Intel-based PCs.

Judge Jackson defended his decision to define the relevant market in this way by pointing to the high cost consumers face in switching to the available alternatives. For one, unlike Windows, the Mac OS is licensed only for use on Apple’s own hardware—the two products are sold as a bundle. The owner of an Intel- compatible PC is therefore required to purchase an Apple machine to take advantage of that substitution possibility. In addition, because it is considerably more expensive at current prices, the Mac OS option places few competitive constraints on the market for Intel-compatible PC operating systems.³

The applications barrier to entry provides a second justification for a narrow market definition, according to Judge Jackson. Even if consumers could easily overcome the costs of installing and learning to use an alternative operating system, they would still be deterred from switching because of the relatively small number of application programs written for those operating systems.

Defining the relevant market as including only Intel-compatible PC operating systems leads inescapably to the conclusion that Microsoft enjoys substantial market power. That conclusion is reinforced in Judge Jack son’s mind by the testimony of Frederick Warren-Boulton, one of the government’s economists, suggesting that, because the price of an operating system represents a small fraction of the total cost of a personal computer, which consists of a central processing unit and peripheral devices such as a keyboard, a monitor, and a printer, consumers are not very sensitive to changes in the price of Windows; that is, Microsoft confronts a demand for its flagship product that is highly inelastic. And since Microsoft could raise the price of Windows substantially without inducing large numbers of consumers to switch to alternatives, the company possesses market power.

It is a well-established principle of the neoclassical theory of the firm that a monopolist will never sell at a price where demand is inelastic. That is because a seller facing an inelastic demand can increase its total revenues and profits by reducing output and raising price. Hence, saying that the demand for Windows is inelastic at the current price is the same thing as saying that Microsoft has failed to maximize its profits. Bill Gates has been accused of many things, but making less money when he could have made more is not one of them.

Relevant market definition is supposed to be about the problem of identifying the alternatives to which consumers could turn if confronted with an increase in the price of the product at issue. Judge Jackson’s stated concern that there seem to be few good substitutes for Windows available at current prices misses that important point. The key question is not whether there are good substitutes for Windows at the current price, but rather whether substitution possibilities exist in the event that Microsoft attempts to charge a monopoly price. The fact that a substantial number of alternative Intel-compatible PC operating systems are on the market presently—a list that includes, among others, IBM, Oracle, Sun, AT&T, Hewlett Packard, NEXT, Xerox, Wang, Be, Linux, DEC, Psion, 3COM, Geos, and GEM—points to the conclusion that Microsoft’s pricing discretion is tightly constrained.4 Taken together with the assertion of inelastic demand, the observation that the sellers of alternative Intel-compatible PC operating systems collectively now account for only about 10 percent of sales suggests that the current price is the competitive price. In other words, Microsoft sells Windows at a price where demand is inelastic because it thinks its sales would fall dramatically if price were increased.

Microsoft’s Contracting Practices

But in any case, Microsoft is not accused of exploiting its market power by raising its prices and profits at consumers’ expense. Nor is it charged with acquiring that power by unlawful means. Instead, Judge Jackson contends that the company has used its dominant position in the market for Intel-compatible PC operating systems to force the manufacturers of PC hardware, writers of software applications, Internet access providers, Internet content providers, and other customers and competitors to agree to contractual terms calculated to protect its monopoly by reinforcing the so-called applications barrier to entry. Put differently, Microsoft has forgone profits it could have earned now and in the past to forestall the entry of new software application platforms (or the expansion of sales of existing platforms), thereby preventing the erosion of future profits. By offering discounts, free advertising, access to pre-release (“beta”) versions of software, placement on the Windows desktop, and other valuable considerations, Microsoft “beguiled” customers into entering a “special relationship” with the company.⁵

Stripped to its essentials, the government’s case against Microsoft is one of unlawful predation. And while the list of the alleged victims of Microsoft’s anticompetitive business practices is long, judged by the amount of narrative space devoted to describing how Microsoft attempted to destroy its competition, the case is all about Netscape. According to Judge Jackson, Microsoft employed a variety of tactics to stave off the threat posed by Navigator. That threat, to reiterate, was the prospect that Navigator would expose interfaces to software developers that would eventually weaken the applications barrier to entry by providing an alternative platform for running application programs and, hence, jeopardize the Windows operating system “monopoly.”

Among other tactics, beginning with the release of Windows 95, Microsoft integrated Web-browsing functionality within its operating system and adopted a policy of giving subsequent versions of Internet Explorer away for free. It also took steps to exclude Navigator from important distribution channels. Microsoft did so by preventing the hardware manufacturers to which it had licensed Windows from modifying the sequence of screens seen by consumers on first use, including deleting Internet Explorer from the start sequence, and, if it had also been pre-installed, from giving more favorable placement to Navigator’s desktop icon on the Windows default screen than given to the Internet Explorer icon. With similar ends in mind, Microsoft also entered into agreements with Internet access providers, such as America Online, to bundle its browser with their proprietary client software and to endorse and promote it, paying bounties for every subscriber converted from Navigator to Internet Explorer.⁶ Microsoft likewise entered into agreements with the Disney Channel and other Internet content providers to promote Internet Explorer in return for favorable placement of their own icons on the Windows default screen.

Such policies, in Judge Jackson’s words, “guaranteed the presence of Internet Explorer on every new Windows PC system.”⁷ Netscape was thereby forced into less desirable distribution channels, such as “carpet bombing” consumers with free disks they must install themselves, a task supposedly daunting to novice computer users.⁸ And indeed, Microsoft’s browser strategy ultimately paid dividends: “from early 1996 to the late summer of 1998, Navigator’s share of all browser usage fell from above seventy percent to around fifty percent, while Internet Explorer’s share rose from about five percent to around fifty percent.”⁹ It is worth noting, however, that Microsoft did not begin taking market share away from Netscape until it had invested considerable resources in improving a product that, at the time of its initial introduction, was widely seen as inferior to that of its rival. In particular, “from 1995 onward, Microsoft spent more than $100 million each year developing Internet Explorer. The firm’s management gradually increased the number of developers working on Internet Explorer from five or six in early 1995 to more than one thousand in 1999.”¹⁰

When Internet Explorer 4.0 was released in late 1997, Microsoft had caught up technologically with Netscape, and so, explanations for Internet Explorer’s success based on the company’s restrictive contractual practices cannot be disentangled easily from those based on technological improvements. There is no doubt, however, that Netscape’s own business decisions contributed to its loss of market share.

In September 1996, Microsoft announced the introduction of the “Internet Explorer Access Kit” (IEAK), which allowed Internet access providers conveniently to download Internet Explorer at no charge and to modify certain of its features to create distinctive identities for their services. Indeed, Microsoft allowed Internet access providers to preset the browser’s default home page so that customers would be taken directly to the access provider’s Website whenever they logged on to the Internet, thereby supplying them with valuable advertising and promotional opportunities. “Netscape, by contrast, refused to allow its [Internet access provider] licensees to move Navigator’s home page from Netscape’s NetCenter portal site.”¹¹ It was slow to follow Microsoft’s lead, continuing to charge Internet access providers between $15 and $20 for every copy of Navigator they distributed, and failing to offer them a tool comparable to Microsoft’s access kit. When Netscape finally released a tool kit, known as “Mission Control,” with the same capabilities as IEAK in June 1997, it charged $1,995 per copy. Too much, too late. By that time, about 2,500 Internet access providers had executed licensing agreements for the IEAK.¹²

Whatever else may be said about Microsoft’s allegedly unlawful contractual practices, it is clear they did not work. Netscape continues to distribute copies of Navigator in droves—primarily by allowing consumers to download the program from its Website—despite Microsoft’s attempts to limit its access to important distribution channels. Moreover, although Netscape has been forced by Microsoft’s policy of giving Internet Explorer away for free to also charge a zero price for Navigator—the same strategy Netscape had employed initially to build a dominant market share in browsing software—Netscape’s continued economic viability was assured when, while the trial was still underway, America Online agreed to purchase the company in a stock transaction worth $10 billion.

In sum, if Microsoft’s contractual practices were intended predatorily to drive Netscape from the market, thereby preserving the applications barrier to entry, that strategy failed utterly, as economic theory would predict.¹³ A plausible theory of predation requires a reasonable expectation that the predator, once successful in eliminating the competition, will be able to recoup his losses. The ability to recoup, in turn, presupposes that at some point in the future the predator will raise his price above competitive levels. Sustaining an above-competitive price in the future requires the existence of barriers to entry that prevent rivals from undercutting that price, thereby eroding the above-normal profits that make the whole sequence of events pay off.

The difficulty of recouping predatory losses in general, a point that the Supreme Court has recognized increasingly in recent years,¹⁴ is even more manifest in the case at hand. Because no competitor in the “browser wars” has been eliminated, any attempt on Microsoft’s part to raise the price of Internet Explorer would quickly be defeated. A zero price for browsers is more plausibly seen, not as predatory, but as competitive. This conclusion follows by recognizing that, like the newspaper business, money is to be made by charging advertisers for the right to access consumers rather than by charging consumers for the right to access the Web.

If it were shown to be sound, of course, Judge Jackson’s finding of an applications barrier to entry might provide the basis for a predatory scenario underlying Microsoft’s actions. But a careful study of software markets by economists Stan Liebowitz and Stephen Margolis raises doubts about the existence of “network externalities” said to raise barriers to the entry of new competitors.¹⁵ (A “network good,” such as a telephone, a fax machine, or a computer operating system, is one whose value increases with the number of units in use.) Examining the evolution of market shares of the sellers of a variety of computer software applications, they report three key findings. First, Microsoft’s software applications begin to capture significant shares of total sales only when they garner favorable quality ratings by independent experts. Second, successful entry by Microsoft into software markets brings prices down. Third, there is no evidence of a “tipping point”—a threshold market share that, when reached, allows the dominant seller quickly to gain control of the balance of the market because his product has achieved the status of an industry standard. The market for Web browsers is a case in point: Netscape’s 80 percent market share evidently did not deter Microsoft’s successful entry. The growing popularity of Linux, a cross-platform operating system freely available for download, casts further doubt on the proposition that entry is hopelessly barred.¹⁶

Microsoft’s Defense and Beyond

If Judge Jackson’s prose reads like a carbon copy of the government’s proposed findings of fact, Microsoft has no one but itself to blame. Its defense team apparently adopted a strategy of retaining economists who had spent their careers defending antitrust in the hope that, if they questioned the charges against Microsoft, their credibility would be enhanced. That strategy backfired. Forced on the stand to explain at great length why their opinions in the case at hand were so inconsistent with their previous writings, the substance of their testimony got lost.

Microsoft’s leading expert witness was MIT Professor Richard Schmalensee, the architect of the infamous “shared monopoly” case brought by the FTC against the nation’s leading producers of ready-to-eat breakfast cereals in the early 1970s.¹⁷ That case, which the commission ultimately dismissed “with prejudice,” was based on the novel and untested theory that Kellogg, General Foods, and General Mills had deterred the entry of new rivals by adopting a strategy of “brand proliferation,” pre-empting scarce grocery store shelf space by taking advantage of every opportunity to introduce new cereal brands and, hence, making it more difficult for new entrants to fill niches in the market. On record in support of the notion that it is somehow less desirable to have consumers’ wants satisfied by existing firms than by new entrants, Professor Schmalensee’s testimony was immediately suspect. Not surprisingly, except to disparage data relied on by him to calculate usage shares for different brands of Web-browsing software, Judge Jackson ignored Microsoft’s experts completely.

The findings of fact leave little room for doubting that Microsoft will be found guilty of using its monopoly of Intel-compatible PC operating systems unlawfully. Moreover, Judge Jackson provides some fairly clear signals about the remedies he has in mind. For one, despite the D.C. Circuit’s 1998 decision to the contrary, he seems to think that there is “no technical justification”¹⁸ for Microsoft’s refusal “to supply a version of Windows 98 that does not provide the ability to browse the web, to which users could add a browser of their choice.”¹⁹ The findings of fact also hint at remedies aimed at undoing Microsoft’s restrictions on the ability of computer manufacturers to modify the Windows boot sequence and permitting Internet access providers to promote and distribute Netscape’s Navigator.²⁰ Observers have speculated that in addition to prohibiting Microsoft from imposing certain restrictive contractual terms, Judge Jackson may go so far as to order the company to separate its operating system business from its software applications or declare the operating system an “essential facility,” granting access to Windows proprietary code to all who wish it.

Remedies are always an afterthought in antitrust proceedings, and the difficulties of crafting one in this case that will not interfere unduly with Microsoft’s ability to innovate are great. Judge Jackson’s unexpected appointment of Judge Richard Posner to have a go at mediating a settlement acceptable to Microsoft and the Department of Justice suggests that neither Jackson nor the government has a clear idea about what is to be done. The real tragedy is that Judge Jackson’s findings of fact will provide Microsoft’s rivals with a legal basis to sue for compensation for the damages caused by its allegedly unlawful behavior. Two such claims have already been filed. These follow-on private antitrust suits promise to entangle Microsoft in litigation for years to come, distracting Bill Gates from doing what he does best, namely putting useful new products in the hands of consumers. Aided and abetted by the Department of Justice, that is what Microsoft’s rivals wanted all along and, unfortunately, they seem to have won the first battle.

It is common knowledge that Netscape, Sun Microsystems, and other interested parties lobbied hard to convince the government to battle Microsoft on their behalf, thereby shifting the cost of litigation from shareholders to taxpayers. In the beginning, Microsoft tried to ignore the powerful political forces arrayed against it, hunkering down in Redmond, Washington, to focus on its core businesses. Guess which side prevailed. The moral of the story is that antitrust is not about market efficiency or the welfare of consumers. It is about interest-group politics.21 As such, antitrust is no different from any other modern public policy process in which the police powers of the state can be mobilized selectively to benefit one group at the expense of others.

Notes:

1. Findings of Fact, United States v. Microsoft Corp., Civil No. 98-1232 (TPJ), November 5, 1999, p. 69. Hereinafter referred to as Findings.

2. Quoted in David S. Evans, “Antitrust on Internet Time: Whatever Happened to the Government’s Case in United States v. Microsoft?,” National Economic Research Associates, September 17, 1999, p. 11.

3. Judge Jackson does say, however, that including the Mac OS in the relevant market would not lead him to alter his conclusions. Findings, p. 21.

4. See Richard B. McKenzie and William F. Shughart II, “Is Microsoft a Monopolist?,” The Independent Review, Fall 1998, pp. 165-97.

5. Findings, p. 252.

6. Ibid., p. 139.

7. Ibid., p. 158.

8. Ibid., p. 147.

9. Ibid., p. 372.

10. Ibid., p. 135.

11. Ibid., p. 249.

12. Ibid., pp. 250-51.

13. Recent evidence casts even more doubt on the logic underlying theories of predation. See John R. Lott, Jr., Are Predatory Commitments Credible? Who Should the Courts Believe? (Chicago: University of Chicago Press, 1999).

14. See, for example, Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 113 St. 2578 (1993) (predatory pricing is “rarely tried and even more rarely successful”).

15. Stan J. Liebowitz and Stephen E. Margolis, Winners, Losers & Microsoft: Competition and Antitrust in High Technology (Oakland, Calif.: The Independent Institute, 1999).

16. As the trial was coming to a close, Red Hat, a major distributor of Linux, went public. The company had a market value of $8 billion as of September 10, 1999. Evans, “Antitrust on Internet Time,” p. 15.

17. In the Matter of Kellogg Co. et al., FTC Docket No. 8883 (1972). The theory underpinning the case is laid out in Richard L. Schmalensee, “Entry Deterrence in the Ready-to-Eat Cereal Industry,” Bell Journal of Economics and Management Science, Autumn 1978, pp. 305-27.

18. Findings, p. 177.

19. Ibid., p. 187.

20. Ibid., pp. 218, 247.

21. See, for example, William F. Shughart II, Antitrust Policy and Interest-Group Politics (New York: Quorum, 1990) and Fred S. McChesney and William F. Shughart II, eds., The Causes and Consequences of Antitrust: The Public-Choice Perspective (Chicago: University of Chicago Press, 1995).