by Thomas M. Lenard, Ph.D.
Vice President for Research
The Progress & Freedom Foundation

January 2000

TABLE OF CONTENTS

EXECUTIVE SUMMARY

I. INTRODUCTION

II. MICROSOFT'S OPERATING SYSTEM MONOPOLY

III. NEW MARKET DEVELOPMENTS

IV. ANTICOMPETITIVE ACTS: THE NETSCAPE BROWSER

V. OTHER ANTICOMPETITIVE ACTS

VI. HARM TO CONSUMERS

VII. ALTERNATIVE REMEDIES

VIII. THE HYBRID STRUCTURAL REMEDY

IX. BENEFITS OF THE HYBRID REMEDY

X. THE OPERATING SYSTEM FRAGMENTATION ISSUE

XI. CONCLUSION

Acknowledgements

ABOUT THE AUTHOR

ENDNOTES

CREATING COMPETITION IN THE
MARKET FOR OPERATING SYSTEMS:
A STRUCTURAL REMEDY FOR MICROSOFT
EXECUTIVE SUMMARY

The case against Microsoft raises fundamental questions about the role of antitrust in the digital economy.¹ The government² has presented testimony showing that Microsoft is guilty of serious antitrust violations. The district court has now issued comprehensive Findings of Fact that leave little doubt that the government has proved its case.³ The evidence convincingly establishes that Microsoft possesses monopoly power in the market for Personal Computer (PC) operating systems and that it has engaged in a broad campaign to protect and extend this monopoly through anticompetitive acts in violation of Section 2 of the Sherman Act.

Microsoft's behavior is not just a case of a business practice or two that strays over the line. The district court found that Microsoft engaged in a wide-ranging effort to protect its operating system monopoly, utilizing a full array of exclusionary practices, including exclusive contracts, tying, market-division proposals, and other forms of predatory conduct. Microsoft aimed its artillery at any product or firm that presented even a remote threat to its monopoly power.

Given the court's Findings, it is a virtual certainty that Microsoft will be subject to remedial action of some form. The range of anticompetitive behavior documented by the court, the importance of Microsoft to the computer industry, and the importance of the computer industry to the economy all argue for a serious remedy that will be effective in promoting competition. The remedy should not only address the illegitimate practices Microsoft employed to maintain its operating system monopoly. It should also try to create conditions where Microsoft is not able to leverage its monopoly beyond the desktop into new phases of computing. Whether the chosen remedy is effective in promoting competition in the software industry will have much to say about whether antitrust is viewed as having a constructive role to play in the digital economy.

The parties and the court have an extensive array of potential remedial options at their disposal. These options can be grouped into two general categories — conduct remedies and structural remedies, with intellectual property remedies straddling both these categories.

Conduct remedies would leave Microsoft intact and attempt to constrain its anticompetitive behavior by imposing what would likely be a very detailed set of behavioral requirements — essentially, a regulatory regime tailor-made for one firm. Microsoft's structure — and, importantly, its incentives — would remain the same.⁴ Given those incentives, the challenge for the decree court would be to develop rules that deter Microsoft's anticompetitive behavior and, at the same time, permit Microsoft to be an innovative, aggressive, value-creating competitor in the software industry.

Structural relief takes a different approach, and there are several different ways this could be done in the Microsoft case. In contrast to behavioral rules, a structural solution can change the incentive structure facing the firm, and thereby be much more effective in promoting competition, which, as Richard Posner as written, "is the proper purpose of the antitrust laws."⁵

There is no perfect solution, and choosing among the available alternatives requires a careful weighing of their benefits and costs. Structural remedies will generally be more disruptive and impose greater initial costs than behavioral relief. However, the ongoing costs of regulatory oversight associated with detailed behavioral relief can be very large. Subjecting Microsoft's business, and even its technical decisions, to ongoing regulatory scrutiny by the court and the Department of Justice would be harmful for Microsoft and for consumers as well.

THE HYBRID STRUCTURAL SOLUTION

This paper argues that the best of the available remedies involves restructuring Microsoft so as to create competition in the operating system market. Our proposed plan is a "hybrid" structural solution, so-called because it combines the best features of the frequently discussed "functional divestiture" and "full division" structural remedies:

The hybrid remedy would work in the following way:

The Hybrid Structural Remedy

(click image to see larger version)

The hybrid solution offers a number of advantages relative to the other structural remedies. In contrast to the functional divestiture, which leaves the operating system monopoly in place, the hybrid solution creates direct operating system competition. This is the primary issue of the government's case against Microsoft and the underlying basis for Microsoft's ability to successfully employ anticompetitive tactics. It should, therefore, be the primary remedial objective. The hybrid solution is also less disruptive than the full division remedy, because it only requires a division of operating systems businesses, not the applications products, which are subject to greater competition and were not the focus of the government's case. The applications businesses would, therefore, not be divided, unless Microsoft proposed to do so.

MINIMAL ONGOING SUPERVISION REQUIRED

While the most important benefit of the hybrid solution is that it creates operating system competition, the second most-important benefit is that it does not require the type of detailed ongoing supervision by the decree court that characterized the AT&T settlement. The hybrid remedy would require only a limited number of short-term restrictions whose purpose would be solely to prevent the successor companies from undoing the decree. The successor companies would be prohibited from acquiring one another. They would also be prohibited from entering into joint marketing or development agreements with each other that might have effects similar to a merger, and they would not be allowed to hire each other's employees.

The lines between permissible and impermissible behavior would be clear, unambiguous and easily enforceable. In contrast to the AT&T decree, Microsoft successor companies would not be subject to line-of-business restrictions. Successor companies could develop new product lines and compete with other successor companies without restriction. Obviously, mergers and acquisitions involving non-successor companies would be subject to normal antitrust merger review.

The restrictions on the successor companies are intended to be short-term, lasting for only three to five years. Given the rapid pace of change in the software industry and the expected boost to competition expected from the hybrid remedy, the competitive landscape will look very different at the end of a three-to-five-year period. At the end of that period, the restrictions would no longer be necessary and the decree would be terminated. The Microsoft successor companies would be treated like any other company subject to the normal constraints of antitrust law.

THE FRAGMENTATION ISSUE

Perhaps the major concern expressed with respect to operating system competition is that it would "fragment" the operating system standard. Windows supposedly would evolve into incompatible operating systems and consumers would lose the benefits of standardization. Consumers would incur either the costs of incompatibility between applications and the new operating systems or the costs of having applications developers perform expensive "ports," or rewrites, of their software in order to make their existing applications work with the new operating systems.

The fragmentation argument is ultimately an argument against the premise on which our economic system and the antitrust laws is based, which is that competition best serves the interests of consumers. The fears of fragmentation of the Windows standard are unwarranted, because they are inconsistent with the fundamental economics that would characterize the post-breakup operating system market. The three Windows companies would have very strong incentives to retain pre-existing network externalities. Simply put, consumers want access to a large pool of applications; software developers, in turn, want access to a large pool of consumers. To not maintain compatibility would risk losses with both these groups, which none of the Windows companies would want to do. The costs of switching and the benefits of network effects create powerful incentives for consumers to stay with their existing operating system standard and for the new operating system companies to retain compatibility with the installed base and each other.

REMEDY WOULD REPLACE MONOPOLY WITH COMPETITION

The issue at the core of the Microsoft antitrust case is the Windows operating system monopoly. The solution to that problem is to create operating system competition, which is what the hybrid solution is designed to do. A major additional benefit is that ongoing oversight of Microsoft by the decree court is largely unnecessary.

Other remedies under consideration leave the existing incentive structure intact and, therefore, must rely on a variety of conduct rules to curtail the behavior expected to result from those incentives. These other remedies would be less effective at achieving competition, and would also subject Microsoft to an intrusive and damaging regulatory regime.

The creation of three Windows companies would immediately replace monopoly with competition in the market for operating systems. The Windows companies would compete on the basis of price, reliability, quality and other features. Competition in the operating system market is also likely to spur competition and innovation in related software and hardware markets. Implementation of the hybrid solution can reinvigorate competition in all these critical sectors, which are so important to the overall health of our economy.

I.  INTRODUCTION

The case against Microsoft raises fundamental questions about the role of antitrust law in the digital economy. The government has presented testimony showing that Microsoft is guilty of serious antitrust violations. The district court has now issued comprehensive Findings of Fact that leave little doubt that the government has proved its case. The evidence convincingly establishes that Microsoft possesses monopoly power in the market for Personal Computer (PC) operating systems and that it has engaged in a broad campaign to protect and extend this monopoly through anticompetitive acts in violation of Section 2 of the Sherman Act.

Microsoft's behavior is not just a case of a business practice or two that strays over the line. The district court found that Microsoft engaged in a wide-ranging effort to protect its operating system monopoly, utilizing a full array of exclusionary practices, including exclusive contracts, tying, market-division proposals, and other forms of predatory conduct. Microsoft aimed its artillery at any product or firm that presented even a remote threat to its monopoly power.

Given the court's Findings, it is a virtual certainty that Microsoft will be subject to remedial action of some form. Whether the chosen remedy is effective in promoting competition in the software industry will have much to say about whether antitrust is viewed as having a constructive role to play in the digital economy.

The parties and the court have an extensive array of potential remedial options at their disposal. These options can be grouped into two general categories — conduct remedies and structural remedies, with intellectual property remedies straddling both these categories.

Conduct remedies would leave Microsoft intact and attempt to constrain its anticompetitive behavior by imposing what would likely be a very detailed set of behavioral requirements — essentially, a regulatory regime tailor-made for one firm. Microsoft's structure — and, importantly, its incentives — would remain the same. Given those incentives, the challenge for the decree court would be to develop rules that deter Microsoft's anticompetitive behavior and, at the same time, permit Microsoft to be an innovative, aggressive and value-creating competitor in the software industry.

Structural relief takes a different approach, and there are several different ways this could be done in the Microsoft case. In contrast to behavioral rules, a structural solution can change the incentive structure facing the firm, and thereby be much more effective in promoting competition, which, as Richard Posner as written, "is the proper purpose of the antitrust laws."

There is no perfect solution, and choosing among the available alternatives requires a careful weighing of their benefits and costs. Structural remedies will generally be more disruptive and impose greater initial costs than behavioral relief. However, the ongoing costs of regulatory oversight associated with detailed behavioral relief can be very large. Subjecting Microsoft's business, and even its technical decisions, to ongoing regulatory scrutiny by the court and the Department of Justice would be harmful to Microsoft and to consumers as well.

This paper argues that the best of the available remedies involves restructuring Microsoft so as to create competition in the operating system market. Our proposed plan — a "hybrid" structural solution — would separate the operating system portion of Microsoft from the rest of the company and then subdivide that portion into three equivalent operating system companies, each with full ownership over the relevant intellectual property and an equal share of employees, contracts and other resources to go with it.

This remedy would immediately replace monopoly with competition in the market for operating systems. Competition in the operating system market would, in turn, promote competition in adjacent software markets and perhaps in hardware markets as well. The proposed remedy would require only minimal oversight during a relatively brief transition period, after which no extraordinary oversight at all would be required. In sum, it would avoid the type of continuing supervision that turned the AT&T decree court into a mini-Federal Communications Commission for years following the AT&T settlement.

The hybrid solution we propose is designed to create an incentive structure that is conducive to competition. This is why ongoing oversight of Microsoft by the decree court is largely unnecessary. Other remedies under consideration leave the existing incentive structure intact and, therefore, must rely on a variety of conduct rules to curtail the behavior expected to result from those incentives.

II.  MICROSOFT'S OPERATING SYSTEM MONOPOLY

Most of the activities in which Microsoft has engaged would not be problematic if Microsoft did not have a monopoly and its customers had someplace else to go. There are, however, no viable operating system alternatives available. The district court's Findings leave no serious doubt that Microsoft has monopoly power in the market for PC operating systems. Findings ¶ 34.

• Microsoft's operating system monopoly is protected by significant barriers to entry. Findings ¶¶ 36-52.

• Microsoft's market share, combined with the absence of any new entry, means that there are no "realistic commercial alternatives" to Windows. Findings ¶ 53.

A high market share does not necessarily imply market power. Indeed, it is possible for a market to be competitive even if dominated by a single firm — if new entry is easy. But, this is clearly not the case in the market for PC operating systems.

Once a dominant firm, such as Microsoft, becomes established, the economics of software markets make entry difficult. Like many other software markets, the operating system market is characterized by pervasive network effects (also called demand-side economies of scale)⁶. Users of compatible programs — for example, an operating system and an applications program or two compatible applications programs — are on the same network. The value of a program increases with the number of users on the network, for a variety of reasons. Windows 90-percent market share, for example, supports a large base of applications programs and assures its users that developers will continue to put resources into developing programs that are compatible with Windows.

These network effects create virtually insurmountable problems for potential entrants. Even if applications programs were available to support a new operating system (which they are not), users of the dominant system would face significant costs if they were to switch. These would include the costs of transferring or reentering data and files to the new system, and the costs of learning the new system.

Applications programs to support a new operating system will not be available in any quantity, however, because it is typically not profitable for developers to devote resources to developing programs for an alternative operating system that only has a small share of the market. In addition, software is characterized by large costs of development ("first-copy" costs) that are typically "sunk", and low costs of replication and distribution. This cost structure further diminishes the incentive to develop software for markets that are not well established. The lack of an existing base of compatible applications programs, or even the prospect that such programs will be forthcoming, makes it very difficult for a new operating system to enter the market. The court referred to this as the "applications barrier to entry." Findings ¶¶ 30-31, 36-44. The difficulties encountered by IBM with its OS/2 operating system provide a powerful example of how high the barriers to entry are, even for large, well-capitalized companies. Findings ¶ 46.

Because of Microsoft's monopoly, and because entry on any significant scale is unlikely, the original equipment manufacturers (OEMs) who are Microsoft's principal customers have no viable alternatives. The court found that "[w]ithout significant exception, all OEMs pre-install Windows on the vast majority of PCs that they sell, and they uniformly are of a mind that there exists no commercially viable alternative to which they could switch in response to a substantial and sustained price increase or its equivalent by Microsoft." Findings ¶ 54.

The court also concluded that Microsoft's monopoly allows it great latitude in pricing. Microsoft's position allows it to price Windows "without consider[ing] the prices of other vendors' Intel-compatible operating systems." The court found this "probative of monopoly power." Findings ¶ 62.

Finally, Microsoft's astronomical profit margins are indicative of monopoly power. The most recent figures show Microsoft with a firm-wide profit margin of 51.9 percent, a figure that is far in excess of other industry leaders. Recent profit margins for other industry leaders were: Cisco, (27.8 percent), IBM (12.6 percent), Intel (28.1 percent), Oracle, (15.8 percent), and Sun (12.1 percent)⁷. Microsoft's ability to earn profits far above the norm (and to do so for an extended period of time) without attracting substantial new entry is also a strong indication of market power.

III.  NEW MARKET DEVELOPMENTS

Critics of the government's case have suggested that the market, if left to its own devices, will erode Microsoft's monopoly position over time, and point to five specific developments to argue that this is already happening: the acquisition of Netscape by America Online (AOL); the subsequent merger of AOL with Time Warner; the development of the Linux operating system; the growing popularity of hand-held information appliances, such as Palm computers; and, the growth in Web-based applications.

If developments such as these did provide meaningful competition to Microsoft, then the case for a significant remedy would be weakened. However, each of the developments cited (with the exception of the AOL/Time Warner merger, which was announced after the Findings of Fact were issued) was considered in detail by the trial court. In each case, the court found no discernable threat to the Microsoft monopoly. More generally, the court was unable to find anything on the horizon that was likely to erode Microsoft's dominance in the foreseeable future.

A.  America Online/Netscape Merger

AOL's acquisition of Netscape Communications occurred during the course of the trial. AOL's core business is "content," specifically the presentation of online content to consumers. AOL is not in the business of developing, licensing, or supporting software. At the time of the acquisition, Netscape had three core businesses — the NetCenter portal, the browser, and a suite of server software.

The AOL/Netscape merger does nothing to affect competition in the market for PC operating systems, which is the subject of the Microsoft case. When AOL acquired Netscape, it purchased Netscape's content business — its NetCenter portal — and turned Netscape's server software business over to Sun Microsystems. The merger does not appear to have done anything to stop the decline in Netscape's browser market share or to reinvigorate competition in that market. In fact, even AOL, which has continued to use Internet Explorer, is not providing a market for the Netscape browser.

It is true, of course, that Microsoft has products — for example, its MSN Internet access service and its MSN portal and other content — that are competitive with AOL. In particular, Microsoft has recently stepped up its promotion of its Internet access service, which is directly competitive with AOL. The existence of such competition, however, does not imply that AOL produces (or is ever likely to produce) software that is competitive with Microsoft's Windows and other operating system products, which are the source of its market power.

B.  America Online/ Time Warner Merger

Similarly, it has been suggested that the subsequent merger of AOL with Time Warner undermines Microsoft's market position and somehow obviates the need for a significant remedy.⁸ Since neither AOL nor Time Warner produces software or operates in the markets at issue in the antitrust case, it is difficult to see the rationale for this argument.

AOL is an Internet service provider whose core business is content, as noted above. Time Warner, an entertainment/publishing conglomerate, is also a content provider. Time Warner can also, through its cable holdings, provide broadband access to approximately one-fifth of the U.S. cable market. The merger will enable AOL to improve its offerings to customers with respect to both content and high-speed access.

AOL's purchase of Time Warner undoubtedly strengthens its position against potential competitors for its core Internet access and content businesses, including Microsoft. But since neither AOL nor Time Warner competes in the market for PC operating systems, it is difficult to see how their merger is relevant to the antitrust case. Indeed, the ability of AOL and other firms to provide improved offerings over the Internet will increase the demand (not undermine the market) for PCs equipped with Windows and with AOL's preferred browser, Internet Explorer.

C.  Linux

Although the Linux operating system has received considerable publicity, it remains a secondary operating system used almost exclusively for server computers that operate computer networks. Linux has virtually no presence on the desktop. There is no indication Linux is able to overcome the applications barrier to entry in that market. The court found that Linux presents no significant challenge to Microsoft's monopoly, or at least cannot do so for many years. See Findings ¶ 50.

D.  Information Appliances

It has been suggested that the introduction of small, hand-held devices that can perform some computing tasks and can be used to access the Internet presents a challenge to the PC.

The trial court addressed this issue and concluded that "no single type of information appliance, nor even all types in the aggregate, provides all of the features that most consumers have come to rely on in their PC systems and in the applications that run on them. Thus, most of those who buy information appliances will do so in addition to, rather than instead of, buying an Intel-compatible PC system." The court concluded that "for the foreseeable future" hand-held computers and other limited function devices simply pose no threat to Microsoft's Windows monopoly. Findings ¶ 23. This sentiment was echoed by Microsoft Chairman Bill Gates in a May 1999 Newsweek article:

E.  Web-Based Computing

It has been argued that the availability of Web-based applications, which allow PCs to perform tasks using software that is accessed through the Internet, undermines the Windows operating system monopoly. As the court observed, Web-based computing has yet to attract substantial consumer interest. "In part, this is because PC systems, which can store and process data locally as well as communicate with a server, have decreased so much in price as to call into question the value proposition of buying a network computer system." In addition, "few consumers are in a position to turn from PC systems to network computer systems without making substantial sacrifices." The court concluded that the day when network computing becomes a viable alternative to PC-based computing does not "appear imminent." Findings ¶ 26.

If and when that day comes, Microsoft will be well positioned, as a result of its near 100-percent share of the market for newly installed Web browsers (see below). If Microsoft controls the browser that is the gateway for Web-based applications, Microsoft retains the standard-setting role and, in addition, can exercise control over the server software on which Web-based applications depend.

IV.  ANTICOMPETITIVE ACTS: THE NETSCAPE BROWSER

The evidence presented at trial establishes that Microsoft engaged in a wide range of anticompetitive practices to maintain its monopoly and forestall the emergence of any technology that might compete with Windows. As the district court made clear, "Microsoft's business strategy" is to "direct[ ] its monopoly power toward inducing other companies to abandon projects that threaten Microsoft and toward punishing those companies that resist." Findings ¶ 132. "Microsoft's corporate practice" is "to pressure other firms to halt software development that either shows the potential to weaken the applications barrier to entry or competes directly with Microsoft's most cherished software products." Findings ¶ 93.

Microsoft was especially concerned about technologies, such as Netscape's browser and Java, that could support platform-independent computing and thereby erode Microsoft's market position.

The threat that Netscape posed was to provide a platform for applications that was operating-system neutral. In other words, applications could be written for the Netscape browser and run on a variety of operating systems, not just Windows. Microsoft had a clear incentive to protect its Windows monopoly and the resulting profits against the threat that Netscape posed. It also had the ability — by leveraging its Windows monopoly — to do so. Microsoft addressed these threats with an anticompetitive repertoire that included a variety of exclusive dealing arrangements, bundling and tying, proposals to divide the market, predatory coercion and threats.

Netscape introduced Navigator in December 1994. The following May, Bill Gates wrote a memo warning his colleagues at Microsoft that Netscape was "pursuing a multi-platform strategy where they move the key API into the client to commoditize the underlying operating system."¹⁰ Findings ¶ 72. Any development that "commoditized the operating system" would neutralize the power of the Windows monopoly both as a leading source of Microsoft's extraordinary profits and as a source of leverage into other markets.

In response to the Netscape threat, Microsoft undertook a broad array of anticompetitive practices to increase the market share of its Internet Explorer. Despite the opportunity to make money by charging a positive price for Internet Explorer (which Netscape was doing for its browser), Microsoft "paid huge sums of money, and sacrificed many millions more in lost revenue every year, in order to induce firms to take actions that would help increase Internet Explorer's share of browser usage at Navigator's expense." Findings ¶ 139.

As the court recognized, these efforts were ultimately successful. Findings ¶¶ 359-376. Internet Explorer's market share, measured by actual Internet usage ("hits" on leading Web sites), is now about 75 percent.¹¹ In the key corporate market, Microsoft has a 65-percent market share. As users continue to upgrade to the newer versions of Windows, which have Internet Explorer integrated into the operating system, the market share of Internet Explorer will continue to rise.

A.  Market Division Proposal

Microsoft first proposed to divide the browser market with Netscape, retaining the Windows market for itself. See Findings ¶¶ 79-89. As the district court observed:

At the time Microsoft presented its proposal, Navigator was the only browser product with a significant share of the market and thus the only one with the potential to weaken the application barrier to entry. Thus, had it convinced Netscape to accept its offer of a "special relationship," Microsoft quickly would have gained such control over the extensions and standards that network-centric applications (including Web sites) employ as to make it all but impossible for any future browser rival to lure appreciable developer interest away from Microsoft's platform. Findings ¶ 89.

When Netscape did not accept the Microsoft proposal, Microsoft began the broad campaign described below.

B.  Exclusive Arrangements with Original Equipment Manufacturers (OEMs)

Most consumers obtain their Internet browser either preinstalled on their computer or bundled together with the installation software provided by their Internet Access Provider (IAP). These two channels are efficient and effortless ways for users to gain access to an installed browser. Microsoft was able to block Netscape from these two principal installation channels through a variety of exclusive arrangements with OEMs and IAPs. Findings ¶¶ 144-148.

Given Netscape's head start, Microsoft concluded that it could not defeat the Netscape threat without, in the words of a Microsoft executive, "leveraging the OS asset to make people use IE instead of Navigator." Findings ¶ 169. Microsoft did this by tying Internet Explorer to Windows. Since July 1995 (with the exception of a few months in 1997) no OEM has been able to license a copy of Windows that did not include Internet Explorer. Findings ¶ 202. The tying of Internet Explorer to Windows has functioned as a de facto exclusive arrangement, because OEMs have little incentive to install multiple software applications that perform the same function.

Microsoft required OEMs to license and distribute Internet Explorer on every PC sold with Windows. Findings ¶ 155. "This policy has guaranteed the presence of Internet Explorer on every new Windows PC system." Findings ¶ 158. For Windows 98, the contractual tying was replaced with technological tying, which Microsoft achieved by integrating Internet Explorer into Windows. In doing this, Microsoft also, in the words of one of its executives, made "running any other browser…a jolting experience." Findings ¶¶ 160-161. The court found "no technical justification" for technically integrating Internet Explorer into Windows. Findings ¶¶ 177, 180, 186, 191.¹²

Microsoft also imposed license restrictions on OEMs that prohibited them from altering Microsoft's prescribed boot-up sequence, including the "first screen" that the user sees. Findings ¶¶ 202-229. Microsoft forbade OEMs from removing or obscuring Internet Explorer and threatened to penalize OEMs that preinstalled and promoted Navigator. For example, Microsoft threatened to terminate Compaq's license for Windows 95 when it learned that Compaq planned to remove the Internet Explorer icon from the desktop of its Presario computers and replace it with an icon that represented Navigator. Findings ¶¶ 205-206.

Finally, even though Internet Explorer was already included with Windows, Microsoft offered OEMs positive incentives in the form of favorable prices for Windows in exchange for commitments to promote Internet Explorer exclusively. In sum, Microsoft made its software prices and its access to technical information and assistance for Windows dependent on OEMs' agreements not to deliver Netscape. See generally Findings ¶¶ 230-241.

The court found that these efforts largely succeeded in excluding Navigator from the crucial OEM distribution channel. Before 1996, Navigator had enjoyed a substantial and growing presence on the desktop of new PCs. By the beginning of 1998, a Microsoft executive was able to report that of the 60 OEM subchannels (15 OEMs each with the following four subchannels — corporate desktop, consumer/small business, notebook and workstation PCs), Navigator was being shipped on only four, mostly with the icon not on the desktop. Sony featured Navigator in a folder rather than on the desktop. And Gateway shipped a separate CD-ROM, rather than pre-installing Navigator. By 1999, Navigator was present in only a tiny fraction of PCs shipped. Findings ¶ 239.

C.  Exclusive Arrangements with Internet Access Providers (IAPs)

Microsoft gave the leading IAPs and Online Service Providers valuable access to the Windows desktop in exchange for their commitment to distribute and promote Internet Explorer and to refrain from promoting any non-Microsoft browser. In doing so, Microsoft sacrificed substantial revenues, all to promote a product it was giving away for free. See generally Findings ¶¶ 242-310.

In exchange for distributing AOL's access software in Windows and placing an AOL icon in a favorable position on the desktop, Microsoft obtained "virtual exclusivity" from AOL, prohibiting AOL from promoting any browser other than Internet Explorer and limiting AOL's shipments of other browsers to 15 percent or less. The agreement also precluded AOL from informing its subscribers that they could download Netscape's browser unless a subscriber specifically made such a request.

Microsoft obtained similar arrangements with most of the other large Internet Access Providers. Findings ¶¶ 273-304. These were successful in dramatically reducing the distribution of Navigator. The court found that 14 of the 15 largest access providers were subject to "the most severe restrictions" concerning distribution of non-Microsoft browsers. See Findings ¶¶ 307-310.

Microsoft also entered into exclusionary arrangements with Internet Content Providers (ICPs). For example, Microsoft gave Disney.com exclusive placement on Microsoft's channel bar in return for a commitment from Disney.com to not promote Netscape or offering Netscape any compensation. See generally Findings ¶¶ 311-336.

V.  OTHER ANTICOMPETITIVE ACTS

The campaign against the Netscape browser was the most prominent, but not the only case of anticompetitive behavior on the part of Microsoft. The court found that Microsoft was ready to use similar tactics, and in particular to use the leverage provided by its operating system monopoly, against a variety of market developments it viewed as threatening. 

A.  Java

Java is a programming language developed by Sun Microsystems that allows applications to run on different operating systems without being ported. Microsoft's anticompetitive acts in the browser market accomplished the dual purpose of also hindering the development of Java. See Findings ¶¶ 387-406.

Like the Netscape Navigator, Java promised interoperability. Programmers could write software that could run on different operating systems. This not only would attenuate Microsoft's monopoly profits from the operating system, but also would remove the competitive advantages enjoyed by its complementary products — i.e., applications and server products.

Microsoft's actions in the browser market slowed the spread of Java, because Navigator was the key means of distributing Java. But Microsoft also worked to impede the cross-platform characteristics of Java by distributing a Microsoft version of Java that was dependent on Windows and other proprietary Microsoft technology. See Findings ¶¶ 387-394. This, at a minimum, delayed the competitive threat posed by Java and the benefits of operating-system neutral applications.

The court found that "[h]ad Microsoft not been committed to protecting and enhancing the applications barrier to entry,…Microsoft would not have taken efforts to maximize the difficulty of porting Java applications written to its implementation and to drastically limit the ability of developers to write Java applications that would run in both Microsoft's version of the Windows runtime environment and versions complying with Sun's standards." Microsoft's actions "resulted in fewer applications being able to run on Windows than otherwise would have" because "Microsoft felt it was worth obstructing the development of Windows-compatible applications where those applications would have been easy to port to other platforms." Findings ¶ 407.

B.  Intel

Microsoft was able to induce a company as large and powerful as Intel to simply stop developing software that was competitive with Microsoft. Intel was developing software called Native Signal Processing (NSP) that would enhance the video and graphics performance of Intel processors. See Findings ¶¶ 94-103. This software would be provided directly to OEMs and other purchasers of Intel chips, bypassing the Microsoft operating system. If more software came with, or was built into, the hardware rather than being dependent on the operating system, the Windows monopoly might have been weakened.

Microsoft was successful in getting Intel to halt the project, to the detriment of consumers. The court found that "[e]ven as late as the end of 1998,…Microsoft still had not implemented key capabilities that Intel had been poised to offer consumers in 1995." Findings ¶ 101.

According to the court, "Microsoft was not content to merely quash Intel's NSP software." Subsequently,

C.  IBM

IBM produced two software products that were directly competitive with Microsoft — OS/2, which competed with Windows, and Lotus SmartSuite, which competes with Microsoft Office. IBM is also a PC company, and like all other OEMs, needs to sell its PCs pre-installed with Windows. Microsoft attempted to use its leverage on the hardware side to persuade IBM to stop competing on the software side. When IBM refused, Microsoft punished the PC arm of IBM with higher prices, a delayed license for Windows 95, and the withholding of technical and marketing support. The court's Findings show that IBM incurred substantial costs for its recalcitrance. See Findings ¶¶ 115-132.

VI.  HARM TO CONSUMERS

Microsoft's anticompetitive conduct has substantially harmed competition and consumers in many ways. As the district court found, "[m]any of [Microsoft's] actions have harmed consumers in ways that are immediate and easily discernible," while others have "caused less direct, but nevertheless serious and far-reaching, consumer harm by distorting competition." Findings ¶ 409.

The operating system monopoly has enabled Microsoft to charge higher prices than would a competitive firm. The court concluded clearly that "Microsoft's actual pricing behavior is consistent with the proposition that the firm enjoys monopoly power in the market for Intel-compatible PC operating systems." See Findings ¶ 62.

Because its monopoly is so profitable, it pays Microsoft to expend substantial resources to maintain its monopoly position. The court found that "Microsoft expends a significant portion of its monopoly power" on measures "that augment and prolong that monopoly power." Findings ¶ 66. These resources, which could be spent on innovation or other productive activities, or returned to consumers in the form of lower prices, represent a dead-weight loss of productive resources for society.¹³

The court pointed to a number of specific instances in which Microsoft restricted consumer choice:

• Microsoft forced consumers who preferred Navigator to take "both browser products at the cost of increased confusion, degraded system performance, and restricted memory." Findings ¶ 410.

• Microsoft's forced inclusion of Internet Explorer in the operating system "unjustifiably jeopardized the stability and security of the operating system." Findings ¶ 174.

• Microsoft's desktop restrictions precluded OEMs from using tutorials and registration programs and other consumer-friendly features that the OEMs had spent millions of dollars developing. The OEMs "believed that the new restrictions [imposed by Microsoft] would make their PC systems more difficult and more confusing to use, and thus less acceptable to consumers" and that they "would increase product returns and support costs and generally lower the value of their machines." Findings ¶ 214.

It is impossible to quantify the innovation that did not take place or the new entrants deterred by Microsoft's aggressive behavior. The nearly complete lack of differentiation between the offerings of PC manufacturers reflects the absence of competition and the restrictions imposed by Microsoft to preserve its monopoly position.

The lack of competition delays consumer access to a variety of innovations, not only in operating system functionality, but in applications software and hardware as well. This is the case because Microsoft's dominant position allows it to determine the pace of innovation for both applications and computing hardware, both of which need to be compatible with the operating system.

Much of the government's case concerned the suppression of the Netscape browser and Java innovations that might have permitted applications to be operating-system neutral. At the very least, Microsoft has been successful in delaying the development of such cross-platform solutions.

The promise of operating-system neutral computing was also that it would inject competition into the market for operating systems, which would foster innovation throughout the industry. The court found that "Microsoft has retarded, and perhaps altogether extinguished, the process by which these two middleware technologies [i.e., Navigator and Java] could have facilitated the introduction of competition into [the] important market" for PC operating systems. Findings ¶ 411.

The conduct proved at trial could suppress competition in new markets beyond the desktop operating system market. Microsoft's campaign against Netscape gives it effective control over the browser, which is the client software for the Internet. This control over the client could be leveraged onto server software markets. If, for example, Internet Explorer, which will be on every desktop PC in every business in America, does not properly run non-Microsoft corporate applications running on the server, enterprise customers would have an incentive to switch to Microsoft software, which presumably would run optimally. This would give Microsoft enterprise applications software and Microsoft's server operating system an artificial advantage and would deprive consumers of the benefits of competition in these emerging markets.

Finally, and perhaps most importantly, Microsoft's repeated anticompetitive conduct has had and, unless effectively checked, will continue to have a chilling effect on the entire industry. As the court concluded:

VII.  ALTERNATIVE REMEDIES

The range of anticompetitive behavior documented by the district court, the importance of Microsoft to the computer industry, and the importance of the computer industry to the economy all argue for a serious remedy that will be effective in promoting competition. The remedy should not only address the illegitimate practices Microsoft employed to maintain its operating system monopoly. It should also try to create conditions where Microsoft is not able to leverage its monopoly beyond the desktop into new phases of computing. Microsoft's campaign against the Netscape browser was effective not only in defending its Windows monopoly. It also will give Microsoft control over the browser, which is the gateway to the Internet. As discussed in the last section, this could give Microsoft the ability to leverage its monopoly onto the server and the server software market.

There are two general classes of remedies that could be employed — conduct remedies and structural remedies.¹⁴ Intellectual property remedies straddle both of these categories.

Conduct remedies would leave Microsoft intact and attempt to constrain its anticompetitive behavior by imposing a set of behavioral requirements — essentially, a regulatory regime tailor-made for one firm. Microsoft's structure — and, importantly, its incentives — would remain the same.¹⁵ The challenge for the decree court would be to develop rules that effectively deter anticompetitive behavior, given that such behavior would continue to be in Microsoft's interest.

Structural relief takes a different approach. As the name suggests, structural relief involves restructuring the company. There are several different ways in which this could be done, as we discuss below. Structural relief will generally be more disruptive and involve greater initial costs than conduct relief. However, a structural solution can change the incentives facing the firm, and thereby be much more effective than behavioral restrictions in promoting competition.¹⁶ 

A.  Conduct Remedies

In a case as complex as this, conduct relief needs to prohibit a wide variety of violations in which Microsoft has engaged, as well as anticipate conduct that might be substituted in order to get around the new rules. This is not a case where one or two practices are at issue. Given the range of illegitimate behavior documented by the court and the complexity of the software industry, a lengthy list of conduct restrictions and requirements would be called for. Some of the measures that have been suggested for Microsoft include the following:

• a ban on exclusive arrangements, such as the contracts Microsoft had with OEMs restricting their ability to distribute another party's products;

• restrictions on pricing and other contract provisions that might not be exclusionary on their face, but would provide strong incentives to achieve the same result;

• a requirement for transparent, non-discriminatory pricing for Windows;

• a requirement for non-discriminatory provision to OEMs of technical information and support concerning Windows and other products;

• a prohibition against tying applications to the operating system;

• a prohibition on boot-up or first-screen restrictions; and,

• requirements for non-discriminatory access for independent software vendors (ISVs) to Microsoft technical information and support.

Conduct prohibitions as complex as these would be difficult to enforce and could likely be circumvented in unpredictable ways. The decree court and the Department of Justice would function as regulatory agencies, as they did in the AT&T case. They would have to monitor the operations of a firm with 30,000 employees producing dozens of technologically sophisticated products. Because enforcement of conduct restrictions would involve ongoing oversight of virtually all of Microsoft's operations, including new product introductions, it would inevitably interfere with Microsoft's ability to develop new products and compete. And, because Microsoft has dealings throughout the software industry, oversight of Microsoft by the decree court might well indirectly mean oversight of other firms as well. In sum, the imposition of behavioral restrictions on Microsoft might have the unintended effect of impeding innovation, at least from Microsoft, and perhaps from other firms as well. 

B.  Structural Remedies

Structural remedies involve greater up-front disruption, but have the benefit of lower ongoing costs. Two general types of restructuring have been proposed as the basis for a Microsoft remedy.

• The functional divestiture remedy would divide Microsoft along product lines, into an operating systems company and an applications company that controls Microsoft's non-operating system product portfolio.

• The full division remedy would divide the company into several identical, integrated firms, each with full access to all of Microsoft's intellectual property and full rights to sell every product.¹⁷

There is a third alternative — a hybrid structural remedy — that combines the full division and the functional remedies. The hybrid remedy offers the advantages of these two approaches without many of the disadvantages. We discuss this remedy in detail in the next section

The functional divestiture remedy involves breaking up Microsoft into separate operating systems and applications companies. See Figure 1. Its fatal flaw is that it does nothing to limit Microsoft's monopoly power in the operating systems market, which is what the current case is all about.

(click image to see larger version)

Figure 1: Functional Divestiture

Dividing Microsoft along functional lines is analogous to the AT&T settlement, which divided the Bell System into seven non-competing local exchange regional monopolies and one long-distance carrier. The logic of the AT&T situation does not, however, apply to the Microsoft case. In AT&T, the purpose was to structurally separate the competitive product — long distance — from the product that was to remain a regulated monopoly — local exchange service. The goal of the divestiture was to promote long-distance competition by removing the incentive and ability to leverage the local-exchange monopoly power onto the long distance market.

Microsoft does not present an analogous situation. The operating system monopoly is unregulated and no one is suggesting public utility regulation as a serious alternative. An AT&T-type remedy would simply leave the operating system monopoly in place. This is the competitive problem at the core of the Microsoft case.

Moreover, ongoing regulation would be required to maintain the initial line-of-business boundaries, as it was for AT&T. Otherwise, the monopoly operating system company could leverage its market power to enter non-operating system markets. Line-of-business restrictions were used to prevent this in the AT&T case, but this involved pervasive ongoing regulation. The prospect of a court continually drawing lines between the operating system and applications in a field as rapidly moving as software development is troublesome. As was the case with AT&T, prohibitions on discrimination also would be needed to prevent de facto integration with favored "partners." Thus, this remedy would impose the up-front costs of restructuring, combined with the ongoing costs of conduct relief — arguably, the worst of both worlds.

2.  Full Division Remedy

The full division remedy would divide Microsoft into three identical vertically integrated companies. See Figure 2.

(click image to see larger version)

Figure 2. Full Division Remedy

Each of the new companies would have full rights to all of Microsoft's intellectual property, including any internal documentation and all development work currently underway. The companies would also obtain the rights to Microsoft's trademarks, such as the use of the Windows name. Each of the new companies would also receive an equal share of Microsoft's employees, sales contracts, investments and cash assets.

Each share of Microsoft stock would be exchanged for one share in each of the new Windows companies. Top managers would be required to divest shares in competing entities to avoid conflict of interest.

The full division remedy would require only narrow restrictions on the future conduct of the resulting new integrated Windows competitors. They would be prohibited from acquiring one another or contractually recombining through joint marketing or development agreements.

The full division remedy would create direct competition in the market for operating systems, which is the primary remedial goal. By avoiding functional divisions, the full division remedy also maintains efficiencies of vertical integration. On the other hand, by splitting up all of the existing Microsoft, with its 30,000 employees, the full division remedy is more disruptive than necessary.

Moreover, this remedy divides up a range of applications products that have not been at issue in the current case. Microsoft does not have a dominant position in some of these products, and to subdivide them among three firms may diminish their viability. This would not be in the interest of competition.

3.  One-Time Licensing Auction

A one-time auction of Microsoft's intellectual property is a variant of the full division remedy that attempts to create Windows competitors by giving other parties the opportunity to purchase the Windows intellectual property. While the full division remedy creates competitors that have Microsoft's intellectual property and an equal share of its other assets — including employees, sales contracts and financial assets — the licensing auction creates competitors that, at least initially, only have the intellectual property.

The licensing remedy avoids the costs associated with breaking up Microsoft. But it is questionable whether meaningful competition would emerge. In particular, the licensees would likely be at a serious disadvantage vis-à-vis Microsoft, which would be fully staffed with employees who are knowledgeable about Microsoft's products. Indeed, much of Microsoft's intellectual property is probably in the minds of its employees. Whether a new licensee can obtain a knowledgeable work force, which would have to include employees from Microsoft, quickly enough to be a viable competitor is questionable.¹⁸

The "hybrid" structural remedy combines the best features of the functional divestiture and full division remedies.¹⁹ Of all the available remedies, the hybrid remedy is best able effectively to terminate Microsoft's monopoly and create an incentive structure that is conducive to competition.

The hybrid remedy would work in the following way (see Figure 3):

(click image to see larger version)

The hybrid solution offers a number of advantages relative to the other structural remedies. In contrast to the functional divestiture, which leaves the operating system monopoly in place, the hybrid solution creates direct operating system competition immediately. This is the primary issue in the government's case against Microsoft and should, therefore, be the primary remedial objective. The hybrid solution is, however, less disruptive than the full division remedy, because it only requires a division of operating systems businesses, not the applications products, which were not the focus of the government's case. The applications businesses would not be divided unless Microsoft proposed to do so.

A.  The Minimum Scope of the Windows Company

The new Windows companies would, at a minimum, include the different versions of Microsoft's Windows operating system:

• the Windows "9x" series, including Windows 95, Windows 98, and the successor version currently under development, "Millennium";

• the Windows NT series, including Windows NT 4.0 and Windows 2000; and

• Windows CE.

It should also include Internet Explorer.

The Windows 9x series is the foundation of Microsoft's operating system business. But, Microsoft is in the process of bringing Windows 9x and Windows NT together. Windows NT/2000 is expected to replace Windows 98 on business and consumer desktops over the next few years. Therefore, to exclude Windows NT/2000 from a forward-looking remedy would make little sense.

Windows CE also should be included in the new Windows operating system companies. Windows CE has much in common with both Windows 9x and Windows NT. It is currently used in hand-held computing devices, but could be developed into a full desktop operating system. This might increase competition. But, it might also enable the Applications company to leverage its dominant position in the Office applications suite and try to reestablish an operating system monopoly.

The Internet Explorer web browser should also be included in the operating systems companies. The integration of Internet Explorer into the leading Windows products is now a fait accompli. Indeed, by the time any decree becomes effective, the majority of computers will be running Windows 98 or Windows NT/2000, which have Internet Explorer technically integrated into Windows. A court-ordered removal of Internet Explorer from Windows would, at this stage, be disruptive for software developers and consumers.

Moreover, including Internet Explorer in the divested Windows companies would create competition in the market for web browsers, which is itself a legitimate remedial goal. Indeed, much of the government's case addressed the attempted monopolization of the Internet browser market.

B.  The Applications Company

The Applications Company would retain the remainder of Microsoft's product lines:

• The Microsoft Office Suite, which includes Word, Excel, Powerpoint and other desktop applications.

• The Microsoft BackOffice Suite of server-based applications, which includes SiteServer, SQL Server, Exchange Server, SNA Server, and Proxy Server.

• Online businesses, such as Microsoft Network (the ISP), Hotmail, and instant messaging; the Internet content businesses such as MSN, Expedia, MoneyCentral, CarPoint, and HomeAdvisor.

• Consumer software, such as Microsoft Money, Microsoft Golf, Microsoft Flight Simulator, other games and home software applications.

• Software development tools and languages, such as VisualStudio, which includes such products as Visual Basic, Visual C++, Visual J++.

• Hardware products, such as the Microsoft mouse and keyboard.

C.  Addition of Products into the Windows Companies

As discussed, Microsoft would be able to add as many additional products as it wants to the core operating system products that will make up the Windows Companies. For example, as shown in Figure 4, Microsoft could add Office if it wanted to do so.

(click image to see larger version)

Figure 4. Hybrid Remedy with Office Added

The ability to add products has several benefits. First, it benefits Microsoft's shareholders, since Microsoft would only add products to the Windows companies if doing so increased shareholder value. Second, it would be procompetitive, since the added products would be divided among the three companies, which would be in competition with each other.

D.  Other Operational Issues

In this section, we discuss the major operational issues that need to be addressed to implement the hybrid remedy.

1. Shareholders

Each share of Microsoft stock would be exchanged for four new shares — one in each of the successor companies. This is analogous to what happened in the AT&T divestiture, where owners of AT&T shares were given shares in each of the new companies. Vested employee stock options could be exercised before the reorganization became effective, or, along with unvested options, converted to options in the employee's new company.

No employee of any of the three Windows competitors should receive shares in any other Windows competitor. Employees going to the new Windows companies should instead receive three times the number of their current Microsoft shares (or options) in the new employer (along with a share in the Applications company). These employees would not be disadvantaged, because shares of the Windows competitors would have identical value on the day the restructuring becomes effective. Because the distribution of employees among the three Windows competitors would be equal, presumably the distribution of employee stockholdings and options would be nearly the same. Microsoft's cash assets should be sufficient for any adjustments that might be necessary.

Additional restrictions on cross-ownership for top management would probably be necessary for a period of three to five years after the decree became effective. Individuals covered by the cross-ownership restrictions should be required, within a reasonable period of time, to divest securities in successor companies other than the company of his or her employment.

Specific individuals with large holdings — for example, Bill Gates, Paul Allen and Steve Ballmer — present a special problem. Gates owns about 15 percent of Microsoft, at last report, and thus would own about 15 percent of each successor company. If Gates, Allen or Ballmer remained involved with one of the successor companies, they would fall under the cross-ownership restrictions and be required to divest shares in the companies they were not associated with, over some reasonable period of time.

Although individuals with large holdings might be concerned that divesting their shares will depress the market price, a study by Nobel Laureate Myron Scholes established long ago that an individual shareholder's holdings "though a relatively large percentage of the outstanding shares of the firm, can be sold at approximately the prevailing market price without suffering financial loss in the event of a necessary sale.²⁰"

If studies such as these are not sufficient to alleviate concerns, special provisions could address the situations of Gates and other very large shareholders. For example, Microsoft could set up a temporary clearinghouse for employees to execute private exchanges of successor companies' shares. In addition, a portion of the cash assets of the pre-divestiture Microsoft Corporation could be used to purchase shares of large shareholders. These measures would at least reduce the amount of stock that large shareholders would be required to divest on the open market.

2. Intellectual Property

Each of the new Windows companies would be given full ownership rights over all of Microsoft's operating systems intellectual property, plus the intellectual property for any other products that Microsoft chose to divest to the Windows companies. This would also apply to intellectual property in the pipeline. Microsoft would be required to provide full internal documentation and any other relevant intellectual property, including rights to all relevant trademarks, such as the Windows name.

3. Employees

Each new Windows company would receive an equal share of Microsoft employees that were dedicated wholly or substantially to any of the products the company had the right to produce. Each new Windows competitor would receive a proportional number of each category of employee, for example, development and marketing personnel.

The division of employees along functional lines between the Windows operating system companies and the Applications company would presumably follow the existing business units and be relatively straightforward. Dividing the operating system employees among the three Windows companies in an equitable manner might be more difficult. Assuming it has the right incentives, Microsoft is in the best position to determine how best to create equivalent groups of employees. One promising method would be to choose "team leaders" and let them sequentially choose employees. Another method would follow the "divide-and-choose" or "cake-cutting" model, in which the person responsible for dividing a unit into equal pieces chooses his piece last. For example, Steve Ballmer could divide up Microsoft's employees into groups, but the company in which he retains a financial interest would get the last choice of the groups. This would provide an incentive to create equivalent groups of employees.

With Windows 2000 scheduled for release in February 2000, the division of Windows employees into three groups would come at a good time. Current staffing levels for the initial release — some 4800 developers — will not be needed going forward. Many of these developers would have to be released or shifted to other projects. The availability of these additional developers makes it easier to staff three competing Windows companies without having to hire additional employees. Indeed, the creation of three companies may provide new opportunities for employees. Many talented people work at Microsoft. Dividing the Windows development staff would allow more of these people to rise to the top — as happened after the AT&T divestiture.

4. Contracts

Each new Windows competitor would be allocated an equal share of existing sales contracts for the products they can sell. However, customers also would be given the right to reopen their contracts so that they could receive immediate benefits from competition among the new firms.

5. Pecuniary Assets and Investments

Microsoft has substantial cash on hand — over $18 billion according to its most recent SEC filing — and a substantial portfolio of investments in other companies. In some cases, WebTV for example, Microsoft has acquired total ownership of companies. There are a variety of possible ways to divide Microsoft's cash assets and investments in other companies.

A relatively simple solution for the cash assets would be to allocate them evenly among the four successors, so that the Windows Companies in the aggregate would receive 75 percent of those assets (in equal 25-percent shares). This would assure that each of the successor companies had sufficient cash on hand to meet any initial organizational costs and to compete vigorously from the start.

With respect to the non-cash securities, several alternatives suggest themselves. First, they could be distributed equally to the four companies in the same manner as the cash. Alternatively, as with other Microsoft assets that were not the subject of the government's case, they could simply be given to the Applications company. Finally, they could be spun off into a separate (fifth) company whose shares would be distributed to Microsoft's current shareholders.

E.  Measures to Preserve the Hybrid Structural Remedy

One of the principal benefits of the hybrid remedy is that it does not require detailed ongoing supervision by the decree court, as happened with the AT&T settlement. The hybrid remedy would require only a limited number of temporary restrictions on the future conduct of the resulting new Windows competitors and the Applications Company. The purpose of these restrictions would be only to prevent the successor companies from undoing the decree — by recombining through merger, joint venture, or informal favoritism. The lines between permissible and impermissible behavior would be clear, unambiguous and easily enforceable. The new companies would not face the kinds of line-of-business restrictions that made the AT&T decree so difficult to enforce and comply with. Any successor company could develop its own products to compete with products divested to another successor company. Acquisitions of new products from other, non-successor companies would be subject to normal antitrust review.

The restrictions on the successor companies are intended to be short-term, lasting for only three to five years. Given the rapid pace of change in the software industry and the expected boost to competition expected from the hybrid remedy, the competitive landscape will look very different at the end of a three-to-five-year period. At the end of that period, the restrictions would no longer be necessary and the decree would be terminated. The Microsoft successor companies would be treated like any other company subject to the normal constraints of antitrust law.

The type of restrictions that would be necessary in the interim to assure that the decree is not nullified or seriously weakened are straightforward. The successor companies would be prohibited from acquiring one another. They would also be prohibited from entering into joint marketing or development agreements with each other, because such arrangements could have effects similar to a merger.

For a fixed period — say, three years — successor companies would not be allowed to hire each other's employees. Such movement of employees might be a way to get around the first set of restrictions (described above). Because this restriction does involve restrictions on individuals' mobility, it should be effective for as brief a time as possible.

Exceptions to the prohibitions on cooperative arrangements between successor companies should be permitted in order to facilitate the sharing of technical information and the maintenance of compatibility between competing versions of Windows.

For example, the Applications company and the Windows companies should be permitted to disclose technical specifications to each other. Such disclosures would generally need to be available to all the companies on an equal basis or they would be inconsistent with the restrictions on "recombining through informal favoritism" discussed above.

In addition, the Windows Companies should be permitted to establish standard-setting bodies for the purpose of maintaining compatibility. While it will be in the interests of the companies to maintain compatibility on their own (see discussion in Section X below), standard-setting bodies may be helpful and are routinely used in many industries.

Obviously, the antitrust laws would continue to apply to the successor companies. For example, acquisitions of outside companies would be subject to merger review and challenge by the antitrust agencies or private parties. After the decree was phased out, all potential acquisitions — including among the successor companies — would be subject to these procedures. Other business practices likewise would be subject to antitrust scrutiny if they posed anticompetitive threats.

IX.  BENEFITS OF THE HYBRID REMEDY

The creation of three Windows competitors would immediately replace monopoly with competition in the market for operating systems. The Windows companies would compete on the basis of price, reliability, quality and other features. In addition, competition in the operating system market is likely to spur competition and innovation in related software and hardware markets.

Competition would lead to all sorts of new dynamics in the industry that are impossible to predict at the outset. The Windows competitors could seek market alliances with non-Microsoft vendors. These could be linkages with hardware platforms (following the model of Sun or Apple), or with applications packages (like the current link between Windows and Microsoft Office), or with Internet service capabilities (like the new AOL/Time Warner). In turn, in pursuing alliances or developing market niches, the Windows companies inevitably would compete with the Suns, Apples, AOLs, and Oracles of the world, helping consumers and promoting competition. The new Windows competitors should have little difficulty raising capital or maintaining the value of their equity. The market value of the Windows companies would be sustained by the value of the potential revenue stream from the massive installed base of Windows users, from the intellectual capital of the Windows employees, and from the possibilities for growth.

The most common concern expressed about a remedy that involves operating system competition is that it will lead to fragmentation of the operating system standard. Obviously, this viewpoint is at odds with purpose of the antitrust laws, which is to promote competition. Moreover, it is not likely to happen, because the new Windows companies would have strong incentives to maintain compatibility rather than add incompatible features and functions. The fragmentation issue is discussed in detail in Section X.

A. Operating Systems Competition

Alternative suppliers of the Windows operating system would compete on price. Indeed, at the outset, that would be the only basis for competition among the Windows companies. Microsoft's extremely high profit margins suggest that there should be considerable flexibility for each Windows company to offer competitive prices below what Microsoft currently charges today.

Competition also should stimulate improvements in quality, reliability, ease of use and other non-price attributes. Currently, the rate of technical progress in the operating system market is determined by Microsoft alone. The absence of competition permits Microsoft to undertake product rollouts at a pace that may delay innovations that otherwise would become available sooner. In addition, Microsoft devotes much of its design and marketing efforts to maintaining and extending its monopoly rather than to directly meeting customer needs.

By contrast, three competing Windows companies would have to become customer-focused very quickly in order to retain and expand their customer base. For example, as a monopolist, Microsoft has little incentive to produce customized versions of its products, because each one increases the costs of support and development. With competitive Windows companies, there is a higher likelihood that operating systems will be developed to meet specific user needs. A Windows company might decide to emphasize the consumer segment, or the enterprise segment. Such emphasis would maximize the attractiveness of the operating systems product line to a particular type of user — while also maintaining compatibility with the other versions of Windows.

B.  Competition in the Applications Market

The hybrid remedy would simultaneously create operating system competition and remove the linkage between the Windows operating system and the principal applications. This would stimulate competition in the applications market.

Entry into applications markets where Microsoft is dominant — office suite software, for example — has been exceedingly difficult because Microsoft allegedly favors its own applications. The competing Windows companies would no longer have an interest in according any one company favorable treatment. In fact, the Windows companies would have an incentive to support as much applications competition as possible, because the availability of more and better applications products increases demand in the operating system market. The Windows companies might well enter the applications market themselves.

Moreover, unlike the current Microsoft, the competing Windows companies would have the incentive to keep the channels of distribution open to applications competition. This would also stimulate applications development, provide greater choice for consumers, and increase the demand for operating systems.

The three Windows companies would have incentives to standardize on an open set of APIs so that application developers can consistently develop applications for each of the operating systems. Because the current applications programs would be in a separate company, the Windows companies would derive no advantage from a set of secret APIs or, more generally, from anything that impeded applications development.

C.  Competition in Hardware Platforms

Competition in the market for operating systems would support innovations in hardware technology as well. Today, Intel and other chip manufacturers are wholly dependent on Microsoft's decisions concerning new product rollouts. Intel's hardware innovations, which might be able to move at a faster pace, are limited by Microsoft's schedule. With operating system competition, this dependency would end. Each new Windows company would have strong incentives to support new technologies, knowing that if it did not, one of its competitors would. An operating system company that did not support new technologies would lose its customer base.

Competition in the operating systems market also would provide bargaining leverage to OEMs. They would be able to choose among alternative suppliers, rather than having terms dictated to them. The type of exclusionary conduct that forced OEMs to carry specific products and precluded them from carrying others would be a thing of the past. Likewise, no Windows competitor would be able to dictate to the OEMs their boot-up sequences, what their first screen should look like, or which icons should be placed on their desktops. By eliminating these practices, the hybrid remedy would enable OEMs to provide more variety and be more responsive to their customers.

The new Windows companies might also partner with various hardware vendors to create an alternative to the "Wintel" platform. Unlike Windows 95 and Windows 98, Windows NT/2000 and Windows CE are both relatively platform-independent. Competition to the Wintel platform might offer better performance or reliability, or lower prices.

D.  Browser Competition

One result of Microsoft's anticompetitive acts has been to slow innovation in the browser market. Three separate Windows companies, each with its own browser, would obviously reinvigorate competition in this critical area. This would speed the development of Web-based computing that would compete directly against PC-resident applications.

X.  THE OPERATING SYSTEM FRAGMENTATION ISSUE

Perhaps the major concern expressed with respect to a structural solution that creates operating system competition is that it would "fragment" the operating system standard. Windows supposedly would evolve into incompatible operating systems and consumers would lose the benefits of standardization. Consumers would incur either the costs of incompatibility between applications programs and the new operating systems, or the costs of having applications developers perform expensive "ports," or rewrites, of their software in order to make their existing applications work with the new operating systems.²¹

The fragmentation argument is ultimately an argument against the premise on which our economic system and the antitrust laws is based, which is that competition best serves the interests of consumers. The fears of fragmentation of the Windows standard are unwarranted, because they are inconsistent with the fundamental economics that would characterize the post-breakup operating system market. The three Windows companies would have very strong incentives to retain the pre-existing network externalities. Simply put, consumers want access to a large pool of applications; software developers, in turn, want access to a large pool of consumers. To not maintain compatibility would risk losses with both these groups, which none of the Windows companies would want to do. The costs of switching and the benefits of network effects create powerful incentives for consumers to stay with their existing operating system standard, and for the new operating system companies to retain compatibility with the installed base and each other.

A.  Short-Run Compatibility

Initially, each of the new operating systems competitors would be marketing identical, fully compatible versions of the Windows operating systems. Porting costs would, therefore, not be an issue. During this initial period, the new companies would compete primarily on the basis of lower prices and better customer service.

B.  Long-Run Compatibility

In the long run, the incentives to maintain compatibility will be strong. Indeed, the claim that a structural remedy would create significant porting costs relies on the erroneous assumption that the new Windows competitors would find it in their interest to develop mutually incompatible operating systems. This assumption is contrary to the economics of operating system markets, which are characterized by significant network externalities and high switching costs. These characteristics give the Windows competitors strong incentives to remain compatible. Competition would be within the Windows standard, rather than for a new standard.

1. Incentives to retain backward compatibility

Over time, the new companies created by the hybrid remedy would find it in their interest to maintain compatibility with the installed base of existing applications. Any operating system competitor that chose to be incompatible with existing applications would face the task of convincing the installed base of Windows applications users to incur the very significant costs of replacing their existing applications and being unable to use new applications developed for the established standard. This would be an impossible task. In short, a decision on the part of any of the Windows competitors to forego backward compatibility would be suicidal.

The Windows operating system vendors would still be able to engage in quality competition by improving and differentiating their products over time. They could add new features and APIs to the base Windows operating system while retaining the existing Windows APIs. But, it would be in the interest of each company to maintain full compatibility with applications written to the common Windows APIs. Thus, whatever product differentiation occurred would not create incompatibilities that require applications developers to incur significant porting costs.²²

Because new features would represent incremental additions to the existing Windows APIs, the cost of adapting applications to use these features would be far smaller than the cost of a complete port to an entirely different operating system. Moreover, changes to applications programs that took advantage of the new operating system features would constitute genuine improvements. The costs of these improvements would not constitute costs of complying with an antitrust decree. Rather, they would be the costs of providing better products. These costs would not be incurred unless they were outweighed by benefits to consumers.

2. Incentives to maintain compatibility with each other

As discussed above, the operating system companies will have a very strong incentive to maintain compatibility with the installed base of applications programs. If they maintain compatibility with a common base of programs, it is logical to assume they will maintain compatibility with each other.

In addition, competition among the new operating system companies is likely to expand the core Windows standard, rather than fragment it. The new companies will compete by adding new features that improve functionality. This will appeal to applications developers who will find it their interest to support these new features. Competing operating systems firms will have the incentive to imitate the most successful new features to attract or keep customers. Through this process, the core Windows standard will be expanded and improved.

C. Cooperative Standard-Setting

Cooperative standard-setting bodies are common in the economy and also in the software industry. For example, the Open Software Foundation and the IEEE are responsible for some software standards. Programming languages such as FORTRAN have been standardized under the auspices of the American National Standards Institute (ANSI). The new operating system competitors may want to supplement the economic incentives described above by cooperating explicitly to retain compatibility over time.

The new Windows competitors would have an incentive to cooperate in order to expand the set of new features shared by all versions of Windows. (See also the discussion in Section IX of the benefits of operating system competition for the applications market). This would enable applications developers to develop improved products that take advantage of the new features. Improved applications functionality, in turn, increases the demand for the newest Windows operating systems, both as upgrades and with new PCs.

Cooperative standard-setting bodies might also work to create "bridges" that allow the competing operating systems to take advantage of competitors' proprietary extensions. Bridging technologies could also be developed in order to mitigate the effects of whatever "fragmentation" does arise following the breakup of the Microsoft monopoly.

D. Porting Cost Estimates

Professor Stan Liebowitz has estimated that applications producers would face $14.5 billion in porting costs over three years for each new operating system that is created by a structural remedy.²³ These estimates simply assume that applications developers will need to rewrite their programs at substantial cost. The analysis takes no account of the economics of software markets and the powerful incentives that exist for producers to maintain compatibility, as described above.

Liebowitz' critique of a structural remedy that creates operating system competition is, more generally, a defense of monopoly and a critique of competition in software markets. Competition and product differentiation do involve some costs, though not of the magnitude of Liebowitz' estimates. Ultimately, competition provides consumers with choices they otherwise would not have. The benefits of this competition are clearly sufficient to outweigh the costs.

XI.  CONCLUSION

The hybrid structural solution achieves the principal goals that a remedy in this case should strive for. First, it immediately replaces the existing operating system monopoly with a competitive market. Competition will lower prices, improve quality and spur innovation in the operating system market itself, as well as in adjacent software and hardware markets.

Second, the hybrid solution is the least regulatory of the available alternatives. Behavioral restrictions would be minimal and short term. The new Microsoft companies would be able to develop software and compete unencumbered by a decree court looking over their shoulders.

The major cost of the hybrid solution is the disruption associated with creating three new operating system competitors out of one existing monopoly. This cost is not trivial. But it is also temporary. The benefits to consumers of reinvigorated competition and innovation throughout the computer sector will be long lasting and well worth this short-run cost.

ACKNOWLEDGEMENTS

The sheer complexity of the Microsoft case and the market for personal computer software makes it perhaps even more necessary than usual to seek counsel and expertise during the course of any study of the matter. I am therefore more than usually grateful to the many people in think tanks, academia and the private sector who have made themselves available to answer questions, provide information, make helpful suggestions and offer constructive criticism. While I cannot possibly list all those who have helped, I am especially indebted to Jeff Eisenach, Ken Glueck, Michael Katz, Robert Litan, James C. Miller III, R. Craig Romaine, Steve Salop, Garth Saloner, Andrew B. Steinberg, Hal Varian, Larry White, Robert Willig and Rick Warren-Boulton,

I also want to express my thanks to Catherine E. Gilliam, Program Associate, and Robert P. Frommer, Research Associate, who worked above and beyond the call of duty checking data, preparing charts and editing text.

While I am very grateful to the many people who assisted in producing this volume, I am solely responsible for its content, including any remaining errors.

Thomas M. Lenard
Vice President for Research
The Progress & Freedom Foundation
Washington, DC
January 2000

ABOUT THE AUTHOR

Thomas M. Lenard is Vice President for Research and Senior Fellow at The Progress & Freedom Foundation. Dr. Lenard joined the Foundation in 1995 as Senior Fellow and Director of Regulatory Studies.

Prior to joining the Foundation, Dr. Lenard was Vice President of a Washington, DC-based economics consulting firm. Dr. Lenard has also served in senior government positions at the Office of Management and Budget, the Federal Trade Commission and the Council on Wage and Price Stability.

Dr. Lenard was a member of the economics faculty at the University of California, Davis, and has been a visiting economist at the Brookings Institution. He has published and testified before regulatory agencies and congressional committees on a variety of economic issues, including competition in the computer industry, electricity regulation, regulation of drugs and medical devices, and postal competition. Recent publications include Competition, Innovation and the Microsoft Monopoly: Antitrust in the Digital Marketplace, The Digital Economy Fact Book and Deregulating Electricity: The Federal Role.

Dr. Lenard received his B.A. from the University of Wisconsin and his Ph.D. in Economics from Brown University. He currently serves as Vice President for Programs of the National Economists Club in Washington D.C.

ENDNOTES

1.For a review of the economic issues involved in the case, see Jeffrey A. Eisenach and Thomas M. Lenard, eds., Competition, Innovation and the Microsoft Monopoly: Antitrust in the Digital Marketplace, The Progress & Freedom Foundation (Kluwer Academic Publishers: 1999).

2.The Department of Justice and 20 States and the District of Columbia (collectively, "the government") filed a complaint against Microsoft on May 18, 1998. Subsequently, South Carolina withdrew, leaving 19 states still party to the suit.

3.Findings of Fact in U.S. v. Microsoft Corporation, Civil Action No. 98-1232 (TPJ) and State of New York, ex re. Attorney General Eliot Spitzer et al., v. Microsoft Corporation, Civil Action No. 98-1233 (TPJ), November 5, 1999 (hereinafter "Findings of Fact" or "Findings").

4.Of course, any entity's incentives are changed to the extent it faces legal penalties. In the Microsoft case, those penalties would have be very large, indeed, for them to have a significant effect.

5.Richard A. Posner, Antitrust Law: An Economic Perspective, University of Chicago Press, 1976, p. ix. Judge Posner was asked to mediate settlement talks by Judge Thomas Penfield Jackson, the trial court judge. By changing the "incentive structure," we mean changing both the firm's incentives and its ability to act on those incentives.

6.For a discussion, see Michael L. Katz and Carl Shapiro, "Antitrust in Software Markets," in Eisenach and Lenard.

7.Profit margins for Microsoft, Sun, Intel and IBM are for the quarter ending September 1999. For Oracle, the figure is for the quarter ending August 1999. And for Cisco, the figure is for the quarter ending May 1999.

8.See Peter Huber, "The Death of Old Media," The Wall Street Journal, January 11, 2000, and Stan Liebowitz, "Hey, Remember Microsoft?," The Wall Street Journal, January 14, 2000.

9.Bill Gates, "Why the PC Will Not Die," Newsweek, May 31, 1999.

10.Applications program interfaces (APIs) are the language and message format by which an applications program such as a word processor communicates with the underlying operating system in order to display data, write to memory, and use peripheral devices.

11.See Microsoft Dominating Browser War <www.statmarket.com>. See also Websidestory's Statmarket.com Reports Browser War All But Over, Aug. 9, 1999, <www.websidestory.com>; Netscape's Share Keeps Shrinking, Industry Standard, Aug. 9, 1999 <www.thestandard.com/articles/display/0,1449,5841,00.html>.

12.The preferred remedy we discuss below makes it unnecessary for courts to make such determinations.

13.The phenomenon of a monopolist spending real resources to maintain its monopoly was first discussed by Gordon Tullock, "The Welfare Cost of Tariffs, Monopolies, and Theft," Western Economic Journal, 1967, vol. 5, pp. 224-32."

14.For a review of the available remedies, see R. Craig Romaine and Steven C. Salop, "Slap Their Wrists? Tie Their Hands? Slice Them Into Pieces? Alternative Remedies for Monopolization in the Microsoft Case," Antitrust, Vol. 13, No. 3, Summer 1999.

15.As noted earlier, any entity's incentives are changed to the extent it faces legal penalties. In the Microsoft case, those penalties would have be very large, indeed, for them to have a significant effect.

16.By changing the firm's "incentive structure," we incorporate changing both the firm's incentives and its ability to act on those incentives.

17.A variant of the full division remedy has also been proposed, involving a one-time licensing auction of Microsoft's intellectual property in order to create competitors. This remedy would try to create competition in operating systems, or across-the-board, by requiring Microsoft to license its relevant intellectual property on a one-time basis to one or more other companies, at a price set by an auction process. For reasons discussed below, this is an inferior alternative for restoring competition.

18.For it to have any chance of success, a licensing remedy would have to open up employment contracts that either preclude Microsoft's employees from moving to competitors, or make it financially disadvantageous for them to do so.

19.The hybrid remedy is discussed in Romaine and Salop.

20.See Myron Scholes, "The Market for Securities: Substitution versus Price Pressure and the Effects of Information on Share Prices," Journal of Business, Vol. 45, No.2, April 1972.

21.Stan Liebowitz, Breaking Windows: Estimating the Cost of Breaking up Microsoft Windows (Association for Competitive Technology and the ASCII Group, Inc., April 30, 1999).

22.For example, Windows 98 added features not supported by the original release of Windows 95. The transition from Windows 95 to Windows 98 created few, if any, incompatibilities for existing applications. Thus applications developers were not forced to rewrite their code to run on the new system.

23. Liebowitz, p.11.

Thomas M. Lenard is Vice President for Research at The Progress & Freedom Foundation. The views expressed here are the authors and do not necessarily reflect those of The Progress & Freedom Foundation, its Board, Officers or Staff.