Speech
Stanley Liebowitz, University of Texas at Dallas
The Law and Economics of
United States v. Microsoft Conference
May 5, 2000
First, I would like to thank Nicholas Economides and organizers of this conference for inviting me here.
Before I get going I would like to hold up my book up for a minute so the publisher will be happy. The book, co-authored with Stephen Margolis, is entitled Winners, Losers, & Microsoft: Competition and Antitrust in High Technology. I think that the book provides more information about software markets and the effect of Microsoft on consumers than any other source, including the trial transcripts. I also think you'll find in the book an interesting discussion of much of the economics underlying this case. In particular, a discussion of economics related to what the judge refers to this the application barrier to entry, a theme I will come back to in a few minutes.
Let me start the talk by saying that there are four points I would like to make but that I can’t adequately address in my fifteen minutes. Hopefully we can come back to them during question-and-answer session. [In these notes that I'm cheating a bit and pretending I can talk faster than I can].
Here are the four points.
Why do I believe that the economics in the case is flawed? The answer has to do with the central role in the Judge’s theory of the case played by the concept of applications-barrier-to-entry, sometimes referred to by the judge as the intractable chicken and egg problem. The term in the economics literature that describes the type of problem the judge refers to is ‘lock-in’.
Lock-in theories claim that an incumbent’s product can continue to dominate a market even the face of superior rivals if certain conditions exist. Those conditions include network effects and economies of scale, both of which are asserted to and presumably do exist in many software markets.
The judge points out that Windows survives and continues to thrive even when some competitors try to enter, such as OS/2 or perhaps Linux. The problem identified by the government and the judge is one where the supposed monopoly, Microsoft, keeps its monopoly not on the basis of its superior performance but on the basis of the application barrier to entry. The Judge views virtually all of Microsoft’s behaviors as an attempt to enhance this monopoly barrier.
So what's wrong with all of this? Quite simply, the theory that the judge has swallowed has no real world evidence to support it. The examination of that issue forms the basis of the work I conducted with Stephen Margolis over the last decade, which is detailed in the book. The claim of lock-in that so concerns the Judge turns out to be nothing but a theoretical curiosity. The famous examples that have been put forward to support these claims, typewriter keyboards and videorecorder formats, turn out to be examples of poor scholarship, not lock-in. Additionally, my research with Margolis indicates that market shares for software applications rapidly respond to changes in product quality, quite the opposite of what the Judge has accepted.
The judge also puts forward ‘facts’ that won't hold up under any reasonable standard. For example, The Judge points out that even though IBM spent perhaps $1 billion on OS/2, it failed to make any serious inroads. Why? The judge blames the application-barrier-to-entry. But this is just not in accord with actual events. OS/2’s failure wasn’t due to any lack of software support. In fact, there were at first, more companies providing software for OS/2 than for Windows. Particularly the big name producers like Lotus. The actual explanation for OS/2’s failure was that it was a half-baked operating system for consumer use. We shouldn’t expect an inferior challenger to unseat the incumbent, with or without application barriers to entry. OS/2 failed to take off for several reasons: it was much more expensive; it was almost impossible to install; it required twice as much memory and a faster processor; it wasn’t compatible with many video and sound cards. And the networking version of OS/2 was only going to run on IBM machines with the Micro-Channel architecture—talk about attempted monopolization!
Another very serious problem with the judge’s claims is the role that he gives to the market share needed by Netscape in order, according to the judge, to compete successfully with Windows as a (middleware) operating system. To ordinary users like you and me, Netscape might appear to be just a browser, but the judge claims that combined with Java, it really was a threat to Windows. According to this thinking, writers of software could write code that would run within the Netscape browser.
The claim was that software developers could write software to run in the Netscape browser and since Netscape ran on all machines and operating systems, there would be no reason for consumers to prefer any given operating system. Netscape would become the defacto operating system and Windows would lose its ‘monopoly’. The judge claims that it was to deter this threat that Microsoft decided to compete with Netscape. He did not believe that Microsoft perhaps wanted to be seen as the most important Internet firm, a position then held by Netscape. He did not believe that Microsoft wanted to sell advertising time on its web sites to which browser users were steered. Nor did he accept that Microsoft might wish to generate revenues from selling servers where the sales pitch might have included the fact that the server would be expected to work transparently with the leading browser, a Microsoft browser. Instead the Judge bought the far-fetched claim that Netscape was an operating system Trojan horse.
Even if you accept the Netscape/Java middleware platform as a reasonable idea, the rest of Judge Jackson’s logic makes no sense. Netscape has had and continues to have a very impressive base of users—perhaps not as large as Windows, but far larger than the Macintosh—certainly large enough to be an inviting platform for those who wish to write software to run on many computers. But ask yourself how many applications were written for the Netscape-Java ‘platform’? The answer is "very few".
If Netscape's market share was and continues to be large enough to make a viable platform competitor then why aren’t consumer programs being written for it? The answer is simply that it is not viable because programs in Java run more slowly than they would in Windows. The reason for this is that in Java the program does not run directly but instead run on a virtual machine sitting on top of a real hardware machine translating the code. The translation takes time so the programs run more slowly. Corel was trying to write the WordPerfect Suite to work in Java and came out with some early versions. It was trashed by reviewers as being too slow. This is the reason Java and Netscape are not usurping Window’s position in the operating system world, and it has little or nothing to do with the Microsoft's behavior.
My second point is that the government failed to demonstrate consumer harm from Microsoft’s behavior.
This should be apparent to anyone who reads the judge's finding of fact. He devotes virtually no space discussing consumer harm. When he does discuss consumer harm the discussion is trivial or ludicrous. Under the trivial classification he discusses the fact that Netscape users will have wasted space on their hard drive taken up by Internet Explorer as if it were uncommon for users to fail to take advantage all the functionality contained in programs. Another trivial harm is the extra costs of installing Netscape. We’re talking five minutes here, ladies and gentlemen.
The final harm the judge identifies is a far-fetched harm. It's certainly possible to imagine a better world where the software market provides more choice and better products at lower prices than those we currently have. We can also imagine a world without pain, without death, without war. Imagination, after all, knows no limits. The judge claims there would be more innovation if Microsoft did not act as it did.
Although Microsoft critics and the government’s lawyers constantly assert that the market would be more innovative but for Microsoft, there is no support for this view. How could you even test such an assertion? There is neither theory or empirical support for the claim that firms with large market shares innovate less than firms with small market shares. If innovation brings profits, we would expect all size firms to try to innovate. If we look abroad to see how our domestic software industry does in comparison, things seem to be going pretty well.
The government apparently did not conduct an examination Microsoft's overall impact on consumers. That's where my book provides more information. We looked at different software applications on both the PC and the Macintosh, year by year. We looks at prices, quality, and market share. We discovered that those products considered to be the best had the large increases in market shares and that when Microsoft (or anyone else) came to dominate a market it was only when they had the better product. We also discovered that the lower software prices that so many of us take for granted did not occur across the board. Instead, the fall seems to be due in large part to Microsoft.
Examine this chart which indicates what happens to prices in software markets for two categories of software markets, markets where Microsoft competes and markets where it doesn’t. The chart very clearly demonstrates that in markets where Microsoft did not participate prices fell hardly at all, perhaps 12% over the decade. But in markets where Microsoft competes, prices fell by over 60%. Therefore it's a reasonable to conclude that Microsoft is in fact responsible for much of the price decline. Plus there are documented instances where Microsoft lower its price dramatically even after achieving very significant market share.
Given these results, it is fairly clear that a verdict of ‘not guilty’ would be the best outcome. Even if you believe in the government’s case, however, its proposed remedy does not make any sense.
The government proposes to divide Microsoft into two companies, one based on applications and the other on the operating system. The rationale that is being put forward for this remedy contradicts the theory and facts put forward by the judge.
The problem, of course, is that if Microsoft has a monopoly operating system, this remedy does nothing to remove it. Now the government is claiming that if Microsoft Office were ported to another operating system such as Linux, then Linux would become a more viable competitor and perhaps it would then be able to overcome Windows’ application barrier to entry. But this claim cannot be true according to the Judge’s findings.
Microsoft Office consists of 5 or 6 products. The Judge clearly stated that it takes thousands and thousands of programs to be written for an operating system before it could overcome the application barrier to entry and become viable. The Macintosh has 10,000 to 15,000 applications, including Microsoft Office, yet the judge claims it isn’t a viable competitor.
The fact that the government claims that this remedy might make Linux viable indicates that it doesn’t believe its own application-barrier-to-entry theory.
Of course, the other argument being used to support this remedy is related to the ludicrous middleware idea. The claim is that Microsoft Office might become an alternative operating system to compete with Windows. How exactly programmers are going to write programs to run in Office hasn’t been explained, but I presume it has to do with Visual Basic. I doubt that the government will be able to sell this proposition to many computer users. And I suspect that typical computer users would be shocked if they understood that the government’s case revolved around the fiction that Netscape was a cloaked operating system.
This brings me to my final point. I believe that the end result of the government action and the judge’s verdict, if upheld upon appeal, will hurt consumers because it helps Microsoft's competitors and doesn't enhance competition.
In this case the firms most being helped by the government action and the firms most strongly pushing for dismemberment of Microsoft are not the current direct competitors to Microsoft in the Windows market or in closely related markets. The competitors most benefited by this action are Sun, Oracle, IBM, and others that have a large stake in high-end servers and workstations.
Their motives are straightforward. Microsoft, with Windows NT (now Windows 2000) poses a serious threat to the workstation market which is currently dominated by these other firms. The purpose of politicking for this lawsuit is simply to hamstring Microsoft and reduce the level of competition that Microsoft will bring to the workstation market.
Even the government’s bonehead remedy has the fingerprints of these firms on it. Not only are breakup costs imposed on Microsoft, but more importantly, the government is asking that Microsoft’s server software be put in the application group and not with the operating system where is so obviously belongs. This would weaken the Microsoft workstation product, with the benefits redounding to firms like Sun, not to consumers.
That's why this judgment will reduce competition instead of increasing it. The real political focus of the case is in the workstation/server market, but case itself only talks about the desktop market. This is a clever sleight of hand, hiding the market where competition will be reduced.
I think that this is a misguided case that is likely to harm desktop consumers and even more importantly, workstation consumers. The best we can hope for is that the appeals process overturns this very bad decision.