Dow Jones
The Dismal Science
Illustration by Robert Neubecker Soft Microeconomics
The squishy case against you-know-who.

By Paul Krugman
(posted Thursday, April 23)

       I wrote this piece in WordPerfect, not Word, mainly because Word's equation editor is awful (are you listening, Mr. Myhrvold?), and I may as well use the same software for plain English articles and professional gobbledygook. I surf with Netscape Navigator and check e-mail in Eudora. So, I am not a fan of Microsoft's products. Moreover, for reasons explained below, it is in the public interest to have Bill Gates always running a bit scared of the Justice Department. Nonetheless, the more anti-MS propaganda I read, the more pro-MS I get. There is a case against Microsoft, but it is not the one you hear, and I would hate to see crude misunderstandings posing as sophisticated analysis prevail.
       Probably the first thing one ought to say is that the public has no interest in helping Bill Gates' rivals for their own sake. It's easy to think of the people who run software companies other than Microsoft as underdogs fighting Big Bad Bill. But those "little" guys are no more in need of extra money than Gates is, and if allowing him to get an extra billion at Larry Ellison's--or even Marc Andreesen's--expense is good for the rest of us, so be it. Those of us who do not get paid in software stock options should not allow ourselves to become pawns, either way, in struggles among those who do.
       So what is the public interest?

The case for leaving Microsoft alone does not rest on some naive faith in the perfection of free markets. Software is an industry characterized by powerful increasing returns in both production and consumption: The more units Netscape ships, the lower its per unit cost; the more copies of Navigator in use, the more attractive it is to the typical user. These increasing returns make the kind of atomistic, "perfect" competition that prevails in the market for, say, wheat impossible in the market for browsers or word processors. Necessarily, each type of product will in the end be produced by only a few companies, perhaps only one.
       But how does this concentration of production take place? One of the depressing things about public discussion of the Microsoft case, even among supposedly well-informed people, is that much of it has come to be dominated by a basically primitive view about what increasing returns do to markets--namely, that they convey monopoly power purely randomly, on whoever happens to be in the right place at the right time--and that this "path dependence" allows clearly inferior technologies to become "locked in." Now path dependence has its place--but when applied to the Microsoft case it misses the point. After all, high-technology companies are themselves quite aware of increasing returns, and their strategies--above all the prices they charge when they are trying to establish themselves in a market--are very much affected by that awareness.

Illustration by Robert Neubecker
Consider, for example, one particular form of increasing returns: The so-called "learning curve," which says the more units of something you have already produced, the lower the cost of producing the next one. You might think this means that whoever gets into a market first will simply have a snowballing advantage. But as Stanford's Michael Spence pointed out in a classic, 20-year-old paper on this subject, profit-maximizing companies that know they face a learning curve will compete fiercely to move down it more rapidly, selling cheaply in the early stages of a product cycle (and therefore losing money) in the hope of making the money back later.
       The same logic applies to increasing returns on the demand side: As a manufacturer, if I know that a typical customer's choice of browser depends both on the price and on the number of other people using that browser, I will initially make my own browser cheap--maybe even free--so as to build market share. In either case I must incur initial losses that are, in effect, part of the price of entry into the market--an add-on to the cost of developing a product in the first place. And because nobody will want to pay this entry fee without a reasonable hope of earning it back, only a few companies will enter a market subject to strong increasing returns. The point is that the eventual domination of an industry by a few companies--and a high rate of profit on sales for these companies in the later stages of the cycle--doesn't happen because these companies just happened to get a head start. On the contrary, it is precisely because it isn't purely a matter of luck--because everyone competes so fiercely on prices in an effort to get some of those nice increasing returns--that only a few dare enter.

Of course, there is also an element of luck. It's not true that whoever gets a head start always dominates the market--if a company has a small head start but offers a clearly inferior product or has clearly higher costs, rivals can and will overtake it. But nobody can be sure just what an as-yet undeveloped product will cost to produce, or how it will go over with consumers. Thus, competition in a market characterized by increasing returns is--as it must be--a sort of demolition derby in which only some of those who enter cross the finish line. Those who do make it across the finish line will typically make big profits. But this profitability is necessary to the enterprise. Who will enter a demolition derby without the incentive of a prize?
       So this is the case for leaving Microsoft alone: High-tech competition is, necessarily, a competition that ends up being won by a handful of players. Those who make vast fortunes may not always be the most deserving--but so what? For the rest of us, what matters is not who wins or loses, but how they play the game. And if they know or suspect that too much success will be punished--that anyone who does too well will become a target of envy-driven litigation--they will have that much less incentive to play hard.

Illustration by Robert Neubecker Now there is, of course, also a case against Microsoft. Never mind the crude complaint that it is too big, or too profitable, or that nerdy types tend to dislike its products. The real concern is that because Microsoft's victory in an earlier derby happens to have given it control over a particularly strategic part of the industry--because it supplies operating systems--it is in a position to squeeze out rival suppliers of other software. And that is a real concern: If Microsoft had, for example, written Windows 95 in a way that made it hard or even impossible for me to use WordPerfect, the Justice Department would have been absolutely justified in calling out the arsonists.
       But the fact is that MS has been very careful not to use its undoubted power to practice any crude, obvious version of what is known in the trade as "vertical foreclosure." WordPerfect and Netscape work just fine on my Windows-based machine. This restraint may partly reflect Microsoft's market strategy--after all, Microsoft beat Apple partly because Apple did practice vertical foreclosure, and as a result inhibited the development of complementary software (although the main problem was Apple's persistent belief, despite all the evidence to the contrary, that everyone would be willing to pay a premium price for a niftier machine). For sure, however, Microsoft has mainly been restrained by the knowledge that any crude use of its power would indeed land it in court.

And yet, despite all that restraint, Microsoft is in court anyway. Any nontechnologist ventures into the browser wars at his peril, but here is how I understand it: After initially missing the significance of the Internet, Microsoft has gone to the other extreme, designing Windows 95 so that it uses an Internetlike metaphor for everything. It makes sense, then, for a browser that can find both internal and external documents to be an integral part of the system--unless, that is, you regard browsers and operating systems as still basically different things, and view Microsoft as practicing vertical foreclosure under the guise of product enhancement. Well, maybe--but it's a pretty subtle point. Microsoft isn't preventing anyone from using Netscape or charging Netscape for the right of access; it's providing Internet Explorer free, but then that would be normal practice in this kind of industry even if IE wasn't allegedly an integral part of Windows 95. You can argue that Microsoft has stepped across the line on this one--but surely by only a few inches.
       Here's what worries me: Given the subtlety of the real issues here, what is the chance that this stuff will be decided on its merits? When you hear that despite the fact that he has economists who know better, the Justice Department's Joel Klein apparently either believes or chooses to claim that this case is about path dependence, you start to wonder. And when you hear that the anti-Microsoft side has retained the services of that economic and technology expert Bob Dole, you start to despair.


An article in the Feb. 25 Wall Street Journal ("QWERTY Spells a Saga of Market Economics") is a good example of the mistaken view that path dependence is the core of the case against Microsoft. Nicholas Economides, who really understands these things, took the Wall Street Journal to task in a letter that it, of course, did not print. His useful site also contains a good, slightly anti-MS discussion of the vertical foreclosure issue on browsers as well as a set of links to just about everything relevant to the case.

Paul Krugman is a professor of economics at MIT. His new book, The Accidental Theorist and Other Dispatches From the Dismal Science, will be published this month. His home page contains links to many of his other articles and essays.

Illustrations by Robert Neubecker.

Slate contents
Slate Search

More The Dismal Science
More by Paul Krugman
Search Slate: