I have a feeling that we are two ships passing in the night. We may be talking about the same thing, but our different languages make communication difficult.
Changing a core package like the spreadsheet tool drags with it a rat-tail of costs ranging from dealing with archived files, training and support, integrating the product on the desktop, changing the currently active spreadsheets, making sure your customers have the same release as you so that you can convert them, ensuring that people who work at home are upgraded, etc. Time and the risk of error is more important than money in this case.
My point is simply that the cost of switching can be quantified and compared with the cost of staying put. Not being able to switch is definitely a barrier, but it's related to efficiency, not some external factor, like law, for example. "Efficient" is not the same thing as "cheap" or "easy."
Reginald has no idea of what he is talking about.
In general, I agree. ;)
Path dependence and then lock-in through network externalities are feedback mechanisms that may or may not converge to an efficient setup. At least not in the sense of the most efficient or best.
Here, you're talking about macroeconomic efficiency, or "consumer welfare;" when I say that the network effects barriers to entry are efficiency-related, I'm talking about efficiency at the firm level, what Bork calls "productive efficiency." Path dependence and network effects are a characteristic of the market which affects the economies of scale and efficiencies of the firms that make up the market.
As for examples of path dependence leading to a demonstrably less efficient outcome, all I can say is, "show me." Liebowitz and Margolis discuss in detail the fact that there is no documented example of path dependence leading to an unambiguously "inferior" technology being adopted over a "superior" alternative. They thoroughly discredit the QWERTY and VHS vs. Betamax examples and contend that the Mac vs. Windows example is also not a case of path dependence leading to widespread adoption of the "inferior" technology. Liebowitz, Margolis, Should Technology Choice be a Concern for Antitrust Policy?, 9 Harv. J.L. & Tech. 283, 312-316 (1996).
Furthermore, even if you can show me an example of demonstrably inferior technology being adopted due to path dependence, in order to justify governmental intervention to get to the higher efficiency level, you have to also show me that government intervention is more capable than the market of deciding which selection is more efficient and then getting us there. I don't think you can do it, but you are certainly welcome to try.
I don't think path dependence refers necessarily to a company
It refers to the market structure within which the company must operate and must deal with in order to be efficient.
and certainly does not imply "natural" monopoly in the sense of "sheer efficiency".
Oh yes it does.
Go read the literature by people like Margolis and Liebowitz, and Page and Lopatka. (I can't give you specific cites now, but let me know and I'll get them.) They explain in precise detail why the network effects literature has really just created a new terminology to describe good, old fashioned natural monopoly, and criticize it for this reason, among others.
1 - Network externalities and path dependence are not a kind of efficiency, they are the mechanisms that lead to a lock-in condition. This lock-in has allowed MS to be some kind of monopoly on the desktop.
Yes, I agree that's what the government's case says. It's called a "natural monopoly."
2 - However, true efficiency of the marketplace in which Microsoft competes is not determinable because the marketplace itself is under discussion. Therefore there is no straight ahead way of determining if MS is a "natural" monopoly.
All I can say is: go read the DOJ's complaint. They define the monopolized market as the IBM compatible desktop PC Operating System market. Furthermore, this kind of narrow market definition is exactly what one would expect to see from someone schooled in network effects theory.
As for proving that Microsoft is a "natural monopoly," it can be shown through an exposition of the structure of the IBM desktop PC OS market, how Microsoft has 90 percent plus in that market, how size is a proxy for efficiency, how network effects/path dependence literature says a 90 percent plus market share is exactly what one would expect to happen.
At that point, I think the proof of a monopoly will just boil down to how contestable the Microsoft natural monopoly's market really is. I suspect the government will try to argue that market is not contestable, and Microsoft will try to argue that it is.
As for who is right, your guess is as good as mine, maybe better.
I suggest we leave the question of efficiency behind us and address the more concrete issue of whether or not artificial barriers exist. Then propose the converse; since artificial barriers were created, MS is a garden variety monopoly.
What artificial barriers are you talking about? Network effects? If so, those I submit are efficiency-related. If not, then what are they?
And, if the government thinks artificial barriers, as distinguished from efficiency (whether due to network effects or otherwise) are what give Microsoft market power in what would otherwise be a competitive, multifirm industry, why did the government take so much care *not* to attack Microsoft's existing monopoly? Why did it not ask for a breakup of the company?
How can this be done? We need to show the existence of barriers. Barriers can only exist at boundaries, so the DOJ will have to spend some time delineating them.
Yes. The government would have to rewrite its whole complaint, and do an extensive historical analysis of Microsoft's past conduct in order to show that Microsoft's monopoly arose from predation or other "artificial" barriers erected by the firm. My understanding is that Joel Klein is a fan of the "strategic strike" lawsuit where he would go after specific practices instead of getting bogged down in an IBM-style marathon. The natural monopoly theory the government appears to be pursuing is in accordance with this kind of approach. A case based on artificial barriers in an OS market otherwise capable of supporting multiple firms, apart from being inconsistent with the facts as described so aptly in the network effects literature, would be more similar to the IBM style suit. As always, I could be wrong, but it is my impression Joel Klein does not want to go there.
2 - A second argument has to be about distribution channels and the barriers to entry on the Windows screen (icons) as well as pressure on the distributors themselves. This will delineate Pepsi in a Coke 6-pack question.
Well, just on an intuitive basis, I think the "barriers to entry on the Windows screen" are trivial. I mean, what does it take to get on the internet, type in www.yahoo.com in the location field, then type "Netscape" in the search field and press "enter"?
You say, "But, oh, it takes so long to download and install Netscape."
And my response is that this is a result of slow modems, a problem which is associated with government regulation in the telecom industry, not Microsoft, and which is being remedied as the telcos finally get around to deploying ADSL. By the time this case finally gets resolved, this "barrier" may well on the way to gone.
Then you might say, "But, oh, the ignorant consumer will just click on the first thing that shows up on the screen. So Microsoft has an unfair advantage because it gets to decide what consumers see first."
And my response is that I do not think this is such a big advantage, because consumers are not so stupid, and because even if they start out ignorant, they ramp up the learning curve pretty quickly.
To me, the DOJ's complaint reads almost as if the internet didn't exist. I'm sure they know all about the internet's potential as a distribution tool, and maybe they are thinking that it's an issue for Microsoft to raise. But, in any event, I question how much distribution of Netscape's products has really been foreclosed by these desktop icon restrictions. The contractual restrictions on ISP's ICP's and OLS's promoting other products and paying money to Netscape are a different matter, but these desktop icon restrictions just seem trivial to me.
It seems to me that, to prevail on the desktop issue, DOJ has to prove that consumers are ignorant. That's not a position I would want to be in.
Once this structure is in place we can return to the issue of efficiency. If Windows is defined as different from an Internet browser then clearly Netscape was "more efficient" initially because MS had no product. At that point MS started erecting "artificial" barriers to entry, which leads to the conclusion that MS is not a natural monopoly. After all if MS were a natural monopoly they wouldn't have to choke off any air supply.
Well, funny you should mention that.
I just happened to be leafing through Areeda and Hovenkamp on Antitrust today in my search for information on how Antitrust deals with natural monopolies. Areeda and Hovenkamp state:
"Although natural monopoly should be a defense to any claim for divestiture or forced accommodation of multiple firms, we do not believe it should defend against an injunction or damages action for a provable exclusionary practice. Our conclusion rests on the following grounds: First, courts are not good at recognizing natural monopoly and as a result our confidence level in a natural monopoly determination is not always as high as it should be. Second, the fact that a market is a natural monopoly should not operate as a guarantee that a particular incumbent is entitled to be the natural monopolist; a more aggressive rival might be a more efficient occupant of the same position. Third, if a market really is a natural monopoly, exclusionary practices should be unnecessary as a general matter."
3 Areeda & Hovenkamp, Antitrust Law 658b3, at 117 (1996).
The first and third rationales, I would argue, are worse than weak; they actually could support a rule of non-intervention.
"First, courts are not good at recognizing natural monopoly and as a result our confidence level in a natural monopoly determination is not always as high as it should be."
This cuts both ways. If a court has trouble recognizing a natural monopoly, then it makes as much sense to say that the defense should be interpreted overly broadly as that it should be narrowly construed. In fact, in makes more sense, since, as Easterbrook has stated elsewhere, antitrust policy should bend over backwards to leave imperfectly understood markets and business practices alone; it should intervene in the market only when the need for and effects of a proposed remedy are clear.
Furthermore, if a court has trouble recognizing a natural monopoly, why would it have any less difficulty in recognizing that the conduct engaged in by the unrecognized natural monopolist is predatory or exclusionary, especially if it is true that, as argued below and in my other posts, conduct that would be exclusionary or predatory if engaged in by a firm in a competitive market is not necessarily so when engaged in by a monopolist?
As Lopatka and Kleit put it:
"In the end, true anticompetitive foreclosure cases - those involving practices that unambiguously reduce economic welfare - may well exist, but they have yet to be confidently identified. Foreclosure theories are certainly ground for scholarly attention. Just as certainly, however, they need to be refined and tested before they are ripe for practical application. At least for today, these theories are too weak to support a theory of antitrust liability."
Lopatka, Kleit, The Mystery of Lorain Journal and the Quest for Foreclosure in Antitrust, 73 Tex. L. Rev. 1255, 1306 (1995).
"Third, if a market really is a natural monopoly, exclusionary practices should be unnecessary as a general matter."
So true . . . which would suggest that the practice in question is not exclusionary!!!
If a rational natural monopolist would not engage in exclusionary conduct, a court should search long and hard for alternative rationales before concluding that a particular form of conduct by a natural monopolist is exclusionary. Why should a court assume that natural monopolists act irrationally and inefficiently?
That leaves the second rationale, which is the truly substantive rationale, the meat of the argument:
"Second, the fact that a market is a natural monopoly should not operate as a guarantee that a particular incumbent is entitled to be the natural monopolist; a more aggressive rival might be a more efficient occupant of the same position."
This is true.
However, it does not necessarily follow that antitrust law should attempt to prohibit exclusionary or predatory conduct by the natural monopolist. This is because conduct that antitrust theory and precedent would recognize as predatory if engaged in by a firm in a competitive market might actually be welfare enhancing if engaged in by a natural monopolist.
If you accept that, then you will accept that we don't have an efficient software industry.
I'm not saying that the software industry (actually, the desktop PC OS market, to be precise) is optimally efficient. I'm just saying that, in the case of a natural monopoly, a single firm is more efficient than multiple firms at meeting the demand of the market. After all, a natural monopoly is *still* a monopoly. It's just that the policy implications of that fact are different. It is pretty much accepted that antitrust should not be used to impose structural remedies on a natural monopoly. "[S]ociety would not be well served by an antitrust rule that broke up natural monopolies and forced competition. As a result, we can confidently conclude that the proven natural monopoly resting on economies of scale should be immune from antitrust restructuring, whether or not it charges more than several less efficient firms would charge." 3 Areeda & Hovenkamp, Antitrust Law 658b2, at 117. As Liebowitz and Margolis put it: "It may be that some types of software products should be produced only by a single supplier. But this is not the usual venue for antitrust." Liebowitz, Margolis, Should Technology Choice be a Concern for Antitrust Policy?, 9 Harv. J.L. & Tech. at 317.
I'm looking forward to your reply to my rather spirited discussion of your arguments. :) |
