MAKING MICROSOFT PLAY NICE?: Why "Conduct Remedies" Won't Work, And A Breakup Should Be Reconsidered

In June, when the U.S. Court of Appeals for the D.C. Circuit issued a decision setting aside the district court's order to break up Microsoft, it ruled out a "structural remedy" for Microsoft's misbehavior — that is, changing the structure of Microsoft to alter its interactions with other companies.

With structural changes removed from the courts' arsenal, the only remedy left is to ask Microsoft to play nice. As a result, the subject of "conduct remedies" has been a hot news topic lately. The problem is that such remedies have not worked in the past, and they will not work in the future. As the shortcomings of conduct remedies begin to show themselves, the structural remedies the court took off the table will start to look like the only workable option, and we may be back to square one.

The Decision for Plaintiffs: Try to Delay Windows XP?

Within the next few weeks, the plaintiffs in the Microsoft case — the U.S. Department of Justice and eighteen state attorneys general (nineteen minus New Mexico, which recently dropped out) — must make an important decision.

They must decide whether to ask for an injunction to delay the release date for Microsoft's new operating system, Windows XP, which is set for release in October. Their alternative is to ask the court, instead, to force Microsoft to reconfigure the new system before its release.

Windows XP is said to integrate several new features into the operating system, including Internet shopping and instant messaging services. The plaintiffs are concerned that this "bundling" of services will tighten Microsoft's existing stranglehold on the software industry.

Their fear is realistic: In much the same way, Microsoft's integration of its internet browser into Windows 98 all but destroyed Netscape and turned the browser market into another Microsoft monopoly.

An Unclear Ruling from the Appeals Court

Last week, the D.C. Circuit issued a ruling that might have made it easier for the plaintiffs to make this call. Unfortunately, the ruling was far less clear than it might have been.

The Court held that the software industry is not entitled to a special legal standard when a court decides whether "commingling" of code is anticompetitive. Rather, the question whether behavior within the software industry is anticompetitive is to be judged by the same standards that apply in every other area of business.

In other businesses, of course, bundling is not a matter of commingling code, but simply offering two goods for sale as an indivisible package deal. A classic example of this was the movie double feature, in which consumers were induced to pay one premium price for a ticket to two movies. Not surprisingly, this price was designed to force consumers to pay more for the less popular movie, taking advantage of their desire to see the main attraction.

It sounds simple, but the court's decision leaves many ambiguities unresolved. Even knowing that the general test applies, the parties have little clue how it will work in practice. The outcome of the test — which balances any competitive harm deriving from Microsoft's combination of products in Windows XP against the consumer benefits from the combination — is unpredictable.

Unpredictable, too, is whether an injunction will issue even if the balance goes against Windows XP. To grant an injunction, courts must assess whether future harm from the software's release would be irreversible — an inherently speculative inquiry.

And if Microsoft is, indeed, enjoined to uncouple the components of Windows XP, what exactly would that mean? Would it be enough (as some experts assert) for Microsoft to provide users with an "opt out" — a method by which they can remove any bundled features that they do not want? And if so, how clear must the removal option be, and how easy must the deletion process be? Must a court, in essence, create a standard for how Microsoft's code must be written?

The Problem with Conduct Remedies

These issues illustrate the problems with conduct remedies. They are logically indistinguishable from regulation of the industry, and — as conservatives often point out — regulation is the province of the legislature (and, through the legislature, administrative agencies), not the judiciary.

Judges are neither inclined to regulate industries nor generally trained for the task of doing so. As the leading antitrust law casebook remarks, with marked understatement, "antitrust tribunals would not often find it congenial to supervise the price and other terms on which … firm deals with outsiders."

Ironically, while conservatives dislike structural remedies because they intrude on the sphere of private business, it is possible that they should dislike conduct remedies even more, with their flavor of "judicial regulation." Which is worse, the government's telling Microsoft to turn itself into two independent companies, or telling it how to write and market its software?

A Return to Structural Remedies?

Given the difficulties with conduct remedies, might Microsoft want to adopt a voluntary structural solution? During the trial, some observers suggested that Microsoft might even want to split itself in two, voluntarily. A few analysts even suggested that a break-up might increase profits to the two companies' shareholders, much as other companies increase shareholder value by spinning off units.

Faced with the possibility of ongoing judicial scrutiny if conduct remedies are imposed, Microsoft might be better off taking its medicine, agreeing to a breakup, and getting on with its two lives (although Bill Gates so far has categorically rejected this solution).

But would consumers benefit from a breakup of one monopoly into two? Arguably, the answer is no. Here, the two monopolies would have a "vertical" relationship–that is, the operating system company would produce a product that an applications company would rely upon. And a basic economic argument (too lengthy to restate here, but widely taught in undergraduate microeconomics courses) holds that when one monopoly is broken into two, and the two have a vertical relationship, their final product will have lower output and a higher price.

Economists across the ideological spectrum have invoked this idea in opposing a breakup. Paul Krugman, the liberal New York Times commentator, claimed that "standard economic analysis" proves that a breakup would impose "considerable cost" on consumers. And arch-conservative Gary Becker claimed that "economic theory" proves that commingling products benefits consumers.

It is not at all clear how Krugman claims to know that the cost would be considerable, since economic theory tells us nothing about magnitudes. The two-monopoly price could be 0.1 cents higher and output 0.001% lower than the one-monopoly result, or the price could be ten times higher and the quantity 98% lower, depending on a range of technical factors.

Even accepting all of the simplifying assumptions of the orthodox economic analysis, therefore, the most that economists can confidently assert is that the market would be to some degree less beneficial to consumers in the short run.

The software industry is interesting, however, precisely because the future happens so quickly. If Microsoft were broken up, the operating systems company would no longer have any reason to favor certain Microsoft applications. Its incentive would be exactly the contrary — to ensure that all programs run smoothly on its system.

Simultaneously, the separate applications company would have no special incentive to favor Windows-based applications. Most importantly, the possibility of bypassing operating systems entirely (so-called middleware) could re-emerge.

In short, the "two monopoly" problem could quickly turn into a world with no monopolies — and despite economists' criticisms, the structural solution to Microsoft's behavior could end up looking pretty good, in a relatively short amount of time.

Moreover, economic analysis is not the whole story. One of the oft-ignored roles of antitrust policy is to limit concentrations of wealth and power–and it is difficult to imagine more concentrated wealth and power than Microsoft currently possesses. Indeed, it was Microsoft's arrogant abuse of that power that convinced the trial judge to order the break-up remedy in the first place.

Why Conduct Remedies Probably Will Not Work

Suppose conduct remedies are imposed. Can we expect Microsoft fully to comply? Not likely.

Recall that Microsoft already has a pretty miserable track record when it comes to conduct remedies. Before the trial that led to the break-up order, the Department of Justice had attempted to work with Microsoft on a consent decree (an agreement between Microsoft and the government in which Microsoft agrees not to misbehave in certain ways) — to no avail.

The government filed suit only after it concluded that Microsoft was not interested in the consent decree solution. Indeed, Gates even bragged to interested parties that the agreements did nothing to stop Microsoft from doing as it pleased. With this background, it would be the height of optimism–or naivete–to imagine that conduct remedies will do the trick.

If experience is any guide, now that the courts must now regularly decide whether Microsoft is misbehaving, and if so, what to do about it, Microsoft will continue to fight and delay and obfuscate — and ultimately to get away with whatever it wants.

The court of appeals left open the possibility that a breakup could be reconsidered, should further proceedings in this case warrant it. As events play out over the next several years, the odds are that structural remedies will look better and better.