by Ronald A. Cass & Keith N. Hylton

Consider the following developments over the last year. Visa and Mastercard settle antitrust litigation brought by Wal-Mart for $3 billion over the allegation that the card businesses illegally tied use of debit-card and credit-card purchases. The European Union announces a decision to impose a fine of $600 million on Microsoft for, among other things, its incorporation of a Media Player into its Windows computer operating system. “Tying” doctrine, here and overseas, has become the crowbar with which virtually every business now can be threatened. A quick review of the doctrine suggests that although courts have made progress in understanding why businesses combine particular features in the packages they sell to the public, much work remains to be done to bring the law into line with business realities.

Tying doctrine prohibits (under certain circumstances) the connection of two goods for sale as a package. This doctrine began with a simple view of what tying was all about: the firm could use the product that was intensely demanded to generate sales of a less popular product. Courts viewed that tie as inimical to competition and to consumers. And some of the early cases looked, at least on the surface, as though they fit that pattern. Not surprisingly, then, the courts adopted a per se rule making tie-ins illegal when the seller had market power in the intensely-demanded product.

Over time, tying doctrine has come under considerable pressure. Two observations lie at the core of this pressure.

First, commentators observed that the connection of more and less popular (more and less intensely demanded) goods did not, except in the most unusual conditions, confer advantage on the firm. The “one monopoly rent” argument asserts that a firm with market power can extract the benefit of that power only once – by raising the price of the good it has power over – and that the firm can change the form of the rent extraction but not the magnitude. Although that assertion left room for debate about just which conditions allowed a shift of profits to the tying firm, economists generally agreed that the basic argument was correct.

Second, observers recognized that tying is not a rare and suspicious practice. Quite the contrary: it is ubiquitous, commonly done by firms with little or no market power as well as by dominant firms. Hotels bundle rooms with robes and towels, shampoo and chocolates – and no one thinks their hotel should be offering them a choice of competing vendors for the amenities, even if they might prefer another brand or item. Law firms offer the services of their senior lawyers in combination with those of their more junior lawyers – and no one thinks that partners in one firm should be required to offer clients a choice of associates from other firms. [other examples(?): Colleges offer the services of their best teachers along with those of average teachers – and no student thinks that gives him a right to substitute courses from another school. Cars are sold with audio, dvd, and air-conditioning systems. Shoes are sold with laces even though every stores that sells shoes also sells laces independently.] The list of packages – of tie-ins – is endless. Virtually all products and services are sold as combinations of items or features that could be – and often are – also sold separately. Certainly, if everyone ties goods together as packages, even with no market power in either good, the reasons for tying cannot primarily be the extension or leveraging of monopoly power. Instead, efficiencies – lower cost to producers and lower cost or higher value to consumers – provide the general explanation for tying. That point seems lost on the European Union’s antitrust regulators, who seem far more concerned with preventing disadvantage to competing businesses than with promoting efficiencies that benefit consumers.

Over the past two decades, however, U.S. courts responded to the obvious mistake of tying doctrine – of labeling tying as so clearly inimical to consumers to be branded as per se illegal – with various efforts to step back from its more egregious consequences. The Supreme Court’s 1984 Jefferson Parish decision, for instance, attempted to separate instances in which there are obvious, pervasive inefficiencies of separate provision of goods – so pervasive that there was not an independent market for the tied good. The Court retained the per se rule in form, but it moved toward analysis of the markets for the tied and tying goods that functionally moved the doctrine closer to a neutral standard. Four justices would have gone further and adopted a “rule of reason” analysis (an unstructured inquiry into the efficiency benefits and competitive effects of the practice in each specific instance).

The Jefferson Parish decision marks an important step in the development of tying doctrine, but it ultimately does little to reverse the initial mistake. After all, almost all tied goods do have independent markets. There plainly are separate markets for bathrobes and towels and chocolates, even if it is grossly inefficient for hotels to separate those items from the room. Hotels tie the items together when leaving the provision of these items to the traveler reduces the value of the package by more than the cost savings to the hotel. The same is true for joint provision of automobiles and audio systems and for innumerable other bundles. Failure to clearly recognize this point left the law in limbo, with some opportunity to except efficient bundling from tying law but with the onus largely on defendants to carve their conduct out from the presumption that tying was improper.

Another step in the development of tying law came in the U.S. litigation over Microsoft Corporation’s evolving Windows platform for personal computers. The Department of Justice claimed Microsoft violated the antitrust law by bundling its Explorer Web browser with its Windows operating system. The Web browser was one of many features that over time went from options available strictly on a stand-alone basis to integrated parts of the Windows platform. Indeed, Windows was created by bundling the DOS operating system with a (formerly separate) graphical user interface.

The U.S. Court of Appeals for the D.C. Circuit rejected the district judge’s conclusion that the Web browser’s integration into Windows created an illegal tie-in. It rooted this rejection largely in its understanding of the market forces that support bundling. Going beyond Jefferson Parish, the Microsoft III decision (as this one is commonly referred to) recognized that there typically are efficiencies in combining features that could be offered on a stand- alone basis. The court observed that bundling can increase value to consumers and can take advantage of economies of scope that lower costs. It noted that software producers, whether plausibly having market power or not, commonly integrate new features into their software. The court saw that in platform markets especially, integration could help software developers as well as consumers, by reducing the costs of other software that would be added on to the platform.

The court, therefore, adopted a rule of reason approach, under which the government would bear the burden of showing that the harm to competition from integration of features into software outweighs its efficiency benefits. This test, which the Microsoft III court expressly limited to software tying cases, is a far cry from the presumption that tying is per se illegal. It seems far more in keeping than European competition doctrine with an understanding of why all software firms bundle features together rather than selling each one discretely and why software programs grow over time to house more and more features that once were sold separately.

As a step in the general development of tying doctrine, then, Microsoft III should be applauded. But the test also stops well shy of acknowledging that the practice the court was addressing – and the problem with treating the practice as presumptively illicit – was not peculiar to the software business. The court also set forth a test that does not readily distinguish the problematic tie from the ordinary, market-driven integration of features and that leaves a great deal of discretion to the decision maker in the particular case.

Microsoft III’s rejection of the per se illegality approach is clearly the right decision, given the efficiencies of tying. Before software businesses crack open the champagne to celebrate, however, they should think about two things. First, the new rule of reason test for software integration, although more lenient than the hybrid per-se test generally applied to tying, is actually a more restrictive test than the one courts traditionally have applied to the physical integration of existing products. In this sense, Microsoft III will appear to some as a step backward along tying law’s path. Second, software businesses should consider the difficulties that arise under the rule-of-reason test. How will the burden of proof ultimately be allocated by courts? How will anticompetitive harms be determined – by the complaints of rival software firms or by proof of harm to the typical consumer? Will the test accommodate instances in which efficiencies are likely to be realized only after several years have passed? Will allowance be made for “honest mistakes” in business judgment, as when a firm takes a competitive action that harms a rival and the resulting efficiencies are weaker, or further delayed, than anticipated? How much weight should be put on market share in software markets, where consumer herding imparts volatility to market share statistics?

Given the continuing pressures on the per se illegality rule, the future of tying law depends on how courts answer these questions. A rule-of-reason test that makes sense in light of tying’s efficiencies, in software and in other markets, would set a high standard of proof – perhaps requiring “clear and convincing” evidence – and put the burden on the plaintiff. A test that aims for a formal neutrality by imposing significant proof burdens on the defendant would leave many businesses in the position they are in today, uncertain about the liability risks of tying. When the vast majority of ties are efficient, as observation suggests, the proper resting point for tying doctrine is on the other end of the spectrum from its per se illegality starting point: something close in operation to a presumption of legality. We hope that tying doctrine will continue to evolve in this direction, to continue to be guided more by business reality than its original formulation, and to be less dependent on its initial path. But after Wal-Mart and the EU’s Microsoft decision, we fear that the path toward a sensible tying doctrine will be a long one.