Antitrust in the New Economy.
Address by Judge Richard Posner.
Date: September 14, 2000.
Source: Office of Judge Posner. Tech Law Journal converted a MS Word file into HTML.

ANTITRUST IN THE NEW ECONOMY

Richard A. Posner[1]

Concern has been expressed recently that U.S. antitrust law may not be well suited to regulating the “new economy.” Doctrines developed to deal with competition and monopoly in smokestack industries is not well adapted, it is argued, to dealing with the dynamic economy of the twenty-first century. What I shall argue is that there is indeed a problem with the application of antitrust law to the new economy, but that it is not a doctrinal problem; antitrust doctrine is supple enough, and its commitment to economic rationality strong enough, to take in stride the competitive issues presented by the new economy. The real problem lies on the institutional side: the enforcement agencies and the courts do not have adequate technical resources, and do not move fast enough, to cope effectively with a very complex business sector that changes very rapidly. This problem will be extremely difficult to solve; indeed, I cannot even glimpse the solution.

I shall use the term the “new economy” to denote three distinct though related industries. The first is the manufacture of computer software. The second consists of the Internet-based businesses (Internet access providers, Internet service providers, Internet content providers), such as AOL and Amazon. And the third consists of communications services and equipment designed to support the first two markets. There are other candidates for inclusion in the new economy, but these three will do for my purposes.

The principal output of these industries (with the partial exception of communications equipment) is intellectual property, namely computer code, rather than physical goods, although the intellectual property may be shipped on a disk or other physical product (not necessarily: software is increasingly shipped to the purchaser over the Internet). This is obviously the case for computer software, but it is also true to a large extent of the Internet-based businesses. Their ability to take and fill orders and carry out the other operations (such as marketing, billing, handling returns, responding to customer questions and complaints, and allaying the customer’s privacy and security concerns) required to give their customers whatever goods or services the business provides is a function to a large extent both of the sophistication of the business’s computer software and also of its trademarks and copyrights.

Intellectual property is characterized by heavy fixed costs relative to marginal costs. It is expensive to create but once created the cost of making additional copies is low, dramatically so in the case of software, where it is only a slight overstatement to speak of marginal cost as zero. Without legal protection, the creator of intellectual property may be unable to recoup his investment, because competitors can free ride on it; and so legal protection can expand, rather than as the usual case with monopoly contract, output. But with legal protection, output may be artificially constrained and consumers deflected to more costly substitutes. The owner of the patent or copyright will charge a positive price for copying, even though the marginal cost may be zero; the price will deflect some consumers to substitutes. To prevent these defections would require perfect price discrimination, which is infeasible because (administrative costs to one side) it would require the seller to have complete information about the elasticity of the demand for his product by all his customers and potential customers.

The patent and copyright laws try to strike the output-maximizing balance by giving the creator of intellectual property some but not complete protection from competition. Copyright confers protection for a longer period than patent does because, traditionally, less was protected—just the form in which the composer, painter, writer, etc. had chosen to express his ideas, and just copying, not independent discovery. The extension of copyright to software has been controversial, many arguing that it confers excessive protection because of difficulties of inventing around an innovative code. Even worse, it is argued, the methods of distributing software often enable the creator to obtain by contract even more protection than copyright law gives him. Copyright gives the holder a property right in his intellectual property even when it is in the hands of a person with whom he has no contract, such as the purchaser of a copyrighted book from a bookseller. To the extent that the creator of software contracts directly with the ultimate purchaser, he can impose via contract more restrictions than the copyright law would allow him to do in the absence of contract; for example, he can forbid the purchaser to make an extra copy for his own use, as copyright law permits. It is true that a contract is unlikely to have the same duration as copyright protection, but length of protection is academic in the case of software, which becomes obsolete long before the copyright on it expires.

The possibility that the combination of copyright and contract gives software manufacturers too much monopoly power in the economic sense, that is, causes a lessening rather than an increase in the output of the intellectual property in question, creates a natural concern with any further practice or circumstance that might increase the manufacturer’s power over the price of his software. It is at this point that another feature of the new economy that I mentioned, economies of scale in consumption, becomes troublesome. Economies of scale in manufacture are familiar; up to a point, the longer the production run, the lower average cost is. Economies of scale in consumption refer to the situation in which the larger the firm’s output is (up to some point), the more valuable that output is to its customers. The traditional example was the telephone. Telephone service is worthless if there is only one subscriber; he has no one to talk to. The more subscribers, the more valuable the service is to each one, or at least to many of them. Interactive services, such as email and online auctions, are similar. Likewise the sharing of computer programs, as where two or more academics collaborate on writing a scholarly article by means of word processing and spreadsheet programs. Literal networking or sharing to one side, computer programs tend to be more valuable the more people use them because training, support by IT personnel, and standardization of equipment and procedures are facilitated. It is the same reason that the typewriter keyboard is standardized.

These economies of consumption presuppose uniformity rather than common source. The international telephone system is a single network, but its components are owned by a vast number of separate firms and individuals. The components have, however, been standardized to assure interoperability, in the same way that the gauge of the railroad track has been standardized. A firm that manufactures one of the essential components of a network (a term I shall use to denote any situation in which there are economies of scale in consumption) would prefer to be the exclusive source of that component rather than be required to disclose the information that would enable competitors to duplicate it. If the component is subject to intellectual-property protection through patent, copyright, or contract (or can be held as a trade secret), then the requisite uniformity is more likely to be achieved by monopoly provision than by standardization.

Networks are not valuable to the consumer in themselves; they are conduits for the services that the consumer values. This is one point at which vertical integration enters the new economy. An operating system is a platform for software applications, and so the writer of operating-system software may decide to write software applications to ride on it, in much the same way that AT&T manufactured the terminal equipment attached to its telephone lines. Modern operating systems are themselves composites of separate programs which could be and sometimes are provided by different companies; there is an analogy to AT&T’s practice of manufacturing the switching equipment for its telephone system as well as the telephone lines themselves. Firms that provide dial-up connections and other facilities for accessing and browsing the World Wide Web can integrate forward into the provision of Web-based services such as shopping and video.

The features of the new economy that I have been describing, all but the last (vertical integration, of which more later), tug it toward monopoly yet also, oddly, toward competition. The paradox dissolves by a reminder that competition to obtain a monopoly is an important form of competition. The more protection from competition the firm that succeeds in obtaining a monopoly will enjoy, the more competition there will be to become that monopolist; and provided that the only feasible or permitted means of obtaining the monopoly are socially productive, this competition may be wholly desirable. A firm that will have the protection both of intellectual-property law and of economies of scale in consumption if it is the first to come up with an essential component of a new-economy product or service will have a lucrative monopoly, and this prospect should accelerate the rate of innovation, in just the same way that, other things being equal, the more valuable a horde of buried treasure is, the more rapidly it will be recovered. What is more, the successful monopolist is likely to be a firm that initially charges a very low price for the new product that it has created. Think back to the telephone. Since every new subscriber increases the value of the service to the existing subscribers, a telephone company has an incentive to provide price inducements to new subscribers, as the money it will lose on them may be more than made up by the higher price that existing subscribers will pay for access to a larger network. This is especially likely if the network will be a natural monopoly, in the sense that no competitor would find it feasible to duplicate it—then the faster the network reaches maturity the longer the monopolist will be protected from challenges to his monopoly. The prospect of a network monopoly should thus induce not only a high rate of innovation but also a low-price strategy that induces early joining and compensates the early joiners for the fact that eventually the network entrepreneur may be able to charge a monopoly price.

I emphasize “may be able” in the preceding sentence. Traditional networks such as the telephone system and the railroads required enormous capital investments and were therefore difficult to duplicate. If you owned such a network, or an important part of it, you had a pretty secure monopoly. The less capital investment the creation of a substitute network involves, the less secure the network monopolist’s monopoly is. Because of the extraordinary rate of innovation not only in computer software but also in communications technology, the extraordinary amount of capital that is available worldwide for investment in new enterprises, and the rapidity with which new networks that are primarily electronic can be put into service, the networks that have emerged in the new economy do not seem particularly secure against competition. We have seen all manner of firms rise and fall in this industry—falling sometimes from what had seemed a secure monopoly position. The gale of creative destruction that Schumpeter described, in which a sequence of temporary monopolies operates to maximize innovation that confers social benefits far in excess of the social costs of the short-lived monopoly prices that the process also gives rise to, may be the reality of the new economy.

The feasibility of challenging an existing network monopolist is critical. Even if the only way to become a network monopolist in the new economy is to be the first to come up with a new technology that benefits consumers, the existence of the monopoly may discourage subsequent technological innovation by other firms. If network externalities are large, they may give the monopolist a cost advantage that exceeds the benefit of a superior new technology. This is the issue of “path dependence”: an industry may be stuck with an inferior technology because of the cost advantage of the existing network.

Appearances to the contrary notwithstanding, the antitrust laws are not much concerned with monopoly as such. It is not a violation of those laws to acquire a monopoly by lawful means, and those means include innovations protected from competition by the intellectual-property laws. If copyright protection of computer software is too broad, that is a matter to take up with Congress. Nor is it a violation of antitrust law to charge a monopoly price, or to price discriminate in an effort to maximize monopoly profits. And now that the Alcoa doctrine is thoroughly discredited, it is understood that a monopolist is free to compete, whether against the competitive fringe in its monopoly market or against potential competitors, as vigorously as a firm in an ordinary competitive market would be. The fact that a monopolist buttressed by network externalities may be hard to dislodge even by a firm with a superior technology has no antitrust significance in itself. What is true is that a firm is forbidden to enter into a price-fixing agreement with its competitors, or to acquire a competitor if the acquisition will alter the structure of the market to make it much more conducive to price fixing, but these rules do not require any modification to deal with mergers and price fixing in new-economy industries. Nor is there a flat rule against communication among competitors on matters such as standards setting where consumers may benefit from a degree of interfirm cooperation.

The focus of concern with the application of antitrust law to the new economy is on the methods by which a firm that has a monopoly share of some market in a new-economy industry might seek to ward off new entrants. The lawyers and economists who express this concern are fearful lest a “Chicago school” approach to antitrust deny the possibility that a single firm, without collaborating with competitors or potential competitors (thus inviting application of the rules against price-fixing and large horizontal mergers), can, at least under new-economy conditions though probably more generally as well, prevent efficient challenges to its monopoly. If the Chicago-school approach so understood is law, these critics want it modified to do service in new-economy antitrust cases.

This is a misunderstanding of the Chicago school approach, at least as I have been articulating it for the past quarter century or so.[2] The approach is skeptical—but no stronger word would be correct—about the danger to competition that is posed by unilateral firm action, unilateral in the special sense that it does not require cooperation with competitors (it usually requires cooperation with customers or suppliers). The approach emphasizes both the difficulty of squashing competition by such means and the danger that heavy-handed antitrust enforcement may suppress a practice that seems anticompetitive but actually is efficient, or at least neutral, from the broader social standpoint. A classic example is the tying agreement, which used to be thought a means by which a firm having a monopoly of one market (the market for the tying product) could obtain a second monopoly (over the market for the tied product). Since by definition the two products are complements (hammers and nails, for example), an attempt to increase the price of one will reduce the demand for the other. Owning both monopolies will produce a net increase in the monopolist’s profits only if the second monopoly enables the monopolist to engage in price discrimination more easily, with sales of the tied product being used to monitor the intensity of the consumer’s demand for the service (such as hammering nails) that the products jointly produce. Although price discrimination has no general tendency to increase efficiency,[3] banning one form is unlikely to do any good, since there is no general antitrust prohibition against price discrimination.

Or consider exclusive dealing. A manufacturer of some consumer product requires its distributors to agree not to carry any potential competitor’s products, thus increasing (he hopes) the cost of distribution to potential competitors. With an important exception noted below for the case where there are economies of scale in distribution, the manufacturer will have to compensate its distributors for agreeing to the restriction, and this will increase its costs, making entry more attractive. If a potential competitor has a promising product, other distributors will be delighted to carry it; if there are no other distributors, new ones will appear. Even if the potential competitor has to do its own distribution, and thus enter on two levels, manufacturing and distribution, the fact that the capital requirements for entry are now greater should not be an obstacle to entry, since there is no shortage of capital for promising new ventures. For these reasons, exclusive dealing is unlikely to be an effective means of forestalling entry. At the same time, it is easy to see how exclusive dealing might promote efficiency by increasing the likelihood that a distributor will use his best efforts to promote the manufacturer’s goods. Exclusive dealing may also help a seller of intellectual property to prevent piracy, a serious concern in intellectual-property markets.

But skepticism about unilateral monopolizing actions is not the same as denial. Antitrust analysts such as Robert Bork and myself pointed to the Standard Fashion case[4] as a case that involved a unilateral action that might well have increased the defendant’s monopoly power. The defendant manufactured a line of women’s clothing that was very popular. Retailers thought it essential to be able to sell the line. The defendant required retailers to agree not to carry competing lines. Competing manufacturers could in principle create their own retail outlets, but who would shop there if the most popular brand could not be found? Competing manufacturers would have to create a line as long and as popular as Standard Fashion’s line, and that would be difficult, maybe impossible, to do.

What distinguished Standard Fashion from a garden-variety exclusive-dealing case was the existence of economies of scale at the distribution level. Consumers didn’t want to traipse from store to store. They wanted a full line in each store, so anyone entering the women’s clothing business had to provide the full line if it was excluded from stores that carried the dominant firm’s line. Restricting its retailers no doubt cost Standard Fashion something. But it is plausible that the cost was less than the increase in its expected monopoly profits from forestalling new entry by compelling any prospective entrant to enter on a full-line basis.

The point is not that the new entrant would have to invest more capital, as in the previous example. The point is that it would have to embark on a much riskier undertaking, that of creating not a single successful product but a whole line of such products. It’s as if one couldn’t make commercial aircraft without making military aircraft as well. In such a case, as in Standard Fashion itself, exclusive dealing, while it would increase the defendant’s net costs of distribution (if it were indeed adopted for anticompetitive purposes and had no efficiency rationale), might increase the costs (not the capital requirements) of new entrants much more.

The analogy to a new-economy network externality should be plain. The network corresponds to the full-line retail store in Standard Fashion. A firm may wish to enter the market by producing one component of the network or one value-added service, but if a competitor by virtue of owning or having an exclusive-dealing contract with the network refuses to cooperate with the firm, the firm won’t be able to duplicate the network in order to get distribution of its component or service.

Notice that the monopolist would have no incentive to engage in exclusionary conduct unless his monopoly were fragile, that is, vulnerable to new entry. The more vulnerable it is, however, the less likely it is to endure even if the monopolist does resort to such conduct.

A complicating factor in both Standard Fashion and new-economy cases is the utility of exclusive dealing as a means of dealing with piracy. Standard Fashion may have been worrying about the piracy of its fashion designs by competing manufacturers, and attempting to prevent this by denying outlets to the pirates. To anticipate the second half of my discussion, whenever an antitrust court is called on to balance efficiency against monopoly, there is trouble; legal uncertainty, and the likelihood of error, soar.

Exclusive dealing, by the way, is analytically the same as tying. Exclusive dealing ties distribution to manufacture; equivalently, tying is exclusive dealing in the tied product. All you need is sensible law on exclusive dealing to deal sensibly with tying cases. If there are economies of scale in the tied product, a firm that to enter the market for the tying product is forced by the tying arrangement to produce the tied product as well will have higher costs than the monopolist and this will reduce the expected gain to him of entering the market for the tying product. Notice, finally, that exclusive dealing and tying can be accomplished in a variety of different ways, including vertical integration, contract, product design (in the case of tying), and bundling (charging a zero price for the tied product). And, speaking of bundling, while documented cases of predatory pricing are rare, I have long argued that such pricing is not always an irrational method of deterring entry. Especially when it takes the form of area price discrimination or “fighting brands,” so that the predator does not have to lose more money than the new entrant by lowering prices throughout his market, a monopolist may maximize his profits by “investing” in a reputation for predatory response to threatened entry.

Well, the plaintiff won in Standard Fashion, and the decision remains good law; likewise the illegality of predatory pricing is settled antitrust doctrine. I conclude that, as I said at the outset, existing antitrust doctrine is sufficiently supple, and sufficiently informed by economic theory, to cope effectively with the distinctive-seeming antitrust problems that the new economy presents.

What is troublesome is the institutional structure of antitrust enforcement. To begin with, cases in the new economy present unusually difficult questions of fact because of the technical complexity of the products and services produced by new-economy industries. Such questions can be central rather than peripheral to the antitrust issues, as where a plaintiff complains that the defendant has altered a technical protocol or interface to make it more difficult for the plaintiff’s product to work with the network, or a defendant contends that disclosure of a protocol would enable its competitors by reverse engineering to deprive it of a valuable trade secret that it cannot feasibly protect by means of copyright or patent law. The second example is an aspect of two broader questions, both also very technical and very difficult—the relative merits of monopoly provision and standardization as methods of optimizing a network, and the adequacy in fact of copyright or patent law relative to trade-secret law to protect investments in software and other new-economy products. Similar questions have arisen in other antitrust network settings, notably in the telephone industry, but they simply are much more difficult for a lay person to understand in the new-economy context.

A further complication is that it is difficult to find truly neutral competent experts to advise the lawyers, judges, and enforcement agencies on technical questions in the new economy. There aren’t that many competent experts, and almost all of them are employed by or have other financial ties to firms involved in or potentially affected by antitrust litigation in this sector. The Antitrust Division, I believe, does not employ any computer scientists or electrical engineers, but is wholly dependent on consultants, as are also, I believe, all the state antitrust offices; and, as I say, it is difficult to find a consultant in the new economy who is both competent and disinterested.

We deal with technical questions in the judiciary not by having judges or jurors who have the requisite technical knowledge or by giving them technical assistants but by having technical experts present evidence which the judge and jury (if it is a jury case) is expected somehow to assimilate. This system does not work as badly as its critics maintain; but the more technical the area of litigation and the fewer experts are reasonably disinterested, the worse it is apt to work. Computer science and communications technology are much more technical and difficult areas than the average body of scientific or engineering knowledge that lay judges and jurors are asked to absorb en route to rendering a decision.

The difficult factual questions presented by new-economy antitrust cases are not limited to technical areas, moreover. The combination of intellectual property, network externalities, and rapid growth in consumer demand creates difficult questions involving the ascertainment and measurement of monopoly. Suppose a firm has 100 percent of some new-economy product market and charges a price for it that is greatly in excess of marginal cost. Suppose also that the product is one that involves economies of scale of consumption (that is, it is a network market) and the demand for it is growing very rapidly. This is a common new-economy scenario. Does the firm have a monopoly in an economically relevant sense? The ratio of price to marginal cost is meaningless, since pricing intellectual property at marginal cost is nonremunerative and leads to bankruptcy. What about the market share? But one characteristic of intellectual property is its durability, and the more durable a product is, the fewer repeat sales the manufacturer will have. (It is different if the firm leases rather than sells its product.) New-economy firms that don’t lease get around this problem with frequent upgrades, but the installed base acts as a drag on the price that can be charged for an upgrade, since the prospective customer for the upgrade has the option of sticking with the existing product. If demand is growing rapidly, moreover, the firm in deciding on a price must trade off current against future profits, since a higher price today will slow the growth in demand. And by the assumption that it is a network market, the firm will have a strong incentive to charge a very low price in order to increase usage because that will enable it to charge a much higher price in the future—unless rapidity of innovation makes the future wholly uncertain, in which event it may abandon the hope of a network monopoly and charge a high price now. Still another factor is that a vertically integrated firm competes against itself: a firm that owns a network but also sells value-added services may want to keep the price of network access low to increase sales of those services. The result may be that a firm has a monopoly market share only because it is not charging a monopoly price.

And while it is possible to argue that monopoly can have other bad effects besides the limitation of output over the competitive level that monopoly pricing brings about, namely a reduction in the rate or distortion in the direction of innovation, economic theory and empirical evidence have yet to generate a consensus on whether monopoly is on balance good or bad for innovation. The possible effect of network externalities in discouraging subsequent innovation (the “path dependence” problem) not only is speculative, but is operative even if the monopolist is passive, in which event there would be no arguable antitrust violation.

The peculiarities of new-economy markets that I have been describing are apt to make the trial of a new-economy case a daunting challenge to the factfinding capacity of the judiciary.

The rapidity of innovation in the new economy has another very important institutional implication. Federal courts are highly efficient by the standards of the American legal system. The federal court queue is short, and strong district judges can move even complex cases along briskly. But this is speaking relatively. Antitrust litigation moves very slowly relative to the new economy. Law time is not real time. The law is committed to principles of due process that limit the scope for summary proceedings, and the fact that litigation is conducted by lawyers before tribunals that are not technically trained or experienced inevitably slows the process.

The mismatch between law time and new-economy real time is troubling in two respects. First, an antitrust case involving a new-economy firm may drag on for so long relative to the changing conditions of the industry as to become irrelevant, ineffectual. That was a problem even in the old economy. One recalls for example that by the time the monopolization case against Alcoa completed its journey through the courts, Alcoa had lost its monopoly for reasons unrelated to the litigation; as a result, the decree finally entered against Alcoa offered little more than nominal relief. This type of problem is likely to be more frequent in the new economy.

Second, even if the case is not obsoleted by passage of time, its pendency may cast a pall over parties to and affected by the litigation, making investment riskier and complicating business planning.

These problems are aggravated by the tendency of antitrust litigation to create multiple lawsuits out of a single dispute; call this the cluster-bomb effect. No sooner does the Antitrust Division bring a case, but the states and now the European Union are likely to join the fray, followed at a distance by the antitrust plaintiffs’ class-action bar. The effect is to lengthen out the original lawsuit, complicate settlement, magnify and protract the uncertainty engendered by the litigation, and increase litigation costs.

I wish I had a solution to these problems, or at least some of them, but I don’t. As minor palliatives, I would like to see, first, the Antitrust Division and the Federal Trade Commission given the necessary appropriations to enable each of these agencies to hire a competent technical staff—which won’t be easy, given the salaries that competent new-economy scientists and engineers command in the private market. And I would like to see the states forbidden to bring antitrust suits except under circumstances in which a private firm would be able to sue, as where the state is suing firms that are fixing the prices of goods or services that they sell to the state. States do not have the resources to do more than free ride on federal antitrust litigation, complicating its resolution; in addition they are too subject to influence by interest groups that may represent a potential antitrust defendant’s competitors.

But these measures, even in the unlikely event they are adopted, would not do much to correct what seems to me a very serious mismatch between the conditions of the new economy and the institutional structure of antitrust enforcement.

Footnotes

[1] Judge, U.S. Court of Appeals for the Seventh Circuit; Senior Lecturer, University of Chicago Law School. This is the text of a paper to be delivered in New York on September 14, 2000, at a conference on antitrust sponsored by the American Law Institute-American Bar Association Committee on Continuing Professional Education. I thank Dennis Carlton and William Landes for their very helpful comments on a previous draft.

[2] See Richard A. Posner, Antitrust Law in an Economic Perspective, ch. 8 (1976).

[3] Perfect price discrimination would bring about the same output as under competition, because, as noted earlier, no customer willing to pay the seller’s marginal cost would be turned away. But perfect price discrimination is infeasible, and imperfect price discrimination can result in a lower output than under competition. See Richard A. Posner, Economic Analysis of Law 306 (5th ed. 1998).

[4] Standard Fashion Co. v. Magrane-Houston Co., 258 U.S. 346 (1922).