TCS Daily

Everything Old Is New Again (So Let's Get It Right This Time)

By James V. DeLong - February 4, 2008 12:00 AM

Arguments over net neutrality and other issues of access to telecom facilities roil the FCC and the press; the European Union is after Microsoft (again) and other U.S. companies; railroads and shippers are at their usual loggerheads; several firms are playing hardball with Qualcomm over use of its patented cell phone technologies; and dozens of companies have a fearful eye on Google.

A common thread unites these varied disputes. They involve situations in which one company controls a resource upon which other firms must rely to pursue their own businesses. The resource may be a transportation network, such as a railroad, or a telecommunications network, such as access to the Internet, upon which the firm must depend for distribution. Or it may be a platform, such as a computer operating system (Windows or Linux) for which an application writer provides a complement, or a distribution network, such as eBay, on which numerous small businesses have staked their futures, which has the characteristics of both.

Commonly, the network/platform is characterized by high fixed costs and low marginal costs, a structure that makes entry difficult, and/or by a need for interoperability, which creates strong pressures for the industry to coalesce around a single standard. Thus, the proprietor of the resource has considerable market power, in that a dependent company cannot easily shift its reliance because of the inherent economic or technological characteristics of the situation or because of intellectual property.

Each dependent fears that the network/platform, given its druthers, will raise the price of access to the network/platform to a level that extracts all the value from the total enterprise and leaves the dependent with only crumbs. They also know that the network/platform has a strong interest in turning the complements into cheap commodities, so as to increase overall demand for its product and improve its leverage over the dependents. In the worst case, the network/platform will wait until others commit resources, then pounce.

Each network/platform company also has a problem. It must spend substantial resources to attain its position, but is wary of investing to the optimum because it fears that a political response organized by the dependent companies will result in confiscation once its capital is sunk.

The fears of both sides are well founded, and, for reasons explored below, neither a "let the free market do it" nor a "regulate" solution provides a totally satisfactory resolution. Nor do current legal doctrines, such as antitrust or common carrier regulation, provide an answer, since both are barnacled by the mistakes of past decades and by the rent-seeking propensities of the legal profession.

The result is a shortage of institutional mechanisms and legal rules that can accommodate smoothly the needs of both platform/network companies and their dependents. Because the issues are so important to so many, government policy becomes crucial, but that policy tends to lurch back and forth, favoring now one group and then the other, to the long-term detriment of both, and of the public.

To explore the problem in more depth, and to think about solutions, start with a simple but historically realistic model: a 19th century railroad running past a factory or a farm. The railroad needs factories and farms en masse to provide revenue, but it does not need any single one of them. So, in the pre-truck days, the railroad could set a rate that captured most of the value of the combination of factory+railroad or farm+railroad, allowing the producer a return that paid only marginal costs, forcing it to forgo any return on capital.

Furthermore, the railroad could set different rates for every factory along its lines, extracting maximum value from each. A plant making a commodity product with low margins would pay a low rate; one making specialized goods with high demand would pay a high rate. But each could be charged so much as to be left with only the pittance needed to keep it in business

Shift the hypothetical and add a second railroad. Now the factory's situation improves greatly, because it can force the railroads to bid against each other. Indeed, if the factory is big enough, it can reverse the bargaining power, and use the railroad's high-fixed, low-marginal-cost structure against it, driving rail rates down to the railroad's marginal cost, and forcing it to eat capital costs and retain only a pittance. If the shipper is big enough, such as John D. Rockefeller's Standard Oil, it can force the railroad to charge Standard's competitors a high price and then rebate part of the revenue to Standard, a two-fer of disrupting the competitors and reducing the railroad's returns.

Obvious ways exist for the participants to try to break out of these dilemmas, and they often work pretty well. Build the factory where roads, rails, and water all coincide, and take your pick of transport modes, for example. Or don't build a factory unless the railroad guarantees reasonable rates.

But these work only approximately. Sometimes the location of production facilities is constrained by geography, and in many cases we are talking about serious long-term capital investments, and it is hard to see how either side can commit to anything sufficiently specific. On the other hand, if the factory is located first, how does it persuade the railroad to build? Does it promise never to use a different railroad? Not to use water transportation? Again, we are talking long term investments under conditions of great uncertainty, and it is difficult to write the contracts that would be required. A chronic problem with the libertarian mantra of "do it by contract" is that it requires contract writers with an unlimited legal budget and a level of foresight that would be the envy of a psychic.

Another possible solution to the problem is integration. If the railroad owns the factory, then allocating the total revenues it receives from the factory+railroad becomes a question of internal accounting. This approach, too, has precedents; it is a variation of the land grants given to railroads in the 19th century, which enabled them to finance construction by cashing in on the increase in the value of the land triggered by the building of the railroad, and by the ability to sell land to raise crops which the railroad could then haul.

But there are also some problems with integration. If the railroad owns one factory, then what about the competitors of that factory? The railroad has an incentive to charge them more or provide them with inferior service. A modern example: The wars over Microsoft's relations with the providers of applications that compete with Microsoft's own offerings and run on Windows.

One could take it a step further, and conclude that the railroad should own everything that relies on shipping. This is getting into reductio ad absurdum territory, but even as a technical argument, it fails. To allocate its capital among factories and transportation, the railroad would have to solve the same set of intractable problems that led to the integration in the first place. Putting the issues into a big black box called a corporation does not make them easier. And think of the inefficiency of having every decision flow through a central hierarchy.

Besides, if society values innovation, a world of large integrated firms is a poor way to promote it. For many reasons, such entities tend to be uncreative. The railroads enabled the great catalogue mail order businesses that transformed rural America, but they did not invent them. A great concern of the contemporary tech world is that innovators need access to platforms on reasonable and predictable terms if the full creativity of the society is to be tapped.

What about the opposite approach - forbid integration, and impose a legal rule against discrimination on price or other terms? This breaks down as a matter of elementary economics, which says that the capacity of a network/platform is not infinite, that it must be rationed and traffic prioritized, and that the price system is the best way to accomplish these ends.

Besides, the dependent companies themselves need price discrimination. If I own a factory, and another shipper can pay only a widow's mite for transportation, then I want the railroad to give that shipper a deal so that it contributes to the railroad's capital costs, even if it pays less than I do. If the railroad does not cut price far enough to capture this firm's business, then all the costs will fall on me. Of course, if the other factory is a competitor of mine, then my calculus changes; I want to share the costs, but not at the price of seeing a competitor get a cost break. (This is to some degree the oil rebate case, in which Standard's view was that it was simply recapturing some of the economic benefits that its volume conferred on a railroad, and that the other oil shippers should be grateful to Standard for financing so much of the capital costs of the railroad.)

What about the competitive solution of ensuring that there are many railroads? This runs into the marginal cost problem. Excess capacity develops, at which time competition drives the price down to the marginal cost of the most efficient, which puts the others out of business, and restores a monopoly situation, at which point prices are raised and the dependents are once again in trouble, until a new technology or new entry saves them. Economists look at this great wheel of life with equanimity; actual businesses and governments are less sanguine.

Besides, monopoly, rather than revived competition, may be the logical outcome. In some cases the existence and persistence of multiple entities is not a realistic option. Historically, regulation was based on the concept of "natural monopoly," which could be natural as a matter of economics as well as technology.

In a more recent twist, economic literature is now full of discussions of winner-take-all markets, which coalesce around a single technology, such as Windows on the desktop computer, and, most recently it would appear, Blu-Ray as the hi-def DVD disk. The assumption is that new technologies will ultimately disrupt such empires in the long run, but dependent companies are not interested in waiting for the long run, so there will be immediate political pressure for countervailing action right now.

The same basic scenarios between networks/platforms and dependents play out repeatedly, not just with shippers of physical goods but with developers of software applications for Windows, with internet companies dependent on telecoms, with cell phone manufacturers using patent-protected chips. Indeed, in the tech area, the whole "openness" philosophical movement, whether in the context of software or net neutrality or anti-copyright, traces at least part of its lineage to this 19th century problem of shippers and railroads, and to the desire to avoid dependency. Old railroad battles over long-haul vs short-haul rates, oil tariff rebates, price discrimination, value-of-service vs cost-of-service rate-making, lines built solely for blackmail, and other issues of those thrilling days of yesteryear are re-enacted in the guise of Google battling Verizon or Comcast over net neutrality, or the EU forcing Microsoft to open up server technology.

The possible variations on the theme of network/platform companies and dependents are many. But one rule seems constant: whichever first commits its resources, loses. The counterparty can then threaten to withhold its investment unless it gets the major share of the gains. The situation becomes like the game in which one person decides how to share out a payment with a fellow player, with the latter limited to the binary choice of accepting the proposal, in which case the money is so divided, or rejecting it, in which case neither party gets anything. If the decider takes a large chunk for himself, then the other party is left with the choice of accepting and getting something or rejecting and getting nothing except the satisfaction of knowing that the decider also gets nothing. Experiments indicate that excessive greed by the decider will produce a the-hell-with-it anger and rejection of the offer, but that point is well short of an even split.

As a result of the first-mover disadvantages inherent in committing resources, there is, inevitably, pressure for the political system to address the problems. Unfortunately, our earlier efforts at achieving regulatory solutions did not end happily. Almost every railroad went bankrupt in the late 19th Century, after a descent into marginal cost hell, and Grangers and investment bankers eventually co-operated to produce the Interstate Commerce Commission. Thereafter, policy oscillated between encouraging monopoly and exploiting the railroads for political gain, and the result was a long, slow decline.

Telecommunications were also regulated, as a result of similar Baptists and Bootleggers cooperation, and the history of the Federal Communications Commission became a sad saga of suppression of technology and investment.

Nor has the world learned. If the EU pursues its present bent, Microsoft and its fellow techies are not likely to do very well as regulated industries, and some pending proposals for patent reform incorporate concepts of compulsory licensing at bargain rates.

So how do we avoid repeating these experiences? The usual complaint is that regulatory regimes allow cartelization and monopoly. This may be, but, as the history of the railroads and telecommunications shows, the pressure to protect complementary dependent industries also leads to suppression of investment in the network/platform companies, to the detriment of all.

There is no rule of the world that says that all problems have neat solutions, and this is set for which no theoretical answers emerge. But a number of practical approaches look useful.

First, it is important to understand what is not relevant. Antitrust law is fairly useless. It focuses on competition among the network/platform companies themselves, and deals with their relations with dependent companies only in the limited context of "tying." This is an intellectual dead-end. The antitrust mavens also think that a "competitive price" means marginal cost pricing, which is total gibberish when applied to high-fixed-cost industries.

Antitrust doctrine is also focused on harm to consumers, not to intermediate dependent producers. If New York diners have access to meat from Argentina and Poland as well Nebraska, antitrust should not, in theory, care if a railroad is exploiting a dominant position to extract most of the value of the meat shipped from Nebraska because there is no consumer harm. Competition from overseas will keep prices down. The Nebraska farmers, and the U.S. political system, care a great deal.

The EU doctrine of "abuse of a dominant position," has potentially more relevance, but it has not been developed in useful directions. The only part of antitrust doctrine that seems relevant is the maligned Robinson-Patman Act, which is directed at the fundamental question of the relations between networks/platforms and their dependents.

Common carrier law as spelled out in generations of regulatory pronouncements should also be avoided. But the more fundamental, and limited, common carrier principles as they developed in the pre-regulatory era are quite interesting. These consisted of openness, availability, advance notice of terms, and inhibitions (but not prohibitions) on price discrimination. Reading 19th Century cases is good exercise, as one thinks through the issues with the Supreme Court in ICC v. B&O RR (1892) or in the old grain elevator classic, Munn v. Illinois (1877).

There is some good wisdom in those rules. Three of the dependent companies' biggest problems are ambush, in which its resources are committed and then the network/platform company raises the rates; uncertainty, as when an innovator is in the impossible position of asking a large carrier, "if we invented an X, what would you charge to attach it?"; and negotiation costs, of trying to get an answer from a large bureaucracy. Advance notice and transparency are powerful tools.

But it is equally important to recognize the limitations of any common carrier-like approach. "No integration" may seem attractive, but it deprives consumers of the advantages of efficient packages. In general, Integration often has some powerful benefits, especially in allowing consumers to buy packages of integrated products and services. No one wants to buy an automobile one part at time, and the same is true increasingly of computer programs. We want plug and play. Should Apple be required to unbundle its hardware/operating system/applications package? It is precisely the smooth integration that makes the Mac appealing to so many.

Also, over time, a no-integration rule that a network or platform must be nothing but stand-alone dumb pipe triggers a cascade of problems that must lead to regulation, price control, and technological stasis. Imagine a telecom future in which every household has ready access to cable, fiberoptic lines, broadband over power lines, and satellite. All the basic infrastructure has been put in place and the only gap is connection to the premises. Now, suppose that each of these is a dumb pipe, forced to carry all content. The obvious result would be a war which drove prices down to marginal cost, which would starve the carriers of the ability to service their capital or maintain their plant, and send them into bankruptcy. We know this, because this is the 19th century railroad case. Perhaps the only way to maintain competition among carriers in the long term, and thus avoid the evils of regulatory structures, is to allow integration.

It is not easy to find an optimum path between the limited doctrines of historic common law common carrier doctrines and full blown regulatory regimes. However, network/platform companies can influence their own fate by contractual obligations or public announcements, especially if these constitute binding obligations, either practically or legally.

In 2005, the Federal Communications Commission did the telecom companies a favor by issuing four principles supporting openness on the Internet, designed to assure the community generally that Internet service providers would not be allowed to exploit their position to exclude others' content or devices. Consumers are entitled to access the lawful Internet content of their choice, to run applications and use services of their choice, to connect their choice of legal devices that do not harm the network, and to competition among network providers, application and service providers, and content providers.

The FCC had no particular authority to adopt these principles, and their legal status remains uncertain. They are important, thought, because the fear of dominance by ISPs is great, and telecom companies might well forestall unpleasant regulation by promising not to assert their maximum position. In fact, no telecom company dares contradict the four principles, and a large part of their counter to demands for net neutrality rests on the proposition that the four principles make any such regulations unnecessary.

On this issue, the telecoms strategic withdrawal is dead right. A third way between old regulatory models and total telecom freedom is needed, and there is no reason for the telecoms to wait for it to be imposed legislatively. They have little to lose, because any effort to claim a right of total control over Internet traffic carries heavy risk of a reaction that deprives the carriers of all discretion.

In 2006, Microsoft developed its Windows Principles, which are designed to assure applications developers of their ability to capture the gains of their enterprise. It acted under the lash of the Justice Department, admittedly, but the result is important. For one thing, it makes clear that the real axis of the antitrust case against Microsoft revolved around its relations with complements, not with the bogus issue of creating competitors to the Windows operating system.

The development of the open source Linux operating system serves the same function as Microsoft's Principles - that of assuring developers of complements that they cannot be exploited after they have committed their resources. Tech expert George Gilder once attributed the rise of the open source software movement to the fact that in the late 1990s Microsoft's share of the total value of Windows+application systems sharply increased, thus irking the developers, who became receptive to the idea of a new platform.

The most important thing in approaching these issues is a sense of history. Philosopher George Santayana said: "Those who cannot learn from history are doomed to repeat it." Historian Crane Brinton added that, unfortunately, those who can learn from history are doomed to repeat it with them. The great regulatory regimes of the 19th and 20th centuries were established because of real and difficult problems centered around the nature of high-fixed, low-marginal cost network/platform providers and their relationships with other producers that are dependent on them.

These regulatory regimes failed in many ways, and the trend toward deregulation that took place from the 1970s onward unchained the economy. But the old problems are now appearing in new areas, and sometimes in the same old areas, and it is time to think in terms of new hybrid solutions based on contract and public commitment rather than repeat the old debate of total laissez faire markets vs total regulation.


TCS Daily Archives