How to disentangle competition policy in the Net's domains

THE INTERNET was essentially invented in 1994. It is expected to have around 200 million users worldwide by 2004. No wonder there is more and more analysis of the "new economy", and no wonder governments are starting to contemplate what, if anything, they should be doing to earn the "e-" prefix that has turned into the kitemark of modernity.

THE INTERNET was essentially invented in 1994. It is expected to have around 200 million users worldwide by 2004. No wonder there is more and more analysis of the "new economy", and no wonder governments are starting to contemplate what, if anything, they should be doing to earn the "e-" prefix that has turned into the kitemark of modernity.

There are many aspects of policy that need updating for the e-economy - for example, getting government online so companies and individuals can file their tax or claim their benefits electronically. But some of the knottiest issues concern competition policy.

Joel Klein, the US Justice Department's combative assistant attorney-general, was right when he said in a speech last year: "It is no mere coincidence, I would suggest, that at a time when our economy is the most competitive in the world, it is also the strongest."

Yet there are several reasons why competition policy is getting more difficult. The reasons arise from typical characteristics of much high-technology business. These include the presence of network effects and more broadly of increasing returns to scale - both set out in a handy new survey of the e-economy from Credit Suisse First Boston.

The classic example of a network effect is the fax machine. These are not much use if few people have them. The more users, the more useful they are, and once there is a critical mass of users the machines can become cheaper and bought by even more users.

Many new products and new technologies have this characteristic, and it is certainly true of software. One of the reasons Microsoft's Windows operating system has become so dominant is, that with so many people using it, it is easier for everyone else to do so. Network effects make it easy to create a technical "lock-in" whereby one product sets the standard everybody else has to follow.

More broadly, increasing returns to scale are becoming prevalent. Network effects certainly contribute to economies of scale. But these can also arise from high start-up costs, such as the investment in R&D needed for a new drug or biotech product, or from near-zero marginal costs - say, any piece of software that can be copied for nothing and so costs the same for a million customers as for 10,000.

These features need not change the basics of competition policy. As Hal Varian and Carl Shapiro point out in Information Rules , their instant classic on doing business in the new economy: "Competition policy is intended to ensure a fair fight, not to punish winners or protect losers."

The catch is that it is getting harder to tell whether the fight is fair. For given the nature of hi-tech markets, it will often be the case that there is a dominant company, the one which got there first and garnered the "winner-take-all" network benefits.

What's more, as Varian and Shapiro point out, an attempt to boost competition by setting industry-wide technical standards could easily be interpreted as collusive behaviour designed to keep out new entrants, the familiar and understood obstructing potential technical and competitive improvements.

Mr Klein argued in his speech that there is nothing analytically new and challenging in all of this. For example, the Justice Department's case against Microsoft hinges on the allegation that the company abused its monopoly power to force manufacturers of PCs to bundle its Explorer Internet browser with Windows when they installed the basic operating system, thereby squeezing Netscape's competing browser out.

Although the facts might be harder to resolve, there is little difference in principle from a case the department brought against IBM in 1936 for bundling tabulating machines and punch cards. The same anti-competition laws apply, and their application depends on the facts of the case, he concluded.

But Mr Klein's confidence in the robustness of existing competition law is perhaps overdone. The economic benefits of competition arise from the fact that it is the one sure way of ensuring consumers get better quality products and wider choice at better prices. It is hard to argue with the proposition that Microsoft has delivered continual product improvements and lower prices and hard to be sure that it has done so at the expense of wider choice. What would the technological and business landscape have looked like without Microsoft?

To attack it on competition grounds - as I believe you can - requires a new focus on the process of competition over time. In an era of rapid technological change, competition is serial rather than parallel. A dominant company will have to be overthrown by a technologically-superior new dominant company.

After all, that is precisely what Bill Gates says he is frightened of, so why not take him seriously? Linux, the free Internet-based, open access operating system could yet do him more damage than the Justice Department.

In that case, regulators need to focus specifically on barriers to entry and barriers to innovation. Classic monopoly indicators such as current market share or ratio of price to marginal cost are next to useless in the e-economy. Instead, competition authorities could start to focus on practices such as the acquisition of competitors.

As Professor Michael Porter, the Harvard Business School guru has argued, it is hard to think of any convincing reason why it should be a good thing for the economy if such takeovers go ahead.

Standard-setting could also become a key competition tool. Although industry practitioners obviously need to be involved in setting technical standards for practical reasons, there is a new public good aspect in the creation of standards that keep the industry open to newcomers rather than cementing the advantages of incumbents.

Varian and Shapiro suggest there is also a good case for government support for new basic technology projects until they can reach critical mass, as long as this does not degenerate into the old bad habit of picking winners. The Internet itself would never have reached the stage of commercial exploitation without extensive government-funded research.

Such approaches will be more productive than a competition policy that runs headlong into the basic economics of ever more widespread increasing returns and network effects.

The purpose of competition policy is the same as ever, but it still needs to have its "e-" attached.