07/07/2001

The Economist

Finance and Economics

A bundle of trouble: The rights and wrongs of bundling: Behind antitrust actions against Microsoft and General Electric lie concerns about "bundling" different products together. Is bundling really so bad?

IN ENGLISH schools, "bundling" is what happens when boys knock a fellow pupil to the ground and then pile on top of him. Economists use the term to describe something less rough-and-tumble, yet, some fear, equally suffocating to the victim: bundling two or more products together as a package, and selling it for a single price.

The world's two biggest companies, by market capitalisation, are under fire from antitrust authorities, in part because of concerns about bundling. On June 28th, an American appeals court upheld parts of an earlier judgment that, by bundling its Internet Explorer browser with its Windows operating system, Microsoft was seeking to preserve, or even extend, a monopoly position in operating systems. On July 3rd, the European Commission blocked General Electric (GE) from acquiring Honeywell, because it feared that the combined firm, by bundling together engines, electrical components and trade financing, would increasingly dominate the aircraft-manufacturing business.

In the 1940s, America's Supreme Court concluded that "tying agreements serve hardly any purpose beyond the suppression of competition." Economists have long challenged that absolutism. Many now think that, though there are exceptions, selling bundles of goods together as a package can be a source of economic efficiency.

Bundling covers many things. Two or more of the same product might be sold as a package - a "buy one, get the second at half-price" deal, say, or a railway season ticket. A camera might be sold in a box with a free film; a hotel room might come with accompanying breakfast.

A product can also be bundled together with a loan. Financial bundling has become so widespread that three economists at Morgan Stanley - Steven Galbraith, Mary Viviano and Elmer Huh - suggest in a recent report that, as manufacturers such as GE, General Motors and Lucent grow ever more involved in providing finance, so "manufacturing is becoming the loss-leader of the profit chain for many companies." In other words, give away the product; make money on the lending that is bundled with it.

Bundling can be good for consumers. It can reduce "search costs" (the bundled goods are in the same place), as well as the producer's distribution costs. There are lower "transaction costs" (because a single purchase is cheaper to carry out than multiple ones). And the producer may be a more efficient bundler than the customer: few of us choose, after all, to buy the individual parts of a car to assemble them ourselves.

In perfectly competitive markets, bundling should happen only if it is more efficient than selling the products separately. Where there is less than perfect competition - that is, most markets - economic models suggest that bundling sometimes benefits consumers and sometimes producers. Nicholas Economides, of New York University, says that when firms have a measure of market power, they can engage in price discrimination, charging different prices to different customers. Bundling can play a part in price discrimination, as different bundles of goods and prices may appeal to different customers.

Price discrimination in general, and bundling in particular, is usually a profit-maximising strategy for a producer that enjoys substantial market power, says Mr Economides. But if there is a deal of competition, two rivals selling bundled goods may see their margins fall even more readily than if they were not bundling.

Wherever there is market power, the antitrust authorities have reason to be watchful. Bundling a monopoly product with one that is competitively provided may result in the competitive market being distorted. But it would be wrong to assume that bundling is inevitably bad in those circumstances. If there is some loss of consumer choice, the cost may be outweighed by efficiencies from bundling. Each case must be judged on its merits.

The merits are harder to assess where there is "pure bundling", involving products available only as a package, than where there is "mixed bundling", with products available both as a package and individually.

Microsoft's package of Explorer and Windows is a pure bundle. On the face of it, requiring Microsoft to sell Windows and Explorer as separate products as well as bundled ones would end the arguments - except that Microsoft claims they are not a bundle at all, rather a single product incapable of being broken into parts. In the past, such claims have been resolved by assessing whether a viable market exists for each product on its own: if it does, the product is a bundle.

Yet, as the appeals court noted, this does not work well in Microsoft's case, because any innovation that improves an integrated product might be stifled. Again, the costs and benefits of any such innovation, and whether it constitutes a genuinely new, integrated product, not a bundle, should be judged case by case.

Product definition was also an issue in the proposed merger of GE and Honeywell. The European Commission concluded that GE's aircraft-leasing arm would favour GE/Honeywell's competitively priced bundles of engines and components for aircraft it financed, over (perhaps unbundled) alternatives sold by competitors. Maybe. But aircraft financed by GE could equally be regarded as a bundled product in competition with other, bundled aircraft. As long as a merger of GE and Honeywell did not reduce competition in this end market, and it appears it would not have done, should the antitrust authorities have a say in who makes the components that GE chooses to include in its bundle?