Conference Summary

A conference on the economics of broadband communications was held by the AEI-Brookings Joint Center for Regulatory Studies at Brookings on October 4-5. The conference dealt with public policy issues involved in deploying the requisite facilities to deliver this service to large numbers of households and small business.

The focus of the conference was the evolution of the Internet from a service that large numbers of subscribers access over ordinary telephone lines through low-speed modems to a service that is accessed at high speeds (200 kb/s or more) through cable modems, digital subscriber lines (DSL), or various wireless and satellite services. These new high-speed services allow users to navigate Internet web sites more rapidly and to use the Internet for services, such as video streaming, that are not practicable at low speeds.

The opening session focused on general approaches to regulating a new service such as broadband communications. The first paper by Bruce Owen, president of Economists, Inc., acknowledged the difference in regulatory approaches to cable modems and DSL, but noted that the difference largely derived from the historical concern over “bottleneck” facilities of telephone companies and the political economy of regulation. Owen considered the discussion of barriers to the deployment of broadband as one of several “mysteries” in the current discussion. He concluded that there is no evidence that broadband facilities are not being deployed rapidly and that there is similarly little evidence that there are regulatory barriers to such deployment. He admits, however, that these conclusions may be premature, given the early stages of broadband development.

Howard Shelanski, Professor of Law at the University of California (Berkeley) presented his paper, detailing the various characteristics of network industries and the propensity of these industries to evolve into a monopoly structure. He then detailed the potential policy responses to a non-competitive market structure in such industries, including common carrier obligations and access to “essential facilities” under the antitrust laws. He pointed out that the regulatory environment finds cable systems under pressure only to open their cable-modems to competitive Internet Service Providers (ISPs) while telephone companies are required to lease their network facilities to rivals. He raises the “threshold” question of why competition from DSL and other services is sufficient to make it unnecessary to regulate cable modem services when similar competition from cable modems is not sufficient for regulatory authorities to decline to regulate telephone company DSL services. This threshold question would emerge as the dominant, but ultimately unanswered question of the conference.

In a lively discussion of these papers, Harold Furchtgott-Roth of AEI offered the view that the uncertainty of the current regulatory climate for broadband is a critical problem, one that economists cannot influence because they are so badly used in Washington. Michael Kellogg, of the Kellogg, Huber law firm, argued that the current asymmetrical regulatory structure penalizes the incumbent telephone companies because their profits are “socialized” while the risks remain with the telephone companies.

Professor Glen Robinson of the University of Virginia Law School offered the observation that the “killer applications” for broadband do not yet exist and are being delayed by the failure to solve copyright problems. Robinson also observed that the “essential facilities” argument assumes a natural monopoly that clearly does not exist in broadband. Why, then, does the current regulatory regime require sharing or “unbundling” of incumbent facilities? Robinson thinks that such unbundling is both unpromising and likely to reduce incentives for investment.

On the other hand, Robert Hall of Stanford University opined that such unbundling reflects a “delicate” intervention that is required because of the incumbents’ “last-mile monopoly” of the local copper-wire subscriber lines. Hall also offered the observation that investment incentives in broadband might be maximized under conditions of monopoly given the large potential network externalities generated by broadband deployment.

Session 2:  Empirical Evidence on the Demand for Broadband Services

The next session considered the demand for broadband services. Paul Rappoport of Temple University presented a paper he co-authored with Lester Taylor and Donald Kridel on the demand for and use of the Internet. Using a new source of “click-stream” data, he showed that broadband subscribers use the Internet more intensively than narrowband subscribers, but nevertheless the usage patterns of these two groups are remarkably similar. As with other telecommunications services, household use of the Internet is heavily skewed--a significant number of extremely heavy users are on line, even in the early hours of the morning. A preliminary discrete-continuous choice model provided confirmation of the importance of age, income, and education in the choice to subscribe to broadband services.

Hal Varian of the University of California, Berkeley, presented a paper summarizing the results of an experimental study on the demand for high-speed Internet access, conducted at Berkeley in 1998-99. Subjects were provided with access at various speeds and prices per minute  that varied with these speeds. Alternatively they were offered fixed prices for a given quantity of weekly use. Econometric analysis of their purchase decisions revealed that the various speeds offered--8 kb/s to 128 kb/s--were substitutes for each other and that the own-price elasticity of demand was quite high, particularly for higher speeds. The implicit value of time saved by higher speed access was astonishingly low, perhaps because respondents could use the time required for downloading email, for example, in performing other tasks.

Timothy Tardiff of NERA suggested that these results are reflective of an earlier time when the uses of the Internet did not require high-speed access. If the value of broadband is so low, he asked, why has broadband penetration grown from virtually zero in 1998 to about 7 percent today?

Clifford Winston of Brookings commented on the Rappaport paper, suggesting improvements in a future revision of the paper. He observed that the low current penetration suggests that the marginal value of the service is less than its cost for most consumers. He would like to see a demand model that accounts for non-price attributes of the services as they change over time or as they vary over carriers. One could then obtain estimates of the consumer’s willingness to pay for the service.

Session 3:  Broadband Technology and Competition

The next session dealt with intramodal competition in the delivery of broadband services--wireless versus wires and DSL versus cable modems. Jerry Hausman’s paper offered a number of perspectives. First, he provided evidence that wireless services have already begun to compete aggressively with traditional wireline voice services. Second, he demonstrated that broadband competition is occurring in Korea without wholesale regulation of the incumbent telephone company. Third, he provided a five-part test for the need to regulate incumbent services that leads to the conclusion that regulation of incumbent telephone companies’ provision of DSL is not necessary.  Fourth, he stated that the unbundling regime in the United States that has been established to facilitate entry is inefficient because rates are set at TELRIC. He suggested that this regime might even be responsible for the wave of CLEC failures that began in early 2000. Finally, he concluded that competition in broadband services from commercial wireless providers through the new “3G” technology will be slow to develop and particularly slow in the United States because regulators have moved too slowly in freeing spectrum for it.

In the second paper of the session, Charles Jackson provided a technical analysis of the various broadband delivery systems. He considered fiber to the home, DSL from a central office or a set of remote terminals, and cable modem service (fiber-coaxial cable). He spelled out the disadvantages and advantages of each. For instance, the speed of DSL service attenuates sharply with increasing loop length while cable modem service suffers congestion costs due to sharing a common path to the fiber node. He pointed out that discussions of deliberate carrier delays in deploying these technologies are misguided because the available technology was not even developed until the mid 1990’s, and the prices of modems were in excess of $1,000 in the early 1990’s. It was only when modem prices fell to the $100-$250 range in the latter part of the decade that widespread broadband deployment became possible.

The first discussant, Bob Hall, pointed out that Korea’s broadband market is dominated by the incumbent telephone company, but U.S. telephone carriers have not been as successful because they fear cannibalizing, i.e., reducing the demand for, other high speed services. The U.S. therefore needs competitive DSL providers (DLECs) and an unbundling policy for telephone companies, but he could not offer a reason why a similar policy is not necessary for the “last mile” of cable companies.

A second discussant, Keith Bernard of Hughes Network Services offered the view that the DLECs and fixed wireless operators are failing or have failed in no small part because they tied their fortunes to DSL. The remaining carriers with a fixed wireless strategy, including WorldCom and AT&T, are using it only as a complementary “fill-in” strategy where other systems are not available. He offered the view that his company, the owner of the DirecTV satellite service, will soon have new Ka-band satellites with hundreds of spot beams that can offer millions of households and businesses two-way broadband Internet connections.

In further discussion, Hausman offered the view that further vertical separation of telephone company operations would be disastrous. Hausman offered the view that open access for ISPs on both the telephone-company and cable networks is a good idea, a view supported by Hall. Gerald Faulhaber offered the view that there is no reason for regulation of any new facilities developed for broadband.

Session 4:  Regulatory Policy under the 1996 Telecommunications Act

The next session dealt with the current regulatory regimes for broadband services with particular emphasis on vertical integration. Gerald Faulhaber began with a paper that asked: Why has broadband roll-out been so slow? In fact, he concluded that broadband has grown more rapidly than wireless telephone service had two decades ago. If there are problems, he surmised, they are in the supply side rather than inadequate demand. But he concluded that the supply problems are probably not due to regulation, but to difficulties that the telephone and cable companies have had in extending broadband service to many neighborhoods and installing it once it is ordered by a customer. These problems occurred because the Internet developed in an unpredictable manner and therefore required the cable and telephone carriers to adapt their networks rapidly to the delivery of a new broadband service. He suggested that new technologies are likely to render DSL services obsolete rather soon. His view is that regulatory costs for the incumbent local telephone companies are not likely to be very great and that the companies are rolling out DSL as fast as they can to compete with cable. Nevertheless, he reiterated his view that unbundling and line sharing should be limited to legacy facilities and not imposed on new facilities.

Thomas Hazlett’s paper observed that requiring the telephone companies to allow unbundled access to their networks for broadband competitors (DLECs) destroys the benefits from vertical integration and thereby reduces investment incentives. Pressing for even more vertical dis-integration through structural separation--as AT&T has in various state proceedings--only exacerbates these problems. Given the fact that there are competitive suppliers of broadband in most geographic markets, there is no “bottleneck” problem to regulate. Indeed, he showed that the “q ratio”, the ratio of market value to reproduction cost of assets, is much greater for cable companies than for telephone companies. Who, he asked, has the monopoly power? His paper observed that the demand for regulation of ILEC facilities and open-access requirements for cable television systems is simply a reflection of rivals’ desire to impose costs on the competitors’ facilities.