Essential Facility and One-Way Access: Policy

General Issues for the Design of Access Policies

Access pricing rules are necessarily imperfect for several reasons, which we have regrouped into three categories. Any access pricing rule is an instrument of regulation of the bottleneck owner's rate of return on the bottleneck investment or for that matter of the owner's overall rate of return. As such, it governs the owner's incentives to build and maintain the bottleneck. There is then a trade-off between two considerations: "ex post efficiency," which goes in the direction of fostering competition through access beyond the level that would be spontaneously permitted by the bottleneck owner, and "ex ante efficiency," which suggests giving the bottleneck owner flexibility in exploiting the bottleneck.

The second snag in designing a good access policy comes from the difficulty in defining the notion of "service," and the third relates to the problems associated with the monitoring of compliance of the access policy once services have been defined. The last two difficulties echo those faced in the regulation of retail prices, and so we will content ourselves with illustrating them with wholesale pricing examples.

1 Liberalization and Deregulatory Takings

A central issue in the creation of competition is whether it breaches the "regulatory contract" between the incumbent operator and the regulators and thus constitutes a taking of the incumbent's property. This issue has already taken center stage in the debates on the deregulation of the power and gas industries in the United States. For example, in the wake of the deregulation initiated by the Public Utility Regulatory Policies Act (PURPA) of 1978 and accelerated by the Energy Policy Act of 1992, which mandates wholesale wheeling, that is, access to a power utility's transmission grid, and eventually retail wheeling, vertically integrated utilities might be left with more than $200 billion of "stranded assets" in generation (the competitive segment). The utilities' earlier investments in nuclear power have proved very costly, and utilities had also been forced by regulators after PURPA to enter into costly supply agreements with "qualifying facilities" (cogenerators and small power producers). Electricity purchasers (large industrial users, distributors) were reluctant to pay the high embedded costs of utility generation, especially at the time when the price of gas had tumbled and new technological developments such as combined cycle gas turbines offered cheap alternative sources of energy.

The debate has centered on the contention that the utilities would have been able to generate a substantial profit "on the upside" if things had turned differently, given that the opening of competition leaves them with a substantial deficit "on the downside." Answering negatively amounts to admitting that the utilities' investment (here in generation) would not have been committed if utilities had been forewarned of incoming competition, or equivalently that, in the absence of proper compensation, liberalization constitutes a regulatory taking. Adopting this view, the California Public Utility Commission announced in 1995 that power utilities in its state will be entitled to full recovery of their stranded costs through an "electricity service surcharge" to be levied on transport (the equivalent of the access surcharge currently paid by long-distance carriers to local exchange carriers for access to the local loop bottleneck, and due to be eliminated in the wake of the Telecommunications Act). However, some other public utility commissions intend to leave a substantial part of the stranded assets burden on the incumbents' shoulders. And American jurisprudence has not yet brought clear guidance with respect to the treatment of stranded assets.

In a competitive environment various regulatory actions have the potential of constituting a taking: low access prices paid by entrants, of course, but also excessive collocation requirements, entry subsidies, rigidity of regulated prices before a competitive threat, uncompensated requirements to add transmission or switching capacity to accommodate new traffic, line-of-business restrictions, and so forth. Recently, the California Public Utility Commission's adoption of a bill-and-keep system (in which two interconnecting carriers do not pay access charges to each other) for interconnection between ILECs and CLECs was challenged by two incumbent local exchange carriers who claimed that the bill-and-keep rule was a taking on the basis that cross-networks call flows would be unbalanced. Sidak and Spulber (1996) argue that the California Public Utility Commission failed to recognize the existence of an obvious taking, despite a 1913 California Supreme Court decision hostile to a bill-and-keep rule (already in telecommunications) on exactly those grounds.

Of course, incumbents have strong incentives to claim that there is a taking whenever regulators and lawmakers contemplate a competitive move. Quite generally, competition-oriented policies must trade off the benefits of entry against the incumbents' incentives to build. No access pricing policy is likely to strike the right balance. The challenge is therefore to design rules that do not err too much with regard to this trade-off.

2 Definition of Services

As for retail services, the right to define the set of telecommunications wholesale services whose price is to be regulated is by no means obvious. Let us recall a few snags:

• Changing services: The configuration of the local loop (for example, of the feeder part), the switches' vertical features, and other elements of the local networks are changing rapidly with technological progress. Accordingly it is difficult for price regulators to keep up with the ever-changing range of services and elements to which competitors can have access.

• Nonlinear tariffs: Nonlinear tariffs amount to a multiplication of services. A linear price for a given service corresponds to a single service; a two-part tariff corresponds to two services: access to the service and variable consumption of the service; and menus of two-part or more-complex tariffs (such as those in optionalcalling plans) correspond to even more services. Discounted wholesale tariffs have made an appearance as a way of countering bypass; with the development of local competition, local exchange carriers will most likely want to enlarge their wholesale pricing options. These changes, of course, will raise the standard problem that current regulatory methods (backward- or forward-looking incremental cost, price cap) are designed for linear prices and are modified in a basically ad hoc way to reflect the nonlinearity of prices.

• Bundling: Relatedly, carriers may want to offer policies that bundle elements of the network or resale services. Local loop resale is an example of bundling, but one can imagine that local exchange carriers will want to multiply menus of bundles in a competitive environment. Another dimension of bundling is intertemporal bundling. A three- or five-year access contract with volume requirements cannot easily be summarized by a single yearly price.

• Level of deaveraging: The number of access services and of access prices is an important aspect of access policies. The theory's precept is that there should be as many access services as possible: Maximal unbundling allows entrants to buy access only to what they need. This approach, however, ignores the transaction costs of defining and monitoring access charges. (Imagine, for example, the complexity of cost-based rules that would single out each technologically, chronologically, or geographically differentiated piece of a network as a separate element for which a cost must be assessed!) It also ignores the fact that the multiplication of imperfectly determined access charges increases the possibilities open to entrants in arbitraging by purchasing undervalued pieces and leaving others aside. However, access price averaging also creates scope for substantial arbitrage by entrants (if they can buy individual elements or services separately) by putting together in the same pricing formula services or elements that have different costs or values to the entrants. The level of unbundling and access charge aggregation does matter geographically, in view of varying topologies and customer densities. It also is a subject of an unresolved debate for time-of-day pricing of access. Telephone networks' marginal costs are very small off peak and quite high at peak. While economists rightly argue in favor of peak-load pricing of access, the implementation of peakload pricing still has to be carefully designed. Besides the standard issue of computing off-peak and peak marginal costs, the new and interesting issue in a competitive environment with access is that several operators are jointly responsible for the timing of the peak during the day. This fact may induce some gaming on their part, and in any case adds to the difficulty encountered when foreseeing the location of the peak.

3 Information Requirements and Compliance Monitoring

A key feature of access pricing rules is their information requirement. To be realistic, access pricing rules must economize on the collection of information by the authorities that are in charge of enforcing them. All existing rules are information demanding. Cost-based rules require information about marginal costs of various elements and services in the bottleneck segment. They also require information about demand either if they make forecasts of capacity utilization (as is the case for forward-looking long-run incremental costs) or if they attempt to adopt a time-of-day structure. ECPR requires information about marginal costs in the competitive segment. Price caps require information about cost and demand in order to set the weights in the caps properly.

Once the access pricing rule is in place, compliance with it must be ensured. Such compliance may be particularly problematic if the rule creates strong incentives for gaming by the incumbents. The only way to ensure compliance then, as we will argue, is to engage in heavy-handed regulation.

Information requirements and the complexity of compliance monitoring are key dimensions of access pricing rules. And one of our concerns with current regulatory reforms is that, beyond the liberalization and free-market rhetoric, one may be creating an environment that will lead to heavy-handed regulatory intervention.

Two Specific Concerns and Some Common Misperceptions about Ramsey Access Pricing

As we will see, the old practice and the new reforms all depart from the theoretical precepts just described, at the cost of substantial inefficiencies. Indeed, it is fair to say that the participants in the current regulatory debate are on the whole suspicious of Ramsey access pricing. In our view, this mistrust is an amalgam of a legitimate concern that Ramsey pricing of access might be miscast within an otherwise incoherent regulatory framework, and of a misapprehension of its implications.

Two specific objections have been leveled at Ramsey prices: informational requirements and violation of the nondiscrimination rules.

1 Informational Requirements

Academic economists and policymakers both often argue that regulators do not have the information to set Ramsey prices. One leg of the argument, namely, the widespread shortage of relevant information, is correct. Regulatory agencies have a much smaller staff and less contact with markets than telecommunications operators.

But taken as a whole, this argument should look unconvincing to any observer of unregulated businesses. The latter indeed engage in sophisticated marketing strategies. They offer discounts to high-elasticity-of-demand customers, adjust their prices to competitive pressure, and carefully coordinate the pricing of substitutes or complements. The structure of unregulated firms' prices (though not the level if the firms have substantial market power) thus reflects Ramsey-Boiteux precepts. This observation suggests that the most promising alley for implementing Ramsey prices in a regulatory context is to decentralize pricing decisions to the operator.

The idea of decentralizing pricing decisions may be foreign to those who favor heavy regulatory intervention. Yet, a key feature of the regulatory revolution of the 1980s was departure from the detailed setting of individual prices and flexibility to operators to adjust their price structure to demand and competitive pressure conditions. While the implications of this revolution for access prices have been overlooked, we still find it surprising that regulators who routinely design price caps dismiss offhand Ramsey pricing as being informationally infeasible!

2 The Rhetoric of "Fair and Nondiscriminatory" Access Prices

We have observed that optimal access prices are usage based and thus discriminatory. Most experts oppose this implication of Ramsey access pricing and argue in favor of "fair and nondiscriminatory" access prices, a phrase that originated in the competition policy treatment of wholesale markets (e.g., the Robinson-Patman Act of 1936 in the United States) and made its way into almost all regulatory statutes governing the incumbents' wholesale transactions.

The ban on wholesale price discrimination applies to second- and third-degree price discrimination. Second-degree price discrimination consists in offering two different units at two different prices, for example, a two-part tariff involves a fixed fee as well as a variable price and thus implies a steep discount after the first unit of consumption. Menus of two-part tariffs, such as AT&T's retail optional calling plans or the 80–90 percent discounts below retail currently offered by the U.S. long-distance carriers in wholesale transactions, also involve second-degree price discrimination. Third-degree price discrimination in contrast refers to the offering of different tariffs to different categories of users or, by stretching the definition of third-degree price discrimination a bit, for different usages (mobile, data, video, etc.). To be sure, exceptions have been made and have allowed operators to practice some price discrimination at the wholesale level. In the United States, local exchange carriers have been given some flexibility to offer wholesale discounts to respond to competitive access providers (CAPs), after it was perceived that uniform access pricing created some inefficient bypass entry. In the former U.K. regime set up in the early 1990s, BT's long-distance competitor, Mercury, paid higher charges when using BT's local network for international call origination than for domestic call origination (or termination) even though the access service is identical. An instance of third-degree price discrimination in many countries is the differential in access charges for fixed-link and mobile communications.

The ban on third-degree wholesale price discrimination applies both to discrimination between two nonaffiliated carriers in the adjacent, competitive segment, and, if the bottleneck owner is vertically integrated, between a nonaffiliated carrier and the bottleneck owner's division or subsidiary in the competitive segment. The interpretation of the latter nondiscrimination requirement is subject to debate. After all, the pricing of internal transactions is principally an accounting matter, which may have little connection with economic reality. Some experts have understood this form of nondiscrimination as saying that the access price charged to nonaffiliated buyers should be equal to marginal cost (that is, be cost based), since efficient transfer prices for transactions between two divisions of the same firm are equal to their marginal cost. Other experts have given it an "ECPR interpretation" by arguing that access prices that do not exceed the unit profit on the competitive segment (that is, the vertically integrated firm's opportunity cost) allow a competitive subsidiary or a division of the bottleneck owner to break even when it is subject to accounting separation.

The origins of this deep-rooted fear of wholesale price discrimination can be found in competition policy. They are, therefore, worth a short digression into the economics of foreclosure. Let us start with the paradigm of an inventor who holds a patent on an innovation and subcontracts the exploitation of the innovation to some licensees (one can alternatively think of a franchisor subcontracting to franchisees). The licensing contract is similar to an interconnection agreement: The licensees obtain access to the bottleneck (the innovation) at some access price (fixed fee or royalties). How can the inventor make money on her innovation (which most of us would agree is a desirable outcome if society is to provide incentives for innovation)?

Let us start with the wrong way for the licensor to proceed. Suppose that she contacts prospective licensees separately and offers them independent contracts. It is clear that she cannot collect any money in this way, for the following reason: Suppose that she is expected to sign up n licensees. Once she has entered into n licensing contracts, nothing prevents her from contracting an (n + 1)st licensee, thus creating more competition for the existing ones. Anticipating this incentive to "flood the market," existing licensees would not be willing to pay much. Because the total licensing profit is bounded above by the downstream industry profit, stiff downstream competition hurts the licensor. In other words, downstream product market competition destroys the profit that could a priori be gleaned from the upstream monopoly position. (Similarly, McDonald's would be unable to collect any money from franchisees if it could install new franchisees at the doorsteps of existing ones.)

To avoid this disastrous outcome, the licensor has at least three alternatives. First, she can sign an exclusionary contract with, say, one licensor, that stipulates that she will not be able to enter into further licensing agreements. That is, she can create a downstream monopoly contractually and demand the corresponding monopoly profit. Second, she can vertically integrate and undertake the exploitation of the patent herself; she will then have no incentive to engage in excessive licensing because such licensing amounts to expropriating herself. Third, she may ask for (or it may be imposed!) a legal ban on third-degree price discrimination (but not on second-degree price discrimination). In this legal environment, she can offer a fixed-fee/no-royalty licensing contract to all prospective licensees, with the fixed fee equal to the downstream monopoly profit. It is clear that only one potential licensee will enter a licensing agreement, for duopolists in the downstream market would be unable to recoup the high fixed licensing fee.

The economics of wholesale transactions in general are similar to those of licensing and franchising. An upstream bottleneck must find a way to "discipline" the downstream market if she wants to exploit her upstream monopoly power. Whether it is socially optimal to let an upstream bottleneck exploit this monopoly power is another matter. There seems to be intellectual consensus that the owner of an upstream bottleneck should be allowed to exploit some of this monopoly power in order to recoup the fixed investment: An inventor who would be prevented by competition policy from making money on licensing agreements and from integrating downward into development and production would not undertake R&D in the first place. Similarly, incumbent telecommunications operators would not build local loops if they expected not to be able to enjoy some markups when reselling or exploiting the local loops themselves. How much monopoly power should be enjoyed by owners of bottlenecks is a matter of intense debate in competition policy, and antitrust practice (probably rightly) differs substantially across applications. For example, a number of foreclosure practices are well tolerated in the context of licensing and frowned upon in other contexts.

What do those antitrust considerations have to do with the regulated environment of the telecommunications industry? Probably little as such, as they do not take into account the specificities of regulated industries. But there is no denying that such considerations resurface under some regulatory paradigms, and that new issues arise as well. We illustrate these two points in sequence. Let us return to the licensor's third way of reestablishing her market power. We saw that contracts with sharp discounts can enable the licensor to create a downstream monopoly while giving the appearance of nondiscrimination. Transpose this insight to the pricing of the local loop by incumbent local exchange carriers, and recall that ILECs have been allowed to offer substantial discounts to counter inefficient bypass by CAPs. This opportunity to practice nonlinear wholesale pricing raises the concern that an ILEC enters an agreement with a long-distance carrier, AT&T, say, specifying a very high fixed fee and a very low wholesale price. Such a "sweet deal" (a slight misnomer, since the deal would be available to anyone else as well) would not enable rival long-distance companies to compete effectively with AT&T, as they would not have access to the low wholesale price without paying a fixed fee so high that they would be unable to recover it in a competitive long-distance market!

Next, we return to the innovator's second way of restoring her monopoly power. We saw that she can integrate vertically, deny access to nonaffiliated potential licensees, and let her downstream affiliate charge monopoly prices on the applications of her patent. The equivalent situation in telecommunications is provided by a local loop owner serving an adjacent segment, say, long distance. The local loop company has little incentive to create its own competition and is therefore likely to foreclose its rivals on the competitive segment unless the latter are much more cost-efficient or else produce a sufficiently differentiated service.

Suppose now that the vertically integrated innovator is forced to distribute free or cheap licenses to independent licensees. The innovator then has a strong incentive not to cooperate with the licensees in order not to dissipate the downstream profit; for example, she will refuse to supply the missing specifications that makes the invention work or to provide expertise at the development stage. The same holds in telecommunications. As we will later argue, a bottleneck owner who is forced to sell access below or around marginal cost and is otherwise an unregulated competitor in the adjacent segment has an incentive to use nonprice methods to deny access to rivals.

To sum up, one cannot directly apply foreclosure theory as developed for competition policy to regulated environments. Regulatory rules can change the private and social costs and benefits of foreclosure, and further analysis is required before all-encompassing statements such as those related to "fair and nondiscriminatory access" can be made.