Regulated and Deregulated Segments: The Problem of Cross-Subsidies

1 The Theoretical Argument

Incentives for cross-subsidies stem from a differential in the sharing of earnings across product lines. An operator gains from transferring costs from a segment in which it keeps a sizable fraction of its profits to another segment where consumers share a higher fraction of costs.

Cross-subsidies are particularly attractive for the operator when some segment is fully deregulated while a substantial fraction of costs on another segment (for example, regulated through some cost of service methodology) is reimbursed. For example, if 80 percent of the operator's cost on one segment is reimbursed, a $1 cross-subsidy with a deregulated segment increases the operator's profit by 80 cents.

But the incentive for cross-subsidies also exists, albeit in a weaker form, when the regulated segment is subject to a price cap. As is well known, the ratchet effect implies that high profitability today leads to a more stringent cap tomorrow, thus inducing profit sharing. Conversely, there sometimes exists an explicit or implicit regulatory insurance against low profitability, which adds a further mechanism for profit sharing.

2 Examples

There are two main categories of cross-subsidies: accounting cost allocation and managerial cost allocation. (The literature has focused on the first and neglected the second.)

2.1 Accounting Cost Allocation

Telecommunications technology gives rise to many joint and common costs. The allocation of these costs among product lines is by and large arbitrary and may be used to cross-subsidize some product lines. For example, an operator has an incentive to allocate expenses relative to maintenance personnel, product development, marketing, connection between the customer's home and the first switch, and so on to those services for which profit sharing is the greatest.

Cross-subsidies may also have an intertemporal dimension through the depreciation of investment expenditures. For example, a few years ago it was argued that U.S. local exchange carriers could install a fiber-optic network that was then useless in providing plain old telephone services but would later be a valuable asset when introducing new and innovative services such as interactive TV and video on demand. To the extent that the investment is (partly) depreciated before the new services are introduced, there may be a cross-subsidy from (current) regulated services to (future) unregulated ones.

2.2 Managerial Cost Allocation

In the presence of asymmetric profit sharing, the operator also has an incentive to allocate real resources strategically, thus generating social waste. The operator can allocate its best engineers and marketing agents to the competitive segment, and leave its less efficient or yet untrained personnel with its regulated segments.^[7] The CEO and top executive team may devote most of their attention to competitive segments and neglect regulated ones, for which high costs are more lightly sanctioned. Investment choices that jointly affect the marginal costs on competitive and regulated segments may be distorted toward achievement of low cost on competitive segments and high cost on regulated ones.

It is thus clear that managerial decisions are not geared to the minimization of the production cost, but rather to the minimization of the cost as perceived by the operator. Coming back to the theory, it can be shown that, under some assumptions, a uniform power or intensity of the operator's incentive scheme is socially optimal even when detailed supervisory monitoring prevents accounting cost manipulation. That is, the sharing of profit between the operator and the consumers should be based solely on the operator's overall cost and thus make no use of disaggregated information about cost at the product line level

3 Can Cross-Subsidies Be Prevented?

It is by no means easy to prevent cross-subsidies once one has created incentives for them. One can require accounting separation and invest regulatory resources into checking that the actual cost allocation follows clearly defined accounting principles. This accounting supervision, although costly, bars the most flagrant accounting cross-subsidies. Accountants, however, cannot substitute their judgment for business judgment. They have neither the training nor the information necessary to evaluate investment and personnel allocation within the firm. It is thus difficult to measure and prevent cross-subsidies.

The use of benchmarking has been advocated as a way to detect cross-subsidies, in particular in the context of a telecommunications operator's purchase of equipment from a manufacturing affiliate. The basic idea behind benchmarking is to compare the price paid by the regulated operator to the unregulated affiliated equipment manufacturer with external prices, in order to prevent the firm from engaging in cross-subsidization by inflating the price of its equipment. There are two distinct measures of external prices: prices charged by other manufacturers for similar equipment, and prices charged by the manufacturing affiliate to external buyers for the equipment. While benchmarking is useful, it is no panacea.

3.1 Comparison with the Price of Similar Equipment Sold by Nonaffiliated Manufacturers

There are two limits to benchmarking by comparison with the equipment of other manufacturers. First, equipment may be heterogenous across manufacturers. They may differ in capacities, functions, sturdiness, manufacturer's reputation, and the like. Second, pricing has several dimensions, which may also differ among types of buyers: installation cost, financing, training, penalties for delays, prices of spare parts, and so on.

3.2 Comparison with the Price of the Same Equipment Sold to External Buyers

There are three limits to comparing prices to those of equipment sold to others. First, as we just noted, the equipment may not be well-defined "equipment," since its features and pricing may be customized. Second, the internal buyer must not collude with external buyers and induce them to purchase at inflated official prices. Third, forcing such nondiscrimination between internal and external buyers affects the price paid by external buyers, since the supplier may want to forgo transactions with external buyers in order to be able to raise the price and engage in cross-subsidization. Alternatively, the supplier may decide to keep serving external buyers, but benchmarking still raises the price paid the latter, as the link between the two prices creates a new and profitable effect of a price increase through cross-subsidization.

Backward-Looking Cost-Based Pricing of Access

The traditional approach to computing interconnection charges consists in applying the methodology of cost-of-service regulation to the operator's wholesale offerings. There are a variety of possible cost allocations.

A popular cost-of-service methodology is that of additive or usage-proportional markups. Suppose that several services utilize a common element. After the allocation of costs that are attributable to a particular service to the corresponding services, there remains a residual corresponding to the "fixed cost" or "common cost." This unallocated residual is then spread across services, and the additive markup on each price is the same for each service. In other words, a usage-proportional markup is tantamount to a fixed (price-independent) excise tax, whose magnitude is computed so as to cover the unallocated cost.

It is interesting to note that usage-proportional markups satisfy the ECPR. ECPR requires the access price to be equal to the operator's opportunity cost on the competitive segment. The operator's price on the competitive segment is equal to total marginal cost, that is, the marginal cost of access plus the marginal cost of the segment itself, plus the markup. The access charge, which is equal to the marginal cost of access plus the markup, is thus equal to the difference between the price and the marginal cost on the competitive segment, that is, the opportunity cost.

Another popular approach to allocating the fixed cost is that of uniform or price-proportional markups. The markup over the marginal cost of a service is proportional to this marginal cost. The uniform markup is thus akin to a proportional (VAT type) tax. Unlike additive markups, uniform markups do not satisfy ECPR. Because the total marginal cost of the competitive segment exceeds the marginal cost of the access facilities used by this segment, the price of the competitive segment is inflated more than that of the access segment, and so the access charge is set below the operator's opportunity cost. The burden of cost recovery then falls disproportionately on the competitive segments.

The benefit of fully distributed cost pricing is that it commits the regulator to allow the operator to recoup its investments and to break even. Thus, to a large extent, it solves the problem of regulatory takings. In particular, an operator who incurs a large fixed cost to install fiber optics in the local loop or to endow switches with new functions need not be concerned that this investment will later be expropriated by the regulator's setting low access charges, for example.

Despite this advantage, in the context of retail pricing, fully distributed cost pricing has been as frequently decried by economists as it has been used in practice. It has well-known flaws. First, it is determined through a cumbersome process. For example, a rebalancing of access charges must be cost justified, a requirement which is likely to imply a delay of several months in the rebalancing. Second, fully distributed cost pricing is cost based and therefore does not encourage cost minimization. Third, it yields an improper price structure and is a vastly suboptimal way of financing the access deficit. Because it is cost-based, it "subsidizes" inelastic-demand segments to the detriment of elastic-demand ones. In the presence of competition, fully distributed cost pricing tends to create an inefficient amount of entry. For example, under uniform markups, an inefficient entrant producing the same service as the operator in the competitive segment finds it profitable to enter as long as its cost handicap relative to the operator is smaller than the markup on the operator's marginal cost on the competitive segment. Furthermore, under all fully distributed cost methods, the markup on access invites inefficient bypass.

To alleviate the cost of the first and third drawbacks (delays in price revisions, inefficient entry in the competitive and bottleneck segments), the prices set by fully distributed cost methods have sometimes been interpreted as ceilings or caps, providing, in particular, flexibility to respond to competitive threats such as those by competitive access providers. By letting operators respond to competition, this downward flexibility has perhaps brought actual prices closer to Ramsey levels, but fully distributed cost methods still have only limited appeal.