ANNEX TO THE EXECUTIVE SUMMARY
Responses to "List of Questions and Options"
Q. Which services should be subject to price regulation?
A. Price regulation should, in general, be extended to all services for which adequate or effective competition has not yet developed. There are specific tests for determining whether or not existing competition (if any) is effective. Some of those tests pertain to the structure and conduct of the telecom industry and, in particular, whether operators are able to exert market power (i.e., control or influence market prices) over services of interest.
Q. What should be the basis for categorizing services into those subject to price regulation and the other services which should not be so regulated? Should these two categories correspond to, for example, basic/non-basic telecom services, or voice/non-voice telecom services? If basic/non-basic services were used as the criteria then what should be the coverage of basic services in this context?
A. In most circumstances, all "wholesale" services (i.e., those not sold directly to end-users but rather to other operators) such as local and long distance interconnection services are the likeliest candidates for price regulation. Typically, services excluded from such regulation are "retail" services (i.e., those sold directly to end-users). Of these, if competition has yet to develop for, say, local service (or, more specifically, the residential customer segment of that service), then price regulation of that service would be appropriate. In general, distinctions between voice, data, or video services are irrelevant; the test of effective competition alone should apply. It is appropriate to adopt a basic/non-basic classification. One possibility is to classify all services into one of three categories: (i) basic (generally, interconnection, essential, or single-provider services), (ii) non-basic (generally, retail services which are not competitively provided, e.g., long distance service from a single operator, but which could become competitive before long), and (iii) competitive (which, by definition, should be exempted from price regulation).
Q. Should the services not subject to price regulation be those services which are considered not essential? Or, should non-essential services be categorized into those subject to weaker price regulation and others not subject to such regulation? If yes for the latter query, then on what criteria should such a categorization take place? What should be the definition of essential services in this context?
A. By definition, essential services are those for which effective competition does not exist even remotely. While ultimately the definition of essential services emerges from existing legal doctrine, one possible definition in the telecom context is as follows:
An essential service or facility has the following characteristics: (i) it is an integral part of (i.e., an input into) the production of a retail (or, end-user directed) telecom service; (ii) it is only available from the incumbent operator that also produces the retail service in which it is used; (iii) competitors for the retail service do not produce the essential service themselves; and (iv) competitors do not have the option of self-supply at reasonable cost, that is, they must depend on the incumbent operator for the essential service at the same time that they compete with the incumbent for the downstream retail service.
This definition implies that essential services should be viewed as akin to basic services and should, therefore, be subject to price regulation and other safeguards. Non-essential services, in contrast, have some measure of existing or emerging competition, or do not engender an exploitable dependence of competitors for the retail service on the incumbent operator. Therefore, they may be subject to the weaker price regulation accorded to non-basic services or exempted from regulation altogether.
A. The purpose of all regulation is (or should be) to secure the benefits and outcomes of competition where market forces are not themselves developed to the point of being able to deliver sustained and reliable competition. As such, as the market for a service becomes increasingly competitive, the level or scope of regulation in that market should be reduced proportionately. Therefore, the level of price (or any) regulation should be inversely related to the degree of competition. That is the basis for the prescription that weaker price regulation should prevail when competition is incipient, but withdrawn altogether when competition is stronger or effective.
Q. What is adequate competition? Should one use a thumb-rule that three or more operators in the market result in adequate competition for reconsidering the pricing methodology to be applied to the service provided by these operators?
A. "Effective competition" (more a term of art than "adequate competition") has to be inferred from the objective conditions of the market. There can be no hard and fast rule for establishing the presence of such competition. For example, while textbook notions of "perfect competition" often serve as a baseline paradigm by which to judge market performance, any such baseline is likely to be highly unsuitable for typical telecom markets. Unlike atomistic markets which can be perfectly competitive (because there is a single well-defined and homogeneous product, a "large" number of suppliers each with an insignificant share of the market, and a "large" number of well-informed and market-responsive customers), telecom markets tend to be highly capital-intensive and their fixed-to-operating cost ratios tend to be very high. These features are associated with economies of scale and, where multiple services are provided from common or shared facilities, economies of scope or density. In such markets, the minimum efficient scale, i.e., the minimum volume of production which results in the greatest cost-efficiency, is typically a rather large fraction of the total volumes of service in the market. Therefore, even vigorous competition in such markets may only occur among relatively few, large operators (even though the market shares of those operators would, by traditional standards, seem too high to be competitive).
In yet another scenario, competition may well be quite effective despite the existence of one very large firm with, say, 90% market share and a competitive fringe of much smaller firms sharing the other 10%. This could be a market where the incumbent operator is emerging from its monopoly status and is contending with competitive entry by much smaller firms. In highly capital-intensive industries, the biggest barrier to entry is usually the significant capital needs of the entrant. However, if unbundling of the incumbent’s facilities allow new entrants to combine their own resources with those of the incumbent’s (leased at fair prices), then successful competitive entry may occur without the need to first incur huge capital expenditures. Moreover, if those competitors could underprice the incumbent, take quick profits and, if necessary, exit the market at little or no cost, then the incumbent even with 90% of the market would be unable to exert any market power over the prices of its services. When entry barriers are lowered by the absence of significant sunk costs, the market becomes contestable (a suitable proxy for "effective competition") even though, by traditional textbook paradigms, the market may not appear to be competitive.
It is, therefore, prudent to use measures of market share (as indicators of competition) with great caution. The industry’s structural conditions (particularly, non-traditional features such as unbundling) must be factored into the determination of whether competition is effective. This implies eschewing either market share targets or counts of operators ("three or more"). In fact, experience shows that such rules of thumb may still produce an effectively cartelized industry. For example, in the U.S., 14 years after the old monopolistic Bell System was broken up, there are only four long distance companies today that operate with their own facilities. While nearly a thousand other companies resell their services, the Big Four effectively control nearly 85% of the long distance market (even though the share of AT&T—by far the largest— has shrunk to about 50%). As a result, there is considerable evidence that AT&T has held a "price umbrella" above the other three, resulting in price increases in lockstep by all four. This is a form of tacit price coordination that is hard to challenge legally; however, the rule of thumb does not work.
In the ultimate analysis, effective competition must always be judged by the presence or absence of market power in the pricing actions of the biggest firms. This usually means comparisons of price-marginal cost margins with those expected in industries with high capital-intensity.
A. It is unusual for price regulation to set specific price levels. Instead, once the "going-in" prices have been determined, price regulation limits any further price increases to only the adjustments made necessary by inflation, productivity growth, and other exogenous factors (e.g., changes in tax policy or in accounting practices). These adjustments are usually guided by formulas.
Q. Should only a price floor and ceiling be specified for certain services. If yes, then which services are these?
A. The weakest form of price regulation (typically reserved for emerging competitive services) only impose price ceilings and floors. Mainly, non-basic services are candidates for such regulation.
Q. Should price flexibility be greater than that provided by floor and ceiling prices, i.e., should only a price cap be provided and not the floor, or vice versa?
A. A price ceiling exists for the protection of consumers (particularly, in the absence of competition). A price floor, in contrast, exists for the protection of competitors. Typically, regulators are more concerned with the exploitation of hapless consumers and, therefore, expend more effort in designing price "caps." Pricing below the de facto price floor (the long run incremental cost) causes losses in the short run, but may be a successful strategy if such below-cost pricing drives equally or less efficient competitors from the market, thereby freeing the firm to raise its price to monopolistic levels later. There are very few documented cases of successful use of such an anti-competitive pricing strategy. In general, if greater pricing flexibility is desired, price floors may be relaxed before price ceilings.
Q. Should a combination of different types of regulation be used to regulate the same service, e.g., should the cost-based price be supplemented by a price cap mechanism?
A. No, combinations of regulatory instruments are, in general, a bad idea. Recalling that the purpose of enlightened regulation is to emulate the effects of market competition, the proper basis for regulation should always be minimalism. That is, instead of combining instruments in order to make regulation more onerous, the idea should always to be do what is barely necessary to generate the desired effect. A cost-based price is already pegged to the underlying cost, and any movement in that price would be linked to movement in that cost. Imposing the additional layer of a price cap mechanism is unlikely to generate any incremental benefit and only result in overkill. Moreover, regulation should never be viewed as a permanent state of affairs; its purpose should always be to effect a transition to a fully competitive market. The best way to do that is to maximize the incentives faced by firms to become more efficient and productive. Several layers of regulation would only hamstring those incentives.
Q. If a price cap mechanism were to be used, then which services (and methodologies) should it apply to? For example, should it apply to both tariffs specified in terms of a specific level as well as tariffs specified in terms of floor or ceiling?
A. By definition, a price cap mechanism establishes the manner in which service prices may be increased over time. This is usually accomplished by one or more formulas that take into account the rate of inflation, changes in productivity, and other exogenous factors. Any adjustment according to this mechanism must, therefore, occur to an existing level of price. Logically, such adjustments are not made to price floors and ceilings (unless their cost bases themselves change). Rather, floors and ceilings are only used to establish a range of feasible and acceptable prices.
Q. In an overall price cap, which telecom services should be subject to sub-caps, and what should be emphasized in these sub-categories when the specific price cap is being decided for them?
A. Price cap regulation is often designed at two levels. Services are first assigned to different groups or "baskets," (e.g., basic, non-basic, and competitive). Next, for the baskets which qualify for regulation (namely, the first two), the first level of price cap ensures that the weighted average of prices of all services within a basket respects a cap. That is, as prices of the constituent services are changed, their average is not allowed to exceed a certain bound, say, 3.5%. This may mean that in order to raise prices for one set of constituent prices by more than 3.5%, the prices of the other constituent services may have to be lowered, so as to allow the average to comply with the overall cap.
A second level of capping may occur within a basket. The price of a "sensitive" service may be limited to only small changes within a year (or, at least, smaller than those allowed for the other services). For example, in the U.S., it is not unusual for the price of basic local service sold to residential customers to be frozen for a number of years. On the other hand, other price-capped services may be allowed to change within more liberal bounds. For example, interconnection services sold to long distance companies are free to vary over a wider range. Sub-caps are, therefore, used to isolate services that vary by their degree of importance or sensitivity within a basket.
Q. Should a price floor be defined for all services in order to address the issues of unfair competition? If not, for which types of services should a price floor be provided? If such a price floor were to be specified, should a rebuttable presumption be that prices lower than the floor result in unfair competition?
A. As stated before, price floors are used to protect competitors from unfair or predatory pricing. As such, instances of proven predatory intent are extremely rare. As a practical matter, therefore, whether or not price floors should be established is up to regulatory discretion. In any event, the contention that prices below their floor necessarily result in successful predation (and recoupment of the short-run losses from predation) should always be treated as a rebuttable presumption.
A. Yes, long run incremental costs are the appropriate basis for determining cost-based prices. However, economic theory dictates that simple long run incremental cost (LRIC, a concept analogous to marginal cost for a firm that provides multiple services) be used for pricing while TSLRIC is more appropriately the basis for testing whether a cross-subsidy exists between services.
By definition, the LRIC is the incremental cost associated with the next unit of a service. There is no reason a priori to believe that LRIC remains constant from one unit of service to the next. For example, because of economies of scale, the LRIC could decline continuously as volume increases. Or, it may decrease, bottom out, and increase eventually. Because it is analogous to marginal cost, LRIC does not include the fixed costs specific to the service. In principle, this allows one to adhere to the basic economic principle of efficient pricing, that the price of any given unit of a product be equal to what it costs to produce that unit. The fixed costs must be incurred in order to start production and must, therefore, be part of the LRIC for the first unit of service, but those costs do not figure in the LRIC of, say, the 100th unit of service.
In contrast, TSLRIC represents the total cost that would be saved when a firm that provides multiple services stopped providing the service of interest altogether. TSLRIC, therefore, includes both service-specific fixed costs and variable costs. At any given volume of service, it is calculated by dividing the combined such costs by the number of units produced. There is no reason to believe that TSLRIC and LRIC are equal, except for the first unit of the service. TSLRIC is best used in testing for cross-subsidy. When a firm that produces multiple services exactly breaks even (i.e., its total revenues from all services equals its total costs from all services), for a service to provide a cross-subsidy it must earn more in revenue than its TSLRIC. Correspondingly, the service that receives the cross-subsidy must earn less in revenue than its TSLRIC. Therefore, detecting whether a cross-subsidy exists is simply a matter of determining whether each service earns enough revenue to equal or exceed its TSLRIC. Given that unauthorized cross-subsidies in pricing can be a form of unfair competition, the utility of the TSLRIC-based test is clear.
Despite these distinctions from economic theory, however, TSLRIC is sometimes adopted in place of LRIC as the appropriate price floor.
A. Demand-based pricing is, in general, impractical for implementing a price regulation scheme. Price regulation aims to ensure that changes in prices for regulated services have a justifiable cost basis. Only prices for fully competitive services—which, by definition, should be exempt from regulation of any sort—should be determined by market forces such as market demand and supply.
Q. Should the floor and ceiling prices be linked to certain concepts of costs? If yes, which concepts should be used for this purpose (e.g., certain version of long run incremental costs, stand-alone costs)? If not, what should be the basis for choosing floor and ceiling prices?
A. The price floor should always be LRIC (or TSLRIC). This should always be the minimum price that a firm is allowed to charge. But, in order to recover all costs (including shared and common costs), markups above LRIC (or TSLRIC) may be necessary. The price ceiling should always be the stand-alone cost.
Q. Does the fact that new entrants would be coming into the market and that there will be a substantial increase in telecom capacity in particular imply that TSLRIC is an appropriate concept for cost-based pricing?
A. No, there is no link between the capacity increase that new entry will cause and the appropriate concept of cost for pricing. Strictly speaking, efficient prices should be based on LRIC. However, TSLRIC has been adopted for that purpose in certain quarters.
Q. Which type of markup should be used? For example, should it be a reasonable commercial return or Ramsey markup?
A. It is unclear exactly how the term "reasonable commercial return" should be defined. The ultimate goal should be to allow a telecom operator to recover its legitimate total cost (including the "normal profits" which represents a reasonable return on its capital). Given the relatively large magnitude of that operator’s shared and common fixed costs, pricing all of its services at their respective LRICs (or TSLRICs) will cause a revenue deficit. Therefore, a markup on service prices would be needed to recover the deficit. In economic theory, the most efficient way to do so (i.e., with the minimum loss of allocative efficiency and social welfare) is Ramsey pricing which sets markups for services in inverse relationship to their respective elasticities of demand. Since the services can vary in their elasticities (and the degree to which they face competition), the markups could easily vary by service as well. A single fixed markup (representing the reasonable commercial return), on the other hand, is administratively simpler but is not necessarily fully efficient. Therefore, the trade-off is between administrative simplicity (or ease of implementation) and maximum allocative efficiency.
Q. Should certain services which provide greater use or value to the subscribers be charged a higher markup on cost, i.e., a markup which reflects a demand-based price?
A. If the sole concern is with maximizing economic efficiency, then the markup scheme suggested here should be used. Higher use value is associated with lower demand elasticity, hence, the markup should be inversely related too. Under unfettered market operation, such markups are likely to ensue in the normal course of business. However, there are often equity or fairness concerns about such a rule because it implies setting the highest prices for services that people most need or depend on. Typically, basic local service taken by residential customers has the lowest demand elasticity and would, therefore, qualify for the greatest markup. Yet, as demonstrated by a strong societal interest in universal service, there is often a policy to subsidize customers of that service, i.e., to mark down the price. If Ramsey pricing is not a feasible pricing option (for this and other reasons), certain forms of nonlinear pricing schemes (e.g., the two-part tariff or volume-based price discrimination) may be better.
A. This is a rather complicated issue, and I first sort through some of the nuances underlying it. Traditionally, the purpose behind calculating the FAC was to set prices which collectively, and at the forecasted levels of demand, would allow for the complete recovery of a telecom operator’s total cost. In other words, FAC was always meant for cost recovery first, and only incidentally as an instrument for pricing. Unfortunately, what may have been effective for cost recovery was not optimal or efficient for pricing, particularly in the presence of competition. FAC-based prices invariably resulted in prices that could not withstand competitive pressures in markets for certain services. However, in pre-competitive days, the distinction between cost recovery and pricing was, from a practical standpoint, immaterial and, therefore, not given much attention.
Under competition, though, the difference between the two does matter. A telecom operator must recover its legitimate total costs if it is to remain financially viable. Even if it is regulated, it must at least have the opportunity to attempt to recover its total cost. However, for efficient competition, it must always set service prices on the basis of LRIC (or some forward-looking variant of it). Past (historical) or sunk costs simply do not matter for efficient pricing. Unfortunately, if all service prices were set precisely at their respective LRICs, the operator would not recover its sizable other costs, namely, its shared and common costs. Therefore, the rule in this environment should be: (i) for efficient pricing, start with price floors based on LRICs, and (ii) for cost recovery, scale up any or all prices with markups which, collectively, recover all non-incremental costs. The markups chosen here may be market-responsive (such as in Ramsey pricing) or, where administrative simplicity is the prime concern, some uniform percentage rate (though full efficiency is not guaranteed this way).
The cardinal point is this: under competition, FAC ceases to have any utility. Under no circumstances, should there be any resort to FAC or FAC-based prices. If the concern here is with determining the deficit that results from prices set equal to their respective LRICs, then it is a simple matter of either (i) directly estimating the remaining shared and common costs which cause that deficit or (ii) inferring those costs from the difference between total booked (or accounting) costs and the sum of all incremental costs.
A word of caution, however, still applies to this exercise. A telecom operator’s total booked costs may include costs that are neither forward-looking incremental costs nor forward-looking shared and common costs. There may be some unrecovered historical costs, some of which may be legitimately recoverable and others which may not be so. For example, past costs that remain unrecovered because of inadequate depreciation or an understated cost of capital may be legitimate cost carryovers ("legacy costs") if, for various reasons, the operator had no control over these factors. On the other hand, inefficiencies and other unjustified costs incurred during the monopoly era cannot, and should not, be considered candidates for recovery in the new competitive environment. These are costs that the operator should be required to absorb or write off. Only legitimate costs or costs of stranded assets under competition should be carried over for recovery. These costs may be then added to the operator’s forward-looking shared and common costs in order to form the total target for cost recovery using markups above LRIC-based prices.
Q. Should the markup be applied before or after the calculation of the difference between fully allocated costs and long run incremental costs?
A. The previous Q&A outlines an alternative approach to recovering the deficit. Because it eschews the use of FAC for this purpose, the correct principle is that the markup should be applied after the gap between total costs (all incremental costs + all forward-looking shared and common costs + all legitimate legacy and stranded costs) and total incremental costs is known.
Q. Regarding the issue of lower telecom tariffs for certain users, there is a need to consider to whom subsidies should be provided, and how much. For example, should there be an upper limit on local call charges?
A. The traditional reason for subsidies is to promote universal service by making it possible for low-income households or those residing in remote, high-cost-to-serve areas to become subscribers to the public switched network. In network industries, increasing subscribership generates both network and call externalities (benefits not accounted for in prices) that, at least in principle, offset the subsidies that are provided to those pivotal subscribers. However, actual practice in some countries has been to extend the benefits of those subsidies to all subscribers, regardless of their income status or serving locations. This is clearly excessive (even for the goal of promoting universal service) and requires subsidies well beyond the efficient level.
Recent debate on the use of subsidies to support universal service has endorsed the continuation of such subsidies, albeit in a targeted manner. However, there is an increasing realization of the fundamental inefficiency of using prices of other services to subsidize the targeted service, basic local residential service. In the U.S., the old system of implicit subsidies (i.e., subsidies embedded in prices)—which is clearly unsustainable under competition—is being replaced by an explicit universal service fund into which all telecom operators (regardless of the services they provide) must contribute a uniform percentage of their revenues. This system does not distort relative prices (as one based on implicit subsidies would), leaving those prices to be efficient and competitive.
Thus, making subsidies explicit and subsidizing only targeted subscribers greatly improves the efficiency of the universal service practice of pricing local calls "low" and even below cost. In order to determine just how much subsidy would be needed, the following exercise should be conducted. First, divide up the entire service territory by switch locations (called "wire centers" in the U.S.) or exchanges. Second, determine the actual forward-looking incremental cost of providing local service in each such location. Third, determine for each location whether the incremental cost for local service exceeds the charge for local service. Fourth, if it does, then calculate the total deficit specific to that location (aggregating over all subscribers there). Finally, calculate the total deficit by aggregating over all locations. This would be the target subsidy for serving high-cost areas. In addition, if specific low-income subscribers are targeted, regardless of where they live, then the extent of subsidy to be provided to them should be calculated and added to the target subsidy previously calculated. That should determine the required size of the universal service fund.
In general, economic efficiency is maximized when prices come as close to underlying LRICs as possible. This principle rules out prolonged periods of below-cost pricing (even for public policy reasons) just as much as it implies prohibiting prices above stand-alone costs (price ceilings). Precisely where the charge for local calling on the range in between is a discretionary or public policy decision.
A. The charge for local service has two components (or a two-part pricing structure): a fixed charge or rental which is paid at fixed intervals (e.g., monthly) and a variable or usage charge which is paid in proportion to the volume of calling.
By paying the rental, a subscriber obtains connectivity to the network for the purpose of both making and receiving calls. Besides the value to the subscriber from such connectivity, there is additional value (the network externality) to the network itself as each individual now gains access to a larger pool of subscribers. If subscribership to the network (hence, the network externality) is to be maximized, then it makes sense to set the rental at a "low" level (possibly with the help of a subsidy). More importantly, the value from a low rental is greatest (as is the network externality) in the earliest stages of network-building, i.e., when the subscribership rate or teledensity is itself quite low. There is less and less value to subsidizing the rental as the teledensity increases and the subscribership rate approaches 100%.
By paying the usage charge, a subscriber is able to actually place a call of a certain duration (in most countries, the subscriber does not pay for calls received, at least for landline calls). However, even the act of calling generates its own value beyond that reflected by the usage charge paid by the caller: the party that is called receives the benefit of the call but doesn’t have to pay for it. Empirically, most residential calling appears to generate return calling of roughly the same total duration, i.e., there is a form of reciprocity in calling behavior. Therefore, the more calls actually placed, the more calls that are returned. This additional value (or, call externality) in the telephone system will be maximized as longer or more frequent calls are placed. That, in turn, can obviously be encouraged by "low" usage charges. More importantly—and this is a rather subtle point—it could also be encouraged by increasing network participation (i.e., increasing the number of potential calling pairs). That, in turn, could be effected by subsidizing rentals, particularly in the early stages of network growth.
In the ultimate analysis, whether rentals should be increased or subsidized depends on (i) the stage of network growth and participation and (ii) the degree of network and call externalities likely to be realized. When subscribership rates are quite low, subsidizing rentals and/or usage charges can help to maximize externality benefits. Subsidies could be phased out or otherwise reduced as subscribership approaches the universal service target level.
The secondary question pertains to setting different rentals for different classes of users ("user groups"). While the same general principle of phased subsidy to rentals should apply to all user groups, their relative elasticities of demand for network access should be used to determine whether rentals vary by user groups. Empirically, those elasticities are lower for business subscribers than for residential subscribers. Hence, the former may pay relatively higher rentals. Less is known about the demand characteristics of rural vs. non-rural subscribers. In general, the more affluent a subscriber or more alternatives to telephony a subscriber has access to, the higher will be his or her demand elasticity. For such a subscriber, a lower rental would be advisable. In theory, just the opposite would be true for subscribers who are less affluent or have fewer substitutes. While such subscribers could be charged higher rentals, it may be more worthwhile to encourage their network participation by directly subsidizing (even waiving) their rental charges.
A. To answer this question, a short digression on capacity constraints is in order. "Small" capacity is, in and of itself, not a problem unless it acts as a constraint on the level of demand that can be satisfied. Thus, a capacity constraint allows demand to be met only up to the point where the capacity is fully exhausted and an excess demand is generated at the price then prevailing. Relieving that capacity constraint would help to meet some or all of the excess demand and normal pricing rules (based on the Ramsey markup or two-part tariffs) could then be restored. However, what happens if that capacity constraint is binding? Policy makers would then have two options: (i) allow the price to rise until the excess demand is completely removed, or (ii) increase capacity so that additional demand can be met at the prevailing price.
In addition, policy makers would be faced with diagnosing the capacity constraint itself: (i) if the constraint is on feeder and distribution loop or outside plant, then demand for network access or connection may be most affected and, secondarily, the demand for usage; (ii) if the constraint is on switch capacity and trunk lines (for transport), then demand for usage would be most affected.
If the constraint is on outside plant, then the only long run solution would be to install more outside plant in order to increase network access. However, in the short run, the artificial scarcity may be removed, or demand served on a rationed basis, by raising the rental charge. This could have the unfortunate effect of at least temporarily creating haves and have-nots for telephone access, not to mention sacrificing network externalities as potential subscribers are shut out or existing subscribers are removed from the network. Raising the rental may still have a favorable longer run impact if the higher supply price induces new telecom operators to enter and add to capacity, thereby shifting the supply curve to the right. However, such a development is also a likely function of factors outside telecommunications (e.g., telecom ownership laws and structures, the general investment climate, tax laws, capital markets, alternative investments, etc.).
If the constraint is on switch capacity, then somewhat more flexibility may be available. This type of constraint affects usage, not network access. Lines and ports may be added to existing switches, or supplemental switches (which cost far less than outside plant) may be installed. Rationing procedures such as peak/off-peak pricing or even interruptible service can be used to ration capacity on the usage side. The rental charge, of course, would not be directly affected in these circumstances.
In principle, the level of the rental charge should reflect demand-side factors such as (i) the network and call externalities generated by network growth and (ii) the elasticity of demand for local service by a given user group. Capacity and exchange size are, in contrast, supply-side factors which, in the absence of capacity constraints that create excess demand at prevailing prices, should not be directly material for setting the level of the rental charge. When binding capacity constraints exist for outside plant (needed for network connectivity), then raising the rental can only relieve short run excess demand. There must still be price and non-price incentives for additional capacity installation, whether from existing or new sources. Moreover, maintaining lower rentals in capacity-constrained areas would neither clear the short run excess demand nor induce additional capacity installation and may, therefore, be counter-productive.
Q. Should there be a reduction in the number of different categories of exchange capacity which are currently used for pricing rentals?
A. The absolute size of, or capacity in, an exchange is not directly relevant; what matters only is whether enough outside plant capacity exists to serve current (and projected) demand for network access. Therefore, there are really only two categories for setting the rental charge: (i) capacity-unconstrained and (ii) capacity-constrained. For the first category, the rental charge should depend only on the demand characteristics of specific user groups. For the second category, temporary rental increases could be used to relieve short run excess demand and/or induce additional capacity installation. However, if such rental increases to subscribers in capacity-constrained areas are viewed as discriminatory or politically unpalatable, then some subsidy scheme could be used to keep the rental at a uniform level for all subscribers but offer a higher supply price which—in the proper investment climate—would induce new capacity generation.
Q. Should there be a differentiation among subscribers only on the basis of whether they are rural or non-rural subscribers, e.g., for purposes of rentals, there would not be any distinction of subscribers on the basis of exchange-capacity?
A. Any differentiation that occurs should ideally reflect differences in the demand characteristics of different user groups. Because subscribers located in both capacity-unconstrained and capacity-constrained areas could have similar demand characteristics, there is no prima facie case for charging them different rentals. Despite the unique problems associated with capacity-constrained exchanges, a selective subsidy to subscribers in those exchanges could both (i) keep a uniform rental for similar user groups and (ii) address the need to relieve the short run excess demand and supply constraint. Of course, if the user groups differ in their demand characteristics (regardless of the types of exchanges from which they receive service), different rentals could be charged.
Q. Would it be preferable to provide a flexible option which combines two possibilities, namely, a high rental together with a lower usage fee, or a lower rental combined with higher usage fee? Should the same flexible options be provided to subscribers in rural and non-rural areas? If not, then what would be the nature of the difference in the options provided to rural and non-rural subscribers?
A. The purpose behind such flexible options (indeed, behind all multi-part tariffs) is that rather than trying to pre-determine the efficient level of prices and markups (such as with Ramsey pricing), those options allow subscribers to self-select into price-usage combinations that best meet their needs, abilities to pay, etc. That flexibility, therefore, allows the market to reveal demand preferences and to replace the imperfectly-informed regulatory mechanism that has the task of setting prices.
The high rental/low usage fee option best suits a user group that has a low elasticity of demand for network access but relatively high elasticity for usage. In contrast, the low rental/high usage fee option best suits the user group that has just the opposite elasticity profiles for network access and usage. While empirical research alone can determine which profile fits which user group, some general observations may nevertheless be made. In many countries, users have to select from two different pricing schemes: (i) flat rate ("high" rental but zero usage fee, i.e., unlimited local calling) and (ii) measured rate ("low" rental but non-zero usage fee). Flat rate pricing schemes tend to be very popular among residential subscribers of local service. The ability to make unlimited local calls provides an "insurance option" against unexpected changes in calling volume and a fixed, pre-determined bill removes uncertainty for many subscribers who make primarily (or only) local calls. This would suggest that such subscribers, regardless of their income, age, or location status have a relatively low demand elasticity for network access, a characteristic that is reinforced by their preference for certainty about the cost of local calling. If local calling is generally affordable, the preference for flat rate pricing schemes is particularly pronounced. Only when local calling becomes "very expensive" does the preference for measured rate pricing manifest itself.
There is still a third option, particularly in the early stages of network growth. That option is to combine a "low" or affordable rental with no usage fee for local calling, and it may be particularly attractive to subscribers of lower income or in remote areas. There should not be any blanket difference in the treatment of rural and non-rural subscribers. Often, offering multiple flexible options and encouraging subscribers to self-select are a more effective way to ensure network growth and use.
A. See Q&A above. If "preference" pertains only to the level of the rental charge, then the discussion above applies. Any preference should be targeted to user groups with specific demand characteristics. Locations of user groups only become relevant for determining whether a subsidy is needed (because the cost to serve exceeds the rental charge that has been set).
A. Any unbundling that is instituted should only pertain to the essential network facilities that the incumbent telecom operator uses to produce and deliver different services. For example, those facilities may include the feeder and distribution loop plant that links communities to serving exchanges and, in turn, links individual premises to communities. Other facilities would include switches and ports, inter-office facilities that provide transport or trunking between exchanges, and signaling features that control the quality of calls carried over those facilities. The idea behind unbundling is to allow easy and relatively low-cost entry by competitors who wish to provide services either entirely over facilities leased from the incumbent or through a combination of leased and self-supplied facilities. Feeder and distribution loop plant represent a disproportionately high percentage of network facility costs. They can be a potential barrier to entry for a competitor that is uncertain about being able to recover the sunk costs of what would be essentially duplicative facilities.
Just as with interconnection, prices charged for unbundled network elements (UNEs) should be based on their underlying LRICs but also include a reasonable profit or contribution toward the network’s shared and common costs. It is, however, extremely important to remember that when the incumbent provides a service by combining two or more UNEs, it is usually able to take advantage of the economies of scope made possible by the sharing of certain assets and their costs. When scope economies exist, the cost of combining the production of two services is less than the combined individual costs when the services are produced on a stand-alone basis. Empirical evidence shows that telephone networks that use a high degree of shared and common assets experience economies of scope. Therefore, the services they produce benefit from scope economies and can be priced below levels that would be needed if those services were produced separately instead.
When an incumbent unbundles its network and provides UNEs to competitors on a stand-alone basis, it sacrifices some of the scope economies associated with sharing and combined production. Therefore, when the competitor separately purchases/leases UNEs from the incumbent and reassembles them in order to form new services, it is quite possible for it to experience a cost that is higher than that experienced by the incumbent. Hence, it is not necessary for the service originally provided by the incumbent and the same service reassembled from UNEs by the competitor to cost the same. In other words, no "consistency" (meaning equality) between their prices need be expected.
A. Absolutely, yes. Escalating tariffs are only used to control or ration use. With as low a teledensity and subscribership rate that India has, the common goal should be network development and greater utilization. In the information age, telecommunications is already key to the development of other sectors of the economy. The greater efficiency needed for information flows and market transactions can only be achieved by building infrastructure that allows and fosters high utilization of communications technologies. Escalating tariffs run exactly contrary to that need.
There is a more subtle economic reason for abandoning escalating tariffs. Tariffs may escalate in a variety of ways: (i) with number of calls, (ii) with length or duration of calls, (iii) with length-of-haul or distance called, and (iv) with time of day. Empirical research has shown that consumers become more price-responsive (i.e., their demand becomes more elastic) as the level of price rises. This would suggest that as that price rises "sufficiently high" and demand gets progressively more elastic (or less inelastic), the likelihood of losing revenue (or gaining revenue at a slower rate) becomes greater. From a revenue earning standpoint, therefore, escalating tariffs could become counter-productive by pushing price sensitivity up strongly. Second, by making additional usage more expensive, an incentive is created (as pointed out in the paper) for subscribers to demand more network access which is only likely to further exacerbate the excess demand problems that exist in capacity-constrained exchanges. Third, each subscriber line would potentially be utilized at a less than optimal rate causing higher per-line costs of providing service. This would also further exacerbate problems due to the capacity constraint. This form of substitution is, therefore, inefficient on a number of fronts.
A. Again, yes, for all of the reasons discussed in the previous Q&A. Volume discounts are a form of multi-part tariff where subscribers are able to self-select their optimal level of calling. Discounts encourage greater calling and generate greater producer surplus, i.e., they are a form of positive price discrimination. A volume discount-based tariff structure also encourages the emergence and growth of resellers. These resellers have no facilities of their own but introduce a greater choice of suppliers and retail competition into the market. They do so by buying large blocks of usage under a declining block tariff and reselling those blocks in smaller quantities to subscribers at prices between the "full" price and the "discounted" price. This form of arbitrage contributes greatly to increases in economic welfare.
Q. Should the basis for timing used for charging for calls be changed? For example, should an initial flat rate tariff be charged for a specified time, followed by a per second charge applied to the time of the call that exceeds the initial flat rate period?
A. Where metering systems permit, it is most efficient to rate a call at very frequent intervals, even down to the sub-minute level. A typical call may have an "initial period" (e.g., the first minute or 30 seconds) and then "additional periods" (e.g., additional minutes or x-second intervals). The first period charge should cover not only the incremental cost of that period (which may include call set-up) but also include a contribution to the shared and common costs of the telecom operator. The additional period charge may be based solely on the incremental cost and may, thus, be lower.
Q. Should the current number of free calls continue to be provided, or should the free calls not be provided at all? If free calls were not to be provided, then should a specified number of initial calls be charged a lower/higher price than subsequent calls? What should this specified number be, and what should be the link between the price of these initial calls and the subsequent calls?
A. Economic welfare can be maximized by offering subscribers a number of different pricing plans (based on the two-part tariff structure) and asking them to self-select. At one extreme, a true flat rate plan would allow unlimited calling with a zero marginal usage charge. The calls are not really "free," there is simply an unlimited allowance, the cost of which is expected to be recovered on average from the high rental charge. As additional, measured rate plans are introduced, the call allowance would be adjusted down as well. For example, if two measured rate plans are designed (with rental charges for both below that for the flat rate plan), then the one with the higher rental of the two should have a higher call allowance and lower usage fee than the other. At the other extreme, the pricing plan may have a zero or trivial rental charge, zero or little call allowance, and a high usage fee. Repeated experimentation with such pricing plans would reveal how best to structure them in the market. Economics alone cannot dictate the actual rupee level of the rental and usage fees or the call allowance—that should be a business decision.
Q. Should the national STD tariff structure be altered to correspond more closely to the difference in costs that arise with a change in distance? Similarly, should the tariffs for international calls be oriented toward costs?
A. With new technologies (fibre replacing copper, digital switches replacing analog or electro-mechanical systems, increasing use of electronics), the distance-sensitivity of the cost of trunking or transport is declining rapidly. Hence, STD tariffs should reflect this important change in underlying costs. As STD prices decline, demand will be stimulated and the incentive to further expand the long distance telephone network will increase. The same principle should apply to international calls as well.
Q. If the national STD distance-based tariff system were to be changed to better reflect costs, should the discrete distance-slabs as in the present structure of tariffs be retained, or should a more continuous and smooth increase in tariffs be used as the distance increases?
A. In theory, the structure of tariffs should track the manner in which costs change with distance. However, from a practical standpoint, discrete distance-slabs should be retained. Of course, the number of slabs will depend on just how sensitive cost is to distance; the less sensitive it is, the fewer the slabs needed.
Q. If the distance-slab system were to be retained, how many slabs should be considered, and over what range of different distance slabs?
A. Engineering and telephone traffic studies should reveal (i) the distance-sensitivity of costs, (ii) trends in STD traffic (especially, lengths of haul), and (iii) apparent communities of interest (defined as zones with the highest concentration of return or reciprocal calling). These and other considerations may be used to determine the number of distance slabs. Typically, those slabs are not of equal size; the shorter distance slabs (closer in to the exchange) tend to be smaller than the longer distance slabs.
Q. Should the number of different time slabs for domestic off-peak tariffs be reduced?
A. Again, that depends on traffic pattern and the optimal management of traffic load. Business and residential subscribers tend to have generally different calling patterns (peaks and valleys, number of peaks in a day, etc.) for lifestyle reasons alone. The form of peak/off-peak pricing policy adopted will further influence those patterns. This interdependence between calling patterns and the peak/off-peak tariffs implies, in effect, that some experimentation may be needed to pick the optimal number and duration of time slabs.
Q. Are the incentives provided by the current structure of off-peak tariffs such that there is a greater likelihood of congestion in the non-peak period adjacent to the peak period? Could a larger number of calls be encouraged during the late night off-peak period? What type of a change in tariff should be used for that purpose?
A. It should be remembered that tariff structure alone cannot influence calling patterns during a 24 hour period (though it might have an important influence). Cultural and lifestyle factors (e.g., time of day that people rise, go to sleep, go to work, return from work, socialize, eat, etc.) are at least equally responsible for observed calling patterns. Both traffic management and demographic studies would be needed to understand and plan for those calling patterns.
Q. Should the tariffs for operator-assisted trunk calls be made more consistent with the national STD tariffs?
A. Adjustments in STD tariffs based on cost, distance, or time of day should also be extended to operator-assisted trunk calls. There should, however, be a premium or surcharge for such calls.
Q. Should the time difference with various countries and the prevailing pattern of calls to these countries be considered for deciding when off-peak rates should be provided for calls to different countries? Alternatively, should the off-peak times be the same for international calls to all countries?
A. Any adjustment for time differences for selecting the off-peak period would only be relevant if traffic congestion in international calling is a particular problem. If it is, then the off-peak periods should be chosen to balance out the traffic load. Those periods so selected may, or may not, coincide with off-peak periods generated by adjusting domestic off-peak periods for time differences. There is no a priori reason to expect off-peak times to be the same for all countries.
Suppose the essential service in question is long distance interconnection, i.e., the access competing long distance companies need in order to sell long distance service to the incumbent’s local subscribers. Suppose also that the incumbent is also interested in competing for the provision of long distance service. Technically, suppose to keep matters simple, that it can provide itself interconnection to its own subscribers at zero additional cost. However, it may charge its competitors, say, Rs. 15 per minute for that interconnection. Now, again for simplicity, assume that the incumbent and its competitors are equally efficient in actually providing the long distance service, e.g., the incremental cost to them both is Rs. 20 per minute. These numbers would then imply that the incremental cost for the incumbent to provide the total service (interconnection plus long distance) is only Rs. 15 per minute, while it is Rs. 35 per minute for its competitors. Since these incremental costs are also the minimum prices that the two may charge, clearly the competitor is at a great competitive disadvantage. Another way to look at this is that the incumbent could charge its subscribers up to Rs. 35 per minute for long distance service (making a profit of up to Rs. 15 per minute) without losing any business to its competitors. Either way, there is no avoiding this apparent competitive advantage to the incumbent because only the incumbent controls interconnection to the desired subscribers. How could regulation overcome this problem?
The ECPR principle answers this question why first gathering the "facts":
Price charged by incumbent to competitor for interconnection = Rs. 15/min.
Price charged by incumbent to itself for interconnection = 0
Incremental cost to incumbent of providing interconnection = 0
Incumbent’s profit from selling interconnection to competitor = Rs. 15/min.
Incremental cost to incumbent of providing long distance service = Rs. 20/min.
Incremental cost to competitor of providing long distance service = Rs. 20/min.
Total incremental cost to incumbent = Rs, 20/min.
Total incremental cost to competitor = Rs. 35/min.
Assume market price for total service (long distance + intercon) = Rs. 50/min.
Incumbent’s profit from total service = Rs. 30/min.
Competitor’s profit from total service = Rs. 15/min.
These numbers depict the competitor’s plight. Because the incumbent makes Rs. 15/min. more in profit, it could simply squeeze out its competitor by lowering its retail price by, say, Rs. 16/min. and forcing the competitor to match that reduction, still make a profit, and restore its monopoly advantage.
Obviously that Rs. 15/min. profit differential (exactly equal to the profit it makes on the interconnection it sells to its competitor) is crucial. Now, if the regulator were to require the incumbent to impute (i.e., add to) the profit differential to its price floor for the total long distance service (which is Rs. 20/min. above), the unfair advantage would be entirely removed. The imputation rule here is very simple: When setting its price for the retail service (total long distance), the incumbent must add to its incremental cost for the retail service the profit it earns from its wholesale service (interconnection). It may not charge a retail price for long distance service that is below this level. There can then be no price squeeze and an equally efficient competitor will be able to compete.
There is another way to understand this. Under this rule, the incumbent’s profit from the wholesale service (interconnection) = Rs. 15/min. - Rs. 0/min = Rs. 15/min. is equalized with its profit from the retail service (total long distance) = Rs. 50/min. – (Rs. 20/min. + Rs. 15/min. imputed) = Rs. 15/min. This is no accident. The rule is precisely that the incumbent’s profit from its retail service must be at least equal but never less than its profit from the essential wholesale service.
The underlying principle is this: the incumbent has a trade-off between selling the retail service or selling the wholesale service. If it sells a minute of long distance to a subscriber, it obviously precludes its competitor from doing so and the competitor never buys that minute of interconnection. On the other hand, if it loses a minute’s sale of long distance to its competitor, it ends up also selling a minute’s interconnection to the competitor. To eliminate any unfair competitive advantage from its control over interconnection, the ECPR rule asks to leave the incumbent financially indifferent between the two situations, i.e., allow it to earn the same profit from either event. The example above shows exactly how this happens. This underlying principle is akin to the concept of opportunity cost. The incumbent’s price of long distance service should at least equal the incremental cost of long distance plus the lost opportunity or profit sacrificed from not selling interconnection to the competitor. Similarly, its price for interconnection should at least equal the incremental cost of interconnection plus the lost opportunity or the profit sacrificed from not selling long distance to the subscriber.
The biggest difficulty various parties have had with the ECPR rule for determining the price of interconnection is determining the size of that "lost opportunity" or the profit sacrificed. Recall that under ECPR, the following formula must apply:
Price of interconnection
= Incremental cost of interconnection + profit from retail service sacrificed
According to this formula, the larger the profit from the retail service gets, the higher the price for interconnection can get. Objectors to ECPR argue that because an incumbent operator enjoys a near-monopoly position in the market, it can control the price of the retail service enough to the point that it makes very large profits. That profit, in turn, can get fed into the price for interconnection and work to the disadvantage of the competitor seeking interconnection.
This objection is misplaced on, at least, two counts. First, if the retail market is competitive (even if interconnection is not), then it is questionable that profits in that market can get arbitrarily large. In fact, the stronger the retail competition, the smaller the lost profit element gets in the price of interconnection. Second, the objection seems to be to how high the absolute level of price for interconnection can get under ECPR. For competitive parity, the absolute level of price for the essential wholesale service does not matter. Rather, it is the margin between the retail price and the wholesale price that matters. If the price of the wholesale service exceeds that margin, then a price squeeze is possible. As long as ECPR prevents that from happening, the incumbent can derive no unfair competitive advantage. Many of the other criticisms of the ECPR are basically derived from the objections discussed above and are, therefore, not well supported in their own right. However, alternatives to ECPR such as cost-based charges (i.e., incremental cost + reasonable profit) are well established. Ironically, such cost-based charges are likely to differ little from those based on ECPR.
The FCC’s view of TELRIC was fraught with several problems. First, it required isolating the demand for each UNE in its entirety. This is never easy to do because the demand that is typically measured is that for retail services, not network elements. Often services can be put together with different combinations of elements, making estimation of even the derived demands for those elements very difficult. Second, the FCC proposed using the current volumes of network elements being used as the demand for UNEs. It then proposed to calculate incremental costs for those UNEs assuming demand to be given by those volumes. This, too, is problematic under competition. With entry by retail competitors, the volume (and market share) of the incumbent operator’s retail service may change. This, in turn, would mean that the volume of UNEs in use may most likely decline. With shrinking volumes (at least initially), certain economies of scale may well be lost. This would mean that the actual cost to provide UNEs under competition could well be higher than that calculated assuming current demand levels would persist. There are several other practical implementation problems with TELRIC (including their treatment of shared and common costs) which renders at least the FCC’s view of it highly suspect. For example, whose costs matter—the incumbent under efficient operation or a fictitious network? How should contributions to shared and common costs be factored in? Should price be set at TELRIC, or only be based on it?