In 1993, INTELSAT began to study its markets and operations with the long-term objective of making fundamental changes to its business structure. Many observers, including INTELSAT itself, view the organizational rigidity imposed by INTELSAT's structure as an international treaty organization as hampering INTELSAT's expansion into new markets. The current range of restructuring options considered by INTELSAT includes minor structural changes (allowing multiple signatories, or de-linking investment and use), formation of subsidiaries (both regional and commercial), and privatization (into either a single entity or multiple entities).
Restructuring proposals, however, raise the fundamental question of whether it is appropriate for an entity such as INTELSAT, with its close links to national governments, to expand into new services outside its core mission. Indeed, perhaps the reverse is true; perhaps INTELSAT should shrink rather than expand its size and scope of activities. This paper addresses this question from the broader perspective of global welfare, not from that of INTELSAT's interests alone.
INTELSAT was created largely by national governments and government owned telecommunications monopolists. We begin with the premise that services should be provided by competitive private entities wherever possible. Government intervention should be confined to those services where there is a clear market failure -- principally natural monopoly, or public goods or other strong positive externalities. Do such conditions characterize INTELSAT's markets today?
Section I of the paper reviews INTELSAT's original mission and history, noting that INTELSAT was formed to provide global telephone links via satellites. Whether or not private firms could have adequately performed this task at the time, section II notes that dramatic changes have taken place since then, especially the growth of fiber optic cable as a substitute to satellites for providing telephone links, and the emergence of private satellite providers. The main barriers now to expansion of private providers seem to be artificial barriers. These stem largely from the special status enjoyed by INTELSAT and include: its ownership of scarce geostationary (GEO) slots, its tax preferences and immunities from competition laws, and, probably most important, the restriction of access to national markets imposed on INTELSAT's potential competitors by governments or their telecommunications monopolists. Section III identifies some of the potential inefficiencies from retaining the privileged status of INTELSAT, and from allowing it to expand into new services. Section IV concludes that it is time for a fundamental reexamination of what is the proper role of INTELSAT in today's world. Such a reexamination should not occur in isolation, but in the context of a broader move towards liberalizing international telecommunications markets so as to permit increased reliance on competitive forces.
The typical Signatory is the dominant telecommunications service provider in that country. With a few exceptions, this is a ministry of the government, such as the Ministry of Post, Telegraph, and Telephone (PTT). Thus, in most countries (86% according to one estimate), the same entity participates in INTELSAT and controls access to the local telecommunications networks. Moreover, while governments (i.e., Parties) in principle have the ultimate oversight authority, in practice this oversight may be weak: in most cases (92% according to one estimate) Parties have assigned the Signatory as their accredited representative in the Assembly of Parties.
2. Financing and Compensation Arrangements
INTELSAT Signatories make capital and pay space segment charges to finance ongoing operations. Capital contribution and compensation are proportional to a Signatory's investment share, which is based on utilization of INTELSAT services and can change over time. INTELSAT sets prices to achieve a target rate of return on capital, chosen by its governing body. From 1973 to 1991 the organization's cumulative rate of return was 15.9%. The target return on INTELSAT was set at 20% in July 1994. INTELSAT has recently earned returns in nearly this high. These returns were paid out to Signatories in proportion to capital invested.
The total rate of return on INTELSAT investments by Signatories probably exceeds the direct compensation for use of capital (i.e. 15.9% annual average rate of return over 1973-1991) because, in most countries, Signatories also sell satellite services to end users, often earning a profit margin also on these sales. These pricing issues are discussed further in Section III.
1. Competition From Fiber Optic Cable
Technological change has produced dramatic increases in the carrying capabilities of fiber optic cables, and allowed them to begin to underprice satellites in providing connections for point to point traffic over the international routes with the largest traffic streams. Moreover, once fiber optic cable is laid, its operating costs are probably no higher than those of satellite transmissions. Fiber optic cable can carry the same transmissions as satellites and is technically superior to satellites for interactive service (e.g. voice) because satellite transmission involves an echo delay. Accordingly, satellites are losing to fiber optic cable much of the communications services that can conveniently be transmitted by cable, especially telephony.
2. Increasing Competition in the
Market for Satellite Services
For many telecommunications services, fiber optic cable transmission is unlikely to become a good substitute for satellite services, at least in the near term. (i) Telephony in remote areas. Fiber optic cable does not reach all portions of the world. While cable will soon connect most major cities in the developed world and most major seashore communities in the less developed world, the interior of less developed countries (and perhaps also smaller cities and rural areas in developed countries) is unlikely to be penetrated by fiber optic cable for many years. (ii) Provision of broadband services to individual users. Even when fiber reaches a region, the circuit must be completed to individual users via the local telephone system. Those "last miles" are commonly ill-suited to carrying broadband (e.g. video and some data) services which require a capacity for data movement that commonly exceeds even that of current fiber optic technology. Consequently, satellite communications will likely be preferred for broadband transmissions until technology improves fiber's ability to carry broadband services. (iii) Point-to-multipoint service. Satellites may have a significant cost advantage over fiber optic cable for point-to-multipoint broadcast or dispatch services because satellites eliminate the need for multiple transmission. In sum, fiber optic cable is not always a good substitute for satellites.
Before concluding that there is a continuing need for INTELSAT, however, one should consider whether provision of satellite services is a natural monopoly. The answer seems to be no. Economies of scale in ground stations have decreased. Initially, only powerful, expensive earth stations were able to receive the weak signals transmitted by satellites. As satellite signal strength has increased, this has allowed the size of antennas to shrink; also, the price of transmitting and receiving equipment has fallen. As a result, it has become cost-effective for a greater number of users to own "earth stations" (sometimes these are antennas only 18 inches in diameter that attach to the side of a house) to serve their communications needs.1 Also, while INTELSAT in the early days might have possessed some unique know-how about operating a satellite system, such know-how is now more widespread, allowing for multiple providers of satellite services.
Independent satellite systems are carrying traffic once handled almost exclusively by INTELSAT. One satellite services provider, PanAmSat, is likely to compete globally with INTELSAT soon in broad band services (with about one-fifth of INTELSAT's capacity). Two other providers, Orion and Columbia, are soon likely to cover 2/3 of the globe. In addition, regional commercial satellite systems and governmental satellite systems compete with INTELSAT in those national or regional areas that they serve.
Thus, examination of the natural monopoly and public good aspects of international telecommunications suggests that, at least with respect to INTELSAT's original core functions, these potential reasons for government intervention have declined over time. Correspondingly, public authorities should consider reducing the scope of INTELSAT's activities to areas where either natural monopoly or public goods arguments are valid, and encourage reliance on competition in the remaining markets. We discuss issues of universal access further in section III.
1. Ownership of Scarce Geostationary Orbital Slots
There are approximately 180 geostationary orbital slots in any satellite frequency band, though the precise feasible number depends in part on system technology and the radio spectrum the satellite employs. Currently, INTELSAT has registered plans for 31 slots and operates satellites in 23 of them.
Technical coordination for new entrants in GEO orbital slots is becoming increasingly difficult as a greater number of nations and regions launch satellites to serve their needs. A satellite system using GEO orbits requires approximately three properly positioned satellites to achieve global coverage. The scarcity of slots over particular regions, such as the Atlantic Ocean, would likely create the greatest difficulty for a new entrant seeking to provide world-wide service in competition with INTELSAT. Despite these concerns, it appears that potential INTELSAT competitors have been able to obtain the necessary GEO slots to operate their networks.
But entrants must coordinate their positioning and operations with incumbents to avoid interference. INTELSAT's widespread presence gives it the potential ability to delay entry by dragging its feet in negotiations and, perhaps more importantly, the negotiations over "coordination" might give it insights into competitors business plans.2
2. Denial of Equal Access by Governments or Signatories
Denial of access on equal terms to national markets may well be the primary remaining barrier to entry for private satellite organizations. In general, a license is required for a satellite services provider to gain access (i.e. acquire landing rights for its signal) to a national market. In addition, to provide switched-voice telephone service, a potential competitor must also obtain access to the national public switched network (PSN), a network often controlled by the INTELSAT Signatory -- typically the dominant telecom provider.
3. Cross Subsidization, Tax Preferences,
INTELSAT's tax preference and certain immunities (e.g. from restrictions imposed by competition laws) give it additional artificial advantages over private potential competitors. In addition, INTELSAT Signatories typically operate also in regulated markets where they are subject to constraints on profits, and may have incentives to circumvent such constraints by cross subsidizing their operations in unregulated markets (thereby shifting costs to the regulated markets and "justifying" higher rates there). Such cross subsidization gives an artificial advantage over competitors, as discussed further below.
INTELSAT, the argument goes, mitigates such overpricing by establishing some "vertical integration," which gives the monopolists over ground segments a direct stake in satellite profits and vice versa. This helps internalize some of the negative externality between pricing of ground and of space segments as compared with having an independent monopolist on the satellite segment. There remains a substantial overpricing incentive due to the fact that there are independent land-facility monopolists in the different countries. Conceivably INTELSAT helps mitigate this problem too, by providing a convenient forum for reaching bilateral quid-pro-quo arrangements whereby both countries reduce access charges. Correspondingly, things could be made worse not better by restructuring INTELSAT to delegate provision of satellite services to independent entities that still possessed considerable market power. Such market power might persist in the short run if satellite capacity ownership and rights to GEO slots remain concentrated.
The goal, however, would not be to replace INTELSAT by private satellite providers with market power, but by competition. Competition would eliminate the need for vertical integration between earth and space segments (integration that INTELSAT only approximates) for the purpose of reducing prices of satellite services.
The problem remains of how to mitigate the over-pricing incentives arising from the existence of separate ground-based monopolists in some countries. The first-best solution is to foster competition in those markets wherever possible. Advances in technology are making competition possible in many sectors once thought to be natural monopolies; often the main remaining barrier consists of government regulations. Some markets -- especially smaller and less densely populated ones -- may remain natural monopolies. The proper policy response there, however, is national regulation of the access fees and terms that such monopolies charge for interconnection to satellite services, but in a manner that assures a globally efficient utilization level.
Reaching such a solution would likely require agreements among governments, agreements that may not be easy to reach.3 But such access agreements would strike at the heart of the major barrier to competition in telecom services more broadly.
Indeed, some would argue that introducing competition into satellite services alone is only a part, perhaps a small part, of the solution. Broader liberalization of telecom access terms is desirable. We share this view. Also, some might argue that there will be transition costs, as private providers learn to make alternative interconnection arrangements to replace those currently provided by INTELSAT. But this is only a transition problem; private providers have proven adept at putting together seamless networks in other contexts. Therefore we should not let such considerations distract us from moving in the right direction, of increased reliance on private sector competition wherever possible, unhampered by regulatory and other governmental distortions.
The argument that cross subsidies are justified as an efficient way to provide global telecom services to developing countries requires careful scrutiny. The cash-equivalent value of such subsidies is probably relatively small; unrestricted direct aid equal to this amount would seem to be a superior route, as it would avoid potentially large distortions to the global telecom industry.
An argument against such unrestricted cash assistance is specific egalitarianism: according to this argument, assistance-in-kind is superior for certain critical goods or services about whose value the recipient is less informed than the donor. It is difficult, however, to make a serious case that developing countries do not recognize the value of global connectivity. With or without cross subsidies, it is likely that most developing countries will find it worthwhile to become connected to global telecommunications. And while global connectivity would not necessarily be provided to all regions within a country, the policy choice of whether and how to encourage such additional internal connectivity would seem to be best left to countries themselves. Finally, to the extent one believes it desirable to encourage further connectivity beyond the level that LDCs would select, restricted cash transfers (earmarked for this purpose) are probably preferable to subsidies via a distorted telecom industry.
A different type of argument for subsidizing global connectivity to developing countries is based on the positive externalities created for users in developed countries: the value of a telecommunications network increases as the number of participants grows. However, private providers can largely capture ("internalize") such externalities by operating in all countries. Indeed, private telephone carriers provide services to numerous developing countries.
In short, advocates of the cross-subsidization rationale should articulate precisely and document which worthwhile services will go unprovided under competition. Evidence of the absence of private provision in the current regime has to be treated carefully -- many developing countries have never welcomed private foreign telecom providers. Hence, it is government regulatory policy rather than "market failure" which often is to blame for the absence of some services. It is quite possible that the truly "life line services" that would go unprovided in a competitive regime would be fairly limited and the "large INTELSAT" response is disproportionate to the size of the underlying problem.
To the extent that certain worthwhile services would not be provided under competition, the question remains whether it is better to finance such unprofitable services through opaque cross subsidies in the context of a quasi-governmental body like INTELSAT than through more transparent mechanisms. (As an aside, it is worth asking whether in fact INTELSAT currently cross subsidizes in this "intended" direction.) Many economists would argue that superior alternatives do exist or could be devised, including, perhaps the creation of a universal service fund targeted to particular countries or services, and which could be used to procure service from the most efficient providers.
The need to find alternatives to cross subsidies (assuming that these would be necessary) will in any case become more acute as competition in global telecom is likely to increase over time, making cross subsidies increasingly less tenable. Indeed, there is a certain inconsistency in some of the arguments in favor of an expanded INTELSAT. One the one hand, it is claimed that INTELSAT will not exercise monopoly power, and will not therefore interfere with the efficient functioning of these markets. On the other hand, it is claimed that the revenues from these activities are required for funding cross subsidization; bu if these new markets will be truly competitive, then, seemingly, there would be no rents to fund cross subsidization.4 We attempt to examine more carefully in subsection C below the likely direction of cross subsidization and resulting competitive distortions.
1. Limiting Access to Competitors
INTELSAT member nations (Parties) may have incentives to limit competing satellite systems' access to their national telecommunications networks in order to protect the profits of their INTELSAT Signatories. Recall that these Signatories are often government-owned PTTs or entities with other strong government affiliations (PTT for brevity). Conceivably, a government may have such incentives even if acting in the national interest rather than that of its Signatory, in those cases where the satellite services market would inherently remain imperfectly competitive for technological reasons. In such cases, denying access may yield more profit to the protected domestic Signatory than it reduces consumer surplus. Such rent shifting, however, harms foreign independent competitors and typically reduces global welfare. A more likely scenario is that denying access does not increase even national welfare, but only that of the PTT.
Severing the link between quasi-governmental entities such as PTTs and the provision of satellite services or other potentially competitive services is likely to reduce governmental incentives to deny access for foreign competitors. First, in many countries satellite services would likely be supplied entirely by foreign providers, and the absence of domestic producers would remove the incentive for inefficient protection. Such a scenario is consistent with the observation that many commodities are cheapest in countries that lack a domestic industry, and therefore refrain from protectionism and instead rely on free trade to deliver these commodities most economically. Second, to the extent that there will be private domestic producers, it is typically more difficult politically to justify protection of such entities than protection of nationalized entities. Of course, this argument suggests that it would be a bad idea to force INTELSAT out of a particular service but allow the PTT later to vertically integrate into such activities.
Indeed, the PTT or other bottleneck monopolist may have the incentive and ability to deny access for competitors even in contravention of government wishes. Certain satellite services, notably switched-voice telephony links, require physical interconnection to a local network controlled by the PTT monopolist. The latter can deny such links to competitors in ways not readily controlled by the less-informed government or its appointed regulators.
The wider the range of activities in which INTELSAT operates, and therefore in which the dominant telecom providers -- the INTELSAT Signatories -- have a financial stake, the greater is the scope for denial of access to competitors, either with or without the assent of national governments.
Correspondingly, expanding INTELSAT's range of activities may not be in the global interest.
2. "Regulatory Evasion" through
Operations in Other Markets
A second class of distortions could arise from the fact that PTTs often are constrained, whether explicitly or implicitly, in the terms they can charge for core monopoly services. Participation in other, potentially competitive markets, such as through INTELSAT could allow them to circumvent such regulation while introducing several inefficiencies.
Cross Subsidization and Discrimination: If the PTT is subjected explicitly or implicitly to cost-based regulation, it has incentives to evade regulation by cross-subsidization of other, unregulated businesses. Some costs of those businesses would be misattributed to "core" regulated functions, and used to justify higher prices there. Such cost shifting would have two deleterious consequences: it would increase prices in the "regulated" markets, thereby aggravating the monopoly-pricing distortions; and it would distort competition in the unregulated, potentially competitive segment, reducing the market share of potentially more efficient independent producers.
Observe that the incentive is to cross subsidize the potentially competitive (unregulated) segments from the "core" regulated natural monopoly segments. This is important, because a main defense of a large INTELSAT is the need to cross subsidize in the other direction, i.e., to use revenue from unregulated sectors to fund deficits in "lifeline" natural monopoly segments.5
Incentives for cost shifting/cross subsidization arise primarily under cost-based regulation, and are less pronounced under other forms of regulation such as price caps.
However, even in such a case there is a danger of allowing the regulated entity to operate in unregulated but technologically related markets. The monopolist could attempt to evade price cap regulation on its core segment, and mandated interconnection terms for competitors in the related markets, by practicing technological discrimination against rivals, of the sort that regulators find difficult to police. Denying access to competitors would then enable the monopolist to extract its profit in those unregulated markets, as AT&T was accused of doing before its 1984 breakup when it allegedly offered only poor connections to long-distance competitors such as MCI in order to steer businesses towards its own long distance operation. Like cost shifting, such technological discrimination against competitors will distort competition in the potentially competitive related markets. But it is more likely than is cost shifting to raise rather than lower prices in the adjacent markets, because the rival's ability to connect is impaired.
Several restructuring proposals being considered by INTELSAT include provision of both new services and traditional services from within INTELSAT. Proposals range from creation of a wholly owned commercial subsidiary that could enter new markets, to full privatization of INTELSAT as a single entity; but all raise the specter of cross-subsidization of new services, or of technological discrimination against competitors.
Technologically Unrelated Markets: Misuse of Immunities, Tax Privileges, and Risk Shifting: Under its existing privileges and immunities, INTELSAT is exempt from paying taxes. In effect, this is a subsidy to INTELSAT to provide global connectivity. Because it is not taxed on earnings, INTELSAT's effective cost of capital is lower than that of its private competitors that pay taxes. It could use this advantage to subsidize its entry into new markets. By discouraging the development of competition in new services, INTELSAT's actions would allow its Signatories to earn abnormal profits in these new markets. In such cases, the inefficiency stems from providing INTELSAT a subsidy larger than is required to achieve the original targeted goal of global connectivity.
Another potential distortion, which shares the flavor of regulatory evasion, can arise from the fact that provision of new services provided by INTELSAT will carry a combination of technical risk and competitive risk. (For example, a service called Satellite Business Service, formed by Comsat, IBM and Aetna Life in 1975, lost over $500 million between 1979 and 1984, and was eventually sold.) Permitting INTELSAT to operate in unregulated markets could provide excessive incentives to invest in risky ventures. If a new service succeeds, INTELSAT and its Signatories keep the profits. But, if the new service fails, INTELSAT might be forced to raise rates for its traditional services to pay for the losses. Many PTTs rely on INTELSAT completely for interconnection to international and might be able to argue to their national regulators have no choice but to pay the higher rates. This asymmetric incentive structure, of unlimited upside potential but limited downside risk, creates incentives to make risky investments, similar to the incentives of savings and loans in the United States in the 1980s. Consequently, if regulators cannot police where INTELSAT's investments are going and are the investor of last resort, INTELSAT and its Signatories would have incentives to game the system and enter markets for riskier services.
The arguments presented here do not preclude the possibility that INTELSAT could spin off an efficient, competitive affiliate. They do, however, indicate that the form of any residual links between the two (or more) entities will determine whether the restructuring can meet the competitive objectives that we believe are of paramount importance.