There are things known, there are things unknown, in between are doors. Jim Morrison ( ) Access to telecommunications networks

Transcription

1 There are things known, there are things unknown, in between are doors Jim Morrison ( ) Access to telecommunications networks Marcel Canoy*, Paul de Bijl**, and Ron Kemp*** * CPB Netherlands Bureau for Economic Policy Analysis, The Hague ** CPB Netherlands Bureau for Economic Policy Analysis, The Hague, currently at the Ministry of Finance *** EIM, Zoetermeer Paper prepared for European Commission, DG Competition Preliminary version September 2002

2

3 Contents 1 Introduction One-way access Introduction Access pricing First-best solution Ramsey pricing ECPR Dynamic pricing rules Backward-looking access pricing rules Forward-looking access pricing rules Two-way access Introduction Access pricing Dynamic considerations Policy Issues Fixed-mobile termination Policy relevance Economic theory Practical experiences Policy conclusions Margin squeeze Policy relevance Economic theory Practical experiences Policy conclusions The allocation of common costs Policy relevance Economic theory Practical experiences Policy conclusions Static versus dynamic efficiency Policy relevance Economic theory Practical experiences Policy conclusions Conclusions References

4 1 Introduction From the very beginning of liberalisation operations in telecommunications markets, access has been a key issue for regulators. Despite some notable successes in various submarkets, even in 2002, more than ten years after the first liberalisation steps, incumbent operators still have strong positions in many aspects of their business and infrastructure competition has not matured in all parts of the industry. Most notably, there has not been much infrastructure competition in the local loop and the unbundling of the local loop progresses slowly (Buigues, 2002). As a result, incumbent operators owning local loops may have the opportunity to engage in anti-competitive behaviour, e.g. by providing access against unfavourable terms. Because ex-post application of the Competition Law can turn out to be less effective in this type of situations, regulators face the task of deciding on setting the terms and conditions of access charges. Although the legitimacy for regulators to intervene in access charges is easily understood, there is considerable difficulty in determining how to set these charges in practice. There are a number of reasons for that. First, economic theory is not always clear-cut on providing optimal ways of setting access charges. Second, even in cases where theory is univocal, practice is not as easy as theory predicts. The difficulty of daily regulation business is not just the usual cliché. Admittedly, the usual problems exist here: required data are often simply not available and time is lacking. But on top of that, access regulation often serves too many goals. Access charges are used to allocate scarce capacity, provide incentives for productive efficiency, promote entry, promote investments by incumbents as well as by entrants, taking a fair allocation of common costs into consideration and meanwhile serving equity goals such as universal service obligation. This is a bit much for an access charge, which is, in its simplest form, a one-dimensional price. A number of these goals can easily conflict. Low access charges can be good for stimulating competition in services, but might hamper investments in infrastructure. The minimum any regulator has to do is to make sure ex ante which goals it wants to reach with the access charge, and, in case of conflicts, which goals have priority and why. Sometimes smart regulation is possible to reach seemingly conflicting goals: an access charge that starts low but rises over time can both promote services competition in the short run (enhancing static efficiency) and provide incentives for entrants to invest in infrastructure. But more often than not, choices have to be made

5 This chapter overviews the theory of access charges, with a clear policy perspective in the back of our minds. Following Armstrong (2001), we distinguish between one- and twoway access. 1 In a situation of one-way access, there is an entrant who needs access to an infrastructure owned by an incumbent operator, but the reverse is not true. Examples are Unbundled Local Loop and Carrier Select. Two-way access refers to network interconnection, that is, operators mutually need access in order to terminate calls on each others networks. We shall see that the policy implications in one- and two-way access can be quite different. Determining access prices implies taking both short and long run into consideration. Both economic theory and regulation practices (in particular the early days) had a strong focus on static efficiency (i.e. the short run). However, dynamic considerations are important and should not be neglected for a number of reasons, one of which is that if competition in infrastructures matures less regulation is needed. That is why this chapter devotes extra attention to dynamic considerations. The main sources used for this paper are surveys of the economic literature on telecommunications by Armstrong (2001) and Laffont and Tirole (2000), and policy-oriented research by Cave et al. (2001) and (to a lesser extent) Bennett et al. (2001), as well as many policy documents, most notably from the European Commission and OFTEL. This paper is organised as follows. Section 2 discusses one-way access. Section 3 discusses two-way access. Section 4 analyses four special topics, i.e. fixed-mobile termination, margin squeeze, the allocation of common costs and static versus dynamic efficiency. Section 5 concludes. 2 One-way access 2.1 Introduction One-way access is characterised by a vertically integrated incumbent, usually the former state-owned monopolist, with a local access network. There are one or more entrants who do not have such a local network, nor are they able to build one in the reasonably short run. 1 Armstrong also mentions competitive bottleneck as a separate category, but we prefer to discuss competitive bottlenecks in the policy chapter

6 Entrants may have their own long-distance network, but this is not necessary 2. However, in order to compete with the incumbent in the retail market for voice telephony and replicate the pattern of offers of the incumbents, they need access to the incumbent s local access network. One-way access is common in the early phase of competition, when entrants have not yet been able to roll out (part of) their own local connections, but is also relevant in mature stages of the market. An important example of one-way access is Unbundled Local Loop (ULL): an entrant gets access to the copper-cable pairs of the incumbent s local network and gets control over the use of (parts of) the total frequency spectrum of the copper-cable pairs. The main implication of this is that it enables the new network operators to offer high bandwidth directly to consumers and henceforth faster competition for high-bandwidth services. 3 Why does ULL get so much policy attention? First of all, ULL allows for a direct and comprehensive relation with the end-consumer without the need of a full rolled-out network. It also creates pressure on operators to offer consumers a whole new range of services, that uses (parts of) the total bandwidth, such as fast Internet, receiving richer content such as video on demand, etc. Although there might be alternative technologies that can provide broadband services, such as cable and UMTS, ULL seems to be deployed most rapidly. At the moment, the roll-out of broadband services (e.g. ADSL) is relatively limited (Buigues, 2002). This might have to do with difficulties new entrants face: behaviour of the incumbents such as excessive pricing, delays in delivery, predatory pricing or price squeezes and refusals to supply the necessary information or space in the incumbent s locations. All these aspects deserve the attention of regulators. Carrier Select is another example of one-way access: an entrant has originating and terminating access to the incumbent s local network. By dialling a prefix (usually consisting of four digits), consumers can indicate that they want the entrant instead of the incumbent to carry a telephone call. After some starting problems and other topics that need attention (such as margin squeeze, or scarcity), this type of access seems to work quite well. Because of the asymmetry in the incumbent s and entrants bargaining positions, as well as in their interests, it is very unlikely that they are able to agree on the level of the 2 Entrants without their own backbone may lease long-distance capacity from the incumbent. Since the incumbent may have spare capacity, this may be in its interest. 3 Only the leasing part of ULL is one way access. The interconnection part is two way access. The interesting regulation issues are however with the leasing part

7 access price. In particular, the incumbent has a strong incentive to set the access price as high as possible, perhaps even to foreclose entry, whereas entrants prefer cheap access. 4 Hence, the interests are strongly opposed, so that regulation of the access price is necessary. In particular, the central question concerns the optimal access policy. 2.2 Access pricing First-best solution The first-best solution, that is the theoretical solution if there are no imperfections, is to set the access price equal to the marginal cost of access. With such an access price, there are no distortions in retail prices, and entrants receive correct signals: they can make positive profits only if they are more efficient than the incumbent. Also, this access price is fair and nondiscriminatory, because it is the same for all entrants and it is not usage-based. In this firstbest world, the incumbent s fixed cost of its network is covered by a lump-sum payment from the state to the incumbent. The first best can rarely (if ever) be implemented in practice. An obvious problem with the first-best solution is that lump-sum transfers from the government are usually not feasible, at least not without creating large distortions elsewhere in the economy and seriously impeding the incentive structure of the incumbent. Therefore, one has to look for an access price above marginal cost, that helps to cover the incumbent s fixed network cost Ramsey pricing The practical problems of setting access prices equal to marginal costs, create a necessity to introduce mark-ups to enable the incumbent to recover its network costs. On a more general level, one can try to set not only the incumbent s access prices, but also its retail prices such that welfare is maximised, subject to the constraint that the incumbent recoups its fixed cost. Note that welfare, defined as the sum of consumers and producers surplus, here refers to static efficiency. The solution to this problem is known as Ramsey pricing. Put differently, the problem is to find the price structure for a multi-product (i.e., access and retail services) regulated incumbent that maximises welfare, given that overall, the incumbent has to break 4 If the incumbent wants to increase traffic on its network, there need not be a conflict of interest (cf. virtual

8 even. A priori, Ramsey pricing implies that there should be mark-ups in the incumbent s access prices as well as retail prices. The central idea is that to maximise welfare, all the incumbent s prices (wholesale and retail prices) should participate in the recovery of fixed costs. In particular, the optimal access price will be equal to the marginal cost of access plus a Ramsey term (i.e., a specific mark-up). By construction, Ramsey prices (wholesale and retail) reflect underlying costs as well as demand characteristics (see Laffont and Tirole, 2000, section 2.2). Hence, Ramsey prices are at the same time cost-based and usage-based. This implies that they are, in principle, compatible with a firm s standard marketing practices, since the latter also takes costs and demand into account. More precisely, it can be shown that welfare-maximizing prices are obtained if the firm maximizes its profits under the constraint that it offers a certain minimum level of surplus to its customers (the "Ramsey-Boiteux" social welfare level). Note that the regulator is supposed to have full information about cost and demand characteristics. Intuitively, if the price elasticity for a specific service is relatively low, then the mark-up in the price for this service can be relatively high. This is optimal for welfare, since the prices of services with higher elasticities can be reduced while still satisfying the incumbent s cost recovery constraint. For instance, it may be optimal to increase the access price above marginal cost in order to reduce retail prices. In general, mark-ups can be higher when they lead to less distortion in the optimal allocation. It is important to remark that there are potential problems with Ramsey pricing (see Laffont and Tirole, 2000, section 3.4). First, the informational requirements of Ramsey pricing are very high. For instance, Ramsey prices require the regulator to have knowledge about cost levels and demand elasticities, information which the regulator usually does not have One should, however, not rule out the possibility to obtain reasonable estimates of these data. Alternatively, the regulator can delegate pricing decisions to the better-informed firm, of course within the constraints imposed by Ramsey pricing. To see this, notice that unregulated firms are usually able to use fine-tuned and sophisticated pricing tactics, which suggests that they have much more information than regulators do. In particular, pricing behaviour tends to reflect cost levels, elasticities, competitive pressure, and so on. Indeed, it can be shown that the structure (not the levels) of an unregulated monopolist s prices is the same as the structure of Ramsey prices. Therefore, a regulator may obtain the Ramsey pricing structure by mobile operators)

9 imposing a global price cap such that (i) access is treated as a final good and included in the price cap, and (ii) the weights of the price cap are exogenously determined based on forecasted quantities. If the weights are set appropriately, then the operator internalises net consumers surplus when it maximises its profits. Although the principles for setting the optimal weights are rather straightforward, the information that is needed (about forecasted quantities) depend on actual costs and demand elasticities and may therefore be difficult to obtain (see Laffont and Tirole, 2000, section 4.7). Nevertheless, this is much less demanding than obtaining the information needed to set an optimal partial price cap. Second, usage-based access prices are discriminatory access prices, and therefore seem to contradict the policy principle of non-discrimination. More precisely, Ramsey pricing prescribes that the charge paid by an entrant must depend on the use of the service. The access price typically depends on the incumbent s price-cost margin in the relevant retail market, demand-side substitution possibilities, supply-side bypass possibilities, and the elasticity of the demand for access. For instance, customers of services that are not very price sensitive contribute more to cost recovery. From an economic perspective, this dependence simply maximises welfare: access prices should be higher when they are used for services for which the demand is less elastic. However, applying Ramsey prices can imply a (very) skewed distribution of prices (see e.g. Jullien 2001) and hence the political feasibility is often dubious. Overall, the problems mentioned above are, from an economic perspective, not necessarily worrisome for policy makers. By construction, Ramsey pricing is the best way to set access and retail prices simultaneously, that is, it is the least-bad departure from firstbest prices. At least, if application turns out to be prohibitively difficult in practice or politically too unattractive, Ramsey pricing can (or should) serve as a useful benchmark for access regulation ECPR Whereas Ramsey pricing aims at choosing optimal access and retail prices, it may be the case that the problem of access regulation is separated from retail pricing. In the latter context, the Efficient Component Pricing Rule (ECPR) is a popular pricing rule, since it provides a link between the access and retail

10 ECPR, which is also known as the Baumol-Willig rule, has a background in the theory of contestable markets. 5 A retail market is said to be contestable if there is potential hit-andrun entry which constrains the incumbent s retail price. That is, the threat of quick entry disciplines the incumbent s pricing behaviour. An underlying assumption is that potential entrants take the incumbent s price as given. Therefore, the idea is that for a fixed price charged by the incumbent, a more efficient firm can enter and take over the market; there is no in-market competition. Another assumption underlying ECPR is that access price regulation is separated from price setting in the retail market. In particular, retail prices are fixed beforehand by the regulator. Therefore, ECPR s prescription is to choose the access price that maximises welfare given the incumbent s retail prices. The regulator is not concerned with overall welfare maximisation, as in the case of Ramsey pricing, but aims at cost recovery and productive efficiency. Assuming that final products are homogeneous (which seems a realistic assumption for voice telephony) and the market is contestable (which is a controversial assumption, as discussed above), ECPR prescribes that access price should not be larger than the incumbent s opportunity cost, which is equal to its marginal cost of access plus its missed retail mark-up. Equivalently, one can state that the access price should not be larger than the incumbent s retail price minus its cost in the competitive retail activity. An attractive feature of ECPR is that entrants receive correct signals, that is, they enter only if they have a cost advantage. Also, it is sometimes argued that another positive feature is that entry is revenue neutral for the incumbent. But it is not clear why one should want to have revenue-neutrality. If the incumbent s profits are excessive, they will remain so under ECPR also if market entry is possible.ecpr and Ramsey pricing do not generally coincide, although this divergence can perhaps be restored if one makes specific assumptions about symmetry and absence of entrants' market power.an important difference is that ECPR neglects that the wholesale market and the retail market are related. ECPR is a partial rule, in contrast to Ramsey pricing. If retail prices are regulated at Ramsey levels, then ECPR is optimal. 5 See Tirole (1988), chapter

11 2.3 Dynamic pricing rules The pricing rules for access, discussed in the previous subsection were based on the implicit assumption that firms do not invest or develop new technologies. The purpose was simply to compensate the incumbent for providing entrants access to its network, in a way that its fixed investments can be recouped. Since telecommunications markets are changing very rapidly exactly because of investments and technological progress, dynamic considerations should also play a role in access regulation. In particular, it is important to note that any access price affects operators (potential) profits, and hence also their incentives to enter the market, to invest in new technologies, to roll out networks, to maintain and upgrade existing networks, and so on. To deal with these types of dynamic issues, regulators have come up with specific access pricing rules, which are discussed below. More generally, when assessing the effects of access regulation on firms incentives to invest, one has to distinguish between: the effects on entrants incentives to roll out networks themselves (versus using an incumbent s existing network), and the effects on an incumbent s incentives to maintain and upgrade its existing network. These effects depend on current as well as expected access prices. For instance, if the access price is low and firms expect it to remain low in the future, then the incumbent is not very eager to invest in its existing network, while entrants feel no urge to roll out their own networks. On the other hand, competition in the short run will be intense, as entrants can easily compete by using the incumbent s network. In the short run this is good for consumers, but it is uncertain if this is also the case in the longer run. If the access price is high and firms expect it to remain high, then it makes more sense for entrants to start rolling out their own networks, which in turn imposes discipline on the incumbent to invest as well, in order to remain competitive. However, in the short run, it is expensive for entrants to start building up market share (e.g., by starting with Carrier Select services as long as their networks are not yet ready), which may negatively affect consumers for a long time. The following subsections discuss the dynamic consequences of access rules

12 2.3.1 Backward-looking access pricing rules Backward-looking cost-based access pricing rules are based on the incumbent s actual or historical costs (also called embedded costs). An example is Embedded Direct Costs (EDC). A backward-looking rule can be called fair in the sense that the incumbent is compensated for its actual network investments. On an ongoing basis though, it may give weak incentives to reduce costs, since the implicit message is that any cost will be reimbursed. Also, backwardlooking access prices will be relatively high, compared to forward-looking prices (see below), which makes it harder for entrants without networks to compete with the incumbent Forward-looking access pricing rules Forward-looking cost-based access pricing rules are based on state-of-the-art, currently available technology. Hence, they explicitly take technological progress into account. If a new network can be rolled out at half the cost of the incumbent s network, than the cost of the new technology serves as the relevant benchmark. Accordingly, forward-looking rules incorporate cost efficiency; they can be used to correct for possible inefficiencies of the incumbent. Arguably, they can be used to mimic competition that is not yet existent. Because of the downward pressure on access prices, it is even possible that an incumbent s access service, using an existing or obsolete technology, becomes a loss-making activity. Hence, in theory forward-looking rules give the incumbent an incentive to keep up with technological progress, and to keep investing in its network. A downside of forward looking rules is that they tend to neglect so-called stranded assets. If an investment has a 50-50% chance of success, then the revenue that is needed in case the investment is a success should account for the chance this it would have failed. If access rules do not take this into consideration, this may lead to underinvestments and risk-averse behaviour. The main example of a forward-looking rule is Long Run Incremental Cost (LRIC). LRIC prices can be substantially lower than the incumbent s actual cost levels. However, it should be noted that LRIC can lead to substantially higher prices as well (e.g. due to increasing labour costs, better but more expensive technology), especially in the case of access to the local loop. LRIC does not seem to be well-suited to incorporate corrections for quality differences between old and new technologies. It may therefore better be suited for

13 interconnection fees (i.e., two-way access prices; see the next section) than for pricing access to the local loop. 6 Despite its intuitive appeal, LRIC has some further drawbacks (see also Laffont and Tirole, 2000, Leo et al 2002). For instance, note that the cheapest technology that is available cannot be derived from standard accounting systems. Hence the regulator has to determine how efficient the incumbent should be (i.e., the regulator does not determine the desired efficiency level, but can put a lower bound on it). Also, LRIC may not allow the incumbent to make a profit margin on access. If this happens, the incumbent has strong incentives to deny access to entrants by using non-price anti-competitive practices, such as refusals to deal, and delays in interconnection. Accordingly, a possible consequence of LRIC (and other forwardlooking rules) is that the regulator has to continue to play a key role in managing entry. In other words, LRIC may imply that regulation remains heavy-handed, which is in contrast to the plan to gradually withdraw regulation as competition matures. The disadvantages of LRIC have their origin in the implicit assumption that markets are contestable. Since telecommunication markets are typically not contestable, LRIC fails to take into account that assets can become stranded and that operators need risk premiums to take care of that possibility. It follows that applying LRIC without taking these dynamic considerations into account can be pretty disastrous. Summarising, forward looking rules such as LRIC are appealing for interconnection. For access to the local loop there are serious dangers, in particular it puts a lot of weight on the quality of regulation. Overall, one cannot draw a clear-cut conclusion on the dynamic considerations. The problem is that access prices, especially in situations of one-way access as we have seen above, often have to perform too many tasks at the same time. Those tasks are possibly conflicting, as indicated above. In the policy chapter 4 we will come back to possible solutions of this problem. 3 Two-way access 6 Applying LRIC to certain services (e.g., wholesale services) and backward-looking prices to others (e.g., retail services) might lead to inconsistencies

14 3.1 Introduction Two-way access refers to a situation where there are two or more operators with their own infrastructure (consisting of long-distance and local access networks), that need mutual links so that any consumer can call anyone else. These operators compete for subscribers to their networks and need each other to offer maximum network benefits to their customers (interoperability). The typical situation of two-way access is network interconnection. Another important example of two-way access is fixed-mobile terminating access: a network operator can charge a consumer for making use of the network (e.g. a two-part tariff). The network operator can also charge other parties to reach their subscribers. There is competition for subscribers but no competition for reaching the subscribers of a network. The characteristics of fixed-mobile call termination are that there is competition for mobile subscribers, but no competition for providing access to mobile customers once they subscribed to a particular mobile network. The main difference with network interconnection is that the services are provided to non-competing operators. For example, in fixed-mobile termination, the mobile operator needs to access the fixed -and conversely- but (to some extent), they are not exactly competing head-to-head against each other. Much of the theory that is discussed in this chapter applies to fixed-mobile termination. That is why we have decided to deviate from the categorisation introduced by Armstrong (2001) who has a separate discussion on competitive bottlenecks (of which fixed-mobile termination is considered to be the main example). Indeed there are a number of interesting and separate policy questions related to fixed-mobile termination. E.g. access to the incumbent s fixed lines was more heavily regulated than access to mobile. This creates some policy questions but does not necessarily warrant new theory. We will discuss the policy aspects of fixedmobile termination in chapter 4. Another situation of two-way access seems to be international network interconnection, which was already relevant before liberalisation started to take off. The crucial difference (and the reason why we do not consider this two-way access) with the examples above is that there is no competition for subscribers. This creates other types of problems, which we will not discuss here. The same applies to international roaming. If a subscriber roams on another network to enable its mobile to function abroad, the visited network operator bills the home network operator for this call and vice versa. Since

15 international roaming is, in principle, a competitive service, and possible problems related to it are beyond the access discussions of this chapter, we will not pursue this in this section. Similar to one-way access, the central question is how high the access price, in this case the terminating access price, should be. A difference is that by definition, there is always more than one access price each operator charges its own terminating access price. Hence, an additional question is whether terminating access prices should be reciprocal, that is, the same for all operators, or not. A third question concerns the so-called missing price. Because of the calling party pays principle (CPP), there is no price for receiving a call. This fact will turn out to be important for the discussion. 3.2 Access pricing There are two questions related to optimal terminating access prices. The first one concerns the optimal level of access prices, and whether they should be reciprocal. The main issue, at least in a mature market, is to set access prices that maximise welfare (or, alternatively, consumers surplus) such that total industry profits are sufficient to cover fixed costs. The second question is whether operators should be able to negotiate on access prices, or whether these prices should be regulated. The results in the literature on the optimal level of access prices in two-way access situations are somewhat scarce, although there are some recent contributions, for instance De Bijl and Peitz (2002 a and b). The main results in the earlier literature focused on mature markets. Nevertheless, a new entrant, starting from scratch and slowly building up its market share, finds itself in a much more difficult position with regard to the incumbent than vice versa. The large asymmetry in market shares leads to asymmetric traffic flows between the networks, and hence asymmetric access payments (depending on the access prices). A result in the literature (Laffont and Tirole, 2000, chapter 4) is that in a mature market with symmetric operators that compete in two-part tariffs, welfare is maximised by setting a reciprocal access price equal to the marginal cost of access. Operators optimally set perminute prices equal to marginal costs of calls, and recoup fixed costs on a per-customer basis through subscription fees

16 A central question in the literature on two-way access is whether the access price is an instrument of tacit collusion. Under specific assumptions, it can indeed be shown that operators profit levels are increasing in the level of the reciprocal terminating access price. The reason for implicit collusion is that high access prices increase the cost of a unilateral retail price cut. This is because such a price cut increases the net outflow of calls to the rival operator, which is costly because of the access mark-up. However, this result of collusion is not very robust. For instance, an underlying assumption is that there are symmetric operators who compete in linear prices, that is, in per-minute prices only and not in subscription fees. This is not very realistic, since entrants are usually quite small compared to the incumbent in the early phase of competition, and moreover, operators typically do not compete in linear prices but in two-part tariffs. Although also this result critically depends on the assumptions of the model (e.g., the demand for subscriptions is inelastic), it is important to note that the tacit-collusion result may easily vanish, depending on the assumptions of the model that is used. Overall, the risk of tacit collusion should not be neglected, but in an immature market, it seems less pressing than protecting small entrants from a high access price charged by the incumbent. On a general level, it seems safe to conclude that in a mature market with roughly symmetric operators, the socially optimal access price is equal to the marginal cost of access. Since operators who compete in two-part tariffs do no benefit from access mark-ups, it is not necessary to regulate the access price, at least not in such a stylised situation. In more realistic cases where operators are asymmetric, access regulation seems to be necessary in order to protect small entrants as well as to prevent a reduction in consumers surplus (See De Bijl and Peitz (2000, 2002a, 2002b), who explore the nature of asymmetric competition between a small entrant and a large incumbent). Moreover, non-reciprocal access prices may be very useful to stimulate competition in the short run. 7 In particular, if entrants are still small compared to the incumbent (in terms of market shares) while there is customer lock-in (e.g. because of switching costs, lack of quality track record for new operators), as in immature markets, it may be optimal to regulate the incumbent s access price at the marginal cost level 7 E.g., Armstrong (1998) and Laffont et al. (1998) address asymmetric competition between an incumbent and an entrant, but do not consider non-reciprocal access regulation. Laffont et al. (1998) do, however, show that non-cooperative access price setting by the operators, which allows for non-reciprocal access prices, may lead to very high (and inefficient) access prices

17 (possibly allowing for a modest mark-up that yields a reasonable return), while allowing entrants to charge an access mark-up. Such an asymmetric access mark-up increases the entrant s profits as well as consumers surplus. Welfare is hardly affected, but the incumbent's profits are obviously reduced. Although this latter effect is undesirable (and should not become too large), it is outweighed by the positive effects on consumers surplus and the entrant's profits. Thus, non-reciprocal access prices may be useful to stimulate competition in an immature market. In the longer run, when the entrant has built up substantial market share, cost-based access prices for both firms are optimal for consumers surplus and welfare. All the preceding discussion assumes that the calling party pays principle (CPP) applies. And indeed CPP applies in all European countries. Recent literature (Laffont et al., 2001) has shown that most of the problems mentioned above may disappear when the CPP principle is abolished. The reason is that under CPP a price is missing, namely a price for receiving a call. As a consequence, all costs have to be incurred by the caller, while technically the costs should be split somehow. One of the consequences is that operators have monopoly power on their incoming calls, which generates problems, e.g. in the fixed-mobile termination traffic (see section 4.1). Laffont et al. (2001) show under quite general assumptions that interconnection charges will be set at the competitive level if the missing price is recovered. In contrast with the CPP case, perfect retail competition can exist and is viable. In the case of perfect competition, the access charge serves to determine how the cost of communication is split between the caller and the callee, with the operators being indifferent (since they achieve zero profits anyhow). However, whenever there is imperfect competition (market power, even limited), operators favoured access charge need not coincide with the socially optimal one; unfortunately, the nature of the operators bias is not that easy to predict. 3.3 Dynamic considerations There are two issues that concern the short and the long run. First, as we have pointed out that asymmetric access regulation is optimal in an infant market, an important question is at what moment the market can be considered to be sufficiently mature, so that asymmetric regulation

18 can be lifted. Second, similar to one-way access situations, any access price affects operators profit levels, and hence their incentives to invest. Concerning the first issue, if the regulator imposes asymmetric access prices to stimulate competition in the short run, a crucial question is: when does the transition phase end, that is, when are entrants sufficiently large so that there is effective competition? For policy purposes, this is mainly an empirical question. Clearly, the 'stopping rule' or 'sunset clause' should be based on observable market outcomes. In practice, policy makers can look at various indicators, such as growth of entrants market shares, reductions in retail prices, quality/price ratios and variety for end-users. There is a risk involved in making the lifting of asymmetric regulation dependent on market outcomes, though. 8 For instance, a criterion based on market share gives both the incumbent and entrants incentives to compete less aggressively by increasing retail prices. To see this, notice that the incumbent wants the entrant to gain market share more quickly so that it no longer incurs the access mark-up charged by the entrant, and the entrant wants to keep enjoying the access mark-up as long as possible. Therefore, various regulators have been investigating the criteria for sun-set clauses: (e.g. Canadian Radio-Television and Telecommunications Commission, 2000, or OFTEL, 1999). However, it is not yet clear what all these dynamic regulation methods have produced. Concerning the second issue, note that the policy prescription of asymmetric access regulation (as discussed above) does not conflict with the possibility that the regulator may want to introduce a bias towards facilities-based entry. Nevertheless, if the regulator explicitly wants to stimulate entrants to roll out networks themselves, it is more direct to make other types of entry (Carrier Select, unbundled access to the local loop) gradually less attractive. This may be sufficient to induce entrants without their own local loops but with experience and market share to start building them. 4 Policy Issues In this section, we shall discuss four policy issues in more detail. The issues are fixed-mobile termination, margin squeeze, the allocation of common costs and static versus dynamic efficiency. The issues are highly relevant for regulators at this stage of development of the telecom market. A lot of policy attention concerns issues of fixed-mobile termination and 8 This is pointed out by De Bijl and Peitz (2002b)

19 margin squeeze. The allocation of common costs and static versus dynamic efficiency issues are important issues in regulation in general. The structure of this chapter is as follows. For each policy issue we first discuss the policy relevance, followed by an overview of what economic theory has to say. Sometimes this implies a summary of earlier chapters. The policy issues proceed with practical experiences and end with some policy conclusions. The policy conclusions combine insights from the earlier chapters with the practical experiences. 4.1 Fixed-mobile termination Policy relevance Fixed-mobile termination receives a lot of policy attention. Quite a number of complaints of fixed network operators concern this issue and also politicians talk about the high fixedmobile termination tariffs. In practice, one can indeed observe that these access prices are usually very high, compared to the cost of access. Also there is a price asymmetry, i.e. the cost of being called is higher than making a call (although they place similar demands on the network). After other mobile operators entered the market, the retail price for making a mobile call decreased whereas the terminating access price decreased much slower, the price asymmetry increased. One might conclude that the fixed networks are subsidising the mobile networks. OPTA (2002) calculates that for 2000 and 2001 the revenue of fixed-mobile termination was about 270 million above revenue based on cost orientation, i.e. a crosssubsidy from about 270 million from the fixed networks to mobile networks. Furthermore, the revenues from termination are very significant. For Telefonica, 75% of mobile revenue was derived from termination and roaming services (OECD, 1999). This might be a reason for intervention of regulators. What are the options for intervention? Regulators can decide not to intervene. One can expect that the situation will remain the same because in the short run no alternatives for mobile termination are to be expected. Fixed subscribers subsidise mobile subscribers. Regulators may decide to intervene and treat fixed-mobile termination as a competitive bottleneck and regulate the terminating access tariffs. Different pricing principles can be applied to calculate cost-oriented tariffs (see section 2). Ramsey pricing and LRIC are

20 most commonly referred to. When Ramsey pricing is applied, the problem might not be solved in a way that satisfy the concerns of the regulators (excessive prices). 9 The common costs will be allocated based on the demand (elasticities). Demand in the retail market is more elastic than in the fixed-mobile terminating market. This implies that the largest part of the common costs will be recovered in the terminating market. The current price asymmetry might therefore be in line with Ramsey pricing. The alternative is LRIC, which is commonly used by regulators in this situation. To recover the common costs, a fair share of these costs is often set on top of the price based on LRIC. The calculation of this fair share is arbitrary. This implies that parts of the common costs that are now recovered in the fixed-mobile access market have to be recovered in another market. If this cannot be recovered because of competitive forces in that market (e.g. the mobile retail market), this implies that companies will make losses. This might result in a shake-out in the market. On the other hand, if the common costs can be recovered, this might imply that retail prices will increase (see also section on common costs). Call termination might be tied to the price of mobile originating, e.g. a termination charge has to be proportional to the originating access charge. While this option might seem appealing, it also comes with a cost, i.e. the price-setting in a competitive market will be influenced by the regulatory action Economic theory Fixed subscribers, when they want to reach mobile users, have no choice but to pay the high access fee (as part of the overall per-minute price). The level of the fixed-mobile termination access price does not directly influence the level of the usage-based mobile retail prices, but part of the fixed cost related to the mobile subscribers (e.g. lower subscription fee or subsidies on hand sets). That is, high access mark-ups imply that fixed customers subsidise mobile users. Mobile users benefit from high fixed-mobile access charges, to the extent that retail prices are subsidised. Under rather standard assumptions, it can be shown that welfare is maximised by setting the terminating access prices equal to marginal costs (see Armstrong, 2001). However, it seems that there is no effective competitive pressure to set terminating prices at marginal costs. The policy discussion focuses on the issue to what extent different mechanisms restrict 9 It is not clear however, if that would constitute a problem in terms of welfare

21 the MNO s to behave to an appreciable extent independently of its competitors, customers and ultimate consumers. In other words, has the MNO a dominant position or not? If the MNO is dominant and (in addition) sets excessive terminating access prices, intervention seems required. To what extent do mobile consumers care about how much it costs others to call them? One might argue that mobile users can be reached in other ways as well, for instance by using a fixed connection. This argument, however, misses the point that one usually tries to reach mobile users at their mobile phone when they are mobile, that is, remote from home or office where there is a fixed line with a known phone number (or remote from other alternatives to contact them). Even when there are substitution possibilities, they may not be strong enough to result in substantial downward pressure on access charges. Furthermore, one might argue that market forces already exert sufficient discipline on mobile operators. For example, suppose that mobile customers derive utility from being called (which is undoubtedly true in many cases). If consumers want to receive more calls, they may care about the termination access price charged by their operator, and hence they may take the costs of inbound calls into account when they choose a mobile subscription. As a consequence, the presence of call externalities gives mobile operators an incentive to reduce fixed-mobile termination access prices, although it is unclear how strong this effect is in reality. Armstrong (2001) actually shows that if mobile subscribers derive utility from incoming calls, then welfare is maximised by setting termination charges below the marginal cost of access. The reason is that access below cost encourages fixed subscribers to make calls to mobile subscribers. Another way in which consumers may take the costs of inbound calls into account occurs if one assumes that they care about the costs incurred by people calling them. In theory, if mobile customers, when they choose a mobile operator, internalise the welfare of people calling them, then again mobile operators have incentives to reduce termination fees (Armstrong, 2001). This story has some realistic content, since people often receive most calls from family members or direct colleagues. Ultimately, it is an empirical question how much mobile customers care about the costs of incoming calls. Existing evidence (e.g. Oftel 2001) together with casual observation (the rates are indeed very high all over Europe) suggests that it is most likely that mobile customers do not care that much for the costs incurred by people calling them when they choose a subscription

22 As the current system of the Calling Party Pays (CPP) and customer pressure does not seem to exert enough downward pressure on the terminating access prices, one might choose changing the payment principle. The alternative is to let mobile operators charge their customers for incoming calls. That is, call recipients pay a per-minute price for receiving calls from fixed subscribers (instead of the CPP principle). This remedy takes away the bottleneck problem (given that the mobile operator does not charge the fixed operator for the termination anymore), since customers who have to pay for incoming calls will be more sensitive to prices for that service. 10 Changing the system may create a distortion, i.e. subscribers pay too little for calls to mobile users if the costs between callers and callees are split for fixed-mobile but not for fixed-fixed. It is not clear how serious this distortion is, but to solve it one may consider to introduce the split also for other services. The conclusion is that policy intervention seems to be necessary. There might be some mechanisms that restrict the behaviour of the MNO s, but the effect is too small to conclude that competition in the fixed-mobile termination market is effective. On the short run, most research indicates that these mechanisms will not increase sufficiently in strength to create a situation of effective competition (see e.g. Armstrong 2001) Practical experiences The central part in the debate on fixed-mobile termination is whether fixed-mobile call termination is a bottleneck and whether Mobile Network Operators (MNO s) can exercise market power. A lot of policy discussions between Regulators and market players concern this point. The principle of the calling party pays is a central point in this discussion. Arguments can be given why there is no (or limited) effect on the height of the terminating access prices versus arguments why there is enough pressure on MNO s to compete on the terminating access prices (see e.g. OFTEL, 2001, IRG, 2002). Despite the results of this discussion, Armstrong showed that in most cases the pressure from these mechanisms is too small to restrict the behaviour of MNOs in setting their termination access prices. The high terminating access prices seem to support this argument. Most regulators also follow this line of reasoning and intervene. OFTEL (2001) discusses different mechanisms that might constrain the behaviour of MNOs. First of all, mobile subscribers might care about the costs of calling them and take the 10 However, it does not solve the originating access from fixed to mobile. These charges should still be

23 termination prices into account when making a choice for a MNO. Their conclusion is that mobile callers put little weight on terminating prices (this might be different for different groups). Second, the effect of closed user groups 11 generates a limited pressure on termination prices. Furthermore, OFTEL concludes that there are no real alternatives and buyers have no countervailing power. The conclusion is that there is insufficient competitive pressure to constrain the terminating prices and that the pressures are expected to remain insufficient. OPTA uses more or less the same arguments as OFTEL and comes to the conclusion that mobile call termination is a bottleneck facility. OPTA decided to intervene and set a timeframe to come to cost-oriented terminating access tariffs in In between, OPTA suggested terminating prices that they considered as fair. For calculating terminating access prices, the regulators use the LRIC pricing principle or are planning to use it in the near future. LRIC is preferred over Ramsey pricing because Ramsey pricing might be difficult to calculate and might result in an unfair allocation of common cost (cross-subsidisation and/or big price asymmetries). Also the common costs seem to be a relatively small problem (see section 4.3 on common costs) Policy conclusions The policy conclusions combine insights from the earlier chapters with the practical experiences from above. Fixed-mobile termination is a hot topic in the telecom sector and receives a lot of attention of the regulators. So far, the conclusion of most regulators is that fixed-mobile termination is (close to) a bottleneck facility. There are too little competitive pressures to restrain the behaviour of the MNOs and therefore regulation is considered to be necessary. It is, however, important to review the developments of the potential mechanisms that might result in effective competition. It is, for example, unclear how closed user groups will develop and to what extent they might restrict the behaviour of MNOs. Another important issue is the potential of alternatives for the calling party pays (CPP) principle. CPP is common practice in Europe. CPP might be exchanged for receiving regulated. 11 Closed user groups are groups which have an interest in how much it costs to call each other and therefore all subscribe to the same network (e.g. familymembers or a company). For these closed user groups special offers are often made