Costing methodologies and incentives to invest in fibre

Transcription

1 Costing methodologies and incentives to invest in fibre Prepared for DG Information Society and Media Avenue de Beaulieu B Brussels Belgium Prepared by Jenny Haydock Gregor Langus Vilen Lipatov Damien Neven Gareth Shier 99 Bishopsgate London EC2M 3XD Date:

2 Disclaimer and its authors make no representation or warranty as to the accuracy or completeness of the material contained in this document and shall have, and accept, no liability for any statements, opinions, information or matters (expressed or implied) arising out of, contained in or derived from this document or any omissions from this document, or any other written or oral communication transmitted or made available to any other party in relation to the subject matter of this document.

3 TABLE OF CONTENTS EXECUTIVE SUMMARY INTRODUCTION THE PRINCIPLES OF ACCESS PRICE REGULATION THE STATIC/DYNAMIC TRADE-OFF REGULATORY COMMITMENT QUALITY UPGRADES AND EXTERNALITIES FROM INVESTMENT CONCLUSIONS TAXONOMY OF COSTING METHODOLOGY CHOICES WHICH COSTS ARE CONSIDERED AS PART OF THE INCREMENT? LRIC LRIC FAC SAC HOW WILL THE ASSET VALUE BE APPRAISED? Historic cost appraisal Current cost appraisal HOW WILL THE ASSET BE DEPRECIATED? Relevant cost in a given period: a building block approach Impact of the depreciation methodology Common approaches to depreciation CONCLUSION DEMAND, TECHNOLOGY DEVELOPMENTS AND COST RECOVERY PROFILES THE PROFILE OF DEMAND REPLACEMENT COSTS AND BUILD OR BUY INCENTIVES CONCLUSIONS COST RECOVERY UNDER UNCERTAINTY THE EFFECTS OF UNCERTAINTY DEMAND SHOCKS REPLACEMENT COST SHOCKS ASSET LIFE SHOCKS CONCLUSIONS FURTHER APPROACHES TO UNCERTAINTY AND RISK Page i

4 6.1. OPTION VALUES RISK SHARING AND CO-INVESTMENT USAGE-BASED VERSUS CAPACITY-BASED APPROACHES RETAIL-PRICE-BASED APPROACHES THREAT OF REGULATION ACCESS ABRIDGEMENT CONCLUSIONS RACE TO INVEST AND DUPLICATE INFRASTRUCTURES RACE TO INVEST REGULATION AND INCENTIVE TO DUPLICATE THE DESIRABILITY OF FACILITY-BASED COMPETITION COPPER AND CABLE AS DUPLICATES GEOGRAPHIC DIFFERENCES TECHNOLOGY NEUTRALITY CONCLUSIONS EXISTING COPPER INFRASTRUCTURE THE TRANSITION FROM COPPER TO FIBRE Introduction The case with no regulation of fibre The case when both OGN and NGN are regulated Relevance to policy FULLY DEPRECIATED COPPER NETWORKS CONTINGENT VALUATION / INCENTIVE PRICING CONCLUSIONS COMPETITION MODEL INTRODUCTION THE MODEL IMPLEMENTATION OF THE MODEL CALIBRATION INITIAL RESULTS CHANGING FIBRE ACCESS PRICE CHANGING VALUATIONS Changing valuation of copper Changing valuation of cable CHANGING MARGINAL COSTS Changing costs of copper Changing costs of fibre Page ii

5 9.9. CHANGING "TRANSPORTATION COSTS" EFFECTS AT WORK PRE-EMPTION POSSIBILITY Fibre investment by an independent entrant Pre-empting new investment SUMMARY CONCLUSION APPENDIX INCENTIVE TO INVEST FOR THE INCUMBENT WITH PARALLEL NETWORKS Copper only Fibre and copper PRE-EMPTION A NOTE ON EQUILIBRIUM COMPUTATION Page iii

6 EXECUTIVE SUMMARY This study considers how alternative costing methodologies for the calibration of access prices affect the incentive to invest in a new infrastructure (with a particular focus on investment in fibre networks), as well as the consequences of alternative methodologies for the development of retail competition and efficient use of the legacy copper network. The study also uses a stylised model of competition to study how alternative cost methodologies and related access prices affect the incentive for fibre adoption. A number of conclusions emerge. We find that a methodology that takes into account sunk and common costs (and not only variable or incremental costs), at least from an ex-ante perspective, is desirable when the goal is to foster investments. We highlight the importance of the amortisation profiles, in particular in relation to the characteristics of possible demand realisation, which may imply that the access price cap may exceed what the market can bear, thus threatening recovery of investment cost. The issue of asset appraisal involves a fundamental choice between current and historic cost appraisal. On the one hand, a current cost appraisal for fibre may not ensure adequate incentives to invest, in particular in the presence of significant technological risk. On the other hand, adopting a historic cost approach for copper assets that have been fully amortised may distort competition between copper and other platforms, in particular cable and mobile. We find strong support for the view that the regulator should make credible allowances for the cost of capital, different sources of risk, and also for the option to wait. Moreover, various forms of risk sharing schemes between incumbents and access seekers can be considered. The appropriate access price may also be affected by the desirability of asset replication, which we find depends on whether a new asset would offer a differentiated product and/or more intense competition. We also find support for a differentiation of access prices across different geographic areas which exhibit different demand densities and scope for platform competition. The issue of joint calibration of copper and fibre access prices and its effect on investment is complex. Whether lower access prices for copper (that would be induced for instance by excluding some fixed and common cost or by using historic rather than current costs) would trigger investment in fibre depends crucially on whether copper and fibre networks are operated in parallel. If copper is switched off at the time of fibre deployment, investment would likely be triggered by lower access prices for copper. If not (as would appear to be realistic), then the effect of copper access prices on incentives to invest in fibre is, in principle, ambiguous. We observe that the incentives to invest are largely determined by the number of access seekers to copper infrastructure that different copper access prices would induce. We find that lower copper access prices increase the number of access seekers and tend to reduce the incentive to invest. These results should be seen as illustrative and should not be applied literally in any particular member state or region. But they clearly suggest that a reduction in copper access prices for the sake of encouraging investment in fibre may not work, or would at least require a level of regulatory oversight that may not be realistic. Page 1

7 1. INTRODUCTION Regulation will undoubtedly play a crucial role in meeting Europe s Digital Agenda goals, in particular with respect to fostering the necessary investments into fibre. The purpose of this report is to describe different costing methodologies used to determine the prices of regulated wholesale access products, and assess these for their potential to foster investment into fibre as well as their consequences for the development of retail competition and the efficient use of the legacy copper network. 1 In this report we proceed on the basis that investments into efficient broadband infrastructure are a necessary condition for effective competition along the whole value chain, including at the level of retail competition. We also assume that the objective of access regulation is to preserve long-term retail competition, once it has been established that infrastructure-based competition is not viable or not sufficient. These two points identify the main challenge of access price regulation: to ensure viable retail competition (by ensuring a sufficiently low access price), but in a way which results in an efficient mix of broadband technologies and which preserves investment incentives over time. A costbased access price is often thought to broadly address this problem and it is this type of approach on which our analysis focuses. We discuss alternatives to cost-based regulation only insofar as costbased access prices may not provide sufficient incentives for efficient investments. We note that, when efficient investments would take place under cost-based access price regulation, departures from cost-based regulation may result in higher retail prices and an associated loss of consumer welfare. We primarily concern ourselves with the incentives of a (vertically-integrated) operator to invest (irreversibly) in a new infrastructure technology, or in improvements to an existing technology. We primarily have in mind geographic areas where fibre is financially viable without subsidy. The first part of the report (Sections 2 to 8) discusses in detail how cost-based access price regulation can impact on regulated telecommunications firms incentives to invest in new infrastructure. We also discuss how these alternative methods would affect retail competition and the efficient use of the legacy copper network. The second part of the report (Section 9) uses a stylised cost model for copper and fibre and a model of competitive interactions to study how alternative costing methodologies and related access prices affect the incentive for fibre adoption. It is important to note that this report does not attempt to be an exhaustive account of all the issues relevant to access pricing. Our remit is related to the impact of different costing methodologies on fibre investment incentives, and although our analysis has inevitably raised some related issues, such topics are dealt with only in the context of our brief. In addition, each methodology will often involve some trade-off with respect to particular objectives and might have conflicting consequences for the achievement of the various objectives being considered. As a consequence, it would be inappropriate to draw definite conclusions with respect to most aspects of the costing methodologies. Finally, the circumstances that are relevant to evaluate the terms of trade-offs implied by alternative costing methodologies will vary across regions and members states. We are thus not, a fortiori, in a position to conclude as to the details of the specific access pricing regime that should be imposed on any given network or operator. 1 We note the European Commission s Recommendation on regulated access to Next Generation Access Networks (NGA) (the NGA Recommendation ), of September 2010, which is clear that the price of access should be cost-oriented. Page 2

8 The report starts with a review of the principles of access price regulation in Section 2. This section first discusses the well-known trade-off between static and dynamic efficiency that the regulator faces in most relevant market circumstances. In order to be able to effectively resolve the inherent conflict between static and dynamic efficiency, the regulator must be able to commit not to expropriate the investors, ex-post, after an irreversible investment is made. In a world of repeated interaction between the regulators and other market players, building reputation for a regulator is an important mechanism which can improve market outcomes. We also consider how regulation will affect the incentives for incumbents to invest in quality upgrades, in particular when this allows access seekers to improve their product offering, and how this relates to externalities from investment. In Section 3 we set out taxonomy of costing methodology choices. In a nutshell, a regulator considering cost-oriented regulation has to decide: 1) which costs to include in its appraisal; 2) whether the costs are appraised according to the costs incurred by the investor or on the basis of hypothetical current costs as would be faced by a new entrant as time evolves; and 3) how the costs should be allocated across the life of the network asset. The differences in these three choices roughly determine the classification of different costing methods. The choices along these dimensions are not neutral with respect to the incentives to invest, as they impact how much an investor can hope to earn from his investment, and when those earnings are made (i.e. the profile of cost recovery). We find in particular that from the ex ante prospective that is relevant for fibre development, there is strong support for a methodology that considers sunk and common costs (and not merely incremental costs). With respect to the methodology of asset appraisal, we emphasise that the anticipation, ex ante, that the regulator will reduce the value of the asset in line with the improvement in technology (the socalled current cost approach) might lead to insufficient recovery of cost. In order to induce investment, the access price might thus have to be based on a level of cost which is higher than the cost incurred by the investor. We also highlight the significance of the amortisation profile for the overall level of allowable revenues and its distribution over time. In Section 4 we focus on the implications of different cost recovery profiles, in particular in terms of how they relate to changes in demand. If allowable cost recovery is relatively high when demand is low (or elastic), the resulting price cap may not bind, which can threaten the investor s ability to recover its investment costs, and hence undermine its incentives to invest in the first place. In particular, it is important that, in order to maintain investment incentives, the regulator takes into account the expectations of prospective investors. We also discuss that aligning the cost recovery profile to movements in the replacement cost of the asset will affect build or buy incentives for access seekers, and this may conflict (or reinforce) with the cost recovery patterns dictated by demand. The need to align cost recovery with demand becomes more complicated when the possibility of uncertainty is introduced, as discussed in Section 5 below. Under uncertainty, cost recovery may also be threatened by shocks in replacement costs through the use of a current cost appraisal approach, by which a regulator attempts to mimic a contestable market by setting the allowable cost recovery to reflect the current replacement cost of the asset, rather than the actual investment cost incurred by the access provider. Compared to a historic cost appraisal approach, a current cost appraisal approach therefore makes the investor subject to technological risk, introducing uncertainty as to whether full cost recovery will be possible, by raising the possibility of over- or under-recovery even without demand shocks (unless some guarantee of full cost recovery is offered). If in the absence of regulation the replacement cost would not affect cost recovery (because the market is in fact not contestable) then this would represent additional risk imposed on the investor by the regulator. Following on from these discussions, in Section 6 we discuss how regulation might deviate from purely cost-based approaches, in order to ensure adequate investment incentives whilst still retaining Page 3

9 regulatory oversight. Broadly, a risk of under-recovery of investment cost will require compensatory over-recovery, either at particular times in the life of the asset or under particular (uncertain) outcomes. Thus, for example, we examine abridgement of certain access obligations; or some form of risk sharing. We note that mandatory access to assets that required risky investment can be seen as offering a free option to access seekers; the incorporation of this issue may justify access prices that exceed what is implied by a simple application of costs alone. In Section 7.1 we note that the presence of more than one possible investor can lead to a race to invest. If there is advantage to being the first investor investment may even take place too soon from society s perspective. In the remainder of Section 7 we introduce two complicating factors that are likely to be relevant to the roll-out of fibre networks: the possibility of more than one fibre network ( duplication ); and the role of other, competing, networks such as cable (which might also be seen as an imperfect form of replication). We discuss literature which indicates that incumbents may price access lower, in order to discourage facility-based competition. Broadly speaking, we note the intuition that if replication is to be encouraged access prices should be higher, whereas if replication is not a concern (perhaps because it is largely impossible in practice) the focus should be on retail competition, and hence lower access prices. The desirability of replication depends on whether a new network offers a differentiated product (that some, or all, consumers will prefer to the existing products), and whether the competition is more intense or effective with facility-based as opposed to service-based competition. We also discuss the fact that geographic differences in terms of the viability of (one or more) fibre networks, and the role of competing technologies such as cable, mean that different regulatory approaches may be needed in different areas: although it is also important to note that with crossregional pricing the effects of competition in one area may be felt in others. The fact that in most areas there already exists a copper network, coupled with the understanding that many fibre investors are likely to be copper incumbents, mean that incentives to invest in fibre will also be affected by the copper access price. In Section 8 we explain how incumbents face a trade-off that could mean that lower or higher copper access prices would encourage investment in fibre. We also discuss the suggestion that access to fully-depreciated copper networks should be priced at variable or operational cost: this may have negative implications for incentives to invest in alternative technologies, including cable. An approach which prices copper at the fibre cost level (a form of current cost appraisal) may be justifiable in cases where overall copper cost recovery is unthreatened. In Section 9 we build a model of competition in broadband provision closely following WIK [2011]. We test the robustness of WIK s results that the access price for copper infrastructure should be set low in order to spur investments into fibre. We also explore how access price regulation of copper and fibre access infrastructures affects fibre investment incentives, when, contrary to WIK s assumptions, copper switch-off is not allowed or not feasible. We find that WIK s results are sensitive to changes in model specification. In particular, we find that the incentive to invest in fibre when the copper network is operated in parallel is largely determined by the changes in market structure induced by the changes in copper access prices. Lower access prices tend to introduce more access seekers (for both fibre and copper) and discourage investment in fibre. We describe and illustrate the mechanisms that drive WIK s and our results and argue that, similarly to the results of the recent theoretical literature, the relation between the copper access price and fibre investment incentives need not be monotonic and depends on a number of different factors, which we discuss in detail. We also consider the possibility of investment in fibre by alternative operators. Our results confirm that the incumbent would have an incentive to pre-empt the investment by alternative operators for a range of access Page 4

10 prices. However, we find that the incentive to invest is overwhelmingly determined by the changes in the number of access seekers induced by different access prices. Section 10 concludes, and the Appendix at Section 11 contains more detailed algebra. Page 5

11 2. THE PRINCIPLES OF ACCESS PRICE REGULATION At a very high level, investment incentives will be affected by the answers to the following questions: What level of costs does an investor expect to recover over the life of the asset? How does the investor expect those costs to be spread over time? How exposed is the investor to risks such as demand shocks? Is there scope for an alternative investor, or is there only one potential access provider? (And could there be more than one actual access provider or are there competing assets?) These considerations mean that among the aspects of cost-based regulation that may affect investment incentives are: The costs included in the measure used to calculate the access price (and in particular the treatment of sunk and common costs) The asset valuation methodology used for long-lived assets: in particular, whether the valuation, and hence the access price calculation, is revisited to adjust for current circumstances and whether full cost recovery is guaranteed The choice of depreciation/annualisation method for investment costs including whether cost recovery is tilted towards earlier or later periods The timeframe of the regulated access regime The impact of these factors will depend on specific circumstances, including: the level of uncertainty regarding future market conditions, including demand and prices; whether more than one firm has the potential to invest, and, if so, whether the infrastructure might be duplicated; and whether there is existing infrastructure that may be substitutable with the new asset and how access to those assets is priced. In this Section, we first discuss some significant high-level issues that we would expect to be of relevance to a regulator. We then discuss the importance of regulatory commitment to induce investment ex ante as well as the appropriate regulation towards quality upgrades (and more generally investments which generate positive externalities). In Section 3 we go on to set out a taxonomy of options from which regulators typically choose when determining a costing methodology for access price regulation. We consider the implications of these choices in the following sections. At the outset, a methodological remark may be in order: the starting point of our discussion is the relevant theoretical economic literature. In general, however, it is necessary to extend the insights therein to produce relevant policy implications for incentives to invest in fibre and we attempt to do this, relying on relevant policy studies (and in particular the contributions made in the context of the Commission s recent public consultation). As noted by Cambini & Jiang [2009] 2 in a recent survey, the current theoretical literature suffers from significant shortcomings, in particular with respect to the 2 Cambini, C., & Jiang, Y. (2009). Broadband investment and regulation: A literature review. Telecommunications Policy, 33, Page 6

12 coverage of some significant policy issues (although some very recent papers have begun to address these): First, current theoretical models typically focus on selected kinds of investment, mostly the start-up of new infrastructure or the enlargement of capacity, whereas the regulatory impact on the overall level of investment is still vague at present. Admittedly local loop unbundling may impair firms incentive to carry out certain investments, but little effort is devoted to discussing the aggregate tendency. In other words, the relationship between access regulation and aggregate investment (for both incumbents and entrants) is a missing point in the current theoretical literature. Second, there is still a lack of convincing theoretical analysis to support the ladder-of-investment theory and to determine how regulatory rules should evolve over time (and eventually in differentiated geographical areas) so as not to prevent new investment in broadband infrastructures by market operators (incumbents and entrants). Third, theoretical analysis does not consider the impact of mixed-case regulation where wholesale regulation is applied only to a subset of access services on investments. On the practical side, some regulators propose the introduction of cost-oriented regulation for specific wholesale services (local loop, bitstream, sub-loop unbundling) but no regulation for others (like the access to FTTH cables in Spain or the access over 3 Mbps connections in France). There does not appear to be any theoretical analysis that focuses on this issue. Finally, a theoretical analysis on the impact of network sharing and co-investment on overall broadband infrastructure is still lacking The static/dynamic trade-off When a broadband infrastructure exhibits features of a natural monopoly and the incumbent has no incentives to provide access to its infrastructure at prices which can support efficient entry at the retail level, a regulator may want to regulate access prices at the wholesale level. In that, the regulator faces a fundamental trade-off: between ensuring that access is provided at costs which promote efficient retail market entry, and the need to ensure that efficient incentives are provided for (sunk, irreversible) investments to be made. A number of academic papers as well as policy-oriented contributions have characterised this trade-off. 3 Regulation in the form of cost-oriented prices for wholesale access aims at improving the prospects for competitive outcomes in the retail markets and in this it necessarily reduces the exercise of monopoly power over an existing infrastructure. This reduces the rents accruing to the investors and at the same time reduces the deadweight loss. From a static point of view, regulation of existing assets which exhibit features of a natural monopoly is thus likely to be socially efficient in many circumstances. However, insofar as the investors expect that new facilities, into which they are considering investing, will also be regulated, they will take into account the effect that regulation will have on the rents that they can extract from their investments and they are often likely to invest less, 3 See for example: Valletti, T. (2003). The theory of access pricing and its linkage with investment incentives. Telecommunications Policy, 27(10 11), ; Gans, J., & Williams, P. (1999). Access regulation and the timing of infrastructure investment. The Economic Record, 75(229), ; Guthrie, G. (2006). Regulating infrastructure: The impact on risk and investment. Journal of Economic Literature, 44(4), ; Graeme, G., Small, J., & Wright, J. (2006). Pricing access: Forward-looking versus backward-looking cost rules. European Economic Review, 50(7), ; Foros, Ø. (2004). Strategic investments with spillovers, vertical integration and foreclosure in the broadband access market. International Journal of Industrial Organization, 22(1), Page 7

13 or later, in response. This trade-off depends on the regulatory regime and as Valletti [2003] 4 puts it,...the nature of ex-post regulation has an impact on ex-ante incentives to invest. We note that lowering prices for access to an existing monopoly infrastructure from high (e.g. monopoly) levels almost always results in an increase in social welfare. At the same time, decreasing prices for an existing monopoly infrastructure will not necessarily reduce investments into new infrastructure (particularly when the new infrastructure is subject to a different regulatory regime). At the same time, a relatively lax regulatory regime on the future infrastructure will in general positively affect investment incentives. We address these issues more extensively in Section 8 below Regulatory commitment Given the trade-off just described, it is significant whether a regulator can commit to an access pricing approach before any irreversible investment is made or committed to. If not, the issue of the static/dynamic trade-off is exacerbated, since, once the investment is sunk, the regulator may be tempted to ignore the investment incentives issue entirely, and price at a level consistent with shortterm, static efficiency (i.e. at some marginal or other variable cost measure). 5 Some academic papers address this issue; Foros [2004] 6 is one example. He argues that, due to the absence of the value of reputation in his model, the regulator would have incentives to behave opportunistically and set an access price for the infrastructure close to marginal costs after the infrastructure was built. He finds that such regulation would result in decreased investment incentives in the case when the services of the incumbent and access seekers are not sufficiently differentiated (when downstream services are differentiated, the incumbent has a strong incentive to invest anyway because access enlarges the demand for the infrastructure). This would, in turn, under many circumstances, result in a decrease in consumer surplus and social welfare. The problem arises because in the presence of sunk costs the investors cannot, without incurring significant costs, disinvest in the face of such opportunistic (and, in the context of Foros analysis, optimal) behaviour. The problem is more general though, and regulatory opportunism can be looked at as a factor influencing the risk that the investor will be deprived of its ability to recover its sunk costs. Many factors will affect this, but one of them is the nature of dynamic interaction between the regulators and the (potential) owners of the infrastructure. Indeed, a number of stakeholders mostly the likely investors, including ETNO and several incumbents in their own national markets, but also the regulators themselves have pointed out the need for the regulators to be able to create and sustain a reputation to be able to resist the temptation to behave opportunistically after the investment costs have been sunk. 7 It appears that in reality the nature of repeated interaction between investors and regulators does indeed resolve part of the problem with regulatory opportunism, as the regulator usually has an interest in building a 4 Valletti, T. (2003). The theory of access pricing and its linkage with investment incentives. Telecommunications Policy, 27(10 11), Note that we generally assume that the investment is wholly sunk. 6 Foros, Ø. (2004). Strategic investments with spillovers, vertical integration and foreclosure in the broadband access market. International Journal of Industrial Organization, 22(1), See e.g. the responses to the public consultation on costing methodologies for key wholesale access prices in electronic communications. Available at: Page 8

14 reputation that it will not behave opportunistically. Proposals to change the methodology for setting access prices to copper that are motivated by the objective to enhance investment into fibre should be considered in this light. The reputation that regulators have accumulated for not behaving opportunistically may be lost. This would in turn enhance the perception of a regulatory risk surrounding future investment and investment in fibre in particular. Investors will naturally expect a risk premium (and hence higher returns) to compensate, which might defeat the purpose of enhancing the investment in fibre in the first place Quality upgrades and externalities from investment One of the reasons for why a regulator might want to actively encourage adoption of fibre is that there are likely to be external social benefits to investment that exceed the returns the potential investor expects to make as has been recognised by the Commission. 8 Besides the benefits associated with the availability of infrastructure that accrue to an investor s rivals and customers there are also likely to be wider social benefits resulting from, for example, improved communications and feasibility of new services. To the extent that these benefits cannot be appropriated through higher retail (or access) prices, the investor will not take them into account in its decision whether, when, and how much to invest. This generally means that the investments are too low (or too late) from a social point of view. A particular form of external effect that may arise is when the development of a network in terms of quality generates benefits for access seekers, which the investor may not appropriate due to regulated access requirements. Regulation can, therefore, affect the rents that accrue to the investor and thus its incentive to invest in quality upgrades. Several academic papers address this issue. The model of Kotakorpi [2006] 9 includes a vertically-integrated incumbent and a fringe of access seekers. Subscribers are heterogenous in their valuation of services. Moreover, the services of the fringe may be vertically differentiated from those of the incumbent. The access provider is considering an investment that would increase consumers valuation of the associated retail services (those provided by the access provider, but also those sold by the access seekers the latter effect is termed spillovers ). There are fixed costs associated with operating the infrastructure, but none associated with the operations at the retail level of the market. If there is regulation, the access price is regulated at a level set after investment has been decided (pre-investment commitments are not credible), but before retail prices are chosen. Thus it is found that the socially-optimal access charge is the marginal cost of network provision. Kotakorpi finds that investments are below the social optimum even when there is no regulation, and access price regulation further reduces investment incentives. Further, while spillovers have a positive effect on investment when there is no regulation, they have a negative effect when there is a regulated access charge. This arises because the regulated access prices do not reflect the cost to the incumbent of providing the increase in quality, and thus access seekers are in a more favourable position in the downstream market. By contrast, absent the access price regulation, the incumbent can indirectly extract extra rents when granting access, to appropriate some portion of the increase in value of services provided by the access seeker due to spillovers. Thus absence of regulation promotes investment for high spillovers. 8 See e.g. COM(2010) 245, Communication from the Commission to the European Parliament, The Council, the European Economic and Social Committee and the Committee of the Regions. 9 Kotakorpi, K. (2006). Access price regulation, investment and entry in telecommunications. International Journal of Industrial Organization, 24(5), Page 9

15 In addition to these results, Kotakorpi also finds that access price regulation need not reduce the likelihood of foreclosure (i.e. an outcome where the access seekers output is zero). Further, while in the absence of regulation foreclosure is most likely when investment spillovers are low (i.e. when investment has only a small positive impact on rivals demand, relative to the impact they have on the incumbent s demand), the reverse is true with regulation. Thus, as Kotakorpi says, the exclusion of competitors is most likely precisely when they would bring highest benefits for consumers. Indeed, even regulation at marginal costs cannot guarantee that the access seekers would not be foreclosed. In particular, in the model, the service providers can only offer a single quality of services and cannot price discriminate. Thus, from the point of view of the incumbent, granting access represents an opportunity to price differentiate between customers with different valuations for different services as well as an opportunity, when the spillovers are very high, to generate and extract, through the access price, the now-larger value that the access seekers deliver through the new infrastructure to retail customers. However, while price differentiation (due to heterogeneity of consumers) is generally desirable from the point of view of the incumbent, it is costly for low levels of spillovers when access seekers services are valued less than the incumbent s, and becomes increasingly costly as the level of spillovers decreases further. Eventually, the trade-off becomes such that foreclosure ensues even absent regulation. Access regulation at marginal cost breaks this trade-off as, with it, the incumbent extracts no profits through provision of access. With regulation and with increasing levels of spillovers, the incumbent suffers from an increasingly negative externality when it invests, which arrives in the form of the more intensive competition it has to face in the retail markets. For some levels of this externality it will simply invest little so that the valuation of the services of its competitors is low and that, as a result, there is no demand for their services. It is important to note that Kotakorpi assumes no possibility of regulatory commitment. He however notes that if credible commitment was possible the regulator would set the access price higher than the upstream marginal cost in order to boost investment incentives. It would however still be the case that some mitigating effect of regulation on investment incentives is likely as long as selling access is less profitable than selling services directly to final customers. In related work, Foros [2004] 10 considers a case where the access provider must decide whether to invest to increase the quality of the product, access to which will then be regulated. In the model, the key consideration is whether the incumbent or the access seeker offers the most value-added services to downstream consumers: that is, which retail provider can make the most of the new quality of input (this is thus similar to Kotakorpi [2006]). Foros finds that if the firms have similar such abilities, access price regulation reduces the vertically integrated firm's investment incentives and, provided the cost of investment is not too convex, leads to lower consumer surplus and lower total welfare. In the model, if the access provider is much better able to offer value-added services, it will use over-investment as a foreclosure mechanism. The foreclosure is possible because the investments in this case benefit the incumbent s downstream arm relatively more than they benefit its rival, eventually causing the rival to make no sales. Foros notes that the regulated access price will be set equal to or close to the marginal cost and that this will remove most of the vertically integrated firm's cost advantage in the retail market compared to 10 Foros, Ø. (2004). Strategic investments with spillovers, vertical integration and foreclosure in the broadband access market. International Journal of Industrial Organization, 22(1), Page 10

16 the case of no regulation. Thus, he argues, access price regulation may imply a second-mover advantage. Though the paper does not focus on entry, Foros asserts that this feature of the regulation will probably discourage facility-based entry. We note that it need not be the case that a higher access price encourages investment. For example, Gayle & Weisman [2007] 11 find that raising the access price can discourage investment. This is because a higher access price means lower output, and thus a lower incentive to invest to reduce costs. However, note that this specifically relates to cost-reducing process innovation. Vareda [2011] 12 finds that, unless the marginal cost of one type of investment is very low, a high access price increases the incentives for quality-upgrading and reduces the incentives for cost-reducing. Vareda concludes that the access price should depend on the relative marginal costs of the two types of investment Conclusions It is clear that investment incentives will be seriously undermined if there is no credible commitment by a regulator to permit the recovery of (at least some) sunk costs. This may be difficult as there is a short-term static incentive to require marginal-cost (or variable-cost) pricing of access: indeed, in a static sense this is the best option for consumer welfare. However, incentives to invest will be damaged if this occurs or is expected to occur, to the ultimate detriment of consumers. The repeated interaction between the firms and the regulators can help in establishing this credibility if the regulators show that they can avoid the temptation to behave opportunistically. This seems to be particularly relevant in the presence of spillovers, for instance those induced by quality upgrades when, in some circumstances, even the absence of regulation would not be sufficient to achieve socially efficient outcomes. Instead, some active policies which further raise the costs of delaying investments may be needed and access prices that are conditional on quality may also be attractive. The literature reviewed in this section also indicates that if the services of different access seekers are likely to be highly differentiated and there are strong spillovers from the investments into infrastructure, in terms of an increase in quality of service of access seekers, a less stringent regulation may be warranted as it is in precisely those circumstances that regulation may be most detrimental to welfare. Intuitively, these are the circumstances in which the network owner and the consumers both benefit most from granting access, meaning that these motives are aligned stringent regulation changes this by undermining the investor s incentive to grant access, in which case the investor might have incentives to foreclose the access seekers. However, not only are the findings of the literature not homogenous, it is also an empirical question whether a given case of investment in fibre or indeed copper networks corresponds to these circumstances Gayle, P., & Weisman, D. (2007). Efficiency trade-off in designing competition policy for the telecommunication industry. Review of Network Economics, 6(3), Vareda, J. (2011). Access regulation and the incumbent investment in quality-upgrades and in cost-reduction. Telecommunications Policy, 34(11), We understand that an NGA investor is typically able to replicate service innovation from access seekers. Further, in case of copper upgrades (VDSL deployment), it is even possible that the access seekers are deprived of an appropriate access product (LLU). This indicates that spillovers from investment may be limited, or even negative. Page 11

17 In any event, the insights from these models (which focus on the upgrade of existing infrastructure) should not be applied literally to the development of a new, competing infrastructure (which may run in parallel), as is the case with fibre investments. The question then becomes how to regulate the two alternative infrastructures efficiently and this introduces other relevant considerations, which we discuss below in Sections 7 and 8. The more general insight is that there is usually a trade-off between static and dynamic efficiency. In a simple world, lower access prices mean greater consumer welfare today but if they are too low the investment will not be made in the first place. Hence access price regulation should aim to achieve the appropriate investment incentives at the lowest cost to static efficiency: which is (very broadly) the logic behind setting access prices on the basis of costs (investors receive just enough from selling access to their asset for them to invest in the first place). However, as we will see below, there are complications to this in practice which may warrant a deviation from purely cost-based access pricing: the lowest access price consistent with adequate investment incentives may offer the investor a higher return, at least under some circumstances. Page 12

18 3. TAXONOMY OF COSTING METHODOLOGY CHOICES The European Commission s NGA Recommendation advocates a cost-oriented approach to determining allowable access charges for a network operator. That is to say, the allowable charges should reflect the cost incurred in providing a unit of access (including a reasonable return on the capital invested). This section focuses on methodologies for assessing that cost. The key issue is how the relevant cost should be defined in any given period. This is complicated by the conglomerated structure of the network operator s business and the longevity of the network. This raises three primary questions: 1. Which of the network operator s costs should be included (i.e. common cost considerations); 2. Should these costs always reflect historic costs (as incurred by the network operator), or should they reflect current costs (as would be incurred by a new entrant); and 3. How should these costs be allocated across the life of the network asset (e.g. sunk cost considerations)? This section considers the nature of these choices and their effect on the profile and total value of access earnings allowed by the regulator. The various profiles created result in different, often conflicting, risks and incentives for the investor. The question of which methodology is optimal is, therefore, dependent upon the objectives of the regulator and the circumstances of the market in which he is operating. This section seeks primarily to describe the outcomes of these methodologies, and their suitability is discussed further in later sections below. Figure 1 presents a diagram of how these choices build to produce a relevant cost measure. Each option is then discussed in turn below. Figure 1: Options for the regulator when determining a costing methodology 1 Which costs are considered part of the increment? Short Run Marginal Costs or Short Run Incremental Costs (Operating Costs) 2 3 LRIC (VC+FC) Actual Investment Cost (historic cost) LRIC+ (LRIC + %CC) FAC (LRIC + %CC) How will the asset value be appraised? How will the asset be depreciated? SAC (LRIC + All CC) Hypothetical Replacement Cost (current cost) Straight-Line Tilted Demand Based Annuity Recoverable Investment Cost + Cost of Capital (asset value x WACC) Relevant Cost Measure Source: CRA Page 13

19 3.1. Which costs are considered as part of the increment? This decision relates to the treatment of sunk and common costs, and with it the regulator affects the level of costs that may be recovered by the investor. As is well known, under perfect competition prices will be competed down to the short-run marginal cost (SRMC) of production. 14 Accordingly, it may be considered that this should be the relevant cost measure for the regulator. A true SRMC measure considers only the cost of providing one additional unit of access. Assuming the network has spare capacity, this cost is approximately zero, which is clearly not viable for pricing. A reasonable alternative may be to consider the Short-Run Incremental Cost (SRIC) of the entire fibre business, i.e. the operating expense (OPEX) of the fibre network in a given period. However, the OPEX for such an access network is negligible compared with the considerable capital expenditure (CAPEX) required. A relevant cost measure considering only OPEX offers no scope for access earnings to contribute to the sunk, CAPEX investment. An expectation of such a costing methodology would prevent the initial investment from taking place. We therefore discount SRMC and SRIC as viable cost measures, and assume at least some proportion of sunk CAPEX cost must be incorporated. We discuss four possible measure of relevant cost: 1. Long-Run Incremental Cost (LRIC); 2. Long-Run Incremental Cost, plus a mark-up (LRIC+); 3. Fully Allocated Costs (FAC); and 4. Stand Alone Costs (SAC). As portrayed in Figure 2, these measures differ only in their treatment of the common costs of the conglomerated investor. In all cases, the aim is to derive a measure of the in scope costs attributable to the regulated activity. This will typically mean an approach which uses a cost measure not lower than LRIC and not higher than FAC. 14 At any price above this, a competing firm can gain by lowering their price by a small amount and capturing the entire demand. At any price below, the firm would gain by not producing the marginal unit. Page 14

20 Figure 2: Comparison of commonly proposed cost measures and their treatment of common costs % mark-up for common costs Allocation of common costs All common costs relevant to the fibre business Direct CAPEX costs to fibre business Direct CAPEX costs to fibre business Direct CAPEX costs to fibre business Direct CAPEX costs to fibre business OPEX OPEX OPEX OPEX OPEX SRMC ( 0) SRIC LRIC LRIC+ FAC SAC Cost of adding 1 unit of access Short-run cost of operating the fibre business Long-run cost of adding the fibre business Long-run cost of adding the fibre business plus a contribution to the common costs of the conglomerate Cost of establishing a new firm for just the fibre business Source: CRA LRIC A LRIC measure considers only those costs that are directly associated with the production of the business increment. In this case, that is the additional (long-run) cost that the firm would incur to provide the increment, holding its other activities constant. This approach therefore excludes costs that are incurred regardless of whether the increment is produced (even if they are relevant to the increment when it is produced i.e. it excludes pre-existing shared costs already employed by the conglomerate business) LRIC+ A LRIC+ measure takes the LRIC measure (as above) but includes a mark-up (e.g. some percentage) to account for a contribution to the conglomerate s common costs by the new business increment FAC An FAC measure includes LRIC (as above) but also some proportion of all common costs. This proportion may be based on a number of different criteria, such as relative revenues, direct costs, or some measure of output. Critically, the sum of FACs across all the firm s activities must equal the firm s total costs (that is, the firms total costs are fully allocated across its activities) SAC A SAC measure includes the costs the firm would incur if it produced only the increment in question (i.e. it includes the entirety of all costs relevant to the production of the increment). This treatment of 15 The + may also refer to e.g. network externalities. Page 15