Why is Europe lagging on next generation access networks?
Fibre-based next generation access (NGA) roll-out across the European Union is one of the goals of the European Commission’s Digital Agenda strategy, however, there remains considerable uncertainty about how the roll-out goal can best be achieved.
Footnotes, references, and the technical annex can be found in the PDF version of this publication.
Fibre-based next generation access (NGA) roll-out across the European Union is one of the goals of the European Commission’s Digital Agenda strategy. By enabling entirely new broadband services, NGA networks have the potential to trigger productivity gains on a massive scale. There remains considerable uncertainty, however, about how the roll-out goal can best be achieved.
The underlying differences between the economics of copper-based and new fibre-based broadband infrastructures should lead to a revision of the regulatory framework for telecommunications markets. While the current regulatory measures have been useful in the past decade to sustain competition and facilitate entry into a market with already-existing infrastructures, the need to create new, much faster broadband networks calls for a rethink of the scope and strictness of regulation.
Next generation access (NGA) networks, a fibre-based high-speed broadband infrastructure, are a general purpose technology with the potential to trigger productivity gains on a massive scale. These gains might take years to accrue, because new applications and new organisational and production designs that use NGA networks need time to be developed. Nevertheless, we consider wide NGA infrastructure roll-out to be welfare enhancing and that it should therefore be an objective of the European Union. This is consistent with the view taken by the European Commission. The Commission’s Digital Single Market strategy, adopted on 6 May 2015, promises that in 2016 an ambitious overhaul of the telecoms regulatory framework will be proposed, and will focus, among other aims, on investment in high-speed broadband infrastructure (European Commission, 2015).
EU markets for electronic communications networks and services are regulated according to the 2002 eCommunications framework. Among its main provisions is the mandated sharing of telecommunications infrastructure, which allows entrants to compete with incumbents. The eCommunications framework was created for copper-based legacy networks, but has been extended to cover NGA networks, which provide users with radically improved broadband access to data, based on fibre-optic cable technologies. Academic research shows that, among various cost and demand side factors that have an impact on the deployment of NGA networks, regulatory access policies play a crucial role. In this Policy Contribution, we discuss how these regulations – devised at EU level and implemented at national level – might affect the deployment of NGA networks.
We start with an analysis of recent NGA trends in EU member states, and assess if the European Commission’s policy goals are being met1. We then review the experience with broadband deployment in the EU and other selected economies. We discuss the differences and similarities in the economics of the ‘old’ broadband (using legacy networks based on copper and coaxial cables) and the ‘new’ broadband (using NGA networks), and assess the extent to which lessons learned about regulation of legacy networks can be transferred to NGA networks. We then discuss the current regulatory framework for NGA networks and in Box 1 on page 9 highlight case studies of EU member states that did particularly well in terms of NGA deployment. This enables us to highlight the key trade-offs involved in regulation of NGA networks and to formulate a set of recommendations to policymakers.
Our key finding is that the underlying differences between the economics of the ‘old’ and the ‘new’ broadband infrastructures should lead to a revision of the current regulatory framework for telecommunications markets. While the regulatory framework for copper-based networks was useful in the past decade to sustain competition in a market with already existing infrastructures, the need to create new broadband networks calls for a rethink.
2 NGA coverage, penetration and take-up rates
Figure 1 shows NGA coverage and NGA penetration for 25 EU member states. NGA coverage is measured by the total number of lines that enable fast broadband internet access that are available to homes or businesses (‘homes passed’2) . Network coverage thus refers to the number of consumers that in principle have access to fast broadband. NGA penetration refers to the actual number of NGA subscribers. Figure 1 captures almost the entire period of NGA deployment in EU member states and shows that the coverage and the penetration follow a more or less dynamic diffusion process. Even though some EU member states do particularly well in terms of NGA deployment, as Figures 1 and 2 show, Europe lags behind a number of non-European nations, including Japan, Korea and the United States (FTTH Council Europe, 2015; Yoo, 2014; OECD, 2013).
In order to stimulate greater NGA coverage and penetration, the European Commission’s Digital Agenda strategy, of which the Digital Single Market plan is a part, said that the EU should “ensure that, by 2020, (i) all Europeans have access to much higher internet speeds of above 30 Mbps and (ii) 50 percent or more of European households subscribe to internet connections above 100 Mbps” (European Commission, 2010). In Figures 1 and 2, the horizontal lines at penetration and coverage values equal to 0.5 and 1 mark the Digital Agenda goals. The mean value of fast broadband coverage already equalled approximately 100 percent in 2014, which on average fulfils goal (i) of the Digital Agenda for the 25 member states in our analysis. Ultra-fast broadband coverage, however, is much lower with an average mean value of only about 35 percent (Figure 2)3.
Figure 1: Fast broadband coverage and penetration rates per household in 25 EU member states for the years from 2005 to 2014
Source: FTTH Council Europe. Note: Data are available from FTTH Council Europe for the group of the EU25 member states (Malta, Cyprus and Croatia are excluded). The coverage for Luxembourg is 2.31 in 2014, which is not reported, because we have restricted the presentation to an upper-bound of 2 for illustrative purposes.
Figure 2: Ultra-fast broadband coverage and penetration rates per household in 25 EU member states for the years from 2005 to 2014
Source: FTTH Council Europe. Note: Data are available from FTTH Council Europe for the group of the EU25 member states (Malta, Cyprus and Croatia are excluded).
There also appears to be a substantial gap in penetration rates. The average NGA penetration rates for fast broadband and ultra-fast broadband are only 25 percent and 10 percent, respectively. Overall, when comparing fast broadband deployment in terms of coverage and penetration with ultra-fast broadband, one finds similar growth patterns but notably lower levels for the latter in most EU states. In some countries (such as Austria, Belgium, Germany and the United Kingdom), ultra-fast broadband deployment has started on very small scale or not at all. Instead, operators focus on much less cost-intense upgrades to the traditional copper and coaxial cable technologies. Hence, the Digital Agenda’s goal (ii) for connection speeds above 100 Mbps is still far from being realised.
In fact, numbers on NGA coverage and penetration overestimate the actual coverage and penetration because of double counting in many member states. In particular, in urban areas, there is double counting of homes passed by cable television operators and traditional telecommunications operators. In addition, business establishments, which promise high returns, might be passed by parallel NGA infrastructures. This is why the coverage levels in Figures 1 and 2 sometimes exceed 100 percent. This is the case for Germany, for example, but in fact only 74.6 percent of all German households actually had access to ≥ 30 Mbit/s fixed-line broadband technologies at the end of 2014 (TÜVRheinland, 2014).
3 Legacy-based infrastructure regulation
The EU regulatory framework for electronic communications markets has established a broad system of cost-based access pricing since voice telephony markets started to be liberalised in 1997-98 (European Commission, 1998; European Commission, 2002). This allowed market entrants to offer narrowband voice telephony products, which substantially reduced average call prices in the first phase of sector-specific regulation. The EU regulatory framework then implemented mandated wholesale access to the local loop (European Commission, 2000), thus enabling market entrants to offer consumers narrowband voice and broadband either as bundled or stand-alone services. Incumbent operators were required to ‘unbundle’ the legacy access infrastructure, ie to rent parts of it to new entrants based on cost-based access charges, and to allow entrants to collocate their infrastructure at a switching office close to the users.
Four access options have provided entrants with a variety of viable business cases, thereby allowing them to effectively compete with the incumbent in offering broadband services:
- Full unbundling, which implies that entrants rent the entire local loop to offer a voice-broadband bundle to consumers;
- Shared access (ie line sharing), which means that entrants rent the upper line’s bandwidth in order to offer broadband only;
- Bitstream access, ie mandated wholesale access to the incumbent’s networks at a more elementary level of the value chain; the access point for bitstream is beyond the local loops and usually closer to the entrants’ core network, but entrants also have less ability to differentiate the broadband services based on bitstream compared to full unbundling or line sharing;
- Simple resale, which requires the least network investment; resale means that the access-seeking operators receive and resell a wholesale input from the incumbent without any scope for technological product differentiation (Briglauer et al, 2015).
Figure 3 shows the hierarchy of access options, as viewed by the European Commission. This view was based on the ladder-of-investment hypothesis, according to which the different access options lead to a gradual increase in infrastructure investment by new entrants over the liberalisation period (Cave, 2006; Cave and Vogelsang, 2003). Thus, by facilitating entry, access regulation was intended to promote broadband take-up by keeping consumer prices in check and simultaneously creating incentives for new infrastructure investment. Ultimately, entrants were expected to deploy their own access infrastructure and engage in fully-fledged facilities-based competition with incumbents4.
Figure 3: Hierarchy of wholesale access remedies
Source: European Commission (2010).
More than a decade of broadband access regulation in Europe has shown, however, that the ladder-of-investment hypothesis works mainly for the lower rungs of the investment ladder. This calls for comparatively low investment requirements, especially, for moves from resale to bitstream access. Empirical evidence suggests that mandatory local loop unbundling has not led entrants to ramp up access network infrastructure as expected. In fact, unbundling might have even reduced total industry investment, meaning that investment by entrants has not been sufficient to offset the unrealised investments of incumbents5. The fact that the ex-ante access obligations did not result in the service-based entrants deploying enough access infrastructure can be interpreted as the “natural outcome of the economics of fixed broadband access” (Vogelsang, 2013). In essence, by depressing the prices of retail broadband services, mandatory wholesale access policies render investment in competing technologies, such as cable broadband, less attractive. Similarly, telecoms entrants often prefer to rent unbundled elements at mandated access rates rather than build their own parallel infrastructure. As access conditions become more favourable, entrants’ opportunity costs for self-deployed access infrastructure increase and “wait and see strategies” (Guthrie, 2006) become more attractive. In our view (see section 5) this is not necessarily bad for economic welfare, especially if the unrealised investment would merely duplicate existing access network infrastructure. But it should be acknowledged as an outcome of mandated wholesale access. This lost investment is also likely to be a by-product of regulation of NGAs, as we discuss in the next section.
In contrast to the EU, in 2005 the US regulator largely removed the unbundling regime, which was rather comprehensive at that time. The decision to deregulate broadband markets was based on the almost nationwide duopoly market structure, which was considered to ensure enough competitive pressure. In fact, facilities-based competition emerged in the US telecommunications market almost entirely because of the cable operators with prior access infrastructure (Vogelsang, 2015)6. With the ladder-of-investment hypothesis, the European Commission took the opposite approach, although it has never been coherently applied (Vogelsang, 2015).
Although the EU broadband access regulations have been only partly successful in facilitating infrastructure investment and facilities-based competition, the eCommunications framework should not be judged as ineffective, especially in international comparison. Table 1 shows that all OECD countries experienced a similar pattern of growth of first-generation fixed-line broadband subscriptions from 2009-13. A country’s position in the ranking highly correlates with its economic development level: the richest countries generally have higher subscription rates per 100 inhabitants. The level of economic development alone does not tell the full story, however. Japan and the US are only slightly above the OECD average. In contrast, the major western European economies, including France, Germany, the Netherlands and the United Kingdom, which are comparable to US in levels of economic development, do much better in terms of ‘old’ broadband penetration. The Scandinavian countries, which have a long tradition of public subsidies for broadband and a high level of consumer ICT-affinity (Briglauer and Gugler, 2013), also beat the US and Japan in the ranking. Overall, the EU access regulations seem to have worked relatively well in facilitating broadband take-up by consumers.
The favourable outcome in terms of European broadband penetration does not mean that the fundamental trade-off inherent in wholesale access policies does not work. Reducing market prices through mandated access fosters the take-up of broadband services, but hinders investment in infrastructure. Note that the broadband penetration figures (Table 1) capture the result of both the demand and the supply factors and thus simply illustrate how this trade-off played out in the past. The underlying differences between the economics of ‘old’ and ‘new’ broadband infrastructures, however, are likely to tilt the trade-off in an unfavourable direction by aggravating the negative impact on investment of access regulation in the case of NGA networks. Investment in NGA networks, in particular ultra-fast broadband, is more likely to suffer from the hold-up problem because investment in the ‘old’ broadband access upgrade was more incremental, while it is very substantial for ‘new’ broadband. Thus, the situation is very different on the supply side of the equation: the software and equipment needed to upgrade the legacy network to support first-generation broadband services were much cheaper to install than new physical access infrastructure such as fibre-optic cables. On the demand side, there is still a significant uncertainty about the willingness of consumers to pay for higher-speed broadband access. Because the regulatory approach to legacy networks has been by and large extended to the NGA networks, we expect underinvestment in the NGA infrastructure in the EU, especially in ultra-fast broadband.
4 NGA infrastructure regulation
The directives of the eCommunications framework have been supplemented by European Commission recommendations on “regulated access to next generation access networks” (European Commission, 2010) and “consistent non-discrimination obligations and costing methodologies to promote competition and enhance the broadband investment environment” (European Commission, 2013) to form the relevant EU regulatory framework for the emerging NGA infrastructure. This framework imposes rather comprehensive and strict access obligations compared to the US (Briglauer and Gugler, 2013). The Commission’s approach to the regulation of NGAs can be seen as a direct extension of the eCommunications framework created for the ‘first-generation’ legacy networks.
The regulatory remedies dealing with access to NGA infrastructure fall into two broad categories, passive and active. The passive remedies grant access to civil engineering infrastructure – in particular ducts – which significantly lowers the costs of laying cables for entrants. To varying degrees, most EU national regulators implement the passive remedies7. Active remedies require the dominant operators to provide access to their physical network infrastructure. This includes wholesale ‘bitstream’ access to NGA infrastructure, which is much less investment intensive for potential access seekers than laying their own cables. In contrast to passive remedies, member states implement the active remedies very differently in terms of timing and scope (Briglauer et al, 2015). Some EU countries started to introduce regulations on wholesale access (such as Belgium, Denmark, Italy, Netherlands or Spain) and fibre unbundling (Finland and the Netherlands) already in 2011 (WIK, 2012). Some, such as Belgium and Germany, have still not decided, at time of writing, on access conditions or prices. The scope of access regulations varies widely depending on the network architecture and the competition conditions in national markets. The national regulatory authorities take also different positions on how to price NGA-specific capital costs. Some (such as Denmark, Hungary and Ireland) have followed the Commission’s recommendation and have applied incremental cost-oriented access prices. Others, such as Bulgaria, Estonia and Spain, so far have not8.
As we have discussed, access regulations can significantly hold-up investment. We expect this problem to be less acute in the case of passive remedies, especially in areas where the ducts have already been built. Active remedies, however, especially when coupled with incremental cost-based access pricing, can substantially impede fibre deployment, because of the hold-up problem.
As well as affecting total investment volumes, active remedies involved in NGA regulation can also be expected to have an impact on which companies invest and which do not. The asymmetric nature of fibre unbundling as required by the eCommunications framework, to a great extent deters investment by incumbents, most of which still hold significant market power, according to the European Commission’s methodology. In fact, it appears that the incumbent operators’ legacy-like market dominance is fading. For instance, as of December 2014, investment by incumbents was responsible for only 27.7 percent of EU homes passed by ultra-fast connections. Among entrants, municipalities and utilities were responsible for 4.9 percent, and alternative operators, mostly cable companies, were responsible for the largest part, 67.4 percent (FTTH Council Europe, 2015; the remaining share was accounted for by other firms such as housing companies). NGA infrastructure owned by entrants is not subject to regulated access.
Similar to the legacy-based infrastructure regulation, the Commission foresees multilevel access to the NGA infrastructure that belongs to operators holding significant market power, with varying degrees of investment for access seekers. The ladder-of-investment approach is again considered as a guiding principle for regulating NGA infrastructure (European Commission, 2010, recital 3: “The appropriate array of remedies imposed by a national regulatory authority should reflect a proportionate application of the ladder of investment principle”). Considering the deployment costs, the duplication of NGA infrastructure is, however, unlikely to be more common than the duplication of legacy networks. Thus, reaching the ultimate goal of infrastructure-based competition on NGA networks (last rung of the ladder-of-investment) is unlikely9. Compared to legacy networks, the unbundled fibre access products are much closer to previous bitstream access products, ie at a lower point of the value chain (OECD, 2011).
Taking the ladder-of-investment perspective, national regulatory authorities should aim to induce optimal movements up both the legacy and the NGA ladders, and optimal incentives for migration between the ladders (Cave, 2010). Even if the NGA infrastructure is not regulated, the copper access charges imposed on the old infrastructure will still exert a crucial impact on the incentives for NGA investment (Bourreau et al, 2012). If a broad NGA roll-out is considered to be welfare enhancing, incumbents and entrants should ideally switch from investing in the ‘old’ infrastructure and requesting legacy-based access products, to investing up the ‘new’ NGA ladder. However, operators are not only confronted with the regulatory access policies imposed under the old and new ladders, but also experience uncertainty about the demand for new services that will be possible once the NGA networks are in place. Thus, the NGA investment decision is far more complex than it was for first-generation broadband on the legacy networks.
The theoretical research on migration from legacy to fibre-based infrastructure, points out that the regulation of legacy networks affects in many ways the incentives to invest in fibre infrastructures (Bourreau, Cambini and Dogan, 2012; Inderst and Peitz, 2012; and Bourreau et al, 2014). Moreover, the expected effects significantly differ for incumbents and new entrants. Incumbents might invest more in new infrastructure when the access price to the old network is low, and thus the opportunity cost of the investment is low. This effect might however be dampened or even reversed, if the low prices discourage customers from switching away from the legacy network. By contrast, entrants will most likely speed up the deployment of new infrastructure when access charges for copper are high. In general, NGA investments ‘cannibalise’ economic profits from first-generation broadband services provided via legacy infrastructure, which might thus reduce profitability and the incentive to invest in NGA infrastructure. This effect is reinforced if conventional broadband services enjoy broad consumer acceptance, which establishes non-negligible switching costs for consumers and hinders migration to the new technology, unless its incremental benefits are sufficiently large and transparent. In sum, there is no clear effect of regulated access to legacy networks on NGA investment.
A social-welfare perspective further suggests that strict regulation to ensure access to legacy networks safeguards consumers against paying higher prices for faster broadband services they might not really need. In areas where NGA networks will replace legacy networks, the legacy-like quality of service (and the related access regulation) could be sustained in order to prevent those operators that lack their own infrastructure from becoming stranded. These operators would then provide the ‘old’ broadband services, but over fibre. Needless to say, as legacy networks will continue to exist in parallel with NGA networks, consumers will still be able to use the ‘old’ broadband if they do not need higher speeds.
Interestingly, 10 out of the top 20 ranked European economies in terms of ultra-fast broadband penetration are eastern European countries (including Russia and Ukraine; FTTH Council Europe 2015). There, the lack of well-established legacy infrastructure meant there was an opportunity to directly deploy high-end network architectures at comparatively low cost (Briglauer and Gugler, 2013; Briglauer, 2014).
In line with the above analysis, the empirical research “indicates a negative impact of ex-ante access regulations on NGA investment incentives” (Briglauer et al, 2015). This corroborates the results of the literature that studies first-generation broadband, surveyed in Cambini and Jiang (2009), which finds similar albeit less-pronounced empirical evidence. There is, however, no empirical research that specifically considers the joint presence of copper-based-access and NGA regulations and their joint impact on investment incentives.
5 Discussion of the key policy issues for NGA deployment
The experience with first-generation broadband deployment in Europe shows that, in general, lowering market prices through mandated access fosters the take-up of broadband services, but might hinder investment in infrastructure. Against this backdrop, the unbundling of legacy networks in Europe has worked rather well. In particular, the penetration rates compare favourably to the leading economies worldwide. Furthermore, the unrealised infrastructure investment, which is the downside of access regulation, might not necessarily have been bad for economic welfare, especially if the investment would have merely led to duplication of the existing access network infrastructure. The European experience shows, however, that access policies suppress fully-fledged facilities competition and, as a result, tend to be permanent rather than transitory. This is also likely to be true for NGA networks and should be acknowledged as an important constraint of mandated fibre access.
Will the qualified success of regulated access be replicated in the future? Our analysis shows that this is doubtful. For NGA networks, the underlying differences between the economics of the ‘old’ and the ‘new’ broadband infrastructures are likely to aggravate the negative impact of access regulation on investment. Investment in NGA networks is more likely to suffer from the hold-up problem, because laying the fibre-optic elements of the local loops needed for ultra-fast broadband is much more expensive than the upgrades needed for the ‘old’ broadband. As a result, under the current regulatory regime we expect continuing underinvestment in NGA infrastructure and therefore underachievement of Digital Agenda goals, especially for the more costly ultra-fast broadband infrastructure.
Although we have focused on sector-specific access regulations, NGA deployment is also influenced by other public policies. Most notably, public funding programmes, housing regulations11 and regulations on net neutrality will have a strong influence on the development of market-based business models and firm profitability and hence on NGA investment. Public subsidies (direct or indirect) for NGA infrastructure are an effective way of stimulating investment, as the examples of Japan and Luxembourg show most clearly12. EU structural funds provided to the eastern European countries might have also played an important role. More generally, subsidies appear essential in areas where private investment is not profitable but NGA roll-out might still be desirable in view of externalities or equity motives. Other forms of subsidy, such as publicly-sponsored eHealth programmes might also give a push to the deployment of fast broadband infrastructure. Net neutrality regulations might render infrastructure investment more profitable if network operators are given more leeway to provide high-quality access to content providers at premium prices.
Our basic assumption is that wide NGA roll-out will be welfare enhancing for Europe because of the potential positive externalities, which are likely to increase over years and decades. This leads us to formulate three recommendations to the regulators.
First, while a nationwide NGA roll-out might be efficient and justified by equity concerns, we recommend de-emphasising fully-fledged facilities-based competition as the ultimate goal of the eCommunications framework. Large-scale facilities based competition has proved unrealistic in most member states and, in fact, not consistent with the economics of mandated fixed-broadband access. Instead, regulators should acknowledge that access regulation is likely to persist for the foreseeable future and should commit to a clear methodology for calculating access prices. The focus of the regulation should shift towards efficient investment, meaning provision of a single infrastructure needed for ultra-fast broadband, especially in areas where the natural market structure is a monopoly. Note that efficient investment might also include mobile broadband to a greater extent in the mid and long term, particularly in areas where no wired NGA infrastructure exists.
Second, the eCommunications framework should provide additional incentives to invest in ultra-fast broadband infrastructure. This could for instance be achieved by relaxing cost-oriented pricing of fibre access for a limited period. At this end of this period, consumers would further benefit from service-based competition enabled by a cost-oriented access policy. Co-investment models might also provide a relevant alternative to prevent infrastructure duplication. We concur with Vogelsang (2013), who argued that incentivising efficient NGA investment requires a shift in the regulatory frontier towards softer regulation, including symmetric regulations that
enhance co-investment models.
Third, we recommend taking a more holistic and evidence-based approach to access regulation. As we have shown, there is a range of alternative public policies that might spur investment in access networks. These policies should be given serious consideration. Also, the regulation of access to the legacy copper networks can be expected to have significant effects on NGA deployment. The migration from copper to fibre thus involves an intertwined set of incentives on the demand and the supply sides. It is not clear whether sustaining incremental-cost-oriented access to the legacy networks helps the NGA roll-out. One important reason, however, to maintain the current regulation of the legacy networks, is to safeguard consumers against paying higher prices for ultra-fast broadband services they do not really need. Overall, more evidence is needed as a basis for sound policy advice in this domain.