Analysis final report for OFTEL: International Benchmarking of DSL Services

CONTENTS

Conclusions

Summary

Introduction

Price Comparisons

Service Rollout

Regulatory and Market Overview

Conclusions

Annex A - DSL prices by country

Annex B - National comparisons of the price level of retail services

Annex C - Current and potential dsl service providers investigated

Glossary - Glossary of terms

Conclusions

Scope

This International Benchmarking Study covers Digital Subscriber Line (DSL) services, providing high speed telecoms connections to consumers, in the UK, France, Germany and the USA. A survey of operators and service providers carried out in March 2000 in these countries has identified a large number of DSL services available to residential and business consumers as well as to service providers. Services on offer in different countries and from different operators have been compared on the basis of price and bandwidth (i.e. speed of connection).

Points to Bear in Mind

Comparing the different service offerings and reaching conclusions regarding the "value for money" offered by providers in the UK compared to those elsewhere is not straightforward:

• To date the services on offer in the UK (with the exception of those from Kingston Communications) are trials, not commercial services. These prices might prove to be unrepresentative of future UK commercial offers.

• Also, "value for money" is affected by other factors than just price, in particular, the "quality of service". While the most obvious measure of service quality, "speed", has been included in the comparisons, there are a number of other quality aspects which may in practice be equally important which we are not able to take into account.

In the report we compare services on the basis of both price and bandwidth. Given that we know relatively little either about the importance that consumers will place on bandwidth, or the extent to which bandwidth can be regarded as a proxy for overall "quality of service", it has not been possible to define a composite measure of "value for money" which enables us unambiguously to compare one service offering with another and determine definitively which is "better". Price and bandwidth are therefore considered separately.

We have considered three market sectors:

• residential consumers
• business users
• service providers (who purchase a wholesale product in order to provide service to the end user).

Results

Retail residential DSL services

Regarding the price of services:

• The DSL services offered by UK operators to residential consumers are comparable in terms of price to those offered elsewhere.
• The service offered by Kingston Communications in fact ranks second cheapest amongst our sample when considered purely in price terms. (Although this is only available to 0.3% of the population).
• There is no clear difference in terms of price between countries – the best offer in each country lies within a narrow price range.

Regarding the bandwidths on offer for services targeted at residential users:

• The bandwidth offered by UK operators varies widely:

• Downstream bandwidth offered in other countries is as follows:

We see that the UK and the USA appear to offer the best choice of, and range of, speeds.

In principle we would expect price to increase with increasing bandwidth both because there are some additional costs in providing higher bandwidth and because it might be expected that consumers would be prepared to pay more for a higher bandwidth service. This price/quality relationship (higher price for higher bandwidth) is not in practice seen either between or within countries. This is probably due both to the immaturity of the market and the fact that bandwidth is only one of several measures of quality.

Retail business DSL services

The range of bandwidths offered is much wider for business compared to residential consumers and services have been grouped into "low", "medium" and "high" bandwidth products to allow more meaningful comparison.

There is only one observation of a DSL service offered to UK business consumers, which falls into the "high" bandwidth group. A single observation makes it very difficult to comment on the position in the UK, however we note:

• this single offer is not out of line in terms of price compared to services elsewhere;
• the bandwidth provided is relatively low compared to other products in the group;
• comparing purely with other services with the same bandwidth, we find the price for the UK service lies between US and German offers.

At lower bandwidths, prices in Europe and the USA are not out of line. However, when we consider higher bandwidths we find that:

• fewer high bandwidth products are on offer in Europe compared to the USA
• where high bandwidth products are in offer in Europe they are at higher (often considerably higher) prices than the equivalent in the USA
• US operators are more likely to offer a "symmetric" service (i.e. the same speed down- and up- stream) - which provides the user with a greater range of business applications.

As for residential consumers, there is no clear correlation between price and quality within each of our bandwidth groups (though the average price of services within the "low" bandwidth group is lower than that for the "medium" bandwidth group etc.).

Wholesale DSL services

We have been able to find wholesale prices for only three operators:

• BT’s prices for wholesale DSL products seem to be in line with those offered by Covad, a US newentrant operator.
• Both BT and Covad offer services at prices significantly below those offered by QSC a newentrant operator in Germany.
• Covad offers a much wider choice of bandwidth than either BT or QSC.

Some caution is required in viewing these results. The service offered by BT is not identical to that offered in the US and Germany (it has somewhat lower functionality) which may impact on the comparisons.

It should also be noted that the "retail mark up" (i.e. the ratio of retail to wholesale prices) appears to be lower in the UK than elsewhere. This may imply that:

• commercial retail prices in the UK will in fact be higher than the trial prices
• UK operators will need to find revenues from sources such as advertising
• retail prices in other countries will come down.

Roll out

As well as considering prices and service quality, service availability is also an important feature for a new service. Comparing the data available on current and projected roll out (and assuming that roll out plans are adhered to) suggests that:

• the UK is currently behind the other countries
• France Telecom will be ahead of all the other countries by the end of 2000
• Deutsche Telecom, BT and the US incumbents considered will have achieved a similar level of service availability by the end of 2000 (i.e. around 40% of households).

Looking ahead

The current report provides a snap shot of services as at February/March 2000. Prices and service levels are still evolving, even in the USA, the most mature of these still immature markets. European prices in general are likely to fall (and service levels are likely to rise) towards the levels currently offered in the USA.

Summary

Scope of report

This report presents the results of an international benchmarking study on DSL telecoms services as at March 2000. The study was carried out for OFTEL by Analysys. DSL technology enables data to be transmitted over an ordinary subscriber line, at much higher speeds than have been possible on the traditional public switched telephony network (PSTN). The study investigated DSL service prices and the rollout of the associated service offerings.

In addition to the UK, we have examined the DSL offerings in France, Germany and the USA. These countries provide a good international benchmark against which to compare UK services. It is the aim of this study to collect information on services which have been commercially launched. However, for the UK it has been necessary to collect information on trial DSL services, because of the limited number of services available commercially.

The report covers wholesale and retail offerings in each country (except wholesale in France). By wholesale DSL services, we mean the provision of a DSL service which is only sold to the public through third party service providers and where the service provider takes responsibility for billing and customer management. We have also gathered data on the regulatory environment for DSL services and on the progress made by operators in rolling out DSL services to customers.

Comparisons of DSL Services

We have gathered a large amount of data about services and prices in the UK, the USA, France, and Germany from sources in the public domain. In order to compare prices we have considered an "equivalent" monthly charge in pounds sterling (£) which takes into account both one-off and ongoing charges.

Services differ with regard to the bandwidth (i.e. the "speed" of connection both to and from the user) offered and price needs to be viewed in the context of the bandwidth provided. Our chosen measure of bandwidth in comparing services is:

• downstream (i.e. to the user) peak bandwidth, for residential users
• downstream peak bandwidth, for "low-end" business users (i.e. users requiring only relatively low bandwidth services)
• a measure which takes into account both downstream and upstream (from the user) bandwidth, for "high-end" business users (i.e. users requiring relatively high bandwidth services).

We note the following important points which limit the quantitative conclusions which can be drawn from this study.

• Comparison of DSL services for the UK with those in other countries is at this stage somewhat tentative, because all the UK services offered (apart from those of Kingston Communications) are trials, not commercial services. These prices might prove to be unrepresentative of future UK commercial offers.

• The user’s needs will be affected by other factors than just price, in particular, the "quality of service". The most obvious measure of service quality is speed, however, there are a number of other quality aspects which may in practice be important:

– In particular, the extent to which the required bandwidth can actually be obtained in busy periods (bandwidth reservation) could be the main contributor to the final service standard. The key parameter is the contention ratio – the ratio of the potential maximum demand to the actual capacity. The higher the contention ratio, the greater the number of users that may be trying to use the actual capacity at any one time, and therefore the lower the effective speed offered, especially at peak times.

In the following, we consider comparisons of DSL services on the basis of both price and bandwidth (note data on other quality aspects such as contention ratio is not consistently available). We present data on price and bandwidth separately, and from this draw qualitative conclusions regarding which service offerings represent the best value for money. Offers with a high speed and a low price (or a low price for the speed offered) are considered to be "good value" compared with the other offerings.

Retail residential DSL services

Exhibits 0.1 to 0.3 below present data on the residential retail DSL services offered in the countries we have studied. Exhibit 0.1 shows the services arranged in order of the monthly charge, Exhibit 0.2 shows the services arranged in order of bandwidth and Exhibit 0.3 presents both price and bandwidth information in a single chart.

Exhibits 0.1 to 0.3 illustrate that:

• there is no clear correlation between price and bandwidth – even within a country low price/high "quality" services may be offered alongside high price/low "quality" products.
• the "best value" deal appears to be the USA SBC offering which provides a relatively high bandwidth service for a low price (note, however, that the bandwidth illustrated is not available to all SBC’s customers).
• In terms of the UK offers:

- Kingston appear to offer a low price but low bandwidth service. (This service is only available to customer’s in Kingston’s operating area, 0.3% of the population)

- UK Demon’s trial service offers a high bandwidth service at a relatively good price. Note, however, that the bandwidth illustrated is in fact the upper end of a range of potential bandwidths. In the context of BT’s wholesale prices, there appears to be some doubt as to whether this could continue to be offered as a commercial product.

- UK Freeserve’s trial service appears to offer an average price, although the bandwidth is in the lower half of the range of bandwidths offered by the services included in our study.

For residential services, US service providers offer commercial services for monthly equivalent prices of around £30 to £75 per month. For these prices the bandwidth provided ranges from 380 to 1500 kbit/s (i.e. there is little consensus on the "speed"). These are flat rate prices ("all you can eat").

European DSL services are, on average, similar in terms of price level to services in the USA – e.g. the best offer in each country lies within a narrow price range - although the range of prices on offer varies more widely in Europe. The range of prices charged by UK service providers, though mainly for trial services, is similar to (though narrower than), those charged in the USA at £30 to £50 per month. As in the USA, the majority of charges are flat rate ("all you can eat"). Exceptions to this are some services providers in Germany which have introduced charges based on the volume of data used or the time spent online. The bandwidth offered by UK operators varies widely with Kingston offering the lowest downstream bandwidth (256kbit/s) of the whole sample and Demon offering the highest (up to 2000kbit/s). This compares to downstream capacity offered in other European countries of: 500 to 512 kbit/s in France, and 768 kbit/s in Germany (although residential consumers could take up a business service if they were willing to pay a higher price). The UK and the USA appear to offer the best choice of and range of speeds. Therefore, in price terms the UK appears comparable to the other countries studied. In terms of speed the UK services differ widely offering speeds both at the top and bottom of the range provided in other countries.

The lack of clear correlation between price and bandwidth both within and between countries is somewhat surprising. There are a number of possible reasons for this.

• This may reflect the immaturity of the market: i.e. consumers are unaware of better value offers or cannot get them in their area; and/or suppliers are at an early stage in pricing their products and are doing so on very different bases.
• This may reflect the fact that bandwidth is in incomplete measure of the service actually provided to the consumer.

Retail business DSL services

Exhibits 0.4 to 0.6 below present all the retail business services arranged in order of the monthly charge and grouped into three bands according to their downstream bandwidth (for ease of analysis): up to and including 512kbit/s; over 512kbit/s to 2Mbit/s, and 2Mbit/s or more. The downstream bandwidth of each service is superimposed on this series for Exhibit 0.4 while our combined downstream/upstream measure has been superimposed on Exhibits 0.5 and 0.6. Note that in order to limit the number of charts we have not included charts showing price and bandwidth separately.

The diagrams show, overall, no direct relationship between price and bandwidth for DSL services. They do, however, allow for easy comparison of service price and service bandwidth. In the first and third cases there is a fairly clear "best value deal" which is at the bottom of the price range and at or towards the top in terms of bandwidth. For example, Exhibit 0.4 shows that World-NET from France offers the cheapest business DSL service in the range 0kbit/s to 512kbit/s and provides a high capacity service in comparison with most of the other services in this bandwidth range. Again we note that that many other factors will affect quality of service.

US service providers offer services at an much wider range of speeds for business customers than for residential customers and many of these services are symmetric, benefiting the higher end business users in particular. Prices are again flat rate prices. There is much less conformity, however, in the prices offered in the USA, particularly for higher speed and asymmetric services (i.e. where the downstream capacity to the customer is much higher than the upstream capacity to the network). Prices range from £40 (without Internet access) to £634 per month for an asymmetric 7Mbit/s service (which includes Internet access). For the symmetric services there is more consensus. For example, the price of a symmetric service at 512kbit/s varies between £140 and £200 per month.

In Europe some operators come close to matching the price levels seen in the USA at lower bandwidths. However, at higher bandwidths some service providers are significantly more expensive. For example, in Germany the charge for an asymmetric service with a peak downstream speed of 6Mbit/s ranges from £1720 to £2500 per month. For the UK we only have data for Easynet’s trial services – at nearly £670 for a 2Mbit/s service, it lies between those prices charged in the USA and in Germany. However, because this is a trial service and we have only one observation for the UK, we believe it would not be prudent to draw any conclusions from this information.

There are also considerable differences in the degree of asymmetry of the DSL services offered in each of the countries studied. Many of the service providers in the USA offer symmetric services, i.e. services with a ratio of downstream to upstream capacity of 1:1. This may reflect the market segment (high-end business users), at which the products are aimed. Asymmetric services are also offered. The picture in Europe is mixed.

• In the UK the ratio is closer to 2:1 which suggests a compromise between the needs of high-end business users and the other segments of the market.
• In France the ratio is 4:1 and this may reflect a target market of residential users and SMEs (as low-end business users).
• Deutsche Telekom’s ratio of 10:1 may reflect the fact that it has other services which it believes are more appropriate for the high-end business market than ADSL.

Wholesale DSL services

Exhibits 0.7 and 0.8 below show the data we were able to collect on wholesale services in Germany (QSC, a newentrant), the UK (BT) and the USA (Covad, a newentrant) arranged in order of the monthly charge and grouped into two bands (for ease of analysis): up to and including 512kbit/s; over 512kbit/s to 2Mbit/s. The downstream bandwidth of each service is superimposed on the first series and the combined measure of downstream and upstream bandwidth is superimposed on the second.

In order to be able to make a consistent comparison, we have chosen BT’s IPStream product which bears the closest resemblance to the wholesale products offered in the USA and Germany. In addition, the inclusion of BT’s DataStream and VideoStream services within this analysis would require us to make assumptions which would be difficult to substantiate. For example, for any one service provider we would need to estimate the number of BT local exchanges from which services are offered, the number of customers served at each local exchange and the distance of each such local exchange from the service provider’s nearest point of presence.

On the basis of the data we managed to collect, UK wholesale services, while offering a lower range of bandwidths, are generally cheaper than in the USA. Services in Germany are more expensive than in the UK and the USA. However, the services on which we collected data in the USA and Germany are symmetric DSL services, as opposed to the asymmetric DSL service provided in the UK (symmetric services offer greater benefit to some users). In addition, the wholesale services in the USA and Germany provide end-users with static IP addresses, which enable them to connect servers to the Internet in order to carry out ecommerce, for example. In contrast, BT’s IPStream provides the end-user with dynamic IP addresses which are unsuitable for connecting servers to the Internet, which we expect to be one of the popular applications for business DSL services.

We can also draw some tentative conclusions on the mark-up that service providers add to their wholesale prices. In the UK, the margin for Freeserve is around 27%, while at a bandwidth of 512kbit/s, Demon’s offer would appear to provide a margin of 5%. However, for Demon’s offer of a 2Mbit/s service, (the upper end of the range it is offering), the trial retail price appears to be significantly below the price of BT’s wholesale product. These prices were announced as trial prices with no guarantee that they will remain static.

In the USA and Germany the mark-up is much higher, on average 60% (adjusting prices pro rata for differences in bandwidths in Germany), which suggests that either service providers in other countries could reduce their mark-ups and still offer profitable services, or that UK service providers will have to increase their mark-up over wholesale prices, unless they have other sources of revenue to subsidise the costs of providing the service.

Regulation and competitive environment

Service providers are faced with a number of options in providing DSL services. They can either:

• provide retail or wholesale DSL services over their own network if they own or build a copper local access network
• lease an operator’s network, through local loop unbundling and provide wholesale services to other service providers, or retail services to the public
• take a wholesale service from an operator or wholesale service provider and provide retail services to the general public.

We believe that it is necessary to discuss the regulatory framework and the competitive environment in order to understand the rollout plans of different service providers. In this section we summarise the regulatory and competitive environment of each country in turn.

France

Competition has been slow to develop, and only a small number of France Telecom’s competitors are offering rival Asymmetric Digital Subscriber Line (ADSL) Internet access services. In the retail sector, these include Club Internet, Easynet and World-NET. In addition to the subscription fee they pay to these ISPs, customers must also subscribe to one of France Telecom’s Netissimo services, which provide ADSL access, but not Internet access. Other ISPs, for example AOL, are running trials of ADSL access. AOL France, for example, began such a trial in Monaco in January 2000 in collaboration with Cegetel and Monaco Telecom.

Wholesale services are now also being offered by France Telecom’s competitors, including ISDnet, which launched its DSL Connect service in December 1999, (although we have been unable to collect definitive data on the service). However, other potential competitors have been reluctant to enter the market until unbundling is initiated.

Germany

The early unbundling of Germany’s local loop has allowed competition to develop in this market considerably earlier than elsewhere, even though many new entrants have claimed that the price of unbundled local loops in Germany is too high. Major pan-European network operators, for example KPNQwest, are also launching DSL services earlier in Germany than in other countries. When announcing the launch of its services, KPNQwest said that the reason for the earlier launch in Germany was the legal requirement, already in place, for Deutsche Telekom to offer access to its local loop infrastructure.

Of the providers leasing local access from Deutsche Telekom, the majority are offering DSL access to the business community. These include prominent ISPs such as TCP/IP GmbH. At the end of November 1999, QSC began offering the first wholesale DSL services in Germany in Cologne by leasing capacity on Deutsche Telekom’s local access network. QSC plans to extend its services nationwide by the end of 2000.

UK

Commercial trials of ADSL access are being offered by a number of ISPs which have purchased BT’s wholesale DataStream or IPStream services. These ISPs include Freeserve, Demon Internet and Easynet. Other service providers have also announced their intention to offer DSL services. Furthermore, BT’s own ISP, BT Internet, is offering ADSL services to consumers and businesses.

There is evidence that BT and its wholesale customers will soon see competition from other local access network operators which plan to trial ADSL access over their own infrastructure. NTL, for example, which will be the largest cable operator in the UK following its merger with C&W, announced in December 1999 that it would run trials of ADSL services over its own local infrastructure to businesses in Surrey. However, design of the cable companies’ networks may limit the scope for launching commercial services. (They do, of course, have the ability to offer broadband access services through cable modems).

Companies which do not possess local access networks, will have the ability to offer DSL services by leasing BT’s network (i.e. purchasing an unbundled local loop), from July 2001, according to the current timetable. This will make service providers less dependent on the rollout of BT’s own DSL infrastructure.

USA

A large number of commercial DSL offerings are available now in major urban areas across the USA. A number of the largest ISPs in the country are now offering DSL services by partnering with network operators. In June 1999, for example, Mindspring announced a partnership with BellSouth Corp, permitting Mindspring to use BellSouth’s DSL facilities for its own launch of DSL access in Atlanta.

The Federal Communications Commission’s (FCC) decision to allow line sharing in November 1999 has prompted more competitive local exchange carriers (CLECs) and ISPs to launch residential-focused DSL services. Prior to the ruling, alternative DSL providers had argued that the cost of installing a second line was a barrier to the extension of their services into the more price-sensitive residential sector. However, following the FCC’s ruling, Northpoint, a wholesale DSL provider, announced that it was trialling a consumer-class DSL service through selected ISPs nationally.

Competition has developed not only in services sold directly to the end-user, but also in wholesale DSL provision, and there are a number of large intermediaries in the USA which have specialised in negotiating with incumbent local exchange carriers (ILECs) for access to their local loop infrastructure and then reselling this access to ISPs. Covad is one such company, and by March 2000 was able to offer wholesale ADSL access services to ISPs in 62 of the largest metropolitan areas in the USA. These intermediaries have also been successful in attracting the business of major ISPs, which benefit from being able to negotiate with one wholesale DSL provider, rather than having to deal with a number of ILECs.

Rollout

The rollout of DSL services in the European countries studied is strongly linked to the rollout of DSL services by the incumbent, although to a lesser extent in Germany, where QSC has been able to purchase unbundled local loop capacity from Deutsche Telekom. By comparison, in the USA wholesale providers using unbundled local loops, such as Covad, Northpoint and Rhythm Networks, have proved a major force in extending the rollout of DSL on a nationwide basis. On the basis of this evidence, it seems that local loop unbundling, particularly where the degree of competition in providing local telecoms infrastructure is low, appears to have promoted a wider spread of DSL services than has occurred in the countries without local loop unbundling.

France appears likely to roll out DSL services most quickly, given France Telecom’s intention to cover 100% of the population by the end of the third quarter of 2000. However, these plans have still to be approved by the ART, the French regulator, which is concerned to ensure that the regulatory safeguards in place protect competition. Apart from France, there appears to be a reasonable degree of convergence in the projected rollout of DSL services. For example by the end of 2000, expected roll out for incumbent operators is:

• SBC – 30%
• Bell Atlantic – 49%
• Deutsche Telecom – 44%.

In the UK, BT had covered 25% of the UK population at the end of March 2000. BT expected to cover 35% of the UK population by end June 2000 and 75% by the end of March 2003 – if we make the simplistic assumption that the quarterly roll out is constant between June 2000 and March 2003, the BT roll out by the end of 2000 would be 39% which is very much in line with the figure for other operators apart from France Telecom. (Note that this comparison is of incumbents and ignores the impact of other players who may be rolling out their own DSL networks).

Therefore, although the rollout and commercial availability of DSL services in the UK has lagged behind the other countries in this area, it appears likely that it will have almost caught up by the end of 2000 (assuming the service providers in each country meet their stated rollout projections).

Introduction

This report presents the results of an international benchmarking study of DSL prices and the rollout of the associated service offerings as at March 2000. The study was carried out for OFTEL by Analysys. In addition to the UK, we have examined the DSL offerings in France, Germany and the USA. The economies and telecoms sectors in these countries are developed to a similar level and so provide a good international benchmark against which to compare UK services (the USA is commonly acknowledged to be a ‘world leader’ in the provision of DSL services). The report covers wholesale and retail offerings in each country except wholesale in France, where data on wholesale DSL services was unavailable. By wholesale DSL services, we mean the provision of a DSL service which is only sold to the public through third party service providers and where the service provider takes responsibility for billing and customer management.

It is the aim of this study to collect information on services which have been commercially launched. However, for the UK we have also collected information on trial DSL services because of the limited number of services available commercially.

Following this introduction, Section 2 gives the results of our price comparisons, explains the assumptions used, and draws qualitative conclusions on the pricing of DSL services in the countries studied. However, price is only one component of a comparison of DSL services, the availability of services and the speed of rollout are also important. Section 3 presents information on the rollout of DSL services in each country and Section 4 examines the regulatory position and the competitive environment in each country to give further insight into the relative strength of the DSL market. Finally, Section 5 provides an overall set of qualitative conclusions on the relative positions of the DSL market. There are four annexes to the report:

• Annex A presents the raw data collected on DSL tariffs in the four countries
• Annex B presents more detailed charts from the price comparisons
• Annex C lists the operators and service providers which had launched or announced the launch of DSL services and for which we tried to find data
• Annex D presents a glossary of terms used in this report.

Price Comparisons

This section explains how we used the data collected to draw comparisons between the countries, before presenting the results and our comments on them.

Variations in the form and structure of DSL charges

The information collected for this study was very heterogeneous, so a number of choices had to be made about how to use it to compare the different service offerings on a common basis. In particular, we had to deal with differences in:

• the combinations of downstream and upstream bandwidth (i.e. the "speed" of connection to and from the user) offered, particularly the degree of symmetry between the two
• installation costs and DSL modem costs to the end-user
• tariff structures in relation to the levying of subscription, volume- and/or duration- based charges
• the application of discounts
• the range of bandwidths offered
• the treatment of value added tax
• the currency in which prices were quoted
• the distinction between wholesale DSL services and resale of DSL services.

Comparing bandwidths

To compare the different combinations of upstream and downstream bandwidth, it is important to establish which applications the target customers will use. Exhibit 2.1 and Exhibit 2.2 below show Analysys’s view of the main applications for residential and business users respectively. These exhibits also illustrate our view of the minimum technical characteristics DSL users will demand once the DSL market has become established. We characterise the bandwidth requirements as Low, Medium or High, corresponding to peak bandwidths of up to 128kbit/s, 128–384kbit/s and greater than 384kbit/s.

Exhibit 2 .1: Technical characteristics of the typical applications for residential DSL users [Source: Analysys, 2000]

Exhibit 2.2: Technical characteristics of the typical applications for business DSL users [Source: Analysys, 2000]

For typical residential applications, the bandwidth needs are highly asymmetric, with "high" downstream bandwidth being the most important characteristic. Therefore typical residential users would generally gain little advantage from using a symmetric service. For this reason, we believe it is appropriate to concentrate on the downstream bandwidth when presenting comparisons of the prices for residential users.

The business user profile is different, with a wide variation in the types of applications used within this group. For example, some customers may use more asymmetric applications such as information research which involves a lot of downloading documents and other information. These users may appear to be similar to residential users in that downstream bandwidth is more important to them than upstream bandwidth (though they may still want more downstream capacity than the typical residential customer). Other business users who are using their DSL lines for connecting Web servers (to carry out ecommerce, to host information and content services, or for communication with their business partners), need equally high downstream and upstream bandwidth.

In summary, therefore, we have chosen the following measures of bandwidth in order to view price in the context of the "service level" provided:

• downstream (i.e. to the user) peak bandwidth, for residential users
• downstream peak bandwidth, for "low-end" business users (i.e. users requiring only relatively low bandwidth services)
• a measure which takes into account both downstream and upstream (from the user) bandwidth, for "high-end" business users (i.e. users requiring relatively high bandwidth services).

In order to choose the best combined measure of downstream and upstream bandwidth, we looked at the results produced by using several different measures. For two services with the same arithmetic mean bandwidth, the more symmetric service will "score more highly" on the basis of the "geometric mean" bandwidth. Because we believe that consumers will place more value on symmetric DSL services, based on the typical applications that business customers will use (such as server hosting) we have used the geometric mean to compare services for high end business users. We tested the sensitivity of the results to using different combined measures and found that the overall conclusions of our work were not greatly affected by the choice of measure for combined downstream and upstream speed.

Installation and modem costs

The charging structures varied for the DSL services in our study. For example, some service providers did not charge explicitly for the DSL modems sited at the customer’s premises and some did not charge an installation fee – choosing to cover these elements in the recurring subscription charge. In order to place all the offers on the same basis, we decided to amortise the charges for modems and installation (where a modem could be rented we added the rental charge to the subscription charge). Where DSL service providers did not quote prices for modem purchase or rental, we used the lowest modem price offered by other service providers from the same country. We chose a period of three years to amortise the charges, calculated the average monthly charge for this period, and added it to the subscription charge. We also checked whether using a longer period of five years to amortise these charges affected our comparison and found that it had no effect on our main conclusions.

Volume and usage charges, and discounts

The vast majority of tariff structures we found for DSL services were based on flat-rate subscription charges. The only operators for which this was not the case were Deutsche Telekom and KKF.net in Germany. In these cases, we derived usage assumptions for two types of customer.

• As a starting point for residential users, we used the current Internet use patterns of 15 hours spent online a month. We then assumed that users with an always-on connection would increase their use 3-fold, bringing their Internet use up to levels equivalent to those seen in the USA, at 45 hours per month.

• For business users, we estimated an average use pattern for an SME user (up to 250 employees), which we took as our typical business user. This was based on Analysys’s research into the relationship between telecoms expenditure, data communications expenditure and bandwidth of data communications links for different sizes of UK business. The main inputs to this work were total telecoms and data communications spending by business and market prices for leased lines. We assumed that the connection was fully utilised on average for four hours per working day during the peak period, at an average bit rate of 512kbit/s. This produced an annual traffic volume of approximately 200 Gigabytes. We tested the sensitivity of our results by varying these assumptions and found they had no impact on our overall conclusions

We would have included data on the discounts offered on DSL tariffs in our study, however this was not publicly available. We found anecdotal evidence that discounts were offered for business services in the USA as volume discounts, or as waived installation charges.

Bandwidth ranges

Each service provider covered in the study has a range of services at different bandwidths which may not match those offered by other service providers. This makes it complicated to compare the prices of these different services. We have therefore looked at them in three ways:

• first, we make a simple price comparison
• secondly, we plot the price of the service against its bandwidth
• thirdly, we plot the price per Mbit/s.

VAT and exchange rates

We included VAT in our comparisons for residential services (illustrated in Exhibit 2.3), but excluded it for business services. This is because we wanted to consider what users finally pay for services. Business users can reclaim VAT, if it is charged, whereas residential users cannot.

For the USA, taxes on telecommunications services vary greatly, by state and by city, from 0% to 10% or more. The information on DSL service provision in the USA in this study has been taken from service providers’ offerings in either New York State or California. In order to relate the data to the real tariffs facing residential customers, we used sales tax rates (state and city) in New York, and the utility users’ tax rate for San Francisco according to the location associated with the DSL service information.

Exhibit 2 .3: VAT and sales tax rates [Source: Analysys, 2000]	Country	Value added, sales or utility taxes
	France	20.6%
	Germany	16%
	UK	17.5%
	USA – New York	8.25%
	USA – San Francisco	7.5%

We converted all the data into pounds sterling using the latest purchasing power parity (PPP) factors calculated by the OECD for November 1999 multiplied by market exchange rates for March 2000 (shown in Exhibit 2.4 below).

Exhibit 2 .4: Purchasing power parity exchange rate [Source: OECD, 1999]	Country	PPP exchange rate
	France	0.1030
	Germany	0.3417
	UK	1
	USA	0.7353

We also carried out a sensitivity test using market exchange rates and found that this had no impact on our overall conclusions.

The distinction between wholesale and resale of DSL services

A number of the DSL services covered in our study can be sold through third parties. Some of these services are wholesale services and others are retail services which are resold by service providers.

We define wholesale DSL services, as the provision of a DSL service which is only sold to the public through third party service providers (not directly to the customer) and where the service provider takes responsibility for billing and customer management. Moreover, the retail price (not including charges for additional services) can differ from the wholesale price.

We define resale of DSL services as the offer of a DSL service to the public (in conjunction with supplementary services) by a third party which is also sold directly to the public as a stand-alone service. Either the service provider or the infrastructure owner may bill the customer for the DSL service.

Examples of Wholesale and Resale of DSL

Covad provides wholesale DSL services which are sold by service providers (including ISPs) to the public. The service provider is responsible for the customer relationship, including setting the tariff billing and customer service (though Covad employees may provide assistance to customers for detailed technical problems). The installation of the line is arranged between the customer and the service provider. The service provider then uses Covad to effect the DSL installation. Covad arranges for a new line to be supplied by the incumbent local operator as an unbundled local loop and will install DSL equipment at the customer’s premises.

Covad supports the provision of DSL services with two further network elements. First, it collocates equipment such as Digital Subscriber Line Access Multiplexers (DSLAMs) at the incumbent operator’s central office. This equipment has to be in place before DSL services are offered in any area. Secondly, Covad operates its own high-speed metropolitan area networks in order to deliver traffic to its partners’ points of presence and maintain greater control over the quality of service it can deliver. Covad services are included in the comparison of wholesale services.

QSC provides services in a similar way to Covad, though it publishes tariffs for its DSL service. The service is only available through ISPs, who may add a charge for Internet access provision. The ISP bills the customer and is responsible for customer service, though QSC also operates a customer help line. QSC arranges for the customer’s existing local loop to be unbundled from Deutsche Telekom and installs DSL equipment at the customer’s premises. QSC collocates equipment e.g. DSLAMs, at Deutsche Telekom’s local exchanges and provides its own national backbone in order to deliver traffic to its partners’ points of presence and enhance its quality of service. QSC services are included in the comparison of wholesale services.

BT’s IPStream, DataStream and VideoStream DSL services are considered wholesale products because they are sold only to service providers and not to the public. The service provider must purchase a connection from BT’s network to its point of presence, and offers a product called BT central for this service. BT’s wholesale prices include cost of the modem and splitter required to provide the DSL service over an existing customer line and BT installs these services at the customer’s premises. It is currently unclear what level of support, if any, is provided to the customer. IPStream is designed to enable the service provider to offer high speed IP connectivity to their end-users, although IPStream itself does not provide Internet access. DataStream is an ATM based service designed to enable service providers to deliver more advanced services than basic Internet access (for example, applications housing and extranet services), to their end-users. VideoStream is a highly asymmetric service designed to allow service providers to deliver video on demand to end users. We have included BT’s IPStream service in the comparison of wholesale services but BT’s DataStream and VideoStream services have not been included.

France Telecom’s Netissimo service is considered to be a resale, not a wholesale service, because it is distinct in several ways from the wholesale services described above. First, both France Telecom and ISPs (indirectly) can sell the Netissimo service directly to the public. This uses the existing line to provide the DSL service, installs the necessary end-user equipment, bills the end users and provides the end user with a starting pack of information on using the service. It also provides a help line directly to the consumer. The end user can choose an ISP to provide Internet access at the relevant DSL speeds, or the ISP can resell the Netissimo service in conjunction with its own Internet access service. France Telecom then routes the subscriber’s DSL traffic to their ISP’s nearest point of presence and charges the ISPs for that service. In some cases, ISPs bill the end user for Internet access and the DSL connection, in other cases the end user receives two bills. Netissimo is included in our retail comparison.

Results

This section presents the results of our comparison of DSL services across France, Germany, the UK and the USA. National comparisons of the price level of retail services are presented in Annex B.

In order to compare services and prices we have calculated an "equivalent" monthly charge calculated in pounds sterling (£) at PPP exchange rates, which takes into account both one-off and ongoing charges. We have divided our analysis of the results into three sections in order to compare the following aspects of the pricing of DSL services:

• retail residential DSL services including price, bandwidth and the price per Mbit/s of bandwidth
• retail business DSL services, including price, bandwidth, the price per Mbit/s of bandwidth and a comparison of incumbent and new entrant DSL services
• comparison of wholesale DSL services.

We note the following important points which limit the quantitative conclusions which can be drawn from this study.

• Comparison of DSL services for the UK with those in other countries is at this stage somewhat tentative, because all the UK services offered (apart from those of Kingston Communications) are trials, not commercial services. These prices might prove to be unrepresentative of future UK commercial offers.

• The user’s needs will be affected by other factors than just price, in particular, the "quality of service". The most obvious measure of service quality is speed, however, there are a number of other quality aspects which may in practice be important:

- In particular, the extent to which the required bandwidth can actually be obtained in busy periods (bandwidth reservation) could be the main contributor to the final service standard. The key parameter is the contention ratio – the ratio of the potential maximum demand to the actual capacity. The higher the contention ratio, the greater the number of users that may be trying to use the actual capacity at any one time, and therefore the lower the effective speed offered, especially at peak times.

- The extent of service availability: some services may be available only in certain geographic areas (e.g. in the USA, service from an individual supplier may only be in the coverage area of one of the "baby Bells"). Within those areas, customers with very long lines may not be able to receive the service.

- The delay in transmission between successive packets of data, or latency, which affects how continuous or disjointed a communication appears to the user.

- Due to the characteristics of DSL, some users will obtain inferior performance. Roughly speaking, the longer the subscriber line, the lower the maximum performance will be.

Retail residential DSL services

In this section, we consider comparisons of residential retail DSL services on the basis of both price and bandwidth. (Note data on other quality aspects is not generally available). We present data on both price and bandwidth separately, and from this draw qualitative conclusions regarding which service offerings represent the best value for money. Offerings with a high speed and a low price (or a low price for the speed offered) are "good value" compared with the other offerings.

Exhibits 2.5 to 2.7 below present data on the retail residential DSL services offered in the countries we have studied. Exhibit 2.5shows the services arranged in order of the monthly charge, Exhibit 2.6 shows the services arranged in order of bandwidth and Exhibit 2.7 presents both price and bandwidth information in a single chart.

Exhibits 2.5 to 2.7 illustrate the following points:

• There is no clear correlation between price and bandwidth - even within a country low price/high quality services may be offered alongside high price/low quality products
• The "best value" deal appears to be the USA SBC offering (quoted for Pacific Bell in San Francisco) which provides a relatively high bandwidth service for a low price. (Note, however, that the downstream maximum bandwidth of 1.5Mbit/s is only available to some customers. SBC specifies that the maximum available bandwidth varies from 384kbit/s to 1.5Mbit/s. We believe that this is dependent on the characteristics and conditions of the subscriber’s line.).

• In terms of the UK offers:

Exhibit 2.8 below presents an alternative comparison of our data on residential DSL services with price plotted against bandwidth for each. It illustrates that service providers in the USA offer commercial services for monthly equivalent prices of around £30 to £75 per month. For these prices the bandwidth provided ranges from 380 to 1500 kbit/s (i.e. there is little consensus on the "speed"). These are flat rate prices ("all you can eat").

European DSL services are, on average, similar in terms of price to services in the USA – e.g. the best offer in each country lies within a narrow price range – although the range of prices on offer varies more widely in Europe, as illustrated by Exhibit 2.8 above. The range of prices charged by UK service providers, though mainly for trial services, is similar to, though narrower than, those charged in the USA at £30 to £50 per month. As in the USA, the majority of charges are flat rate ("all you can eat"). Exceptions to this are some services providers in Germany which have introduced charges based on the volume of data used or the time spent online. The bandwidth offered by UK operators varies widely with Kingston offering the lowest downstream bandwidth (256kbit/s) of the whole sample and Demon offering the highest (2Mbit/s). However, Demon, like SBC, has specified its maximum downstream speed as a range; from 512kbit/s to 2Mbit/s, therefore not all trial customers will enjoy a 2Mbit/s service.

In France the downstream capacities for residential DSL services are very similar in bandwidth, 500 to 512 kbit/s, and price - £30 to £50 per month (the same price range as for the UK). In Germany the DSL services we found information for are on average more expensive than in other countries, ranging from £30 to £80 per month. The German services shown are all offered by Deutsche Telekom and provide a downstream capacity of 768 kbit/s. The least expensive service, provides only basic DSL access and not Internet access. It is intended to be bundled as a part of Deutsche Telekom’s DSL Internet access services. The prices of the two more expensive services include the charges for both Internet access and the basic DSL service. In both countries residential customers can purchase high bandwidth services if they are willing to pay the business price.

The UK and the USA appear to offer the best choice of, and range of speeds. In price terms the UK appears comparable to the other countries studied. In terms of speed the UK offers differ widely offering speeds both at the top and bottom of the range provided in other countries. This reflects the more general absence of a relationship between the price and bandwidth of DSL services. This lack of a clear correlation both within and between countries is somewhat surprising. There are a number of possible reasons for this.

• The immaturity of the market: i.e. consumers are unaware of better value offers or cannot get them in their area; and/or suppliers are at an early stage in pricing their products and are doing so on very different bases.
• Differences in the degree to which the underlying costs of providing DSL services are reflected by service providers given that a service with a maximum downstream bandwidth of 1.5Mbit/s is only marginally higher than providing a maximum available downstream bandwidth of 384kbit/s.
• Bandwidth may be an incomplete measure of the service actually provided to the consumer (contention ratio is also important).

An alternative way of comparing DSL services is to compare the price per unit of capacity (or price per Mbit/s) of each service. Exhibit 2.9 below presents a comparison of the price per unit of capacity versus downstream bandwidth. For each country, the highest and lowest monthly charges per Mbit/s for each bandwidth are shown. If we consider the highest and lowest prices per Mbit/s, it is difficult to see any relationship with bandwidth. However, if we take the cheapest prices only per country, we can see that as bandwidth increases, the monthly charge per Mbit/s falls. Due to the fixed costs involved, the costs to an operator of providing low speed DSL connections are not proportionately lower than the costs of a high speed connection. Therefore, it is possible for operators to offer higher speed (e.g. 1Mbit/s) services which will appear to be much better "value" on an effective price per month per Mbit/s basis.

Retail business DSL services

In this section, we consider comparisons of business retail DSL services on the same bases of both price and bandwidth as for residential services. We also draw qualitative conclusions regarding which service offerings represent the best value for money, as for residential services. Offers with a high speed and a low price (or a low price for the speed offered) are "good value" compared with the other offerings.

Exhibits 2.10 to 2.12 below present all the retail business services arranged in order of the monthly charge and grouped into three bands according to their downstream bandwidth (for ease of analysis): up to and including 512kbit/s; over 512kbit/s to 2Mbit/s, and 2Mbit/s or more. The downstream bandwidth of each service is superimposed on this series for Exhibit 2.10 while our combined downstream/upstream measure has been superimposed on Exhibits 2.11 and 2.12. Note that in order to limit the number of charts we have not included charts showing price and bandwidth separately.

Overall the diagrams show no direct relationship between price and bandwidth for DSL services. They do, however, allow for easy comparison of service price and service bandwidth. In the Exhibit 2.10 and Exhibit 2.11 there is a fairly clear "best value deal" which is at the bottom of the price range and at or towards the top in terms of bandwidth. For example, Exhibit 2.10 above shows that World-NET from France offers the cheapest business DSL service in the range 0kbit/s to 512kbit/s and provides a high capacity service in comparison with most of the other services in this bandwidth range. (However, Bell Atlantic offers two faster and less expensive services outside this bandwidth range). Again we note that many other factors will affect quality of service.

Plotting price for bandwidth by country for retail business DSL services shows that US service providers offer services at an much wider range of speeds for business customers (see Exhibit 2.13 below) than for residential. Many of these business services are symmetric, benefiting the higher end business users in particular. Prices are again flat rate prices. However, there is much less conformity in the prices of business DSL services in the USA, than for residential DSL. This is particularly true for higher speed and asymmetric services. Prices range from £40 (without Internet access) to £634 per month for an asymmetric 7Mbit/s service (which includes Internet access). For the symmetric services there is more consensus. For example, the price of a symmetric service at 512kbit/s varies between £140 and £200 per month. The position of the USA is a little distorted by the fact that, as in the case of its residential services, SBC quotes a range of downstream bandwidths available to its business customers, (for example, 1.5Mbit/s to 6Mbit/s), therefore the top speed of 6Mbit/s is not available to every business customer. However, the overall conclusion is not altered if the lower end of the range is used for the comparison.

In Europe some operators come close to matching the price levels seen in the USA at lower bandwidths. However, at higher bandwidths some European service providers are significantly more expensive. For example, in Germany the charge for an asymmetric service with a peak downstream speed of 6Mbit/s ranges from £1720 to £2500 per month. For the UK we only have data for Easynet’s trial services – at nearly £670 for a 2Mbit/s service, it lies between those prices charged in the USA and in Germany. However, because this is a trial service and we have only one observation for the UK, we believe it would not be prudent to draw any conclusions from this information.

If we consider the combined measure of downstream and upstream bandwidth instead of the downstream bandwidth only there is still no clear relationship between price and bandwidth as shown in Exhibit 2.14 below.

We can also compare the price per unit of capacity (or price per Mbit/s) for business DSL services against their bandwidths, as shown in Exhibits 2.15 and 2.16 below. The first graph presents the comparison in terms of downstream capacity, the second, in terms of the combined downstream and upstream bandwidth. For each country, the highest and lowest monthly charges per Mbit/s, for each bandwidth are shown. Similarly to the comparisons for residential services, there is no clear relationship between bandwidth and price per Mbit/s except when considering the cheapest prices per country. Moreover, if we take the cheapest prices only per country, we can see that as bandwidth increases, the monthly charge per Mbit/s falls.

We conclude that in general service providers in the USA offer the lowest prices per unit of capacity and so are most "cost efficient". Some of the French and German service providers offer similarly cost efficient services to those in the USA, although this is only true for lower bandwidth services.

Once again, the single UK business DSL service makes it difficult to draw general comparisons between the DSL services in the UK and the other countries studied. The one UK service featured in all the exhibits, which is a trial service, does appear less cost efficient than many of the services offered in other countries, but this cannot be taken as a general indication of the relative position of service providers in the UK.

Price, as we have emphasised earlier, is only one consideration in terms of the value that end-users get from DSL services. Price should therefore be qualified by the overall "service level" which is affected by many other factors, some of which are difficult to compare. One particularly relevant measure for high-end business users, however, is the degree of symmetry of the services. Our analysis reveals considerable differences in the degree of symmetry of the DSL services offered in each of the countries studied. Many of the service providers in the USA offer symmetric services (i.e. services with a ratio of downstream to upstream capacity of 1:1). This may reflect the market segment (high-end business users), at which the products are aimed. Asymmetric services are also offered. The picture in Europe is mixed.

• In the UK, the ratio is closer to 2:1 which suggests a comprise between the needs of high-end business users and the other segments of the market.
• In France, the ratio is 4:1 and this may reflect a target market of residential users and SMEs (as low-end business users).
• Deutsche Telekom’s ratio of 10:1 may reflect the fact that it has other services which it believes are more appropriate for the high-end business market than ADSL.

Comparison of incumbent and new entrants pricing

We have also analysed the differences in pricing between incumbents and new entrants for business customers (see Exhibits 2.17 and 2.18 below). For the UK, the fact that we only have a single observation for a newentrant makes comparison both between incumbents and new entrants and between the other European countries in the study difficult. The primary conclusion that we draw is that, in France, there is not a significant difference between the prices charged by incumbents and new entrants for similar bandwidth services, whereas in Germany, prices vary by a factor of two (the incumbent’s prices are approximately twice as high as those of the newentrant). This difference may arise from the fact that the French offers are resale offers of France Telecom’s product, whereas in Germany, local loop unbundling is available.

Exhibit 2.19 and Exhibit 2.20 below show the different prices charged for DSL services by incumbents and new entrants in the USA. The comparison of DSL services according to downstream bandwidths suggests that the incumbents typically offer the cheapest services, with the reservation that, for SBC’s services, the maximum bandwidth available to some customers may be less than that shown in the chart. However, because the upstream bandwidth is also an important criterion for many businesses, as explained above, and because the DSL services offered in the USA are more symmetrical, we feel that the combined bandwidth measure is more appropriate. On that basis, the new entrants’ DSL offers (which are more symmetrical) come closer to those of the incumbents.

Wholesale DSL services

This section looks at the differences in the price of wholesale DSL services across the countries studied. By wholesale, we mean the provision of DSL infrastructure and equipment to a third party which then bills and provides the service to individual customers. Exhibits 2.21 and 2.22 below show the data we were able to collect on wholesale services in Germany (QSC, a newentrant), the UK (BT) and the USA (Covad, a newentrant) arranged in order of the monthly charge and grouped into two bands (for ease of analysis): up to and including 512kbit/s; over 512kbit/s to 2Mbit/s. The downstream bandwidth of each service is superimposed on the first series and the combined measure of downstream and upstream bandwidth is superimposed on the second.

In order to be able to make a consistent comparison, we have chosen BT’s IPStream which bears the closest resemblance to the wholesale products offered in the USA and Germany. In addition, the inclusion of BT’s DataStream and VideoStream services within this analysis would require us to make assumptions which would be difficult to substantiate. For example, for any one service provider we would need to estimate the number of BT local exchanges from which services are offered, the number of customers served at each local exchange and the distance of each such local exchange from the service provider’s nearest point of presence.

On the basis of the data we managed to collect, UK wholesale services, while offering a lower range of bandwidths, are generally cheaper than those in the USA. Services in Germany are more expensive than in the UK and the USA. However, the services on which we collected data in the USA and Germany are symmetric DSL services, as opposed to the asymmetric DSL service provided in the UK. Symmetric services offer greater benefit to some users. In addition, the wholesale services in the USA and Germany provide end-users with static IP addresses, which enable the end-user to connect servers to the Internet in order to carry out ecommerce, for example. In contrast, BT’s IPStream provides the end-user with dynamic IP addresses which are unsuitable for connecting servers to the Internet, which we expect to be one of the popular applications for business DSL services.

The differences between the service available in the UK, the USA and Germany can also be seen in Exhibits 2.23 and 2.24 below. They present the data we collected on wholesale DSL services with price plotted against bandwidth by country. The first exhibit shows a comparison of price versus downstream capacity, the second a comparison of price versus combined downstream and upstream capacity.

We can also draw some tentative conclusions on the mark-up that service providers add to their wholesale prices. In the UK, the margin for Freeserve is around 27%, while at a bandwidth of 512kbit/s, Demon’s offer would appear to provide a margin of 5%. However, for Demon’s offer of a 2Mbit/s service, (the upper end of the range it is offering), the trial retail price appears to be significantly below the price of BT’s wholesale product. These prices were announced as trial prices with no guarantee that they will remain static.

In the USA and Germany the mark-up is much higher, on average 60% (adjusting prices pro rata for differences in bandwidths in Germany), which suggests that either service providers in other countries could reduce their mark-ups and still offer profitable services, or that UK service providers will have to increase their mark-up over wholesale prices, unless they have other sources of revenue to subsidise the costs of providing the service.

Service Rollout

Results

Exhibits 3.1 to 3.4 show the publicly available data on service rollout in the countries we have studied.

Note that in many cases the areas covered will overlap significantly, so it is not possible to "add" the coverages.

As the exhibit shows, in the absence of full local loop unbundling, rollout in the UK and France is limited by the plans of BT and France Telecom respectively. In Germany and the USA, where the local loop has been fully unbundled, rollout of both actual and proposed DSL services is less tied to the plans of the incumbent local operator. It is generally more difficult to obtain data on the number of local exchanges from which services will be available in Europe than it is for the USA, and this has limited the extent of data on rollout and service availability to geographic measures.

Exhibit 3 .1 Summary of rollout of DSL service in the UK [Source: Analysys, 2000]

Provider	Cities or regions	Percentage of population covered	Total customers served	Number of switches from which service is available	Means of service provision
BT	London Manchester Birmingham Coventry Edinburgh Cardiff Glasgow Newcastle by March 2000	25% of the total population at end of March 2000 (6 million homes), 35% end June 2000 and 75% by March 2003	n/a	400 local exchanges will be served initially, followed by a further 100 by the end of 2000	Own network
Kingston Communi-cations	Kingston Communications’s East Yorkshire operating area	0.3% of the total population and all of Kingston’s East Yorkshire operating area	1500	15 local exchanges	Own network, initially, together with BT’s unbundled local loop when available
Demon (Thus) (trial)	13 cities*	Service rollout partially linked to BT’s wholesale ADSL service	500	n/a	Thus’s network is being used for the trial; BT’s wholesale ADSL service and unbundled local loops when available
Telewest (trial)	Croydon	Service rollout partially linked to BT’s wholesale ADSL service	n/a	n/a	Telewest is using its own network for trial; BT’s wholesale ADSL service and unbundled local loops in addition when available
Video Networks (trial)	London	Service rollout linked to BT’s wholesale ADSL service	n/a	n/a	BT’s wholesale ADSL service is being used
Freeserve (trial)	London and Manchester	Service rollout linked to BT’s wholesale ADSL service	150	n/a	BT’s wholesale ADSL service is being used and unbundled local loops will be used in addition when available

 





Exhibit  3.3 Summary of rollout of DSL service in Germany [Source: Analysys, 2000]

Provider	Cities or regions	Percentage of population covered	Total customers served	Number of switches from which service is available	Means of service provision
Deutsche Telekom	43 cities by end 1999, 220 cities covering potentially 17 million households, by end 2000	44% of households by end 2000	10 000 by the end of 1999; 200 000 forecast for the end of 2000)	n/a	Own network
QSC	40 cities and towns by the end of 2000	25% of the total population by the end of 2000	n/a	n/a	Own network
KPNQwest	Cologne, Dusseldorf, Frankfurt am Main and Munich planned for mid 2000	4% of the total population by end 2000	n/a	n/a	Own network

Exhibit 3 .4 Summary of rollout of DSL service in France [Source: Analysys, 2000]

Provider	Cities or regions	Percentage of population covered	Total customers served	Number of switches from which service is available	Means of service provision
France Telecom	Six districts of Paris initially; plans 250 towns by 2002	12% of the total population by the end of 1999; 33% by end of first quarter 2000; 48% by end of second quarter 2000, 100% by end of third quarter 2000	5000 by the end of 1999 (source: Ericom, 3 March 2000)	n/a	Own network
World- NET	All 20 districts of Paris	Currently 4% of the total population; expansion will be linked to France Telecom’s service rollout	n/a	n/a	France Telecom’s DSL service
Easy Net	Paris, Lyon, some cities of the ‘92 district’	Currently 9% of the total population; expansion will be linked to France Telecom’s service rollout	n/a	n/a	Uses France Telecoms DSL service

*Belfast, Birmingham, Cambridge, Cardiff, Coventry and Warwick, Edinburgh, Glasgow, Leeds, London, Manchester, Milton Keynes, Newcastle

**Albuquerque, Atlanta, Austin, Baltimore, Boston, Chicago, Colorado Springs, Dallas, Denver, Detroit, Houston, Los Angeles, Miami, Minneapolis, Norfolk, New York, Philadelphia, Phoenix, Pittsburgh, Portland, Raleigh, Richmond, Sacramento, San Diego, San Francisco, Seattle, Tucson, Washington

*** Austin, Baltimore, Boston, Chicago, Dallas, Denver, Detroit, Houston, Los Angeles, Milwaukee, Minneapolis, New York, Oakland, Orange County, Philadelphia, Sacramento, San Diego, San Francisco, San Jose, Seattle, Washington DC

**** Atlanta, Baltimore, Boston, Chicago, Dallas, Denver, Detroit, Houston, Los Angeles, Miami, New York, Orange County, Philadelphia, Sacramento, San Diego, San Francisco Bay Area, Seattle, Washington DC

***** Residential and business in: Atlanta, Austin, Baltimore, Boston, Chicago, Dallas, Denver, Detroit, Houston, Los Angeles, New York, Philadelphia, Portland, Sacramento, San Diego, San Francisco Bay Area, Seattle, Washington DC. DSL for business only in: Cleveland, Minneapolis, Orlando, Phoenix, Pittsburgh, Raleigh, Salt Lake City, St Louis

Conclusions

The rollout of DSL services in the European countries studied is strongly linked to the rollout of DSL services by the incumbent, although to a lesser extent in Germany, where QSC has been able to purchase unbundled local loop capacity from Deutsche Telekom. By comparison, in the USA, wholesale providers using unbundled local loops, such as Covad (featured in the table), Northpoint and Rhythm Networks, have proved a major force in extending the rollout of DSL on a nationwide basis. On the basis of this evidence, it seems that local loop unbundling, particularly where the degree of competition in providing local telecoms infrastructure is low, appears to have promoted a wider spread of DSL services than in the countries without local loop unbundling.

France appears likely to roll out DSL services most quickly, given France Telecom’s intention to cover 100% of the population by the end of the third quarter of 2000. However, these plans still have to be approved by the l’Autorité de Régulation des Télécommunications (ART), the French regulator, which is concerned to ensure that the regulatory safeguards in place protect competition. Apart from France, there appears to be a reasonable degree of convergence in the projected rollout of DSL services. For example by the end of 2000, expected roll out for incumbent operators is:

• SBC – 30% (of its customer base)
• Bell Atlantic – 49% (of its customer base)
• Deutsche Telecom – 44% (of all German households).

In the UK, BT had covered 25% of the UK population at the end of March 2000. BT expected to cover 35% of the UK population by end June 2000 and 75% by the end of March 2003 – if we make the simplistic assumption that the quarterly roll out is constant between June 2000 and March 2003, the BT roll out by the end of 2000 would be 39% which is very much in line with the figure for other operators apart from France Telecom. (Note that this comparison is of incumbents and ignores the impact of other players who may be rolling out their own DSL networks).

Therefore, although the rollout and commercial availability of DSL services in the UK has lagged behind the other countries in this area, it appears likely that it will have almost caught up by the end of 2000 (assuming the service providers in each country meet their stated rollout projections).

Regulatory and Market Overview

This section addresses those aspects of the regulatory framework in France, Germany, the UK and the USA which pertain to the competitive provision of ADSL services. For each country, we discuss what is permitted under the current legal situation in each country, any forthcoming changes to that legal situation and how these factors have affected the market for service provision.

France

Regulatory framework governing access to bandwidth

No firm decision on local loop unbundling has been taken in France so far – though there are plans to open the local access market here. This first became evident in October 1998, when the national regulator, ART, delivered a discussion paper to the Consultative Commission on competition in local loop provision on networks and telecoms services. As a result, two working groups were established, one to examine the economic ramifications and the other to investigate technical and operational aspects of the issue. In April 1999, a public consultation was initiated, requesting contributions from operators and consumer groups on a set of options: Options 1 to 3 concern unbundling the local loop; Options 4 and 5 concern arrangements for reselling France Telecom’s retail telecoms services.

• Option 1, Access to copper pair. The incumbent, France Telecom, would offer access to its local infrastructure by providing leased copper pair links to new entrant operators, which would then be permitted to install their own equipment at the local exchange. Operators would be permitted to use this local network to supply both broadband and narrow band services.

• Option 2, Access to bit stream. France Telecom would install all equipment within the local loop itself and the new entrant would be responsible for routeing traffic from the local exchange to its end destination.

• Option 3, Access to permanent virtual circuits. France Telecom would be required to provide alternative operators with a virtual circuit, thereby transporting traffic from subscribers to the nearest POP on the alternative operator’s own network infrastructure. This would permit customers to subscribe to an alternative telecoms operator for broadband services, whilst remaining with France Telecom for basic voice telephony.

• Option 4, Reselling local traffic. France Telecom would sell local traffic to alternative operators so that they could offer an integrated package of services including local communications, but the alternative operator would have no power to alter the technical characteristics of the service. France Telecom would be responsible for routeing all calls, but would bill the cost of these calls to the new entrant rather than the end-customer. The new entrant would then have sole responsibility for billing the end-customer.

• Option 5, Reselling subscriptions. This would be similar to Option 4, but include subscriber management functions and the resale of line rental by new entrant operators in addition to traffic charges.

In December 1999, the results of the public consultation were published, revealing that two options in particular were favoured by participants in the debate, namely Options 1 and 3. Option 1 was viewed as necessary to ensure full competition, and Option 3 was viewed as desirable in that it would enable competitors to launch services more quickly.

No firm date has been set – either for the decision about which option will be chosen, or when it will be imposed on France Telecom. However, France Telecom has pledged to open its local loop by the end of 2000. The national Competition Authority believes that it should be possible earlier than this, and in February 2000 asked France Telecom to open its network by mid-April 2000. France Telecom has so far refused to do this, saying that it would not be technically possible in that time frame.

France Telecom launched two ADSL services in November 1999, one providing high-speed ADSL Internet access to subscribers directly (Netissimo) and the other providing a service to other ISPs (Turbo IP). Two Netissimo packages are now available:

• Netissimo 1 offers ADSL connectivity permitting speeds of 500kbit/s downstream and 128kbit/s upstream

• Netissimo 2 offers speeds of 1Mbit/s downstream and 256kbit/s upstream.

Turbo IP is an offer aimed at other ISPs, which incurs a set-up fee of FRF40 000/£4120 – (excluding tax) for a 2Mbit/s connection and FRF80 000 for connections between 34Mbit/s and 155Mbit/s. A monthly fee of FRF19 000/£1957 (2Mbit/s), FRF40 000/£4120(34Mbit/s) or FRF80 000/£8240 (155Mbit/s) is then charged. However, these prices only apply where the alternative ISP’s POP is near to the POI with France Telecom’s network. If the distance is greater, then extra charges are incurred. However, the service merely connects alternative operators to France Telecom’s network and does not give them access to the local loop. It is therefore not a wholesale ADSL offer as usually understood. In addition, France Telecom introduced two further offers aimed at other ISPs in December 1999. First, France Telecom now allows other ISPs to market Netissimo services to their own customers and retain a commission for each sold subscription (between FRF100/£103 – and FRF300/£309 – (excluding tax, according to the type and length of subscription sold). Secondly, other ISPs are now permitted to resell Netissimo services to end users.

ART is currently engaged in an ongoing dispute with France Telecom over its rollout of ADSL services. Following its launch of ADSL services in Paris, France Telecom extended the service to other cities, without obtaining additional authorisation from the regulator. France Telecom asserts that it was told by ART that it did not need to request further authorisation. This dispute has yet to be resolved.

Limitations on access to bandwidth

The introduction of ADSL access in France has not been without its problems, and there are a number of ongoing disputes between alternative telecoms operators and the incumbent operator concerning access to the local loop, which has been allowed to competitors in the absence of unbundling.

In June 1999, Club Internet, one of the leading subscription-based ISPs in France, appealed to France’s competition authority, arguing that France Telecom should be forced to delay the introduction of its Netissimo and Turbo IP services, in order to allow competitors to develop rival ADSL access offers to be launched simultaneously, reselling France Telecom’s Netissimo service. The Competition Authority agreed with Club Internet (a decision which was subsequently reinforced by ART) and required the incumbent to postpone its ADSL launch for 15 weeks until early November. In October 1999, 9 Telecom Réseau issued a complaint against France Telecom to the competition authority, alleging that the incumbent was abusing its dominant position in not allowing other operators unbundled access to its local loop networks in order to offer competitive ADSL services. The action demanded that France Telecom should be obliged to offer unbundled access and allow tests of this access to begin from June 2000, however, the issue has not yet been resolved.

In January 2000, Cegetel and AOL France alleged that the conditions they were being offered for the provision of ADSL access by France Telecom were unacceptable. They complained that the prices charged by France Telecom for access to its ADSL lines were too high, and prevented competitors from offering their subscribers ADSL access at competitive prices. They also claimed that France Telecom was refusing to guarantee a minimum level of access for rival companies’ traffic. Cegetel and AOL France’s complaint follows a decision by Club Internet to take France Telecom to court for its ADSL pricing policy, over allegations that the incumbent is abusing its dominant position to keep rivals out of the Internet market. The legal action is openly supported by both Cegetel and AOL France.

The lack of a definite timetable for local loop unbundling is a cause for concern, both among alternative carriers and with the regulator. In the interim period, France Telecom is benefiting from its position as virtually the only provider with ADSL-enabled local loop infrastructure. For ART, this represents a threat to competition in the market, and it is for this reason that the regulator says it has been adamant in its demands that France Telecom halt the rollout of ADSL in areas not covered by its initial authorisation.

Impact on competitive carriers and ISPs

Competition has been slow to develop, and only a small number of France Telecom’s competitors are offering rival ADSL Internet access services. In the retail sector, these include Club Internet, Easynet and World-NET. In addition to the subscription fee they pay to these ISPs, customers must also subscribe to France Telecom’s Netissimo services. In some cases the customer receives bills from both France Telecom and the ISP, in other cases the ISP bills the customer for the Netissimo service and Internet access jointly. One of the subscription-free providers, Free, also offers ADSL access. Other ISPs are running trials. For example, in January 2000, AOL France began a trial in Monaco, in collaboration with Cegetel and Monaco Telecom. By January 2000, France Telecom claimed that 2300 customers had bought ADSL access services, more than half of whom had chosen Wanadoo (a subsidiary of France Telecom) as their ISP.

Wholesale services are now also being offered by France Telecom’s competitors, including ISDnet, which launched its DSL Connect service in December 1999. Two packages are available, one aimed at ISPs selling to the residential sector (and offering the same bandwidths as France Telecom’s Netissimo 1), and the other aimed at business-focused ISPs (same bandwidths as Netissimo 2). We were unable to establish full details of this service during the course of this study.

However, other potential competitors have been reluctant to enter the market until unbundling is initiated. For example, in November 1999, First Telecom announced that it did not intend to launch its own ADSL services in France until unbundling becomes a reality.

Germany

Regulatory framework governing access to bandwidth

Germany’s local loop was officially unbundled when the country’s telecoms markets were opened to competition in January 1998. However, prices were only set in February 1999. New entrants are permitted to lease local loop infrastructure from the incumbent operator, and to offer both voice and data services. They may collocate equipment in Deutsche Telekom’s local exchanges.

The cost of leasing local access from Deutsche Telekom is determined by the regulator, the Regulierungsbehörde für Telekommunikation und Post (RegTP). In February 1999, this rate was set at DM25.40/£8.70 per month. This price is fixed, regardless of whether the competitive operator is offering both voice and data services or data (typically over DSL) alone. The cost-based model was developed in association with the Germans consultancy Wissenschaftliches Institut für Kommunikationsdienste (WIK). The decision on the lease price was based on a cost model for the local loop, taking into account nominal capital cost and the depreciation period for copper cable (fixed at 20 years). The price is also independent of the area in which a competing operator wishes to lease access, and is based on a geographical average of the costs involved. In addition to the monthly fee, the regulator ruled that Deutsche Telekom can charge a one-off fee of between DM191.64/£65.50 and DM337.17/£115.20 for switching a local loop customer to a rival service provider. The exact sum depends on whether the new entrant is taking over an existing line or a new line needs to be installed, and whether Deutsche Telekom employees are required to do any work at the customer’s premises. Both the monthly fee and the one-off switching fee, are scheduled to remain in force until March 2001.

Limitations on access to bandwidth

Many ISPs have complained that the lease price for Deutsche Telekom’s local access network is too high. The cost of leasing an unbundled local loop in Germany is estimated to be DM304.80 (£104.15) per year, which is 20% higher than the line rental charge which would be paid to Deutsche Telekom by the end-user. The high prices will make alternative operators’ DSL services less competitive than existing options such as Deutsche Telekom’s ISDN access services.

Unlike elsewhere, the incumbent is offering ADSL over Integrated Services Digital Network (ISDN), not over Public Switched Telephony Network (PSTN). Because of this, ADSL access is only available at frequencies of 200khz and above, whereas when ADSL is run over POTS, it can access frequencies above 20khz. Data transfer rates may therefore potentially suffer according to some observers.

Deutsche Telekom launched business-focused ADSL services in April 1999, and followed this with an ADSL service aimed at residential consumers in August 1999. A rapid rollout of services is planned for 2000, and Deutsche Telekom plans to make the service available in 220 towns nationwide, covering 44% of households, by the end of the year. By the end of February 2000, 61 towns had already been connected.

Impact on competitive carriers and ISPs

The early unbundling of Germany’s local loop has allowed competition to develop in this market considerably earlier than has occurred elsewhere, even though many new entrants have claimed that the price of unbundled local loops in Germany is too high. By autumn 1999, RegTP reported that 51 providers had signed contracts with Deutsche Telekom AG for local loop access. Major pan-European network operators are also launching DSL services earlier in Germany than in other countries. For example, in December 1999, KPNQwest announced that it would launch retail business-focused DSL services in nine European countries. Germany would be the first country to get the services, which would be introduced there by mid-2000. When announcing the launch of its services, KPNQwest said that the reason for the earlier launch in Germany was the legal requirement already in place for Deutsche Telekom to offer access to its local loop infrastructure.

Of the providers leasing local access from Deutsche Telekom (including prominent ISPs such as TCP/IP GmbH), the majority are offering DSL access to the business community. In contrast, at the end of November 1999, QSC began offering wholesale services in Cologne by leasing capacity on Deutsche Telekom’s local access network. The company plans to extend its services nationwide by the end of 2000. Unlike Deutsche Telekom, QSC offers SDSL (symmetric digital subscriber loop) services at bandwidths up to 2.3Mbit/s.

UK

Regulatory framework governing access to bandwidth

OFTEL conducted a public consultation on the options for unbundling beginning in December 1998. In July 1999, it published the findings of this consultation process and set out its own preliminary conclusions. The regulator favoured the introduction of full unbundling, with the incumbent making local loop infrastructure available to new entrants as leased circuits. OFTEL suggested that BT should be required to do this by July 2001. A further round of consultations was conducted, the results of which were published in November 1999, and showed majority support for OFTEL’s proposals. Most also agreed with the suggestion that the leased circuit should be priced at long-run incremental cost, plus an allowance for the costs common to the line and other BT services. Unlike in Germany, there will not necessarily be a single lease price for the whole of the country, but rather BT will be allowed to establish different prices for different regions, so that individual prices will more closely reflect costs.

BT began trials of its ADSL services in limited areas in October and November 1999, and the trials were extended to several UK cities in early 2000. As required by its licence, BT is making wholesale ADSL services available to its competitors on the same terms as are available to its own service provider. BT’s wholesale service provides alternative service providers with a permanent virtual circuit (PVC), allowing them to market their own ADSL services to the end-user. The wholesale products currently being trialled were expected to be launched commercially in late March 2000, but BT applied to OFTEL for an extension to its trial until 30 June 2000, to which OFTEL has agreed. A number of alternative service providers are participating in the trials in addition to the incumbent-owned BT Internet, including Freeserve, Demon Internet and Easynet.

BT is offering three related wholesale services: VideoStream, DataStream and IPStream. VideoStream permits the provision of video-on-demand services over ADSL operating at a bandwidth of 2Mbit/s. DataStream enables alternative operators to establish IP networks for businesses, in order to provide as corporate intranets or ecommerce services. IPStream enables alternative providers to offer high speed Internet access over ADSL.

OFTEL is concerned that the provision of PVC access alone to the local loop (as is currently being trailed by BT) is not sufficient to foster competition in the local access market, given that only BT is able to roll out DSL technology. OFTEL therefore plans to mandate full local loop unbundling, and has stated that it would like to see this introduced prior to July 2001. The requirement for unbundling will be set out in a new condition to be added to BT’s service licence. This condition is expected to be formulated in April 2000. OFTEL plans to draw up guidelines on collocation and issue recommendations for spectrum management to ensure minimal interference between lines in time to allow trials of a fully unbundled service to begin by the end of 2000.

Full unbundling will consist of the provision to new entrants of a partial baseband leased circuit, over which the operator may offer voice and/or data services. All operators with interconnection rights under Schedule 2 of the Telecommunications (Interconnection) Regulations 1997, will be permitted to lease local loop infrastructure. Furthermore, new entrants will be allowed to physically collocate their equipment in BT’s local exchanges, where space is available. Distant collocation (where space is rented at a nearby managed site connected to BT’s local exchange) and virtual collocation will also be open to alternative providers, should they require it. If the customer wishes to retain BT’s services for voice, but purchase DSL connectivity for data from an alternative provider, then local loop access will be provided to the alternative provider using spare pairs in BT’s access network. If no such spare pair is available, then OFTEL is considering the possibility of allowing alternative operators to lease the entire existing circuit, but permitting them to resell BT’s voice services. Line sharing, as is now permitted in the USA, is therefore not included in the current proposals.

The only exception to local loop unbundling will be in Hull, where Kingston Communications holds a monopoly on local loop access. Since there has been no significant demand as yet for access to Kingston’s infrastructure, OFTEL has stated that the unbundling requirements on BT are unlikely to be replicated for Kingston Communications at least for the time being.

Limitations on access to bandwidth

OFTEL’s consultation process revealed general dissatisfaction from alternative providers at the July 2001 date for the introduction of full unbundling. However, OFTEL has said that this date represents the latest possible point for the introduction of full unbundling, and has said that it may be possible for BT to unbundle its local loop at an earlier date.

Rollout of DSL technology is proceeding, and by the time BT’s wholesale ADSL services are commercially launched, the incumbent expects to be able to offer service to 35% of the population. Further expansion of ADSL services will depend on the success of this initial rollout. OFTEL has expressed its satisfaction that rollout will make ADSL available to a significant proportion of the population in a relatively short time.

There has been concern from some service providers over the level of BT’s prices for its trial wholesale ADSL services, including from BT’s own subsidiary, BT Internet Business Services. It is as yet uncertain what retail prices will be charged for DSL services to the end-user because BT has only recently announced actual prices for its range of wholesale services. However, unless service providers can gain revenue from alternative sources such as online advertising and ecommerce, these are likely to be higher than the wholesale prices offered by BT. As a result, one of the UK’s largest corporate Internet access providers, UUNET announced in February 2000 that it was considering cancelling its launch of ADSL services because of what it called "prohibitive" pricing by BT.

Impact on competitive carriers and ISPs

Commercial trials of ADSL access are being offered by a number of service providers which have purchased BT’s wholesale DataStream or IPStream services, including Freeserve, Demon Internet and Easynet. Furthermore, BT’s own service provider, BT Internet is offering ADSL services to consumers and businesses. In its East Yorkshire operating area, the Kingston Communications Group offers ADSL access to customers through its subsidiaries, Torch and Kingston Internet, and has a monopoly on DSL services in the Hull area.

There is evidence that BT and its wholesale customers will soon see competition from other local access network operators which plan to trial ADSL access over their own infrastructures. For example, NTL, which will be the largest cable operator in the UK, following its merger with C&W, announced in December 1999 that it would run trials of ADSL services over its own local infrastructure to businesses in Surrey.

USA

Regulatory framework governing access to bandwidth

The 1996 Telecommunications Act required incumbent local exchange carriers (ILECs) in the USA to provide unbundled network elements (UNEs) to competitive local exchange carriers (CLECs). Network elements are defined as facilities used to provide telecoms services and include transmission between central offices, information for billing and collection and routeing of calls. The Act aimed to ensure that alternative carriers were able to purchase only those elements they needed to provide local telephony services. In addition, the Act mandated that alternative carriers be permitted to collocate their own equipment in ILECs’ local exchanges in order to gain access to unbundled local network features. However, so as to ensure that voice services were not disrupted, DSL providers were not allowed to use existing copper lines, and were required to install new infrastructure.

In the ensuing years, there were numerous complaints from CLECs, who claimed that the ILECs were not fulfilling their legal requirements. This led the regulator in the USA, the Federal Communications Commission (FCC), to issue several new rulings in March 1999. First, guidelines were issued for the collocation of equipment in central offices. This was in response to complaints that ILECs were slow to allocate available space. Secondly, a ruling on spectrum compatibility was issued, requiring CLECs simply to prove that the advanced technologies they wished to implement (such as DSL) would not interfere with existing services, prior to gaining permission to install necessary equipment. This was a reduction from previous requirements on CLECs which often resulted in considerable delays before collocation was permitted. Finally, the FCC ruled that DSL providers would thenceforth be permitted to use existing phone lines for their services, and would no longer be required to install new copper.

There has been a continuing debate on what are considered to be UNEs. For example, the ILECs have begun to deploy DSL equipment in digital loop carriers (DLCs). These act as a bridge between the subscriber and the central office and are more numerous and densely spaced than the offices themselves, but are only present in approximately 15% of local access networks in the USA. Only customers within a three mile radius of a DSLAM can be provided with DSL services. Therefore, by installing DSL equipment in DLCs where they are available, ILECs are able to offer DSL services to a greater number of their end-customers in some areas. In September 1999, the FCC ordered that sub-loops should be considered UNEs, therefore allowing CLECs access to any accessible point, such as a DLC, on a sub-loop. At the same time, the FCC ordered that ILECs would not be required to offer access to their own DSLAMs, except when CLECs are unable to install their own, in which case, the CLEC should be able to lease DSLAMs from the ILEC.

In November 1999, the FCC issued a further ruling, which ordered ILECs to share their lines with CLECs. As a result, CLECs were permitted to offer high-speed data services over a line while the ILEC could continue to offer voice services over the same line. Prior to this ruling, many CLEC customers had been forced to order the installation of a second line if they wished to buy voice and data services from two different providers. In contrast, ILECs were already using line-splitters to separate voice traffic from data traffic for their own customers, and were thus permitting their own high-speed data services and voice services to "share" the same line. However, they did not permit the same sharing arrangement to involve CLECs providing merely data services. The FCC ordered that prices for the line-sharing service should be based on the charges that ILECs charge their own data service units for sharing lines.

There are considerable incentives for ILECs to abide by the FCC’s rulings and establish unbundled access for CLECs which apply for it without delay. This is because the ILECs are judged on their success in implementing local loop unbundling legislation when they themselves apply for permission to increase their own service portfolio. A number of ILECs are seeking to offer long-distance voice services. The first to obtain permission for this was Bell Atlantic, which was authorised to sell long-distance services in New York State in December 1999. As part of their application to provide long-distance telephony, ILECs must provide evidence that they are providing non-discriminatory access to their local loop networks. ILECs which are failing to comply with unbundling legislation therefore risk being refused permission to broaden their own range of services. In February 2000, the Justice Department recommended that SBC Communications be denied permission to offer long-distance services in Texas, since it had allegedly failed to adequately open its local loop network to competing ISPs offering DSL services. SBC is accused of missing a significant proportion of due dates on orders for DSL loops from CLECs.

Limitations on access to bandwidth

End-users have repeatedly complained of installation delays, and DSL providers have admitted that waiting lists are long. One of the leading wholesale DSL providers, Covad, has a waiting time for installation of about one month, and the wait for provisioning with Rhythms is reportedly similar. Flashcom, a DSL provider based in Huntingdon Beach, California, had around 15 000 paying DSL customers (from the residential and small business sectors) by February 2000. However, according to the company’s own estimates, it also had between 15 000 and 20 000 customers waiting for service. Line sharing is expected to help reduce these delays, since there will no longer be a need to include the time spent installing a second line for the DSL service to run over. However CLECs have also accused ILECs of imposing longer delays on their rivals than they do on their own DSL service units. SBC Communications for example was accused in the Department of Justice report of missing due dates on 12% of DSL loops ordered by Texan CLECs, compared to delays on only 6.3% of requests from its own DSL retail operation.

In February 2000, the FCC launched a public inquiry on the deployment of advanced telecoms services, inviting comment from all interested parties on whether advanced telecoms capability is now being deployed in a "reasonable and timely manner", and canvassing their opinions on suitable action for the FCC to take if it is found that deployment is being unreasonably delayed.

Geographical availability of DSL services is improving. A number of ILECs have announced accelerated DSL deployment plans. SBC Communications, for example, recently announced that it would deploy DSL services to most of its customers in 13 states in the next three years, committing USD6 billion to the deployment. The initiative, named "Project Pronto", will make DSL services available to 80% of its Ameritech, Nevada Bell, Pacific Bell, SNET and Southwestern Bell customers by the end of 2002. Similarly, Bell Atlantic, which operates in 13 states and the District of Columbia, has doubled its planned deployment of DSL-enabled lines by the last quarter of 2000 to 21 million, equal to 49% of its customer base.

Impact on competitive carriers and ISPs

The effect of the FCC’s additional rulings during 1999 has been to provide a boost to competition in DSL service provision. In Utah, for example, US West began offering DSL services in May 1998, and was the only DSL provider in that state for more than 18 months. However, by the end of 1999, five more companies had announced the launch of competing services. Four of them (Covad, Rhythms, Jato Communications and RMI.net) launched services in December 1999, and a fifth, Northpoint, is due to launch its service in 2000.

There is a large number of commercial DSL offers available now in major urban areas across the USA. A number of the largest ISPs in the country are now offering DSL services by partnering with network operators. In June 1999, for example, Mindspring announced a partnership with BellSouth Corp, permitting Mindspring to use BellSouth’s DSL facilities for its own launch of DSL access in Atlanta.

The FCC’s decision to allow line sharing in November 1999, has prompted more CLECs and ISPs to launch residential-focused DSL services. Prior to the ruling, alternative DSL providers had argued that the cost of installing a second line was a barrier to extension of their services into the more price sensitive residential sector. However, following the FCC’s ruling, Northpoint announced that it was trialling a consumer-class DSL service through selected ISPs nationally. The FCC’s decision had, according to Northpoint, removed the remaining obstacle to a commercial launch of its service. Another wholesale DSL provider, Rhythms, estimated that the line sharing decision could potentially reduce its local loop costs by over 50%. However, the technical, operational and administrative details of implementing the FCC’s decision have yet to be established in many states. Nevertheless, many DSL providers have already been able to introduce services based on line sharing in some areas. For example, by March 2000, Northpoint had launched pilot services in four major metropolitan areas: Chicago, Dallas, New York and San Jose. It expects to be able to roll out full commercial service in these areas in the summer of 2000.

Competition has not only developed in services sold directly to the end-user, but also in wholesale DSL provision. There are a number of large intermediaries in the USA which have specialised in negotiating with ILECs for access to their local loop infrastructure and reselling this access to ISPs. Covad is one such company, and by March 2000, was able to offer wholesale ADSL access services to ISPs in 62 of the largest metropolitan areas in the USA. The service is expected to be available in over 100 such areas by the end of 2000. At that point, Covad’s services would be available to more than 40% of US homes and more than 45% of US businesses. These intermediaries have also been successful in attracting the business of major ISPs, which benefit from being able to negotiate with one wholesale DSL provider, rather than having to deal with a number of ILECs. For example, in January 2000, Northpoint announced that it had reached an agreement with MSN, whereby Northpoint will become the preferred DSL service provider for MSN’s Internet access offerings in areas where Northpoint offers DSL access.

Summary of the regulatory position

Exhibit 4.1 below summarises the regulatory environment in each of the countries studied as at March 2000.

Conclusions

The paucity of commercial DSL offerings in the UK makes it difficult to draw any definitive conclusion on the pricing of DSL services in the UK in comparison to that of other countries. Trial prices do give an indication of what service providers may charge for commercial services, but the wholesale prices for ADSL, on which the trial prices were based, were indicative and did not cover all the charges to the service provider related to providing a DSL service, which creates greater uncertainty as to the prices service providers will charge when they launch commercial services. Given those caveats, we can say that, on the basis of trial ADSL services, UK service providers’ charges for ADSL services are comparable to the other countries in this study.

In more general terms, our results show that DSL services for residential users are comparable between the USA and Europe in terms of price, although services are offered in a greater variety of bandwidths in the USA than in Europe. DSL services aimed at business users, however, are less expensive in the USA than in Europe, and the range of different bandwidths at which service is offered is greater. Although some European service providers are offering services at prices close to those seen in the USA at a downstream bandwidth of up to 2Mbit/s, US service providers are considerably less expensive than their European counterparts for services of a capacity greater than 2Mbit/s. However, 2Mbits/s may be sufficient for most users at the moment.

Prices and service levels are still evolving, even in the USA, the most mature of these still immature markets. With the advent of greater competition in the supply of DSL services, European prices are in general likely to fall (and service levels are likely to rise) towards the levels currently offered in the USA.

Annex A: DSL Prices by Country

This annex presents all the information we have collected on DSL prices and service specifications.

Exhibits A1 to A4 overleaf list the publicly available data as at March 2000 which we have collected on DSL prices by country and detail the service specification in terms of the target market, maximum upstream and downstream bandwidths, and service bundling. All charges are in £ using the PPP exchange rates given in section 2.1.4. Charges for commercial services apply to contracts of one year, except where otherwise stated.

Annex B:  National comparisons of the price level of retail services

This annex compares the data collected on retail DSL charges on a country by country basis, plotted against downstream speed.

Annex C: Current and potential DSL service providers investigated

The information presented in this study covers those service providers for which we were able to find publicly available data on DSL services. For the USA, we limited our sample to six service providers. However, for Europe we tried to gather data on all service providers which had launched or announced the launch of a DSL service. We were not able to find data for all the service providers we identified. Exhibit C1 below, lists the all the service providers considered in this study and indicates for which ones we were able to gather publicly available data on price and rollout.

Glossary of terms