Chapter 3

Internet access – broadband / always-on

Introduction

3.1 This chapter covers Digital Subscriber Line (DSL) and cable modem services. These services provide high-speed telecom services to consumers over the local loop/cable network. These services are becoming increasingly important in providing consumers with access to a range of services, such as fast Internet access and video on demand. This chapter focuses on comparing the cost for different types of users of broadband services providing Internet access. Further information on broadband markets is presented in chapters 4, 5 and 6 of this report.

3.2 Oftel carries out regular benchmarking exercises to allow a time trend to be developed, this provides greater information than a single snap shot. The results in this report will form part of the data input for any future reviews.

3.3 The results provide an update of the results presented in the December 2002 report (based on prices as at August 2002). This study is the sixth in a series, which started with a study carried out by Analysys for Oftel and published in April 2000 (based on prices as at February 2000).

3.4 A full update of prices has been carried out based on tariffs as at February 2003.

3.5 Since services differ in the amount of bandwidth (both downstream to the consumer and upstream from the consumer) that is available, price comparisons presented are viewed in the context of the bandwidth provided.

3.6 There are a wide range of services that are marketed as 'broadband' both in the UK and elsewhere. In particular the availability of services with downstream speeds in the range 128kbit/s to 256kbit/s is increasing. In monitoring take-up, Oftel includes all always-on services with downstream speed of 128kbit/s and above. This is consistent with the approach taken elsewhere and hence aids international comparisons (see Chapter 4).

3.7 In its review of the broadband market Oftel defined the broadband market to include:

3.8 Regarding price comparisons, it is useful, therefore, to distinguish "entry level" services (always-on services with speeds between 128 kbit/s and 256kbit/s) from services that offer the consumer speeds in excess of 256kbit/s, and results are presented for both definitions.

3.9 This chapter covers:

• a summary of the methodology used;
• results for a range of residential and business baskets, showing:
• price indices and detailed results;
• price movements over the period August 2001 to February 2003;
• charts which show the spread of offers by country;
• sensitivity analysis;
• comparison of relative prices of dial-up and broadband internet access; and
• conclusions.

3.10 Detailed results are shown in Annex C to this report.

Methodology

3.11 The basic methodology is as used in the December 2001 study and the details are not repeated here. The methodology takes into account any one-off charges for connection or modem purchase and it is assumed these costs will be written off over three years. However, following Oftel’s market review of broadband services the parameters of the baskets used have been changed to be consistent with the definition used in the review. Results are presented for two market segments: residential and business. The baskets are split as follows;

• for the residential market, the minimum speed of the residential broadband basket has been increased from 129kbit/s to 257kit/s. An additional basket has also been included to capture prices of "entry level" always-on services ie services 128kbit/s and above; and
• for the business market, results are split into three baskets; low, medium and high bandwidth services. The minimum speed of the low basket has also been increased from 129kbit/s to 257kbit/s, the other baskets remain the same as previous reports.

3.12 It is noted that there is not necessarily a clear dividing line between residential and business services. A significant number of home workers and small businesses may well find a service, aimed primarily at residential consumers, adequate for their needs.

3.13 Table 3.1 provides a summary of the comparisons undertaken. Usage per month in GigaByte (GB) was estimated for businesses in December 2002 study by analysing the pricing structure for some German service providers who offered both a flat rate and volume based tariffs. A similar analysis has been carried out based on February 2003 tariffs and the results are consistent with the values used last time, so these have not been changed. For residential consumers, usage in GB per month was previously estimated at 1.2GB per month. This figure continues to be used for the "entry level" basket. However, 3GB per month is used for the residential broadband basket, as shown in table 3.1. This is based on information provided by UK service providers. The effect of varying the usage rates is considered as part of the sensitivities.

Table 3.1: Summary of the comparisons undertaken

Note: GM = geometric mean (this is defined as: square root (downstream bandwidth x upstream bandwidth))

3.14 As in the December 2002 study, price comparisons are presented for both DSL and cable modem services taken together, and for DSL services only. Cable modem services are expected to provide a close substitute for DSL services, especially for residential consumers.

3.15 Price comparisons are presented alongside comparisons of bandwidth. Care needs to be taken in interpreting the results – there are a number of factors which will affect the actual speed experienced by the user when using the service and it has not been possible to take these into account. In particular the "contention ratio" (the ratio of the potential maximum demand to the actual bandwidth) will be an important contributor to the speed obtained in practice. However, this information is not readily available other than in the UK.

3.16 Additional assumptions need to be made to model the prices of particular services. The main assumptions made are as follows:

• for certain German services, DSL is only available over an ISDN (rather than PSTN) line. In these cases the additional cost of an ISDN line, compared to a PSTN line, is included in the DSL price. Where DSL is available over both PSTN and ISDN, the service over PSTN has been modelled;
• in Sweden, a "group" tariff may be available at a lower price than the price for an individual household. To obtain the cheaper tariff, several households within a limited area need to be connected and, householders need to enter into a separate "collaborative" agreement with the DSL provider. The prices for individual households have been used for the main analysis;
• for cable modem services, the price may be lower for customers who subscribe to other services (TV or telecoms) with the cable operator. The cheaper price available to cable customers has been used; and
• services have been designated residential or business based on the market sector that the operator/service provider appears to be targeting. However, while certain "residential" services specifically prohibit business use, others allow it. The distinction between residential and low-end business use is, therefore, not clear-cut.

3.17 Some of these assumptions are varied as part of the sensitivity analysis.

Results for residential users

Price index

3.18 Tables 3.2 and 3.3 below show the results for the price level, price index and (downstream) bandwidth provided for residential services. All these measures are constructed using the average of the two lowest priced residential offers from different operators/service providers included in this study, in each country (it is not possible to include all service providers and this study may not represent the cheapest deals available in every case). Results are presented both with, and without, cable modem offers included.

Table 3.2: Results for price level, price index and (downstream) bandwidth for residential broadband services

3.19 From Table 3.2 it is seen that as at February 2003 for residential broadband services (ie services with speeds greater than 256 kbit/s):

• when cable modems are included:
• UK prices are similar to France, cheaper than Germany and the US. Only in Sweden are prices cheaper; and
• the speed offered is comparable in France, Sweden and the UK, the US services offer lower speed and Germany higher speed; and
• when cable modem services are excluded (ie a comparison of DSL services only):
• UK prices are similar to France, cheaper than Germany and the US and only in Sweden are prices cheaper; and
• the speed offered is comparable in France, Sweden and the UK, but the US offer lower speed services and Germany higher speed services.

Table 3.3: Results for price level, price index and (downstream) bandwidth for residential "entry level" always-on services

3.20 From Table 3.3 it is seen that as at February 2003, for "entry level" services with speeds of 128kbit/s and above:

• when cable modems are included:
• prices in the UK are cheaper than all other countries, Sweden is slightly more expensive and prices in France, Germany and the US are over 11% more expensive.
• the speed offered is less in the UK than in other countries; and
• when cable modem services are excluded (ie a comparison of DSL services only):
• UK prices are similar to France and Sweden and cheaper than Germany and the US; and
• the speed offered is comparable in France and the US, and Germany and Sweden offer a higher speed.

3.21 Price comparisons of 128kbit/s and above show the UK, France and Germany in a more favourable light because in Sweden services with speeds below 257kbit/s are not available. In the US, there are services available below 257kbit/s but these are more expensive than a number of services with a minimum speed of 257kbit/s.

3.22 The results can also be presented by considering a chart which lists the cheapest offers of all packages considered in this study (arranged in price order), with downstream bandwidth superimposed. Figure 3.1 below shows the results for the residential broadband basket run with a minimum speed of 128kbit/s to illustrate all residential services considered in this study (note UK operators/service providers are highlighted using darker coloured bars).

Figure 3.1: Residential DSL and cable modem services in ascending order of price, with "speed" of service (downstream bandwidth) shown separately.

3.23 From Figure 3.1 it is seen that when considering the cheapest residential packages:

• the Swedish services all appear at the bottom end of the range;
• UK and French services are spread throughout the range; and
• German and US services factor only at the top of the range.

Price trend

3.24 It is not possible to directly compare the February 2003 results to the August 2002 results due to changes in the basket parameters, hence the August 2002 benchmarking results were run using the new parameters. A summary of the price changes since August 2002 is provided in Table 3.4 for both the residential broadband basket and the "entry level" basket. Price changes are presented in real terms. Changes in PPP exchange rates have some impact on the results. This table shows:

• for residential broadband services prices fell between 2-6% in the UK, Sweden and Germany and fell more significantly in France (this was a result of new DSL services offered over LLU); and
• for "entry level" services;
• prices fell in the UK and Sweden and fell more significantly in France; and
• price falls in the UK are a result of the introduction of a new 256kbit/s DSL service.

Table 3.4: Percentage change in price levels between August 2002 and February 2003 – residential services

3.25 Charts showing the spread of prices and bandwidths are shown in Figures 3.2 and 3.3 respectively. The price spread chart (Figure 3.2) shows the price for the cheapest package offered by each service provider as a percentage increase over the cheapest price across all countries. The bandwidth spread chart (Figure 3.3) shows the range of bandwidths on offer (based on the lowest and highest bandwidth offered by each service provider).

Figure 3.2: Price spread chart (February 2003)

Figure 3.3: Bandwidth spread chart

3.26 Figures 3.2 and 3.3 show:

• the range of prices in Germany is wider than in all other countries;
• services with speeds below 257kbit/s are available in all countries except Sweden; and
• the US offers the widest range of different speeds.

3.27 Price developments in the residential broadband basket (including cable modem and excluding cable modem) over the research period October 2000 until February 2003 are depicted in figures 3.4 and 3.4. Since the February 2003 results have been based on a different methodology (ie a minimum speed of 257kbit/s), a comparison of the old methodology and the new methodology have been included to ensure a consistent comparison can be made throughout the period. These figure show:

• the change in basket definitions does not have a big impact on the UKs relative position; and
• over the period, October 2000 till February 2003 broadband prices in all countries appear to be converging.

Figure 3.4: Broadband (including cable modem) price developments since October 2000

Figure 3.5: DSL only price developments since October 2000

Results for business users

Price index

3.28 Table 3.5 below shows the results for the price level, price index and bandwidth provided for business services. All these measures are constructed using the average of the two lowest priced business offers (from different operators/service providers considered in this study) in each country.

Table 3.5: Results for price level, price index and bandwidth for business broadband services (cable modem services included)

Note: DS = downstream, GM = geometric mean

3.29 From Table 3.5 it is seen that as at February 2003:

• for the most basic services available to business (ie where the business has a minimum bandwidth requirement of 257kbit/s):
• prices in the UK are similar to Sweden and more expensive than Germany;
• prices in France and the US are 44-70% more expensive than the UK; and
• the speed in the UK is lower than all other countries and France offers the highest speed.
• for business services where a minimum geometric mean (ie square root downstream bandwidth x upstream bandwidth) bandwidth of 500kbit/s is required:

• the UK is significantly more expensive than all other countries; and
• speed is broadly comparable in all countries.
• for business services where a minimum geometric mean bandwidth of 1000kbit/s is required:

• services are only available in Germany, Sweden and the US; and
• prices in Germany are much more expensive than Sweden or the US.

3.30 The results can also be run excluding cable modems. Fewer cable modem services are available to business than residential and generally the cheapest DSL services are cheaper than the cheapest cable modem services. The only country in which a cable modem service features in the index is Sweden – excluding cable modem services, therefore, has a negative effect on relative prices in Sweden.

3.31 Since the last benchmark, the UK relative position for basic business services has deteriorated. Prices in the UK have remained relatively stable but prices in Germany and Sweden have decreased and they are now slightly cheaper than the UK.

3.32 When there is a minimum bandwidth of 500kbit/s, the UK’s relative position has also deteriorated since August 2002 and the UK is now the most expensive country. Over the period UK prices have remained relatively stable and price falls have been significant in France, Germany and Sweden. The UK prices are now higher then those elsewhere. This may change if BT’s price cuts are fed through to the end user (see Chapter 5 – these price changes are currently subject to an Oftel investigation).

3.33 The results can also be presented by considering a number of charts which list the cheapest offers of the operators/service providers considered in this study (arranged in price order), with the relevant measure of bandwidth superimposed. See Figures 3.6 to 3.8 below (note UK operators/service providers are highlighted using darker coloured bars).

Figure 3.6: Business DSL and cable modem services in ascending order of price, with "speed" of service (downstream bandwidth) shown separately – minimum bandwidth 129kbit/s (downstream).

Note: cable modem services have (CM) after the service provider’s name. UK services are shown in bold.

Figure 3.7: Business DSL and cable modem services in ascending order of price, with "speed" of service (geometric mean bandwidth) shown separately – minimum bandwidth 500kbit/s (GM).

Note: cable modem services have (CM) after the service provider’s name. UK services are shown in bold.

Figure 3.8: Business DSL and cable modem services in ascending order of price, with "speed" of service (geometric mean bandwidth) shown separately – minimum bandwidth 1000kbit/s (GM).

Note: cable modem services have (CM) after the service provider’s name.

Price trend

3.34 In this report the minimum speed used for the basic business basket has changed. However, this has no impact on the basket results as all business services included have speeds above 256kbit/s. This means the data can be directly compared with the August 2002 results. A summary of price changes since August 2002 is provided in Table 3.6. Price changes are presented in real terms. Changes in PPP exchange rates have some impact on the results.

Table 3.6: Percentage change in price levels between August 2002 and February 2003 – business services.

3.35 Table 3.6 shows:

• for the most basic services available to business (ie where the business has a minimum bandwidth requirement of 257kbit/s):
• UK prices remained relatively stable; and
• prices in Sweden, France and Germany dropped dramatically. Price falls in all countries are due to the availability of new services at cheaper prices.
• for business services where a minimum geometric mean bandwidth of 500kbit/s is required:
• prices in the UK and the US have remained relatively stable; and
• prices dropped significantly in Sweden, Germany and France. Price falls in all countries are due to the availability of new services at cheaper prices.
• for business services where a minimum geometric mean bandwidth of 1000kbit/s is required:
• prices in Germany and the US remained relatively static and prices in Sweden fell significantly, this was due to new services available at cheaper prices.

Price and bandwidth spread

3.36 Charts showing the spread of prices and bandwidths are shown in Figures 3.9 and 3.10 respectively. The price spread charts (Figure 3.9) show the price for the cheapest package offered by each service provider as a percentage increase over the cheapest price across all countries for each business type. The bandwidth spread chart (Figure 3.10) shows the range of bandwidths on offer (based on the lowest and highest downstream bandwidth offered by each service provider).

Figure 3.9: Price spread charts

Figure 3.10: Bandwidth spread chart

3.37 Figures 3.9 and 3.10 show that:

• for all services Germany has the greatest price spread;
• for low speed services the UK, France and the US have similar spreads;
• for medium speed UK and France have a similar spread and the spread in the US is slightly larger;
• taking all services on offer, the range of speeds on offer in Sweden and in the US is much wider than elsewhere. The range is similar in the UK, Germany and France.

3.38 Price developments over the research period October 2000 until February 2003 are depicted in figures 3.11 and 3.12. Both figures appear to indicate greater convergence in prices over time. However, figure 3.12 shows prices in the other countries benchmarked are decreasing and that the UK has remained relatively stable.

Figure 3.11: Broadband price developments since October 2000

Figure 3.12: Broadband price developments since October 2000

Sensitivity analysis

3.39 There are a number of assumptions that have been made which potentially affect the results. These include:

• the definition of "broadband";
• the usage rate included for tariffs which depend on volume; and
• the treatment of some country specific issues:
• the treatment of ISDN costs in Germany;
• the treatment of "group" tariffs in Sweden;
• the treatment of costs for non-cable subscribers; and
• availability of "residential" services to business customers.

3.40 To test the robustness of the results, a sensitivity analysis has been carried out. The sensitivities for which detailed results are presented are as follows:

1. Change the definition of "broadband", so that the downstream speed must be at least 500kbit/s;
2. Double and half the usage rate;

3.41 The impact of additional costs for non-cable customers and business use of residential tariffs is commented on.

3.42 The results of the sensitivity analysis for residential and business customers are shown in Tables 3.7 and 3.8. Detailed results for business are shown only for "low speed" business services - the impact on consumers requiring higher bandwidth is commented on.

Table 3.7: Sensitivity analysis for price index for residential broadband services

Table 3.8: Sensitivity analysis for price index for business DSL and cable modem services ("low" speed basket)

3.43 The results of the sensitivity analysis for residential services show:

• changing the definition of broadband to include only services with a minimum speed of 500kbit/s downstream speed has almost no effect;
• increasing and halving the usage level has no effect on the results; and
• changing the treatment of some German and Swedish tariffs improves their relative positions.

3.44 For business use ("low" speed basket), the results of the sensitivity analysis show:

• changing the definition of broadband to include only services with a minimum of speed of 500kbit/s downstream speed has no effect on the results;
• doubling and halving the usage rate has no impact on the results;
• taking out the incremental cost of ISDN access in Germany improves the German prices, but does not alter the country ranking.

3.45 Results have also been run for the other business baskets. These show that:

• changing the definition of broadband has no impact on the business baskets (as these baskets already have higher speed thresholds); and
• changing the usage rates has no impact on the results.

3.46 The results of the sensitivity analysis illustrate that changes in assumptions can have an impact on the result for an individual country, but do not alter the overall conclusion regarding the UK's relative position.

3.47 A couple of other areas have also been considered:

• costs for non-cable customers; and
• business use of residential services.

3.48 Full information on the price of cable modem services to non-cable customers has not been collected, and it is not possible to test in detail the effects of including these additional costs. Most (but not all) cable operators appear to charge some additional amount to customers who are not taking other telephony or TV services. As cable modem services feature in the index for all countries for residential services, including these costs would raise prices in all countries.

3.49 A "business" service has been defined as a service that is clearly targeted at the business market. Some businesses may be able to use services that are targeted at residential use. The existence of "home workers" also blurs the distinction between residential and business use. In the UK, prices for the most basic business DSL services are similar to the prices for residential DSL, hence there is little scope for business to reduce costs by using "residential" packages. In other countries the differential is bigger and it is possible that business costs can be reduced in this way.

Comparison of dial-up and broadband prices

3.50 Comparisons can be made of the cost of always-on "broadband" (ie minimum speed of 257kbit/s) services compared to unmetered dial-up for residential and business consumers in each country.

3.51 For residential consumers, Table 3.9 compares the cost of broadband to the cost of "unmetered" dial-up access (based on the average of the two cheapest deals in both cases). Where unmetered dial-up access is not available, the cost used is for 150 hours access (definition as used by OECD). In these cases the cost of dial up exceeds the cost of broadband. Clearly the precise ratio depends on the usage rate assumed.

Table 3.9: Comparison of broadband and dial up cost per annum for residential users

3.52 The equivalent table for business consumers is shown in Table 3.10 (using the "low" speed basket).

Table 3.10: Comparison of broadband and dial up costs per annum for business users

3.53 The figures in tables 3.9 and 3.10 highlight the fact that for consumers in France and Sweden where dial-up unmetered Internet access is not widely available and is expensive where available, broadband is likely to be a cost effective alternative for high usage customers. In Germany residential broadband services are more expensive than dial-up although there is an always-on 64kbit/s cable modem service available which is cheaper than dial-up and for business services broadband is cheaper than dial-up. The usage rate used (150 hours per month) is high, and in practice the difference in price is likely to be lower.

3.54 The availability of unmetered dial-up Internet access may influence the rate of take up of broadband services, discussed in the following chapters.

Conclusions

3.55 Price comparisons have been presented for two residential baskets: residential broadband services (ie a minimum speed of 257kbit/s) and "entry level" always-on services (ie a minimum speed of 128kbit/s). For the residential broadband basket only in Sweden are prices cheaper than in the UK. For the "entry level" basket the UK is cheapest (when cable modems are included). Although a direct comparison with the August 2002 results can not be made, overall the UK's position has improved and prices are now below average for all baskets.

3.56 For business services, the UK’s relative position at August 2002 varies according to the speed requirements of the end user and has deteriorated since the previous benchmark for all baskets. For a "low" speed broadband service UK prices are average and for a "mid" speed service, the UK is now the most expensive. However, overall UK prices have remained stable and the deterioration is due to large price decreases in other countries. BT has recently cut the price of its wholesale services and this may lead to retail price cuts bringing the UK more in line with other countries (see Chapter 5 – BT’s price cuts are currently subject to an Oftel investigation).

  Chapter 4

Internet take-up and availability

Introduction

4.1 This section provides a summary of the Internet options available to consumers in the benchmarked countries and compares take-up, availability and rollout plans. This is based on data from OECD, regulators and service providers for the period December 2002 (except where stated).

Overview of dial-up Internet options

4.2 Table 4.1 provides an overview of unmetered Internet access services available to consumers and comments on the availability of FRIACO.

4.3 FRIACO (Flat Rate Internet Access Call Origination) is an unmetered wholesale product that allows other network operators to offer their own unmetered Internet access products in competition with the incumbent.

Table 4.1 Dial-up Internet options

4.4 The UK was the first European country to offer FRIACO in May 2000 and services using FRIACO are widely available. Unmetered access products are very popular in the UK and Oftel’s market research suggests at least 38% of consumers and 35% of businesses use an unmetered dial-up service when they connect to the Internet.

4.5 A wholesale FRIACO service is available in France and retail services started entering the market in the second half of 2002. So far there has been limited demand from operators given the current price levels. The demand from operators has also been hindered by the development of new retail ADSL products, including an offer for a cheap 128 Kbit/s serv