CONTENTS:
Note 1: Oftel's costing methodology
Note 2: Data requirements for the costing methology
Note 3: The benefits of ubiquity
Note 4: Relevant EC directives
1.1 The net cost of universal service is the difference in the universal service provider's financial performance with and without the universal service obligation. The evaluation of the net cost is divided into two stages. The first, which is the subject of this Note, is the evaluation of which customers (and exchange areas and public call boxes) are `uneconomic' for the universal service provider to serve using an analysis of the relevant costs and revenues associated with that customer. This yields a measure of the universal service costs. But this analysis does not take into account a variety of factors that might enhance the financial performance of the universal service provider. Such factors (eg effects of universal service provision on brand image and corporate reputation, implications of ubiquity) are grouped under the label `benefits'. To derive the net cost of universal service such benefits should be subtracted from the universal service costs. The evaluation of benefits to the universal service provider is not addressed in this note (see the discussion in Chapter 6).
1.2 To evaluate whether or not a customer is uneconomic a comparison should be made between the costs that the universal service provider would avoid if it were to discontinue service to that customer (the long run avoidable costs) and the revenues that it would lose (the revenues forgone). Long run avoidable costs include operating costs, depreciation and a reasonable return on capital employed. The long run is defined as the period over which all assets are replaced. Therefore, the long run view of costs means that all the costs of the capital equipment that the universal service provider would no longer require if it were to discontinue service to the customer will be included in the analysis. This is appropriate because the universal service obligation affects the investment decisions of the universal service provider. Since long run avoidable cost is an economic concept, the appropriate approach to asset valuation should be to use the replacement cost of the asset, as is done for example under Current Cost Accounting (CCA).
1.3 The measure of long run avoidable cost derived from cost information supplied by the universal service provider will reflect that operator's incurred costs. However, the efficient level of avoidable costs should be used in the calculation of the universal service costs. In the event that fund arrangements were to be set up, other operators should not have to pay for the inefficiency of the universal service provider. An efficiency adjustment should therefore be applied to the incurred long run avoidable costs to derive an estimate of the efficient level of costs. In the case of BT, the size of the efficiency adjustment can be informed by the analysis and assumptions used in Oftel's review of BT's price controls (retail and network).
1.4 The concept of revenues foregone is wider than just the revenue received from the customer being analysed. If a customer were to be disconnected, the universal service provider would lose not only the connection, line rental and call charges paid by the customer, but also call charges from incoming calls to that customer. A proportion of calls, both outgoing and incoming, might be made by other means if the customer were disconnected, eg calls to or from call boxes. There should therefore be a downward adjustment to the measured outgoing and incoming call revenues to take this into account.
1.5 The sections below discuss how the long run avoidable costs and revenues foregone should be derived in order to determine the universal service cost of areas, customers and public call boxes. The avoidable cost of an exchange area will include the operating and capital cost of the local loop, including associated indirect costs, the cost of the concentrator centre in the area, the cost of dedicated transmission links to the parent local exchange, and the cost of calls to and from the area. Some exchange areas in the UK may incur a universal service cost, because they are relatively expensive to serve (eg high local loop cost per line due to low density and/or long transmission links due to remoteness) or because they generate relatively low revenue (eg small number of customers in the exchange area with a small proportion of business customers). The avoidable cost of a customer will include the operating and capital costs of that customer's line, including associated indirect costs, and the cost of that customer's outgoing and incoming calls. Some customers may incur a universal service cost, because they are relatively expensive to serve (eg long line from the exchange) or, more typically, because they generate relatively low revenue (eg a low calling customer on the Light User Scheme). The avoidable cost of a public call box will include the operating and capital costs of the call box equipment, including associated indirect costs, the cost of the line to the exchange, and the cost of calls to and from the call box. Some call boxes may incur a universal service cost, because they are relatively expensive to serve (eg frequently vandalised) or because they generate relatively low revenue.
1.6 The text below refers to cost and revenue information that should be, and in some cases may not be, available from BT. The other universal service provider in the UK at present is Kingston Communications. In principle, the same methodology should be adopted to the quantification of Kingston's net costs of universal service. However, if Kingston is unable to supply sufficient cost and revenue information, one approach would be to estimate on the basis of the cost and revenue distributions derived from BT's information, adjusted as appropriate by relevant Kingston data and taking account of the geographical and other known characteristics of Kingston's service area.
This section discusses in detail the approach to estimating the long run avoidable costs of areas (2.1), customers on affordability schemes (2.2), and public call boxes (2.3).
(i) The starting point for the calculation of BT's avoidable costs of areas should be BT's financial statements on a CCA basis for the relevant year. This set of accounts attributes the individual components of BT's operating costs and capital employed among the different businesses and services provided by BT. For each component of the cost and capital employed of each service, the proportion of fully allocated cost that is likely to be avoidable should be assessed, based on the reduction in demand for the relevant asset or activity that would result from ceasing to undertake uneconomic activities, and taking into consideration the relevant economies of scale and scope.
(ii) For this purpose it would be relevant to consider evidence from BT's top-down incremental cost model for network services (after suitable adjustments specified by Oftel) and for retail services, if available. An alternative approach to assessing avoidable costs would be to use a bottom-up engineering model (including the bottom-up incremental cost model for network services). Using a bottom-up model might be one way to derive the efficient level of avoidable costs. However, if this approach were adopted, the results of the bottom-up model used would need to be reconciled to either BT's top-down model or BT's financial statements and adjusted, as appropriate, as a consequence. It should not be assumed that any gap between the top-down and bottom-up models is due solely to efficiency, since a variety of other factors could explain the difference. On the other hand, if the top-down model or BT accounting information were used, consideration would need to be given to the identification of an efficiency adjustment, so that the results reflect the costs of a fully efficient operator.
(iii) The resulting total avoidable costs should then be converted into unit avoidable costs by identifying the relevant cost driver for each cost component and dividing by the associated total cost driver volume. The avoidable costs of each area should then be calculated by multiplying the unit avoidable costs by the cost driver volumes relevant to each area, at the same time taking into account any area-specific cost factors. For example, maintenance costs would be affected by the travel time to the area and the frequency of faults.
(iv) All components of BT's operating costs and capital employed should be included in this process, and hence could potentially contribute to the avoidable cost of an area. However, the extent to which the fully allocated costs and capital employed are avoidable varies from component to component, across the entire range of 0% to 100%. For example, local line capital costs should be 100% avoidable, whereas redundancy and restructuring costs should be 0% avoidable.
(v) The avoidable costs of an area include not only the avoidable costs of access for subscribers in that area, but also the costs of transporting the calls, both outgoing and incoming, that would be lost if service were withdrawn from the area (taking into account any replacement of calls to or from disconnected lines by calls to or from the remaining lines).
(vi) The following cost drivers may be used to calculate the avoidable costs and capital employed for an area:
(viii) Most of these data on the characteristics of areas can be provided by BT, but only for a sample of 299 local exchange areas (BT's Local Line Costing Study, LLCS). Other data for these areas may need to be estimated using national and geotype-specific averages. Geotypes are a classification of local exchange areas according to line density and number of lines used by BT. The geotype of interest when identifying uneconomic areas is geotype 6, which has the lowest line density, although only a small proportion of geotype 6 areas are likely to be uneconomic. It may prove useful to sub-divide geotype 6 according to the number of lines in each area (eg less than 2,000 lines; between 2,000 and 5,000 lines).
(ix) Data on call minutes, both outgoing and incoming, may not be available for individual areas. These data should be estimated making use of the best available evidence. For example, one approach would be to use the number of business and residential telephone connections in each area, national average calling rates for business and residential lines, estimates of the relationship between outgoing and incoming traffic, estimates of the proportions of outgoing and incoming traffic that would be replaced, and geotype-specific factors reflecting evidence that calling rates both from and to low-density areas are different from the national average.
(i) The avoidable costs of a customer include both the avoidable costs of access and the costs of transporting calls made and received by the customer (taking into account any replacement of calls to/from disconnected lines by calls to/from remaining lines).
(ii) The starting point for the calculation of the avoidable costs of residential customers should be BT's CCA financial statements for the relevant year. For each component of the cost and capital employed of each service, the proportion of fully allocated cost that is likely to be avoidable should be assessed, based on the reduction in demand for the relevant asset or activity that would result from ceasing to serve uneconomic customers, and taking into consideration the relevant economies of scale and scope. The resulting total avoidable costs should then be converted into unit avoidable costs by identifying the relevant cost driver for each cost component and dividing by the associated total cost driver volume. An efficiency adjustment should be applied to convert from incurred costs to the efficient level of avoidable costs. As discussed above, an alternative approach would be to use bottom-up models.
(iii) The unit avoidable costs should be used to derive avoidable cost distributions for residential customers, using data on access line length and traffic distributions (both outgoing and incoming). In the absence of specific evidence, the avoidable unit traffic costs may be assumed to be uniform over the country. However, geotype-specific weightings should be applied to the avoidable access cost and information used on the incidence of customers by geotype. These will affect, for example, the cost per kilometre of access duct and cable, and the rate at which duct and cable costs increase with access line length.
(iv) All components of BT's operating costs and capital employed should be included in this process, and hence could potentially contribute to the avoidable cost of a customer. However, the extent to which the fully allocated costs and capital employed are avoidable varies from component to component, across the entire range of 0% to 100%. For example, local line maintenance costs should be 100% avoidable, whereas redundancy and restructuring costs should be 0% avoidable.
(v) The following cost drivers may be used to calculate the avoidable costs and capital employed for a customer:
(vi) One of the main drivers of the differences in access cost between customers is the length of the local loop. Access line length distributions which show the cumulative number of lines which terminate at a distribution point within a certain distance of the main distribution frame are available from BT. Such a distribution can be supplied by BT for six different categories of exchange area, with each category defined in terms of a maximum and minimum number of customer connections (business and residential) and a certain line density. These categories are known as geotypes. BT estimated these distributions based on a detailed line-by-line analysis of a number of areas in each geotype.
(vii) Costs are also incurred by conveying call minutes. Ideally, outgoing and incoming call minutes by customer would be supplied by BT. However, if detailed data on call minutes are not available on a per-customer basis, it will be necessary to construct estimates of outgoing and incoming call minutes. For example, useful for the estimate of outgoing call minutes will be the sample of residential call bills maintained by BT. This sample will allow a residential call revenue distribution to be derived showing the total call bill and the breakdown of expenditure by local, national and international calls, all of these split into peak, standard, cheap and weekend. This revenue shows direct customer billed revenues only and does not, for example, include incoming call revenues, revenues from PRS calls or called-party-pays revenues. Using BT's tariffs and details of how the various option plans affect these tariffs, it is possible to construct estimates for outgoing call minutes by customer. An allowance should also be made for relevant call minutes not included in the sample, such as arising from call-party-pays calls and calls to PRS. The avoidable costs of all call minutes included in the revenues foregone (see below) should be taken into account.
(viii) For incoming call minutes there may be less relevant evidence available. In the absence of direct information on the number of incoming call minutes by type of call and type of day, one approach might be to estimate a functional relationship which, given a customer's call bill for local and national call minutes, estimates the revenue from the incoming local and national calls to that particular customer. To convert incoming call revenue to the number of incoming call minutes a similar approach may be used as suggested above for calculating outgoing call minutes, assuming the same local/national, peak/off-peak distribution as for outgoing call minutes.
(ix) As previously discussed, any assessment of the extent to which fully allocated costs are likely to be avoidable should be based on the reduction in demand for the relevant asset or activity that would result from ceasing to serve uneconomic customers, and taking into account the relevant economies of scale and scope. For example, the extent to which local access cable costs will be avoidable is likely to depend upon the proportion of customers in each type of area that are uneconomic, and possibly also upon their distribution in terms of distance from the exchange. A further complication is that the identification of uneconomic customers itself requires knowledge of their avoidable costs.
(x) The problem is therefore one of finding a stable fixed point simultaneously for the set of uneconomic customers and the corresponding avoidable costs. One approach to solving this problem might be to employ an iterative search technique, such as the following:
(xi) In order to avoid double counting, customers in areas that have been identified as uneconomic must be excluded from the set of individual customers identified as uneconomic when estimating the overall cost of those customers and areas. The precise way that this should be achieved depends upon the detailed approach adopted to the identification of uneconomic customers and areas. One possible approach is as follows. The location of individual customers might not have been linked to individual (identified) areas but to so-called geotypes groups of areas having somewhat similar characteristics (specifically, similar average line density and number of lines). The choice of uneconomic customers to exclude would therefore be based on the proportion of customers in each geotype that had been identified as being in uneconomic areas.
(xii) When estimating the extent to which costs would be avoidable if uneconomic customers and areas were not served, it will be important to identify the relevant level at which to aggregate the resulting reduction in demand. For example:
(xiii) Estimates of the avoidable costs for those elements dependent upon aggregate demand from all customers and areas will need to be based upon the total demand reduction that would result from not serving all uneconomic customers and areas. It may therefore be necessary not only to use an iterative technique to identify the set of uneconomic customers and their corresponding total avoidable cost, but also to undertake an overall iteration to ensure that the results from the uneconomic customer and uneconomic area calculations are in harmony.
(i) The starting point for the calculation of the avoidable costs of public call boxes should be BT's CCA financial statements for the relevant year. For each component of the cost and capital employed of the PCB service, the proportion of fully allocated cost that is likely to be avoidable should be assessed, based on the reduction in demand for the relevant asset or activity that would result from ceasing to provide uneconomic public call boxes, and taking into consideration the relevant economies of scale and scope. The resulting total avoidable costs should then be converted into unit avoidable costs by identifying the relevant cost driver for each cost component and dividing by the associated total cost driver volume. These unit avoidable costs should be used to derive avoidable cost distributions for public call boxes, using data on the distributions of direct maintenance costs and revenues between call boxes. During this process an efficiency adjustment should be applied so that the resulting distributions reflect an estimate of the efficient level of costs.
(ii) All components of BT's operating costs and capital employed are included in this process, and hence could potentially contribute to the avoidable cost of a customer. However, the extent to which the fully allocated costs and capital employed are avoidable varies from component to component, across the entire range of 0% to 100%. For example, PCB direct maintenance costs should be 100% avoidable, whereas redundancy and restructuring costs should be 0% avoidable. The avoidable costs of a PCB include not only the avoidable costs of the PCB itself, but also the costs of access and call conveyance.
(iii) The following cost drivers may be used to calculate the avoidable costs and capital employed for a PCB:
This section discusses in detail the approach to estimating the revenues foregone for areas (3.1), customers (3.2), and call boxes (3.3).
(i) Types of revenue that would be foregone as a result of withdrawing service from an area include:
See Appendix A for a detailed list of relevant types of revenue foregone.
(ii) However, only a proportion of these revenues would be foregone as a result of withdrawing service from an area. Firstly, part of these revenues may have already been lost as a result of withdrawing service from other uneconomic areas. Secondly, some of the demand which generates these revenues may be replaced by increased demand for service in other areas - in particular, calls both from and to disconnected lines may be replaced by calls from or to lines (or call boxes) in other areas. For the latter factor, in the absence of specific evidence to the contrary, revenues foregone should be reduced by proportions in the region of 0% to 20% of outgoing calls replaced and 0% to 10% of incoming calls replaced.
(iii) BT may be unable to supply the required data on the revenues associated with individual areas. If so, the data will need to be estimated. For example, for type 1 revenue one approach would be to produce estimates on the basis of the number of business and residential telephone connections in each area, national average revenues per line for business and residential lines, estimates of the proportion of outgoing traffic that would be replaced, and geotype-specific factors reflecting evidence that calling rates from low-density areas differ from the national average. Similarly, type 2 revenue might be estimated using the estimated relationship between outgoing and incoming traffic, estimates of the proportion of incoming traffic that would be replaced, and geotype-specific factors reflecting evidence that calling rates to low density areas differ from the national average. Allowances for revenues from called-party-pays calls (type 3) and interconnect revenues (type 4) and should be made.
(i) As for areas, there are four types of revenues that would be foregone as a result of withdrawing service from a customer: revenues billed to a customer for both access and calls; incoming call revenues; called-party-pays revenues; and interconnect revenues. A more detailed list of relevant revenues foregone is at Appendix A.
(ii) Not all of this revenue would be foregone as a result of withdrawing service from a customer. Firstly, some of this revenue may have already been lost as a result of withdrawing service from uneconomic areas and other uneconomic residential customers. Secondly, some of the demand generating this revenue may be replaced by increased demand for service from other customers - in particular, calls both from and to disconnected lines may be replaced by calls from or to lines of other customers or call boxes. Revenues foregone should be reduced using an estimate of the proportion of outgoing and incoming traffic that would be replaced. In the absence of specific evidence to the contrary, these proportions should be in the region of 0% to 40% for outgoing calls and 0% to 20% for incoming calls.
(iii) As noted in section 2.2 above, BT maintains a detailed residential revenue distribution, both in terms of the number of call bill segments as well as revenue by destination and time of day. This information should form the basis for the measure of type 1 revenue foregone, ie revenues billed to customers for access and calls. In addition, an allowance should be made where possible for other relevant types of billed revenue, which may be omitted from BT's revenue distribution, such as calls to PRS.
(iv) Type 2 revenue foregone should be estimated using the best available evidence on incoming calls to customers, perhaps using a function describing the estimated relationship between outgoing and incoming call revenues. Allowances should also be made for types 3 and 4 revenues foregone, using the best available information.
(v) The approach outlined above results in a single revenue foregone distribution for all residential customers, regardless of location. However, if there is evidence that calling rates both from and to specific types of area (eg low-density areas) are different from the national average, the national distribution should be adjusted accordingly to derive area or geotype-specific revenue foregone distributions.
(i) Revenues foregone as a result of withdrawing service from a PCB include:
(ii) Both BT and Kingston Communications collect detailed data on the revenues collected from public call boxes. An allowance should also be made for the other components of revenue foregone.
The following sources of revenue should be included in the revenues foregone for areas and customers.
1 The methodology for calculating the cost (before benefits) to universal service providers of providing universal service is based on calculating the costs which would be avoided and the revenues which would be foregone if those operators were not obliged to provide voice telephony services to customers, or to entire areas, when it is uneconomic to do so. Identification of areas and customers which are uneconomic to serve is likewise based on their avoidable costs and revenues foregone.
2 The information required is therefore about the revenues and costs associated with providing telephone service to different categories of customer; the appropriate choice of these customer categories being driven by the need to identify and cost those that are uneconomic. Customer connections are categorised according to the revenues that they generate and the costs incurred in serving them. Because the methodology considers both individual customers and entire areas, customer connections have to be categorised separately according to their individual revenues and costs, and according to the total revenues and costs of the area within which they lie.
3 It is important to emphasise that it is the variation in avoidable costs and revenues foregone between different categories of customers and areas that determines which are economic and which uneconomic to serve, and the financial cost of serving them. The information required by the methodology is therefore not averages but distributions, eg not average revenue and cost per customer connection but the distribution of customer connections by revenue and cost.
4 Information is required about traffic volumes as well as revenues because, whilst certain information may be available in the required form (eg distribution of customer lines by annual revenue), other information may have to be estimated on the basis of proxy variables (eg cost of calls on the basis of traffic volumes, cost of access in part on the basis of local switch size). The methodology therefore requires data on these proxy variables as well as revenues, and the information needs to be disaggregated according to factors that may affect the costs of service provision. It also follows that any additional information about the characteristics of customers or areas which can easily be collected at the same time as the specific information requirements detailed below, should be collected, especially where such information may have a bearing on the avoidable costs or revenues foregone of those customers or areas.
5 Information is required about the connections, calls, revenues and costs of providing the voice telephony services of the universal service provider.
6 Information is required separately for each of the following categories of customer connection. Note that these categories of customer connection have been identified as those most relevant to the identification and costing of uneconomic customers and areas, whilst still being potentially practical to implement. Alternative categorisations may be acceptable, or even desirable.
7 Residential customer connections should be categorised by tariff scheme, annual call bill and type of area. The categorisation of connections by tariff scheme is required so that the number and universal service cost of uneconomic customers using each scheme can be identified separately. To limit the amount of detailed information that needs to be supplied, Oftel has proposed that only the Light User Scheme and its replacements (including Lifeline) would need to be costed and that uneconomic customers only on those schemes would be eligible for funding.
8 The categories of annual call bill for the identified schemes are to be broken down into at least £5 steps. The categorisation of residential customer connections by type of area is intended to allow geographic variations in the cost of providing service to be incorporated into the analysis. This variation is likely to come about as a result of factors such as individual customer line length, average line density, local exchange area size, and geographic remoteness. This information is only useful in the context of additional information on the variation in costs between areas of different types; the number and structure of area types should therefore be determined by an analysis of the geographic variation in costs of service provision by each universal service provider.
9 Residential and business customer connections (separately) should be identified in potentially uneconomic local service areas, separately for each such local service area. The appropriate definition of local service area should be based upon an analysis by each universal service provider of the scale and avoidability of different components of the costs of service provision. For a wireline operator, this analysis typically reveals that a far greater proportion of total costs can be avoided, in the long run, if a concentrator centre and all of its associated local access network is removed, than if only a proportion of customer connections in the area are no longer provided with service. This cut-off point results largely from the difference between traffic-insensitive costs of local access and switching, and the lower traffic-sensitive costs of higher-level transmission and switching. At least for wireline access providers, the appropriate definition of local service area would therefore appear to be the area served from a site at which the first stage of traffic concentration occurs (eg a concentrator centre). However, this should be verified by each universal service provider, on the basis of detailed information about the network architecture employed.
10 Universal service providers are not required to collect information about customer connections in every local service area, only for those areas that they consider likely to be uneconomic to serve. The basis on which they identify such potentially uneconomic areas is at their discretion - previously, for example, such areas have been identified on the basis of the their average line density and total number of lines.
11 To summarise, information is required about the connections, call minutes and revenues for residential customer connections and business customer connections (separately), in each potentially uneconomic service area (separately).
12 The following information is required separately for customer connections in each of the customer categories set out above:
12.1 number of customer connections in category - average for the year
12.2 access revenues received from customers in category - total for the year
12.3 number of call minutes billed to customers in category (used to calculate the costs of transporting outgoing calls paid for by the customer) - total for the year, broken down by charge band and time of day eg:
| Standard | Cheap | Weekend | |
| Local | |||
| Short national | |||
| Long national | |||
| International Charge Band 1 | |||
| etc... |
(Note that this category is intended to cover all calls originating from the relevant customer connections which are not billed to other customers. In particular, free local calls should be included in this category, whether or not they actually appear on customer bills.)
These data are used to estimate the avoidable costs of call transport. The structure of the breakdown of call minutes should therefore reflect the structure of call transport costs as far as possible (eg be distance/interconnect charge/time of day dependant). In particular, the need to break down call minutes by time of day is intended to allow the higher cost of transporting calls during peak periods to be reflected in the avoidable cost estimates.
Obviously, where detailed call transport cost information can be collected directly for customer connections in each category (eg interconnect payments), this is preferable to using call minutes as a proxy. However, it is unlikely that such detailed data will be available for all components of the cost of transporting calls, and so the need for some call minute data is likely to remain.
12.4 revenues received from customers in category for calls billed to customers in category (revenues from outgoing calls made by the customers, excluding called-party-pays and indirect access interconnect calls) - total for the year
A breakdown of these revenues (for example by charge band and time of day as above) is not required by the methodology. However, such a breakdown would be useful in support of, and perhaps even as a proxy for, detailed call minute data.
The revenues quoted should not normally be net of interconnect payments (such payments should be separately identified as a cost of call transport). The only circumstances in which it might be appropriate to quote revenues net of interconnect payments is where the net revenue and cost per call minute are largely independent of the call type, and the gross revenue and cost are not (for example, where an access service provider retains a fixed amount per minute for all calls passed over to a long-distance carrier whatever the final destination of the call), in which case the use of net revenues should be clearly indicated in the final data, and the relevant interconnect payments should be excluded from call transport cost data.
12.5 number of retail call minutes incoming to customers in category (used to calculate the costs of transporting calls received by the customers) - total for the year, broken down by charge band and time of day as above
12.6 revenues received from other customers for retail calls incoming to customers in category (revenues from incoming calls received by the customers) - total for the year, where possible broken down by charge band and time of day as above
12.7 number of called-party-pays call minutes (eg reverse charges, 0800, 0345) from customers in category (used to calculate the costs of transporting called-party-pays calls made by the customers) - total for the year, broken down by call type, charge band and time of day
12.8 revenues received from other customers for called-party-pays calls from customers in category (revenues from outgoing called-party-pays calls made by the customers) - total for the year, where possible broken down by call type, charge band and time of day as above
12.9 number of interconnect call minutes (both outgoing and incoming) for customers in category (used to calculate the costs of transporting interconnect calls made and received by the customers) - total for the year, broken down by call type, charge band and time of day
12.10 revenues received from other customers (including OLOs) for interconnect calls (both outgoing and incoming) for customers in category (revenues from interconnect calls made and received by the customers) - total for the year, where possible broken down by call type, charge band and time of day as above
13 Note that this categorisation of call minutes and call revenues should be taken as indicative of the fact that information is required about every type of call made from or to customers in each category, rather than as a precise definition of the appropriate call categorisation. Note also that these categories are intended to be mutually exclusive, that is to say each call minute (and associated revenues) should be attributed to exactly one of these categories, unless explicitly indicated to the contrary in the final data. For instance, 0345 call minutes might appear both under customer billed call minutes and called-party-pays call minutes with the appropriate revenues attributed to each corresponding revenue category, in which case these call minutes and revenues should be separately identified in the final data and any double counting of call minutes explicitly noted.
14 Estimating the avoidable costs of providing voice telephony services to customers, or to entire areas, when it is uneconomic to do so, requires information about the costs and cost structures of the universal service provider. Where possible, cost data that is specific to individual customers or areas should be used. However, operators generally do not collect cost data at this level of detail and so such costs have instead to be estimated on the basis of aggregate cost data and proxy, cost driver, information.
15 The exact information required therefore depends upon the level of detail at which actual cost data are captured and reported. The following provides an indication of the types of information required when only a limited amount of cost data are captured at any level other than for entire services.
16 Total operating expenses for the year, broken-down by:
17 Total depreciation and capital employed broken-down by:
These data should be produced on a CCA, rather than an HCA basis.
18 Disaggregated information about the characteristics of the network and its operation affecting the costs of serving individual residential customers and entire areas is required to allow disaggregated avoidable cost estimates to be produced from the previously identified aggregate cost data. The information required includes:
19 The following information, necessary to accurately and precisely estimate the costs of universal service provision, was unavailable at the time of the last costing study:
20 The progress that has subsequently been made towards filling these data gaps is now discussed.
21 BT has initiated a study of outgoing and incoming calls using information sampled from a new customer information system. Information is being collected for:
22 Separately information is being collected for samples of residential customers drawn from:
23 It is understood that the information being collected identifies the origin and destination of the call, the time that it commences, and its duration. Using this information BT is able to calculate:
24 Initial results have been produced by BT, demonstrating that the systems used to collect and process the samples work. However, statistically significant results, on the basis of which firm conclusions might be drawn, have not yet been delivered by BT.
25 The new customer information system would appear to provide a valuable and reliable source of data of precisely the types required for the analysis of universal service costs. What is less clear, at this stage, is whether the studies currently being undertaken by BT will be sufficient to provide the data required to estimate universal service costs accurately, or if further work will be required to extract appropriate data from the new customer information system.
26 No progress has been made towards accurately estimating these parameters. It is not certain that it is actually possible to do so without withdrawing service from certain customers and areas - an approach that is clearly unacceptable in practice.
27 A market survey, asking questions of existing low-spending customers and customers in areas likely to be uneconomic, as well as households not currently having a telephone, might provide some evidence about these parameters. However, any estimates produced as a result would still be subject to a large degree of uncertainty.
28 As part of a recent study for Oftel, on factors affecting the geographic variation in telecoms service costs, Analysys undertook a more detailed analysis of the BT Local Line Costing Study (LLCS), the results from which were used during the latest universal service costing work, and developed a statistical model to predict similar results for areas not covered by the LLCS.
29 The statistical model takes as its inputs:
30 The fit of this model to the results from the LLCS is, in general, fairly good (in statistical terms, the R-squareds are close to 1). However, the fit is not particularly good for individual high-cost areas (the residual variance is high for these areas).
31 Using this model, revised estimates of the scope and universal
service cost of uneconomic areas in the UK have been produced,
including maps of their approximate location. However, because
of differences between the results of this model and those from
BT's LLCS, this model identifies as being economic certain sample
areas previously identified as uneconomic, and vice-versa. The
reliability of these results is therefore
still open to question.
32 To produce robust estimates of the universal service cost of uneconomic areas, including the identification of which particular areas of the UK are uneconomic, further work is required either:
33 During Oftel's recent study on the geographic variation in telecoms service costs, data from BT on the characteristics of all remote-local logical transmission links in the UK were obtained and analysed. It is understood that the system from which this information was obtained also holds information about the characteristics of all physical links.
34 Estimating the avoidable costs of remote-local transmission links is a complex process, not least because of the potentially complex relationship between logical and physical links - a single logical link may be carried over many physical links, and each physical link may carry a number of logical links.
35 Now that it is possible to identify the geographic location of all uneconomic areas (albeit using a model of local access costs rather than precise estimates), a more detailed study of the logical and physical transmission systems employed in serving these areas becomes a possibility.
36 To improve the quality of the estimates of the universal service costs of uneconomic areas, further work should be undertaken to:
37 BT has been unable to provide data on the variation in operating costs between different parts of the UK. To improve the quality of the estimates of universal service costs, further work should be undertaken to:
38 Certain information required for the accurate and precise calculation of universal service costs is still unavailable. As a result, universal service costs can only currently be estimated to an accuracy of approximately ±25%.
39 It may prove impossible, even in the long term, to obtain accurate and precise values for all of the missing data. Some of the missing information concerns how customers would behave if they did not have access to the telephone service, either individually or across an entire area - information which could only be obtained accurately if universal service were actually to be withdrawn. There will always, therefore, be some residual uncertainty about the true costs of providing universal service. However, as set out above, the quality of many elements of the required information could be substantially improved.
40 Whilst information concerning the exact costs of serving potentially uneconomic areas is important if the identity of such areas is to become known with any degree of accuracy, the impact of any changes on the overall universal service cost is likely to be relatively low. There are not very many such areas, and the costs of serving them are consequently not the major part of the total cost of providing universal service.
41 Of the information which remains unavailable (and which it should be possible to obtain to a reasonable degree of accuracy), probably the most important in terms of its impact on the overall cost of universal service provision is that concerning incoming call volumes and revenues. A relatively small change in the distribution of incoming call volumes and revenues between different residential customer groups (for example an increase in revenues of 10% for the lowest spending residential customers) can have a significant impact on the number and universal service cost of uneconomic residential customers (for example a reduction in universal service costs of approximately 20%).
1 Every year a proportion of households moves from location to location. All households that move into a new area will be aware of BT as a potential provider of telecoms services in that area since BT, because of the universal service obligation, serves all types of customers in all parts of the UK. The households may not, however, be aware of the existence of an alternative supplier. As a result, a proportion of households will choose BT as its supplier although it would have chosen a competing supplier had it been aware that it served the area. These customers represent one of the gains that BT obtains from ubiquity in a competitive environment.
2 The model that has been developed to estimate the value of this type of benefit to BT has three main elements:-
3 In the UK about 11% of households moves location each year (Source: Housing in England, 1994/5, HMSO). The number of households with a choice of direct access provider (number of lines passed and released for marketing by cable operators) in the middle of the financial year 1996/97 was about 7.8m. Therefore, an estimate of the number of households in new locations with the choice of a supplier competing with BT in 1996/97 is about 858,000 (=7.8m x 11%).
4 But not all of these new households would have been aware of the alternative supplier. It is assumed that 66% of households does not know about the competing supplier. Of these households, it is assumed that 60% chooses BT and would have done so if fully informed. The remaining 40% also takes service from BT, but only because of imperfect information - if these households had known about the competing supplier, they would not have chosen BT. The proportion of new households that BT gains through ubiquity is therefore 26% (= 66% x 40%). This corresponds to around 225,000 lines (= 858,000 x 26%).
5 The approach and assumptions are summarised in the decision tree. The figures in the decision tree were informed by a recent consumer survey commissioned by Oftel. The difficulties of interpretation and the small sample size involved imply that the results of the survey cannot be regarded as conclusive. Nevertheless, given the lack of relevant direct evidence to inform the values of the probabilities in the decision tree, the survey provides useful guidance. The effect of varying these assumptions is reported below in the discussion of the sensitivity analysis.
6 The stock of lines gained through ubiquity will decline over time for two reasons. First, the households will learn about the existence of the competing supplier and some will choose to switch to that supplier (it is assumed that 25% of the original stock of lines switches each year). Second, there is a small probability that the lines will be subject to a further household move and that the new occupant will choose a competing supplier. With these assumptions, the stock of lines gained in the initial year reduces each year and reaches zero after four years.
7 The initial per line net contribution is taken to be the bill of the average residential customer minus the estimated marginal costs to BT of serving such a customer. The estimate is based on the avoidable cost analysis in the universal service costing work. There are two factors which will make the costs of serving the particular customers considered here relatively low: a very small proportion of lines is affected in any area (less than 3% on average), and those lines are retained for a short period of time (2 years on average). In the base case the initial per line net contribution is assumed to be £100 - see below for the effect of varying this assumption.
8 In subsequent years the net contribution per line is assumed to remain approximately constant in real terms, ie increasing in nominal terms (reflecting assumptions about the rate of inflation, price reductions, volume growth and marginal cost reductions). To measure the current value of the stream of future net contributions, a discount factor of 12.5% per annum is applied (BT's nominal cost of capital).
9 The benefit that BT obtains from the stock of customers gained in the initial year is given by the sum of the (declining) stock of lines in each year multiplied by the corresponding (discounted) per line net contribution. With the assumptions set out above, the benefit from ubiquity in the current year would be about £50m.
10 This figure relates only to the benefit from the stock of customers gained in the initial year (1996/97). In all years a set of households moves location, giving rise to a different stock of customers gained by BT by virtue of its ubiquity. On the one hand, there are factors that might lead the value of the benefit to increase over time. The increasing coverage of competing suppliers (reaching 75% in 2000/01) means that a larger number of moving households will be in areas where there is a choice of operator. The value of the benefit (in nominal terms) might also increase, because of a rise in the per line net contribution through time. On the other hand, a general increase in the awareness of competing suppliers will reduce the proportion of moving households gained through ubiquity, by reducing the proportion that does not know about the existence of competing suppliers (see the decision tree). It is assumed that, as the coverage of competing suppliers increases, there are similar increases in the proportion of moving households that knows about the existence of competing suppliers. With the base case assumptions the value of the benefit increases up to about £60m over a couple of years and then falls back to about £50m by 2001.
11 The value of ubiquity benefit is sensitive to the parameter assumptions used. For example, varying the proportion of customers gained through ubiquity that subsequently switch to competing suppliers between 15% per annum and 35% per annum (from the base case assumption of 25%), gives a range for the value of benefits of £40-80m. A similar range would be created by varying the net contribution per line between £80 and £120 (from the base case assumption of £100), or by varying the proportion of moving households that would have chosen the competing supplier if they had been fully informed between 30% and 50% (from the base case assumption of 40%). If these sensitivities to the three parameters were applied cumulatively rather than individually, the resulting range for the value of benefits would be large, £30-130m. But even under quite pessimistic assumptions, there would still be a significant benefit from ubiquity.
12 The illustrative calculation that was put forward by BT resulted in a very low figure (£3m) for two main reasons:-
1 The Full Competition Directive has been adopted and required implementation by Member States by January 1997. A common position on the Interconnection Directive has also been reached in the Council and, although it is now going through a conciliation procedure with the European Parliament, it is likely to be implemented by 31 December 1997. The proposed Amending Voice Telephony Directive (AVTD) will replace the existing Voice Telephony Directive and sets out harmonised conditions for fixed voice telephony and the scope of universal service. Although it is still under discussion, it is expected to be adopted in time to come into force by 1 January 1998. In addition, guidelines from the Commission on costing and funding universal service have recently been issued. (Commission Communication on Assessment Criteria for National Schemes for the Costing and Financing of Universal Service and Guidelines for the Member States on Operation of such Schemes, November 1996)
2 The draft AVTD (COM (96) 419) sets out a series of measures aimed at ensuring that fixed voice telephony is available to all on reasonable request. Licensed operators designated by Member States as having universal service obligations should meet reasonable requests for a fixed network connection, allowing customers to make and receive calls and allowing a fax and modem to operate. PCBs to meet reasonable needs, telephone directories and, where appropriate, specific measures to ensure access to telephone services for users with special needs should also be provided. Member States are required to ensure that access to basic telecoms service remains affordable in rural or high cost areas, and also for vulnerable groups such as disabled people or people who make little use of the phone. Special targeted tariff schemes can be introduced for this purpose.
3 The draft AVTD also requires certain provisions to apply to all organisations providing fixed public telecoms networks and/or publicly available telephone services. These include customer access to directory enquiry services; access to operator assistance; free access to 999/112 emergency services; customer contracts which specify quality levels and compensation available if these levels are not met; and published disconnection procedures. Operators should also provide adequate information about the prices of their services and about their quality.
4 In addition, the draft AVTD requires all organisations providing fixed public telephone networks and/or publicly available telephone services to make tone dialling, itemised billing (free of charge) and selective call barring (so that customers can bar, on request, outgoing calls of defined types or to defined types of numbers) available to most customers by December 1998 and generally available by December 2002. It also requires Member States to authorise specific measures relating to disconnection, including the provision of an `outgoing calls barred service' prior to absolute disconnection in most cases. It should be noted that the precise content of the AVTD may change as negotiations about it continue.
5 Under the draft ICD, in cases where universal service obligations represent a demonstrable net cost to an operator, then National Regulatory Authorities may set in place arrangements for those costs to be shared in a proportionate, transparent and fair way amongst public telecoms operators. The draft ICD restricts the level of universal service that can be funded by such means and sets out the costing methodology to be used. Only costs which a commercial organisation would avoid if it did not have universal service obligations should be included. The approach should be forward looking; be net of revenues; and take account of any benefit to the company derived from universal service provision.
6 Under the FCD, Member States must communicate any universal service financing scheme to the Commission so that it can verify that the scheme is consistent with EC law.
7 The recently published Guidelines from the Commission set out in more detail the approach to be followed in costing universal service. This states that only the unavoidable net losses incurred by an efficient operator in providing universal service should be taken into account. The net cost calculation should therefore identify those customers where, based on the long run avoidable costs less associated revenue, a loss is incurred. Indeed, an operator is entitled to receive funding only in respect of customers which the operator has explicitly declared would not be served but for universal service obligations (or in the case of new provision, would not be prepared to serve on request). Costs associated with maintenance of public security, provision of services outside the scope of universal service, compensation or refund payments, and the costs of providing facilities which all operators must provide should not be included. The Guidelines also state that account should be taken of the intangible benefits of being a universal service provider, such as enhanced brand recognition, benefits of ubiquity, life cycle benefits and marketing benefits.
8 Within this overall framework, Member States will have discretion over how arrangements are implemented. Included amongst the issues for determination at national level are:
9 Member States may determine that universal service financing is not required because the obligations do not result in a net cost or, if they do, that the cost is not sufficiently large to constitute an unfair burden.
10 Member States also have discretion to decide whether or not to impose specific universal service obligations on certain market players or whether market forces can be relied upon to deliver the required service levels; whether to apply specific pricing constraints on elements of universal service (in order to achieve affordability objectives); and whether any obligations should be franchised or competed for in other ways.
The Consumer Panel comprises:
Moira Black - Chairman of the English Advisory Committee on Telecommunications and previously a partner at Price Waterhouse;
John Hughes - Former member of the National Consumer Council where he chaired the National Consumer Council's Economic Policy Committee;
Stephen Locke - Director of Research and Policy at the Consumers' Association and Executive Member of the Bureau of European Consumer Unions;
Richard Thomas - Director of Public Policy at Clifford Chance and previously Director of Consumer Affairs at the Office of Fair Trading; and
Courtenay Thompson - Chairman of the Northern Ireland Advisory Committee on Telecommunications and former member of the Northern Ireland General Consumer Council.
1 The Price Control Review Consumer Panel (`the Panel') is established by the Director General of Telecommunications under S.54(3) of the Telecommunications Act 1984.
2 The remit of the Panel is to:
3 The Panel will meet as and when necessary from September 1995. It will receive administrative and other support from Oftel's Consumer Branch.
Broadband - A service or connection allowing a considerable amount of information to be conveyed, such as television pictures.
Call Diversion - Where calls to a particular number are automatically routed to another number.
Call Level - A Call Level is a set level of call spend (ie excludes line rental) which a new BT customer agrees to per quarter for the first 12 months of service. The advantage of Call Levels is that for new residential customers they have virtually removed the need for a deposit. BT plan to make Call Levels available to those customers who have experienced payment difficulties because along with BT's Call-My-Bill service they will enable customers to have greater control over the size of their bill.
Call Return - BT service which allows customers to find out the number of the last person who called, together with the date and time that they rang.
Call Waiting - BT service which allows a customer to accept a new call while already on an existing call.
Calling Line Identification Display - Where the number of the caller is displayed to the called person prior to the call being established.
Charge Advice - BT service which tells the customer how much the last call cost.
Connection charges - The fee paid by customers for connection to an operator's network.
Consumer Code of Practice - Code giving details about the services and customer care policies of the operator including details about how to complain and how to contact Oftel.
Customer Service Guarantee - Schemes giving financial compensation to customers for specified failures such as not keeping an appointment or disconnecting customers in error.
Fixed link operator - Operators who provide service through fixed rather than mobile phones. These include BT, Mercury, Kingston Communications (who operate in the Hull area only) and a large number of cable operators who have licenses to provide service in limited geographical areas.
Geographic Averaging - The principle under which BT charges customers the same price for a given service wherever they are in the country, subject to its published tariff.
Hard wired telephone connection - Older style installations which do not allow a customer to remove the phone and "plug-in" a phone of their choice.
Incremental revenue - All the revenue that the operator would forgo if it was to discontinue the activity under consideration. For example, for a residential customer it would include the exchange line rental, the bill for outgoing calls, revenue received by the operator for incoming calls (either retail tariff or interconnect payment) and call-party-pays revenue (eg Freephone calls).
Internet - A global network of networks, mainly narrowband, accessed by users with a computer and a modem via a service provider.
ISDN - Integrated Services Digital Network. A network providing end-to-end digital connection, supporting a wider range of services than available across the PSTN.
Itemised billing - An itemised bill shows the cost of each call, or calls above a certain value, line rental and any other charges separately on the bill.
Light User Scheme - A BT scheme which offers a rebate, based on call bill size, for personal customers with low telephone usage. Call charges are payable at the standard residential rate but a rebate on line rental is available to customers where the call bill falls below £10.80 a quarter.
Long run avoidable costs - The costs that the operator would avoid over the long run if it was to discontinue the activity under consideration. For a single customer it would be the marginal cost of the customer's exchange line and calls (both outgoing and incoming.) For an area, it would be the cost of all the exchange lines and calls to and from the area, including the indirect costs and overheads that would be saved if service was no longer provided to the area.
Long run incremental costs- as Long run avoidable costs above.
Modem - A device used to connect digital consumer equipment to analogue telecommunication networks to enable the transmission of digital data via analogue transmission media.
Narrowband - A service or connection allowing only a limited amount of information to be conveyed, such as for telephony. This compares with broadband which allows a considerable amount of information to be conveyed.
Network externality - The benefit derived by the existing users of a network from new users joining the network because the number of users that can be accessed over the network (and thus the usefulness of the network) has increased.
Number portability - In this document, number portability means the capability that enables a customer to transfer from one operator to a second operator and retain the same number provided the customer remains at the same address.
Operator - A body licenced to, and planning to, provide a network service.
PCN - Personal Communications Network. High capacity digital cellular networks. (Orange and One-2-One are the current UK PCN operators.)
Penetration - The percentage of UK households connected to a telecommunications network.
Premium Rate Services - Services, including recorded information and live conversation, run by independent service providers. All calls to these companies are charged at a higher rate than ordinary calls to cover companies' costs in providing the content of the call and the operator's costs for the special network facilities needed.
Priority Fault Repair Scheme - BT scheme giving chronically sick or disabled customers priority when there is a fault on their line.
Protected Service Scheme - BT scheme which prevents the unnecessary disconnection of vulnerable customers.
PSTN - Public Switched Telephone Network. The telecommunications networks of the major operators, on which calls can be made to all customers of all PSTNs.
PTO - Public Telecommunications Operator (see Operator above).
Reminder Call - BT service which allows the customer to programme the phone to ring at a preset time in order to remind them of an appointment.
Selective Call Barring - The ability to programme a Public Switched Network access line, so that outgoing calls to certain related groups of numbers such as special charge rate, international or mobile cannot be made.
Statutory Advisory Committees - The Director General has six committees representing consumer groups which advise him on telecommunication issues. The national advisory committees for England, Scotland, Wales, Northern Ireland, a committee for disabled and elderly people and one representing small business users.
Statutory Consultation - The minimum period, not being less than 28 days, Oftel is obliged to consult on proposed licence amendments. In practice Oftel consults far more widely than this.
Telecommunication - Conveyance of speech, music and other sounds, visual images or signals by electric, magnetic, electro-magnetic, electro-chemical or electro-mechanical means.
Text Phone - A device used by hearing and speech impaired people to communicate over networks in typed text rather than speech (ie. the device is needed at both ends of the call).
Text Relay - A service enabling textphone users to communicate via the network with other customers, by means of an on-line translation service.
Text Users Rebate Scheme - Users receive a refund on phone call charges because on a like for like basis text calls take longer to make than voice calls.
Three way calling - Facility which allows a conversation to take place between three individuals on three separate telephone lines.
Typetalk - a special unit of the Royal National Institute for Deaf People responsible for delivering the text relay service.
Wideband - An intermediate bandwidth without fuller capacity of broadband (see Broadband above).
