Mobile call termination: A simpler price rule for Mobile Call Termination

Start: 16 November 2010
Status: Statement published
End: 14 December 2010

Wholesale mobile voice call termination (MCT) is the service necessary for a network operator to connect a caller with the intended recipient of a call on a different mobile network.

When fixed and mobile operators offer their customers the ability to call UK mobile numbers, they pay mobile communications providers a wholesale charge to complete those calls. The rates that operators pay are called MCT charges or more commonly 'mobile termination rates' (MTRs).

On 31 March 2011, the rules which currently apply to MCT, and which limit MTRs, will expire. We have conducted a market review to consider what rules, if any, should apply after that time.

This statement sets out the conclusions of this review, including our conclusions on market definition, the existence of significant market power (SMP), the detriments likely to arise from the exercise of that SMP and the remedies which should be imposed. In particular we set new rules which limit the MTRs of the four national mobile communication providers (MCPs), and limit all other designated MCPs to "fair and reasonable" rates 1. In most cases, the outcome will be a single wholesale charge for MCPs on different networks, falling sharply each year 2. This simpler regime will benefit consumers by promoting competition, and make it easier for operators to comply. The new rules will apply from 1 April 2011 and end on 31 March 2015.

In our first consultation (published on 20 May 2009), we sought views on different approaches to regulating MTRs, including potentially radical reforms such as removing all rules on call termination or requiring that MCT be priced at zero (termed 'bill and keep'). We set out six options, and asked for comments on these options, or any other option that stakeholders thought we ought to consider.

At the same time the European Commission (EC) issued a Recommendation that fixed and mobile termination rates be limited to the incremental costs of providing call termination to other communications providers (the 2009 EC Recommendation).

We received 30 responses to our May 2009 consultation, from a range of organisations and individuals. Most industry stakeholders, including all of the national MCPs, major MVNOs, BT and other fixed operators, agreed that regulation of MCT was still required and that some form of charge control was likely to be the most appropriate way to regulate charges over the next four years.

Although we had canvassed six options, in fact, almost all respondents supported one of only two choices for the period covered in this market review (that is, to 2015). The first, "LRIC+", involves setting charges using a similar method to that used in 2007 and previous charge controls (which includes a mark-up for joint and common costs, such as the cost of the spectrum used by the network) 3. The second, "pure LRIC" involves setting charges using the method set out in the EC Recommendation 4. Some respondents, including Vodafone, O2, T-Mobile and Orange supported LRIC+, mostly on the grounds that (in their submission) it allocates costs efficiently, allows full cost recovery and is a well understood and proven approach. Others, including BT and H3G, supported pure LRIC citing its role in providing more incentives for innovation and efficiency, and an overall gain in welfare. 'Terminate the Rate' (a group of operators and representative bodies, including BT and H3G, campaigning for lower rates) also preferred this approach.

In our second consultation (published on 1 April 2010) we explained why, having considered the options, and in light of the responses received we thought that capping MTRs, based on some measure of cost, would lead to better outcomes for consumers than alternative approaches. We proposed that we cap MTRs based on the incremental cost of terminating a call (i.e. pure LRIC) and set maximum charges reaching a level set to pure LRIC over four years.

This proposal was a change from previous MCT charge controls, under which MTRs have been set using LRIC+. The change in the way we assess cost makes a significant difference to the expected flows of funds between interconnecting providers: on the basis of charges set using pure LRIC, MTRs would, by 2015, be less than half of the charges calculated on a LRIC+ basis. We considered that adopting pure LRIC would be more likely to promote efficiency, sustainable competition and would confer the greatest possible benefit on consumers.

We received a significant amount of material in response to our April 2010 consultation. We received responses from the four national MCPs, as well as BT, COLT, Cable & Wireless, Asda, Tesco, Virgin Mobile and Lycamobile. We also received responses from 13 smaller CPs, seven trade and consumer groups, and from 'Terminate The Rate' (over 44,000 emails). We also received letters from 43 MPs, who have lodged an early day motion in support of Terminate The Rate's position 5. We also received feedback from the European Commission.

As well as taking into account the submissions made in response to our consultation, we have obtained information and documents (using our statutory powers) from a wide range of affected fixed and mobile communications providers. We have carefully considered this material, which includes commercially confidential data concerning customer spending patterns and behaviour, and forward-looking business plans and information about network volumes and traffic from all four national MCPs.

In this statement, we set out our decision to adopt a charge control for the four national MCPs based on pure LRIC. In deciding to adopt pure LRIC, we have taken the approach we consider will best:

promote efficiency;
promote sustainable competition in the retail mobile market in the UK; and
confer the greatest possible benefits on end-users of public electronic communication services.

In doing so, we also consider whether this approach is objectively justifiable and proportionate 6. Finally our decision to adopt pure LRIC is consistent with the 2009 EC Recommendation.

We have received a small number of requests from stakeholders for clarification of certain aspects of how the 2011 MCT cost model or the charge control mechanism operate. We believe that our explanations may be of assistance to other stakeholders, and we therefore consider it would be helpful to publish these explanations.

These explanations are being given by way of clarification of the operation of the published model to assist stakeholders' understanding, and do not include a discussion of the substance of our decision.

MEA price trends for some assets and the calibration of the MCT model

In response to the 2011 MCT statement, we have received a question relating to the change in the MEA price trend of some assets and the calibration of the MCT model. We have responded to this question in the following way:

In response to the 2010 consultation Vodafone identified that handset costs had been erroneously included in the GBV calibration (see Paragraph A6.169 of the 2011 Statement). The removal of these costs and the recalibration of the model led to a smaller decline in the MEA prices of some assets. The results of the calibration exercise were checked against data provided by the four national MCPs.

The MCT model is calibrated by looking at both asset counts and accounting data. The calibration of GBV figures is not performed on individual elements. Focusing on a single element would be erroneous due to differences in definitions and cost categorisation across operators. The MCT model is calibrated at a higher level and we are satisfied that the GBV is well calibrated on this basis. In addition, any decrease to the unit cost of an individual element would lead to increasing unit costs of other elements to keep the model within its calibration boundaries.

Glide path and RPI-X calculations

In response to the 2011 MCT statement, we have received a number of questions relating to the calculation of the glide-path and the X-value for the RPI-X calculation. We have provided clarifying text as outlined below.

The 2.984ppm nominal cap in 2011/12, is not exactly equivalent to the 2.664 real 2011/12 charge (shown in 2008/09 prices). The main reason for the difference is due to the mechanics of the charge control. The X-value used in the nominal RPI-X calculation undergoes a geometric conversion (which is not applied to the real values) to avoid a mathematical error from the difference between a cap expressed in additive terms (i.e. RPI+X) and the fact that inflation and the required real reduction combine in a multiplicative way (see page 233 of the 2011 MCT statement).

The nominal TAC in 10/11 of 4.428ppm is the starting point (the 4.430ppm shown is rounded as the cap will strictly be to 2 decimal places as per the compliance TACs following the Ofcom Apr 2009 statement following the CAT judgment). Deflating this value to 2008/09 prices gives 4.18ppm (conversion factor = 1/((1+3.5%)*(1+2.5%)))

The 4.18ppm is then aligned to the target of 0.69ppm over 4 years. From this calculation we have the real reduction required. The real reduction then has the geometric conversion factor applied to it (making sure RPI-X works in a multiplicative way) to give the value of X in RPI-X (that is, 37.4%).

Getting the 2011/12 max charge from the 2010/11 TAC is matter of applying lagged inflation (up to December 2010) to the X factor determined above. Lagged inflation is 4.8% (year to December 2010), so the controlling percentage is 4.8% - 37.4% = 32.6%).

The calculation to get the 2011/12 max charge is: (4.428ppm) x (100% - 32.6%) = 2.984ppm, which is the 2011/12 max charge shown in Table 10.1 of the 2011 MCT statement.

We are now reporting to values to 2 decimal places (dp). When rounded to 2dp the actual April 2010 consultation figure was not 4.3ppm but 4.26ppm. Between the April 2010 consultation and the 2011 MCT statement, we corrected the calculation of the 2010/11 nominal TAC so as to make it more precise and this includes using more up to date actual inflation figures. The conversion factor is 1/((1+3.5%)*(1+2.5%)) which applied to the 2010/11 nominal TAC of 4.430ppm, gives the 4.18ppm shown in Table 10.1.

The TAC calculation is constructed for a specific purpose and is not appropriate for more general inflation/deflation calculations. We do not use the inflation series used in the TAC calculation when discounting the 2010/11 nominal price to 08/09 real values. When discounting to 08/09 values we are interested in the charge that is in place in March 2011. We have discounted this value back to 08/09 prices (note the model is in end of FY 08/09 prices). This means discounting for two years of inflation (09/10 and 10/11) using the March inflation values. To be consistent with the MCT model we used RPI inflation (excluding housing and indirect taxes). We do not yet have the 2010/11 inflation figure (i.e. to March 2011) and so we have used a forecast value of 2.5%. Hence we used the inflation values 2.5% and 3.5%.

Treatment of HSPA data efficiency

In response to the 2011 MCT statement, we have received a question on whether the MCT model reflects the higher data delivery efficiency of HSPA.

In response we note that the higher data delivery efficiency of HSPA is taken into account in the MCT model in dimensioning the network and in calculating the cost allocation units used in cost recovery. The MCT model includes a gradual roll-out of HSPA releases (please see para A6.149 of the 2011 MCT statement) to reflect the higher efficiency of HSPA relative to 3G and this is taken into account in mapping HSPA traffic onto the 3G/HSPA shared carrier.

Further clarification on treatment of HSPA efficiency

In response to the 2011 MCT statement, we have received a further question on the treatment of HSPA efficiency in the MCT model.

The question was on the implementation of the higher HSPA efficiency in the model and noted that the 'Params.HSPA.Site.DataDownlift' factor in the model does not flow through to the 'RF_by_element' table. This is correct. However, we note that the model includes a gradual roll-out of HSPA releases, as noted in paragraph 6.149 of the 2011 MCT statement and as reflected in the model parameter 'hspa.efficiency.multiplier'. Please see also footnote 77 in Annex 6 to the 2011 MCT Statement.

It was also suggested that increasing the 'Params.HSPA.Site.DataDownlift' factor results in an increase in the service cost parameter '3G Packet Data (HSPA)'. We note that this effect is not inconsistent with the way that the higher HSPA efficiency would manifest itself in decreased total network costs.

HSPA efficiency and the cost of data services

In response to a question from a stakeholder seeking clarification in respect of the impact of HSPA efficiency on the costs of data services within the MCT cost model, we have provided the following clarification.

The MCT cost model is designed to forecast the unit costs of mobile voice call termination in the context of a network that is also used to provide data and messaging services. The model has not been designed to estimate the efficient split of costs within data services. The model calculation steps of interest are as follows:

i. The model starts from forecasts for various services in the units of relevance to those services (e.g. voice calls are projected in minutes, data services in megabytes (MB) and so on). This traffic is then aggregated by defining a common unit for traffic - specifically busy hour (BH) megabits per second (Mbps).
ii. Within the model, this traffic then feeds into the dimensioning of network elements such as cell sites, 2G equipment, 3G equipment, backhaul, and so on. The network dimensioning takes into account "service routing factors" which map the intensity of network element usage by each service (e.g. how much a 2G incoming voice call uses a 2G MSC and so on). The higher efficiency of HSPA is taken into account when mapping 3G (Release 99) and 3G (HSPA) traffic to shared 3G carriers.
iii. Given the investment cost of each network element and the associated opex, the MCT cost model then determines the present value of network costs to be recovered.
iv. Via the economic depreciation (ED) algorithm the MCT cost model allocates the costs to be recovered per "element output" - i.e. per Mbps for each element. Via the table of routing factors, these costs per element are then aggregated (along with the number of elements required) to calculate the cost per service (e.g. per 2G incoming voice call, per 3G incoming voice call, and so on).

The additional efficiency arising from the use of HSPA technology to deliver 3G data services does not flow into the routing factor table. Therefore, the service cost outputs '3G Packet Data (Release 99)' and '3G Packet Data (HSPA)' (which appear in the 'Service costing' Economic module) have the same value. Instead, the greater efficiency from HSPA is captured by a specific model parameter "hspa.efficiency.multiplier". This in turn shapes the number of dimensioned network elements and the element outputs for cost recovery via the network design algorithms (i.e. step (ii) described above).

Thus in general (for network elements used to provide data services) the greater the HSPA efficiency, the lower the element output (in Mbps) and the fewer the number of network elements, other things equal (e.g. service traffic unchanged from the base case). The fact that network element output reduces should not be interpreted as a reduction in service demand - as noted service demand is assumed to remain constant. Instead the reduction in network element output captures the concept that greater HSPA efficiency enables existing demand to be met with less load or intensity on network elements.

Given the above, once the network algorithms have deployed the necessary network elements and given the necessary capex and opex associated with this, the PV of network costs is determined (i.e. step (iii) described above). From this it can be seen that increased HSPA efficiency will affect the total network costs to be recovered.

In order to achieve cost recovery, step (iv) ensures that total network costs are attributed to services (e.g. 2G incoming voice calls, 3G incoming voice calls, and so on). This is done via the table of routing factors applied to the ED outputs for each network element (coupled with the number of elements used). Since the effect of HSPA efficiency does not feed into the table of routing factors, it follows that the value for "3G Packet Data (HSPA)" (in the "Service costing" worksheet of the workbook "4-Economic) should not be used as a metric to infer the effect of greater HSPA efficiency relative to data services provided via Release 99 (i.e. the value for the variable "3G packet data (Release 99)" in the same worksheet). "3G Packet Data (HSPA)" has the same value as "3G Packet Data (Release 99)" because the higher efficiency of HSPA is not reflected via the table of routing factors.

The "Service costing" worksheet gives the charge in /megabyte for "3G packet data (Release 99)" and "3G packet data (HSPA)". The values of these outputs depend on (a) the present value of total costs for each network element providing that service and (b) the aggregate network element output (specifically the present value of total network element output). Increased HSPA efficiency will reduce each of (a) and (b) above, but if network element output reduces proportionally more than the network element costs, the element unit cost and hence the service unit costs for 3G packet data will go up. The absence of the HSPA efficiency parameter in the routing factors means that it is not appropriate to use the outputs in the "Service costing" worksheet to ascertain the efficiency from HSPA.

HSPA efficiency is taken account of in the calculation of element output unit costs and the voice service routing factors are independent of the HSPA efficiency adjustment.

Therefore, it is inappropriate to use the service costing outputs for "3G packet data (Release 99)" and "3G packet data (HSPA)" to assess the effect of HSPA efficiency on the relative costs of data services provided using Release 99 or HSPA technologies. It is also inappropriate to use these model outputs to ascertain the efficiency of HSPA per se. The impact of increased HSPA efficiency can be verified by looking at the reduction in the present value of total network costs.