Post- och telestyrelsen Att: Torsten Löfvenholm, smp@pts.se Box 5398 102 49 STOCKHOLM REMISS OM BEHOVET ATT REVIDERA HYBRIDMODELLEN FÖR DET FASTA NÄTET BERÄKNINGEN AV KAPITALKOSTNADEN (07-3652, 23 mars 2007) STOCKHOLM DEN 7 MAJ 2007 TELENOR SVERIGE (TNS); (B2 BREDBAND AB 556575-7738, B2 BREDBAND HOLDING AB 556557-5825, TELENOR AB 556511-9822, OCH ANDRA BERÖRDA TELENORBOLAG INOM TELENOR SVERIGE) 1
1 Introduction With reference to PTS paper of May 14 2007 regarding revision of the cost of capital used in the fixed hybrid model, Telenor Sweden (TNS) will in the capacity as a large operator with its own infrastructure provide comments to the cost of capital used in the model 2 Background Sweden has historically been a forerunner in development of the broadband market, but is now however, starting to lag significantly behind European peers. The current (Q4-2006) broadband penetration of 51% is now around the western European average and, according to Analysys, significantly below comparable countries like Denmark (66%), the Netherlands (62%), Finland (57%) and Norway (57%). Furthermore, the percent of households with PCs that also have broadband, according to Merrill Lynch, is ~10%-points below the European average (61% versus 70%). Furthermore, the Swedish broadband market is characterized by a growing divide between households/businesses situated in areas with infrastructure competition and those located in areas with monopolistic infrastructure supply in terms of penetration, end-user prices and product richness. At an overall level, households/businesses in areas with high population density have access to low prices and rich products while households/businesses in less population dense areas must accept higher price (prices are 40-100% higher) and less rich product offering (e.g., in practice no VoIP offering is available which implies higher prices for fixed telephony). (See appendix 1 for a thorough description of the current state of the Swedish broadband market). As a leading provider of broadband, VoIP and IP-TV, TNS is concerned about the lagging development of the Swedish broadband and VoIP market. TNS remains committed to developing the market, but economic and technical boundaries are seriously limiting our ability to invest in and subsequently grow the Swedish broadband and VoIP market. A key economic boundary is the regulated prices determined by the hybrid LRIC model. In the following we argue that the weighted average cost of capital (WACC) currently used in the hybrid LRIC model is too high and thus should be reduced in the revised LRIC model. This will reduce end user prices, increase penetration and increase product richness in substantial areas of Sweden. In the following sections, we therefore propose a set of changes both to the principles underpinning the WACC calculation and to the actual parameter values of the cost of capital used in the hybrid LRIC model to align the model WACC with the WACC of a Swedish fixed line operator. 3 Suggested changes to the principles for calculating the cost of capital TNS fully supports the main principles used for calculating the LRIC cost of capital based on CAPM. To further align the WACC calculations with leading edge economic theory, TNS has the following 4 comments and clarifications to the 16 principles for calculating the cost of capital in the hybrid LRIC model: 2
Suggested new principle: The calculations should normally be undertaken in conjunction with the more thorough reviews of the LRIC pricing methodology, except when there has been a significant change to any of the parameters, in which case it should be done in the annual update. Regardless of when updates occur, if one or more element(s) of the cost of capital calculations is updated, all other elements must also be updated. This is important to ensure a stable and predictable framework for all operators. As an example, in the most recent updates of the LRIC model (e.g., 4.1), the interest rate applied in the WACC calculations was updated whereas the beta calculations were not. Such inconsistent updates risk the stability and predictability of the model. Therefore, it is important that when a revision of the cost of capital is conducted, it is done for all elements in the calculation. Suggested amendment to principle 5: The computation should use an (optimal) target gearing. TNS strongly supports this principle as the LRIC model should be based on an efficient SMP operator, not based on the actual gearing of TS. However, TNS questions if this principle is actually fully reflected in the way the cost of capital currently is calculated. Currently, a low gearing and high gearing WACC is calculated separately and an average is taken. TNS believes that the optimal target gearing is the one implying the lowest cost of capital. In the latest version (4.1) of the LRIC model, this would imply that the high gearing estimate should be used for the cost of capital, not the average. Comment to principle 6: Calculations should be based on a partially divisionalized approach, where the business of providing fixed network services and mobile services are treated separately. PTS states in Konsultationsrapport kommentarer rörande AMI s rapport om kapitalkostnad för svenska SMP-operatörer 15 October 2003 that empirical benchmarking shows that there is a difference in the cost of capital of mobile and fixed operators. An updated benchmarking of the unlevered beta of listed integrated and pure mobile operators shows that the beta of pure mobile players is 25% higher than for integrated players (see appendix 2). The Swedish fixed SMP (TS) has significant emerging market mobile assets (see appendix 3). Given that PTS decides to pursue a partially divisionalised approach, such assets, together with other mobile assets, can be separated from the fixed assets when the WACC is calculated. In the actual parameter value computations, benchmarking could be conducted not only for mobile players with primarily developed market assets, but also for ones with mainly emerging market assets. Comment to principle 14: Beta should be estimated on the basis of daily observations using a time period of 1-3 years. It is the opinion of TNS that this principle is not concrete enough and it is of utmost importance that there is no ambiguity in the way beta is computed. In the latest versions of the LRIC models, the beta is estimated using a 3-year period. The principle should therefore clearly state that a time period of 3 years should be used for the calculation to be consistent over time. A further argument to specify a 3-year period instead of a range is that it is questionable if a period of 1 year is sufficient to calculate the beta, as short term fluctuations not driven by TS specific fundamentals can have significant influence on the estimated beta if the period is as short as one year. 3
4 Suggested changes to the parameter values used in the calculation of the cost of capital TNS to a large extent agrees with the general principles used by PTS for calculating the cost of capital, however, as the financial markets fluctuate, so will many of the input parameters. All input parameters should therefore be re-calculated based on updated data using the same method as in the last full revision of the model. Major inconsistencies in the calculation methods used in different versions of the hybrid model will limit the predictability for operators which is crucial for strategic decisions, e.g., investments. By applying consistent methodology with updated data and the points stated in chapter 3, TNS proposes the following changes which imply that the cost of capital used should change from the current level of 10.8% to 8.1% (which would be approximately aligned with the current level in the Danish model (8.6%) where a reduction of 2.45% was conducted during the last revision of the model 2005): Risk free rate: TNS supports PTS current method of using the 12-month average of a 10-year Swedish government bond. We believe that recent increases in interest rates should be reflected in the cost of capital used. PTS should calculate the precise updated number (in our computations we have used 4.105%, per appendix 4) Risk premium: As previously stated, TNS believes that a market risk premium of 4.5% is too high. As a point of comparison, the levels used in analyst reports on TS are somewhat lower and around ~4.0% (see appendix 5). Furthermore, the market risk premium used in the Danish model is 3.75%. The equity risk premium in the model should therefore not be higher than 4.0%. Beta: General: Given that PTS continues to use the empirical beta of the TS group as a starting point, which we support, all mobile assets including the international assets (i.e., roughly 2/3 of TS value) can be separated from the overall TS beta to isolate the beta of the fixed line assets. Updated TeliaSonera empirical beta: The calculated empirical beta value will differ according to the time period and interval used. In the current version of the hybrid LRIC model, daily observations for a 3-year period using the MSCI world index as a proxy for the portfolio of a marginal investor is used. In TNS opinion, it is critical that no changes are made to the method used to calculate beta. If the method used changes haphazardly between model revisions, the cost of capital used is neither fair nor predictable for the operators in the market. As shown in appendix 6, the unlevered beta for TeliaSonera as a whole is calculated to 0.85 using the current methodology. Separating fixed assets: As stated above and in alignment with PTS general principles for calculating the cost of capital, the mobile assets can be separated out from TS overall beta to isolate the fixed line beta. This can be done by performing international benchmarking of the unlevered beta of integrated players and comparing it with the unlevered beta of pure mobile players. As TS has mobile assets in both emerging and developed markets, a mix is used. The benchmarking shows that the beta of pure mobile players is 25% higher than for integrated players (see appendix 2). Weighting this with the share of TS assets in mobile based on investment banks sum of the parts valuation of TS, the implied unlevered beta of the fixed assets 4
of TS is ~0.5. This is approximately aligned with the actual beta of telcos with mainly domestic fixed line assets. The beta of Swisscom (0.27), Telekom Austria (0.31) and KPN (0.59) confirms that a level of ~0.5 is appropriate (see appendix 7). Gearing: As stated in relation to the general principles, TNS supports PTS decision to use the target optimal gearing of a fixed SMP operator. We believe that the current gearing used in the LRIC model is too conservative and should be increased to 30% in the low gearing case and 50% in the high gearing case (however, as stated regarding principle 5, we think only the most optimal value should be used) based on the following key arguments (which is more closely aligned with the Danish LRIC model which uses a gearing of 35-50%): International benchmarking: TNS has investigated the gearing levels of the integrated Western-European operators that are most similar to TS. Like TS, the benchmark companies have domestic fixed line and mobile assets as well as international mobile assets, some of which are in emerging markets. The average gearing of these companies is 0.43 (see appendix 8) Analyst estimates: According to analysts, the current balance sheet of TS is inefficient, i.e., not optimal (e.g., TeliaSonera s balance sheet is inefficient Societe General 18/4-06). Therefore, the current gearing level should not be the basis for a forward-looking model such as the hybrid LRIC model. Furthermore, analysts believe that the gearing of TS will increase going forward (see appendix 8) Actions from TeliaSonera: TeliaSonera is paying ~30 billion SEK in dividends for 2007, which implies that their gearing will increase going forward (see appendix 8) Debt premium: Given that the gearing level in the LRIC model is increased, the current debt premium still looks somewhat too high. In the Danish LRIC model a debt premium of 1.25% is used for a gearing of 42.5%. PTS should therefore revisit the current debt premium to determine if it is in fact too high (see appendix 9) (Note: we have used 1.0 and 1.4 in our calculations) Tax rate: The tax rate should remain at 28%, as is currently used 5
Total effect: Proposal Current Risk free rate: 4.105% 3.66% Equity risk premium: 4.0% 4.5% Unlevered beta: 0.5 1.0 Gearing: <30%, 50%> <20%, 40%> Debt premium: <1.0, 1.4> <1.0, 1.4> Tax rate: 28% 28% Implied cost of capital: 8.1% 10.8% 20 June 2007 We are, of course, available to PTS to further discuss the questions raised herein including providing detailed data used in the computations Best regards Telenor Fixed Products & Operations Fredrik Helgesson Manager Strategy and Regulatory 6