Estimate of BT s Equity Beta

Transcription

1 Estimate of BT s Equity Beta PREPARED FOR Office of Communications ( Ofcom ) PREPARED BY Richard Caldwell Ilinca Popescu 3 March 2014

2 This report was prepared for the Office of Communications ( Ofcom ). All results and any errors are the responsibility of the authors and do not represent the opinion of The Brattle Group, Inc. or its clients. Acknowledgement: We acknowledge the valuable contributions of many individuals to this report and to the underlying analysis, including members of The Brattle Group for peer review. Copyright 2014 The Brattle Group, Inc.

3 Table of Contents I. Introduction... 1 II. Equity and Asset Beta Estimates... 4 II.A. UK Utility Reference Sample... 4 II.A.1. Up-to-date Equity Beta Estimates... 4 II.A.2. Financial Leverage II.B. UK Telecoms Reference Sample II.C. European Telecoms Reference Sample II.D. Conclusions III. Statistical Reliability III.A. Dimson adjustment III.B. Tests for heteroscedasticity and auto-correlation III.B.1. Heteroscedasticity III.B.2. Auto-correlation III.B.3. Robust regression and Generalised Least Squares III.C. Normality of residuals III.D. Outliers i brattle.com

4 I. Introduction Ofcom has asked us to update our estimate of the equity beta for BT. 1 We understand that Ofcom intends to use the estimate to inform its decisions on the level of access charges for BT s local loop and other wholesale line rental services. We perform various analyses and present equity and asset beta estimates for BT. We also examine equity and asset betas for three reference samples. Our principal reference sample comprises six other publicly traded UK utilities: National Grid (the gas and electricity transmission system operator), a further three water utilities (United Utilities, Severn Trent, and Pennon Group), and two other energy utilities, Centrica and SSE which combine regulated and unregulated activities. 2 All of the companies in the UK utility peer group provide essential services and are subject to regulated price-caps, at least to some extent. A utility peer group subject to UK price regulation represents a natural peer group against which to compare the results of our equity and asset beta calculations for BT and against which to assess the relative riskiness of BT s regulated activities. Indeed, the UK utility peer group was cited in a decision by the Competition Appeals Tribunal, when it assessed the equity and asset beta for BT s local loop and regulated wholesale services. 3 A second reference sample comprises four liquidly traded UK telecommunications stocks. Three of the four companies in our UK telecommunications sample (Talk Talk, Virgin Media and BSkyB) provide various services dependent on the local loop such as fixed line, broadband and pay-tv to customers. The fourth company in the UK telecommunications sample (Colt) provides integrated communications, IT, and network services and solutions to businesses through its own network. The UK telecommunications sample is interesting in part because it largely reflects risks associated with the provision of different telecommunications services dependent on the local loop or wholesale line rental. 4 1 We last provided an update of BT s equity beta in April See Estimate of BT s Equity Beta (April 2013). 2 Our July 2011 update included Northumbrian Water Group in the UK utilities peer group, but it has since been purchased by Cheung Kong Infrastructure Holdings. 3 Competition Commission Determination, The Carphone Warehouse Group plc v Office of Communications, Case 1111/3/3/09, 31 August 2010, p We include Virgin Media in the sample, but we note that Virgin Media was purchased by Liberty Global in brattle.com

5 A third reference sample comprises seven liquidly traded European telecommunications stocks. Six of the seven European companies are incumbent telecommunications providers owning the local loop (Belgacom, KPN, France Telecom, Deutsche Telecom, Telecom Italia, Telefonica). These six companies are fully integrated, combining both regulated and unregulated activities, wireline and wireless, and voice and data. Several of them have significant overseas interests. The seventh company in our sample, Iliad Free, is an entrant combining both a reasonable history of stock trading and a predominant focus on broadband services using the local loop. The European reference sample permits us to benchmark BT against other leading European telecommunications companies. Several important caveats apply when interpreting results: 1. None of the companies examined provide regulated access to the local loop alone, and as a result due consideration is required before direct application of any of our beta estimates to BT s local loop activities. Even the observed beta for BT s stock price may not apply directly to its local loop activities. As a corporation, BT is involved in numerous activities other than the provision of local loop access. For example, retail telecommunications services accounted for 37% of BT s 2013 revenues and 31% of 2013 EBITDA. 5 Likewise, none of the UK or European telecommunications companies provide only regulated access to the local loop, but engage in a variety of retail activities such as the sale of broadband access. For their part, similarity in regulatory regimes makes the UK utilities interesting. Yet despite similarity in regulatory regime, the risk associated with local telecommunications services may differ from those related to the provision of energy or water. For example, developing wireless technology and the introduction of fibre may supplant the need for the existing copper services. Although facing different regulatory regimes, the European peer group at least reflects similar risks due to the interaction with wireless and fibre. Without further analysis, it remains unclear the extent to which the observed betas for BT and the reference samples reflect the particular risks associated with local loop access in the UK. 2. While we examine the statistical robustness of the observed betas and make the necessary adjustments, we do not assess whether the immediate past could be a reliable guide to the future period of interest to Ofcom. 5 See BT Group plc Annual Report, p brattle.com

6 In this report, we adopt the same methodology as in other previous engagements for Ofcom. 6 We calculate daily returns from holding stock in BT and each of the other companies considered, and from holding a broad market index. We examine data for two market indices: the FTSE All-Share reflecting all stocks trading on the London Stock Exchange and the FTSE All-World reflecting a large proportion of publicly traded stocks around the world. As is standard, we perform a regression of the daily returns on each company against the daily returns on the market index. The regression coefficient is the equity beta. We use market data up to and including December 31, Previous work for Ofcom examined beta estimation methods. 7 One issue concerned the frequency with which to measure stock returns: whether to use daily, weekly or even monthly returns. Analysts might use weekly or monthly returns if there is a concern about the liquidity of stock trading. No such concern exists in this case. The major telecoms stocks and utilities under examination are amongst the most liquid stocks around. All of our estimates therefore focus on daily returns. Another methodological choice relates to the duration of the data window. We focus on a two-year window in this report, while also reporting the results from a one-year window. Two-years provides a sizeable sample of daily stock returns without extending so far back in time as to include data from periods before any of the companies made operational changes. Chapter II presents beta estimates for BT, the UK utility reference sample and the UK and European telecoms samples. Chapter III reports the results of several tests of the statistical reliability of the beta estimates. 6 See, for example, Updated Estimate of BT s Equity Beta (October 2008), An Estimate of the Equity Beta of BskyB (March 2009), and Estimate of Equity Beta for UK Mobile Owners (December 2009). 7 See Issues in beta estimation for UK mobile operators, July brattle.com

7 II. Equity and Asset Beta Estimates II.A. UK UTILITY REFERENCE SAMPLE II.A.1. Up-to-date Equity Beta Estimates Table 1 reports up-to-date equity beta estimates for BT and the UK utility reference sample. All of the estimates rely on daily return data. We report separate one and two-year beta estimates as well as separate estimates against the two market indices. A one-year beta is estimated using a regression that relies on the previous year of trading activity. A two-year beta relies on the previous two years. All of the various estimates reflect data up to and including December 31, brattle.com

8 Table 1: Up to date equity beta estimates 8 1 Yr 2 Yr Beta SE Low High Beta SE Low High BT All World All Share UK Utility Peer Group National Grid All World All Share Pennon Group All World All Share Severn Trent All World All Share United Utilities All World All Share Centrica All World All Share SSE All World All Share Utility Peer Group Average All World All Share Note: We report OLS betas except where diagnostic tests indicate the presence of either heteroskedascity or auto correlation. In which case we report robust or GLS betas. We identified autocorrelation or heteroskedasacity for the one and two year equity betas of BT against the FTSE All Share, National Grid, Pennon Group, Severn Trent and United Utilities against both the FTSE All Share and FTSE All World, see Table 9 and Table 11. The most recent data indicate little change in the level of BT s equity beta since our last update (April 2013 and December 2012). Against the FTSE All-Share, we estimate a slightly higher up-to-date one-year equity beta of 1.08, compared with our estimates of 1.03 as of April 2013 and 0.99 as of December We estimate an up-to-date two year equity 8 Low and high refer to the 95% confidence interval and not to the lowest and highest one and twoyear betas observed throughout the year. 5 brattle.com

9 beta of 1.01, compared with our estimates of 1.03 as of April 2013 and 1.01 as of December The changes in the level of the raw equity betas remain well within the range of statistical error. BT equity betas against the FTSE All-World have also seen little change. Figure 1 illustrates the development of BT s equity beta against the FTSE All-Share over time. The plot keeps the duration of the beta estimation window constant through time. It simply shifts the one or two-year data window forward as time passes. It illustrates the relative stability of both the one-year and two-year BT equity betas over the past several years, despite general market volatility at the end of 2008 and the first part of There has been a relative convergence between the one-year and two-year betas since 2009, with both trending slightly upwards. However, the convergence appears to have broken somewhat over the past six months, with a relative increase in the one year beta estimates. Since 2010, the one-year beta has ranged between a low of 0.78 and a high of 1.28, while the two year estimate declined to a low of 0.84 in the first half of 2010 before climbing to 1.08 by mid Figure 1: BT rolling equity betas Yr 2 Yr Table 1 confirms that BT s equity beta comes in higher than the UK utility reference sample. Both BT s one- and two-year equity betas against the FTSE All-Share are now close to double the average of the six UK utilities (1.08 vs 0.61 for the one-year, 1.01 vs 0.52 for the 6 brattle.com

10 two-year). Over the past six months, almost all the companies in the UK utility sample have seen a slight increase in their one-year beta estimates. It is not clear whether this is a temporary phenomenon or whether the recent rise will persist. The two-year estimates have remained broadly stable over the last six months despite the upward movement in the oneyear betas. Figure 2 to Figure 7 plot one and two-year rolling equity betas for the six UK utilities against the FTSE All-Share. National Grid, Pennon Group, Severn Trent, United Utilities and SSE see movement in their one-year equity betas during the last few months of Then in 2009, the one-year estimates fall off dramatically towards the end of the year. The timing may reflect movement of the end of the data window past autumn 2008 and the climax of the credit crisis. The step declines for National Grid, Severn Trent and United Utilities are roughly two standard deviations. In 2010 and the first half of 2011, the one-year equity betas of National Grid, Severn Trent, United Utilities, and Pennon Group trend upwards towards previous levels, before falling back again to the levels seen at the beginning of All four companies have seen their one year equity betas rise back again to early 2011 levels over the past six months. Centrica s and SSE s one-year equity betas have remained more stable. Centrica s witnessing a dip during 2010 and SSE s tracking the rise of some of the other companies during the last six months. Interestingly, the two-year equity beta estimates for the UK utility peer group all display a step rise at the end of 2008 immediately following the collapse of Lehman Brothers, and a corresponding step decline at the end of 2010 as data from autumn 2008 drops out of the data window. Since then, the two-year estimates for the UK utility peer group have remained remarkably stable. Since 2010, National Grid s two-year estimate has varied between a low of 0.38 and a high of 0.53; Pennon Group s between 0.42 and 0.55; Severn Trent s between 0.37 and 0.52; United Utilities between 0.41 and 0.50, Centrica s between 0.49 and 0.69, and SSE s between 0.37 and In Chapter III, we identify which particular data points exert the greatest influence on the one and two- year equity beta estimates and investigate the impact of those particular points on the estimates. We find that the standard OLS betas for BT and the utility reference sample are broadly robust to the exclusion or underweighting of influential data points. 7 brattle.com

11 1.5 Figure 2: National Grid rolling equity betas Yr 2 Yr brattle.com

12 1.5 Figure 3: Pennon Group rolling equity betas 1.3 1Yr 2Yr brattle.com

13 1.5 Figure 4: Severn Trent rolling equity betas Yr 2 Yr Figure 5: United Utilities rolling equity betas Yr 2 Yr brattle.com

14 Figure 6: Centrica rolling equity betas Yr 2 Yr Figure 7: SSE rolling equity betas Yr 2 Yr brattle.com

15 II.A.2. Financial Leverage Equity risk reflects the combination of underlying business risk (principally to do with the cyclicality of revenues and the extent of fixed costs) and financial risk (to do with the presence of fixed debt obligations). Other things equal, the more debt a company has outstanding, the greater the equity risk and the higher the equity beta. In general, extreme changes in financial leverage throughout the measurement window prompt the need for further analyses and checks. We obtained data on the amount of debt outstanding for BT and each of the six publicly traded UK utilities between 2004 and the present. 9 We obtained data from Bloomberg primarily, and filled in any remaining gaps with data from company annual and half-yearly reports and quarterly earnings announcements. We use the data to estimate the companies capital structures at various points in time between 2004 and the present. We focus on market values of equity and debt rather than book values, since market values better indicate earnings power. That being said, we follow the approach adopted in previous reports and assume that the market value of utility debt remained relatively close to its face value throughout the period in question. This assumption appears reasonable given that BT as well as the six UK utilities all maintained investment grade credit ratings throughout the measurement period At the time of writing, none of the UK utility companies had released financial statements for December September 2013 therefore represents our last observation of financial leverage. 10 A possible concern is whether the market price of BT s and the other UK utilities debt could have diverged significantly from face value, most likely during the height of the credit crisis. If a significant market-to-book difference emerged, then a failure to use market values could bias, probably upward, our estimates of the companies financial leverage. For example, as credit spreads spiked during the credit crisis, the price on BT and other UK utility debt may have declined somewhat, reflecting investor concerns about the prospects for the UK and world economy. Incorporating the reduced market price of the debt in the calculation would reduce the appearance of financial leverage. Overstating leverage could lead us to effectively understate BT s overall asset beta, since we would always expect leverage to add to the equity beta. We check the potential impact of the financial crisis on financial leverage by estimating the market price of BT s debt. Much of BT s long-term debt is publicly traded. We obtained available data concerning debt prices and yields. The available data indicates that the market price of BT s debt declined somewhat at the end of 2008 during the height of the crisis, but not to such an extent as to seriously affect our estimates of financial leverage. Adjusting the amount of debt by less than 10% either way could have only a 2.5% impact on BT s apparent leverage ratio, and affect the asset beta by as little as 3%. Suppose we perceived a gearing ratio of 40% on the assumption that the book value of debt were a good proxy for market value. Now, suppose that the market value were 10% less than book value. The true gearing ratio therefore would be 37.5% (36 / (36+60)). Unlevering BT s latest two-year equity beta of 1.03 assuming gearing of 40% and a debt beta of 0.15, we would derive an asset beta Continued on next page 12 brattle.com

16 We compute financial leverage in the same way as in our previous updates, with reference to the face value of outstanding debt 11 and ignoring BT s pension fund deficit. The use of the face value of outstanding debt finds support in a leading corporate finance textbook: first compute working capital (current assets less current liabilities) for each company. If working capital is positive, analysts should zero out short-term debt and estimate financial leverage with reference to long-term debt only. But if working capital is negative, analysts should estimate financial leverage with reference to the sum of long-term plus shortterm debt. Since BT s current liabilities consistently exceed its current assets (including cash), we end-up using the face value of both long-term and short-term debt in the leverage computation. 12 We ignore BT s pension fund deficit in part because the potential impact on the beta is not at all obvious. 13 Figure 8 plots our resulting estimates of financial leverage for the UK utility reference sample. Continued from previous page of Assuming gearing of 37.5%, we would derive an asset beta of 0.70, or only 3% higher than that derived using the book value of debt. 11 As opposed to net debt, which equals the face value of short and long term financial debt less cash. 12 See Brealey, Richard A, Myers, Stewart C, and Allen, Franklin, Principles of Corporate Finance, Ninth Edition, McGraw Hill, (2006), p See Cooper, Ian, The effect of defined benefit pension plans on measurement of the cost of capital for UK regulated companies: A report for Ofcom, (2 September 2009). 13 brattle.com

17 Figure 8: UK utility financial leverage 70% 60% 50% 40% Leverage 30% 20% 10% 0% Sep 04 Aug 05 Jul 06 Jun 07 Apr 08 Mar 09 Feb 10 Jan 11 Nov 11 Oct 12 Sep 13 BT United Utilities Severn Trent Pennon Group National Grid Centrica SSE The black line in the figure indicates that BT witnessed a substantial rise in leverage around BT actually maintained a relatively stable stock of debt over the period, but its share price dropped dramatically during The same level of debt combined with less equity, so that leverage doubled from 30% to just over 60% by the end of The share price has since rebounded, and has prompted a substantial decline in BT s financial leverage. Of the UK utility peer group, only United Utilities witnessed a similar swing in leverage during the recent measurement period. Following the recovery of BT s share price, BT s leverage has returned to its pre credit crisis level and a level somewhat below the rest of the UK utility peer group apart from the two integrated energy utilities, Centrica and SSE. For their part, Centrica and SSE have maintained a relatively lower level of debt than the rest of the UK utility sample throughout. A further table and three figures explore the effect of financial leverage across BT and the UK utility reference sample. Table 2 reports asset beta estimates for BT and the utility sample, illustrating the betas that we would expect if all of the companies maintained only equity financing. We use two separate approaches to un-lever the raw equity beta estimates. The first approach uses the simplest possible un-levering formula and assumes that the debt 14 brattle.com

18 beta is zero. 14 The second approach follows the same approach but is more realistic in that it recognises some correlation between the returns to debt-holders and the broader economy. It assumes a debt beta of Under both approaches, we estimate average leverage across the relevant measurement window for beta. In other words, when focusing on one-year betas, we estimate average leverage across the one-year measurement window. When focusing on twoyear betas, we estimate average leverage across the two-year measurement window. Figure 9 and Figure 10 then plot rolling one and two year asset betas for BT and the utility reference sample. They illustrate a widening of the gap between BT and the other UK utilities for the period after pre-credit crisis data has dropped out of the data window. Asset betas for Centrica and SSE remain below BT, but have been at a premium to the four regulated network companies. Figure 11 compares one- and two-year asset betas for BT. BT s asset beta now exceeds the peak witnessed prior to the collapse of Lehman in We use a standard relevering formula (see Principles of Corporate Finance (8 th edition), Brealey E D Myers and Allen, p. 518): a e d, where βa, βe, and βd represent D E D E asset beta, equity beta and debt beta respectively, and D and E represent the market values of outstanding debt and equity. 15 brattle.com

19 Table 2: UK utility asset betas 1 Year 2 Year β debt = 0 β debt = 0.1 β debt = 0 β debt = 0.1 BT All World All Share UK Utility Peer Group National Grid All World All Share Pennon Group All World All Share Severn Trent All World All Share United Utilities All World All Share Centrica All World All Share SSE All World All Share UK Utility Peer Group Average All World All Share brattle.com

20 Figure 9: One year asset betas FTSE All Share 1.0 BT National Grid Pennon Group Severn Trent United Utilities Centrica SSE Jan 07 Jan 08 Jan 09 Jan 10 Jan 11 Jan 12 Jan 13 Jan 14 Figure 10: Two year asset betas FTSE All Share BT National Grid Pennon Group United Utilities Centrica SSE Severn Trent Jan 08 Jan 09 Jan 10 Jan 11 Jan 12 Dec 12 Dec brattle.com

21 Figure 11: One and two year asset betas BT vs FTSE All Share Yr 2Yr Dec 06 Dec 07 Dec 08 Dec 09 Dec 10 Dec 11 Dec 12 Dec brattle.com

22 II.B. UK TELECOMS REFERENCE SAMPLE In addition to the UK utility reference sample, we examined data for four UK telecoms companies: Talk Talk Group, Virgin Media, BSkyB, and Colt. Talk Talk and Virgin Media now offer quadruple play services to UK consumers, providing fixed line, broadband, mobile and pay-tv. A sizeable proportion of Talk Talk s and Virgin Media s customer base is unbundled at the local loop. BSkyB competes with Talk Talk and Virgin Media, providing fixed line, broadband and TV services to customers, but not mobile. It services many customers through the local loop too. Colt is somewhat distinct from the other three UK telecoms companies, providing telecom and data services to businesses based in many large European cities. Colt s primary operations are in the City of London. Although none of the four telecom companies are directly comparable to BT s local loop and wholesale line rental businesses, they represent a second natural reference sample. Talk Talk, Virgin Media and BSkyB are major users of local loop and wholesale line infrastructure. At the same time, their financial performance inherently reflects the risks and rewards associated with the provision of downstream telecom services to UK consumers and businesses. Using standard techniques, we estimated equity and asset betas for each of the four UK telecoms companies in Table 3 and 4, respectively. 15 Table 4 indicates that BT s up-to-date one- and two- year asset betas against the All-Share exceed those for all four of the reference sample companies and comes in at around 0.2 higher than the average of the UK telecom reference sample. Figure 12 to Figure 13 then plot the development of the one-year and twoyear asset betas against the All-Share for the UK telecoms peer group over time. BT, Talk Talk, and Colt have witnessed an increase in their one-year asset betas over the past year, while BSkyB s one-year asset beta has remained stable. We compute betas for Virgin Media only up to February 2013, when the merger with Liberty Global was announced. Up until then, the one- and two- year asset betas for Virgin Media tracked those for Talk Talk and BSkyB. The roughly 0.2 gap between BT s two year asset beta and the reference sample has remained relatively consistent over the past two years. 15 Like for the UK utility peer group, we measured one-and two-year equity betas for the UK telecommunications sample against the FTSE All-Share and All-World indices. We computed daily returns for each company and the indices and applied standard ordinary least squares without adjustment. We then re-levered the observed OLS betas based on each company s average financial leverage across the one or two year measurement window. We computed financial leverage based on the market value of equity and the book value of outstanding debt. 19 brattle.com

23 Table 3: UK telecoms equity betas 1 Yr 2 Yr Beta SE Low High Beta SE Low High BT All World All Share UK Telecom Peer Group Talk Talk All World All Share B Sky B All World All Share Colt Group All World All Share Virgin Media* All World All Share UK Telecom Peer Group Average All World All Share Notes: * Results as of the date the merger was made public on February 6th We report OLS betas except where diagnostic tests indicate the presence of either heteroskedascity or auto correlation. In which case we report robust or GLS betas. We identified autocorrelation for the one year equity betas of Talk Talk and BSkyB against both the FTSE All Share and FTSE All World, and for the two year equity betas of BSkyB and Colt Group against the FTSE All Share and FTSE All World, and for Virgin Media against the FTSE All Share. See Table 10 and Table brattle.com

24 Table 4: UK telecoms asset betas 1 Year 2 Year β debt = 0 β debt = 0.1 β debt = 0 β debt = 0.1 BT All World All Share UK Telecom Peer Group Talk Talk All World All Share B Sky B All World All Share Colt Group All World All Share Virgin Media All World All Share UK Telecom Peer Group Average All World All Share brattle.com

25 Figure 12: One year asset betas of UK Telecom sample vs FTSE All Share Talk Talk Colt BT Sky Virgin Media 0.0 Jan 08 Jan 09 Jan 10 Jan 11 Jan 12 Jan 13 Jan 14 Figure 13: Two years asset betas of UK Telecom sample vs FTSE All Share Talk Talk Colt BT Sky Virgin Media 0.0 Jan 08 Dec 08 Dec 09 Dec 10 Dec 11 Dec 12 Dec brattle.com

26 II.C. EUROPEAN TELECOMS REFERENCE SAMPLE We also examined data for seven European telecommunications companies: Telefonica, France Telecom, Deutsche Telecom, Telecom Italia, Belgacom, KPN and Iliad Free. Similar to BT, six of the seven companies enjoy leading positions in domestic markets, owning the local loop (Telefonica, France Telecom, Deutsche Telecom, Telecom Italia, Belgacom, and KPN). These companies are all fully integrated, combining both wireline and wireless activities, and data and voice services. Some of the six companies such as Telefonica earn the majority of revenues and income from international activities. 16 Others such as Belgacom focus primarily on domestic activities. 17 The seventh company, Iliad Free, is an entrant in the French telecoms market, obtaining the vast majority of revenues and EBITDA from fixed line broadband and telephony services. Using standard techniques, we estimated asset betas for each of the seven European telecoms companies. 18 We use the six integrated telecoms companies as a benchmark for BT, indicating if investors perceived BT to be of comparable risk overall to the other leading European telecoms companies. We also compare our BT results with Iliad Free, in part because Iliad Free represents the only European telecommunications company combining both a history of stock trading and a predominant focus on wireline services only. 16 In 2012, only 24% of Telefonica s revenues arose in Spain. Almost 50% were generated in Latin America. Telefonica earned only 32% of EBITDA in Spain. Telefonica Form 20-F for 2012, p In 2012, over three quarters of Belgacom s revenues in 2011 and 2012 stemmed from services to domestic customers. 18 We measured one-and two-year equity betas for the European sample against the FTSE All Europe. We computed daily returns for each company and the FTSE All Europe. We applied standard ordinary least squares, and made no adjustments. We then re-levered the observed OLS betas based on each company s average financial leverage across the one or two year measurement window. We computed financial leverage based on the market value of equity and the book value of outstanding debt. 23 brattle.com

27 Table 5: EU telecoms equity betas 1 Yr 2 Yr Beta SE Low High Beta SE Low High BT All World All share EU Telecom Peer Group vs FTSE All Europe Belgacom Deutsche Telekom Orange Iliad KPN Telefonica Telecom Italia EU Telecom Group Average Notes: We report OLS betas except where diagnostic tests indicate the presence of either heteroskedascity or autocorrelation. In which case we report robust or GLS betas. We identified autocorrelation for all the beta estimates of the EU telecom peer group. Table 6: EU telecoms asset betas 1 Year 2 Year β debt = 0 β debt = 0.1 β debt = 0 β debt = 0.1 BT All World All Share EU Telecom Peer Group Vs FTSE All Europe Belgacom Deutsche Telekom Orange Iliad KPN Telefonica Telecom Italia EU Peer Group Average Figure 14 and Figure 15 plot the development of the one and two year asset betas over time for the six integrated telecoms companies against the FTSE All-Europe. BT s two-year asset beta against the FTSE All-Share is roughly 0.1 above that of Telefonica, and at present stands close to The other companies Deutsche Telecom, Orange, KPN, Belgacom and Telecom Italia - display lower two-year asset betas that range around 0.3 to 0.5 and are 24 brattle.com

28 therefore comparable to the average for the UK telecom reference sample, and just above the average for the UK utility reference sample. The one-year betas display the same essential pattern, with the one-year asset beta for BT averaging 0.6 or above over the past two years, while the one-year asset betas for Orange, Deutsche Telekon, KPN, Belgacom Telefonica and Telecom Italia have remained between 0.3 and 0.6 over the same period. BT s current one and two year asset betas are at the top end of both the UK telecom reference sample and the sample of leading integrated European telecoms providers. 1.0 Figure 14: One year asset betas for European integrated telecoms 0.8 Telefonica Deutsche Telecom Belgacom KPN Orange Telecom Italia BT Jan 08 Jul 08 Dec 08 Jul 09 Dec 09 Jul 10 Dec 10 Jul 11 Dec 11 Jun 12 Dec 12 Jun 13 Dec brattle.com

29 Figure 15: Two year asset betas for European integrated telecoms Telefonica Deutsche Telecom Belgacom KPN Orange Telecom Italia BT Jan 08 Jul 08 Dec 08 Jul 09 Dec 09 Jul 10 Dec 10 Jul 11 Dec 11 Jun 12 Dec 12 Jun 13 Dec 13 Figure 16 compares one and two year asset betas for Iliad Free, Orange and BT. Dashed lines represent one-year asset betas, while solid lines represent two-year asset betas. Asset betas for BT and Orange have moved more-or-less in parallel, with BT remaining 0.1 or so higher than Orange between 2008 and end During 2013 the gap widens as BT s asset beta trends slightly upward while Orange s asset beta remains broadly stable. In contrast, the two year asset beta of Iliad Free dropped sharply in 2008 and 2009, and has since remained close to 0.4. Iliad Free s asset beta remained around 0.4 during 2011 and 2012, but has displayed an upward trend over the past year, particularly the one-year asset beta. Nevertheless, a two-year asset beta below 0.5 leaves Iliad Free comparable to Orange, Deutsche Telecom, Belgacom and Telecom Italia, and below BT and Telefonica. 26 brattle.com

30 1.3 Figure 16: One and two year asset betas for Iliad Free Iliad Orange BT Jan 08 Jul 08 Dec 08 Jul 09 Dec 09 Jul 10 Dec 10 Jul 11 Dec 11 Jun 12 Dec 12 Jun 13 Dec 13 II.D. CONCLUSIONS In our June 2011 update, we remained concerned about the large swing in BT s financial leverage over the past few years and its implications for our beta calculations. We have no such concerns this time round. BT s financial leverage has declined by approximately 10% during the latest two year data window. We note that the latest OLS estimates against the FTSE All-Share display some heteroskedascity and auto-correlation (see chapter III), and therefore focus on the GLS estimates. Based on our GLS regressions, the two-year equity beta of BT is 1.01 against the FTSE All-Share. This estimate corresponds with average leverage during the most recent two-year measurement window of 32%, and implies an asset beta for BT of between 0.69 and 0.72 depending on our assumption with respect to the debt beta. The one year equity beta of BT (obtained using GLS, see chapter III) is 1.08 against the FTSE All-Share, corresponding with average leverage of 27%. This implies an asset beta of 0.80 to 0.82, again dependent on the debt beta assumption. What remains clear is that BT s asset beta has not followed the same general trend as those for other UK utility companies since The other UK utilities evidence a step increase in asset beta with the onset of the credit crisis and then a sharp decrease following 27 brattle.com

31 the elimination of pre-crisis data from the analysis. BT s asset beta displays no such drop following the elimination of pre-crisis data. If we attribute such recent movements in the asset betas to a change in perceptions following the credit crisis, then investors appear to have distinguished between BT vis-a-vis National Grid, the water utilities and even Centrica and SSE. Investors apparently considered the other utilities to have represented relative safe havens following the crisis, but have considered BT more risky perhaps because of the poor performance of some of BT s non-core business or uncertainty over the pension fund deficit. BT s most recent asset beta estimates exceed the range indicated by both our UK and European telecoms reference samples. The average one- and two- year asset betas for the UK telecom reference sample are roughly 0.1 to 0.2 higher than those for the UK utility reference sample. BT s asset betas are most similar to Telefonica s and exceed those for Deutsche Telecom, France Telecom, Belgacom, KPN and Telecom Italia. BT s most recent asset beta also exceeds that for Iliad Free, an entrant predominantly focused on wireline internet services. BT s up-to-date asset betas therefore exceed all three reference samples, with asset betas for the UK utility reference sample lower than for the two telecom samples. We normally recommend a range of +/- approximately two standard deviations around our mid-point figures: the standard error being 0.11 for the last year of data, and 0.07 for the last two-years. 19 This implies a range for the one-year equity beta against the FTSE All-Share of 0.87 to 1.29, translating into a range in asset beta of between 0.64 and 0.98 (incorporating the measurement error on the equity beta and depending on whether the debt beta is assumed at zero or 0.10). The equivalent range for the two-year equity beta against the FTSE All-Share is 0.87 to 1.14 translating into a range in asset beta of between 0.59 and 0.81 (again incorporating the measurement error on the equity beta and depending on whether the debt beta is assumed at zero or 0.10). 19 We report a larger standard error for the two year beta than suggested by standard OLS. The larger standard error reflects our use of the GLS results because of the presence of auto-correlation. 28 brattle.com

32 III. Statistical Reliability The use of daily returns data in regressions to estimate equity beta can risk introducing statistical problems, for example in relation to thin trading. We discussed these problems in earlier papers for Ofcom. 20 We perform a number of statistical tests to check for potential problems in this case. Below we report the results of our statistical tests for BT and the UK utility and UK telecom reference samples. We performed exactly the same tests for the betas computed above for the seven EU telecom companies. We confirm the statistical robustness of the European company betas presented above. III.A. DIMSON ADJUSTMENT To test for possible bias relating to trading illiquidity and to assess if time differences 21 caused distortions, we perform the Dimson adjustment to the estimated betas by including a one period lag and a one period lead. For BT and the six UK utilities, two out of 28 lag terms was significantly different from zero (both the one and two year equity betas for Severn Trent against the FTSE All-Share). No lead term was statistically significant. In no case were the Dimson adjustments overall significantly different from zero. A similar picture emerges for the UK Telecom sample, with the Dimson lead adjustments significant in two cases: Talk Talk s one year beta against the FTSE All-World and its two-year beta against the FTSE All- Share. 22 The lag term was significant for Virgin Media one and two year beta estimates against the FTSE All-World. 20 See Issues in beta estimation for UK mobile operators, July The London Stock Exchange closes at 5pm BST, while the markets in other countries may close earlier or later. Broad index data may therefore combine closing prices relating to different time of day. Timing adjustments therefore may be relevant for betas versus the FTSE All-World. 22 We might expect a false positive once in every twenty beta estimates, given the extent of the standard errors. 29 brattle.com

33 Table 7: UK utility peer group, Dimson adjustments up to date data Beta Dimson Beta 1 Yr Dimson SE Significance Beta Dimson Beta 2 Yr Dimson SE Significance BT All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead UK Utility Peer Group National Grid All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead Pennon Group All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead Severn Trent All World Neither lag nor lead Neither lag nor lead All Share Only lag Only lag United Utilities All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead Centrica All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead SSE All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead Notes: Betas are calculated using standard OLS. 30 brattle.com

34 Table 8: UK telecom peer group, Dimson adjustments up to date data Beta Dimson Beta 1 Yr 2 Yr Dimson Dimson Dimson SE Significance Beta Beta SE Significance BT All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead UK Telecom Peer Group Talk Talk All World Only lead Neither lag nor lead All Share Neither lag nor lead Only lead B Sky B All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead Colt Group All World Neither lag nor lead Neither lag nor lead All Share Neither lag nor lead Neither lag nor lead Virgin Media* All World Only lag Only lag All Share Neither lag nor lead Neither lag nor lead Notes: * Results as of the date the merger was made public on February 6th Betas calculated using standard OLS. III.B. TESTS FOR HETEROSCEDASTICITY AND AUTO-CORRELATION We perform a series of diagnostic tests to assess if the equity beta estimates satisfy the standard conditions underlying ordinary least squares regression. The standard conditions are that the error terms in the regression follow a normal distribution and that they do not suffer from heteroscedasticity (differences in variance within sample) or auto-correlation (follow some pattern over time). Failure to meet these conditions would not invalidate the beta estimates, but would have the following consequences: 1. Although OLS is still an unbiased procedure in the presence of heteroscedasticity and/or autocorrelation, it is no longer the best or least variance estimator. 2. In the presence of heteroscedasticity and/or autocorrelation, the standard error calculated in the normal way may understate the true uncertainty of the beta estimate. 3. Heteroscedasticity and/or auto-correlation may indicate that the underlying regression is mis-specified (i.e. we have left out some explanatory variable). 4. Failure of normality does not per se undermine the validity of OLS, but the presence of outliers raises difficult questions about the robustness of the beta estimates. 31 brattle.com

35 III.B.1. Heteroscedasticity Figure 17 to Figure 27 show scatter plots of the residuals against the returns predicted by the regression, for two-year regressions against the FTSE All-Share. We constructed comparable plots for our regressions against the other indices and for our shorter one year beta estimates. Visual inspection does not reveal any obvious pattern - the vertical spread does not appear to change in any systematic way as we move horizontally across the graph. However, there are clearly a number of outliers. Figure 17: BT residuals against fitted values Residuals Fitted values 32 brattle.com

36 Figure 18: National Grid residuals against fitted values Residuals Fitted values Figure 19: Pennon Group residuals against fitted values Residuals Fitted values 33 brattle.com

37 Figure 20: Severn Trent residuals against fitted values Residuals Fitted values Figure 21: United Utilities residuals against fitted values Residuals Fitted values 34 brattle.com

38 Figure 22: Centrica residuals against fitted values Residuals Fitted values Figure 23: SSE residuals against fitted values Residuals Fitted values 35 brattle.com

39 Figure 24: Talk Talk residuals against fitted values Residuals Fitted values Figure 25: BSkyB residuals against fitted values Residuals Fitted values 36 brattle.com

40 Figure 26: Colt Group residuals against fitted values Residuals Fitted values Figure 27: Virgin Media residuals against fitted values Residuals Fitted values 37 brattle.com

41 We also examine whether there is change in the pattern of residuals over time. Figure 28 to Figure 38 do not show an apparent pattern of the residuals for the two-years estimation window. The plots again relate to two-year beta estimates calculated against the FTSE All- Share. Figure 28: BT residuals over time Residuals jan jul jan jul jan2014 tm_bt 38 brattle.com

42 Figure 29: National Grid residuals over time Residuals jan jul jan jul jan2014 tm_ng Figure 30: Pennon Group residuals over time Residuals jan jul jan jul jan2014 tm_pnn 39 brattle.com

43 Figure 31: Severn Trent residuals over time Residuals jan jul jan jul jan2014 tm_svt Figure 32: United Utilities residuals over time Residuals jan jul jan jul jan2014 tm_uu 40 brattle.com

44 Figure 33: Centrica residuals over time Residuals jan jul jan jul jan2014 tm_cent Figure 34: SSE residuals over time Residuals jan jul jan jul jan2014 tm_sse 41 brattle.com

45 Figure 35: Talk Talk residuals over time Residuals jan jul jan jul jan2014 tm_tt Figure 36: BSkyB residuals over time Residuals jan jul jan jul jan2014 tm_sky 42 brattle.com

46 Figure 37: Colt Group residuals over time Residuals jan jul jan jul jan2014 tm_colt Figure 38: Virgin Media residuals over time Residuals jan jul jan jul jan2013 tm_vm 43 brattle.com

47 Even though simple inspection suggests that heteroscedasticity cannot be a major concern, we apply a formal test (White s test) to investigate further. Table 9 and Table 10 report the results of the standard diagnostic test. It indicates the absence of heteroscedasticity in all of the one- and two-year equity beta estimates apart from five: the one- and two- year estimates for National Grid against the FTSE All-Share, the one-year estimates for United Utilities against both the FTSE All-Share and the FTSE All-World, and the two year estimate for Virgin Media against the FTSE All-World. Table 9: White s test for heteroskedasticity up to date data, UK utilities White Stat 1 yr p value White Stat 2 yr p value Heteroskedascity Heteroskedascity BT All World No No All Share No No UK Utility Peer Group National Grid All World No No All Share Yes Yes Pennon Group All World No No All Share No No Severn Trent All World No No All Share No No United Utilities All World Yes No All Share Yes No Centrica All World No No All Share No No SSE All World No No All Share No No 44 brattle.com