E.1 As in previous price control reviews, Oftel will base its proposal on each price cap and sub-cap on forecasts of BT's financial performance, which will be generated using a financial model. Into this model will be fed assumptions about a number of key determinants of BT's profitability over the forecast period, and the modelling process will produce a range of price control formulae that should lead to BT earning no more than an acceptable rate of return on each basket or sub-cap by the end of the price control period (subject to views expressed in consultation on this document on the desirability of a one-off adjustment to prices at the start of the period).
E.2 Since the aim of regulation is to mimic the workings of a competitive market, an acceptable rate of return for BT's price-controlled activities would be equal to its cost of capital on those activities. The next section deals with the derivation of the cost of capital for BT, and presents a range of estimates upon which comments are sought. The subsequent sections deal with the appropriate measure of rate of return that the estimated cost of capital should be applied to within the financial model. In the final section, the basic structure of Oftel's proposed model is discussed in some detail.
E.3 A firm's cost of capital can be defined as the rate ofreturn that could be earned in the capital market on securities of equivalent risk. In general, the higher the riskiness of the firm's activities, the higher its cost of capital, since investors typically require compensation for greater risk.
E.4 In recent price determinations and when setting the standard charges for interconnection to BT's network, Oftel has used a cost of capital of 15%. This rate is in nominal terms, is measured before investors' taxes and has been applied to historic cost asset valuations (or compared to historic cost accounting rates of return).
E.5 As part of the current review, Oftel needs to update its view on the cost of capital that should be used to set the price controls on BT over the following control period. Accordingly, it has sought expert advice on the theoretical foundations of and empirical evidence on the cost of capital, and in particular what this would imply for BT.
E.6 Oftel will announce its conclusions on the cost of capital in the consultative document to be published next March. The following subsections give a brief overview of the general theory behind the calculation of the cost of capital, provide an indicative range of estimates of the cost of capital for BT under various assumptions and suggest how this might differ for those activities which are subject to price control.
E.7 For a firm financed by debt and equity such as BT, the cost of capital will be a weighted average of its cost of capital from both sources. In what follows, general techniques used to derive the cost of equity and debt are first discussed; this is followed by indicative estimates of the components of the weighted average cost of capital for BT, leading to an indicative range for BT's pre-tax weighted average cost of capital, on which Oftel seeks views.
Cost of equity
E.8 Two main methods are typically used to establish a firm's cost of equity. The most widely used model for estimating the equity cost of capital is the Capital Asset Pricing Model (CAPM). The basic premise of this model is that investors require a higher expected rate of return on any investment in order to compensate them for a higher risk of returns on that investment (as measured by the variability of those returns).
E.9 Investors are assumed to be able to reduce risks by holding diversified portfolios of equities. However, there is a degree of systematic risk inherent in even the most diversified portfolio of shares, since the value of the whole stock market can rise or fall, reflecting the risk inherent in the general economy.
E.10 This non-diversifiable risk cannot be eliminated by holding shares in a large number of companies, and is therefore a component of the cost of equity. If the risk-free rate is that rate of return which investors would be able to earn with certainty, the market risk premium is that additional return that investors would require in order to compensate them for holding a share whose returns moved in line with those of the stock market as a whole.
E.11 Returns on shares in some companies will fluctuate in step with, but more widely than, returns to the stock market as a whole. Returns on other types of shares will fluctuate in step with, but less widely than, the stock market as a whole. Others still could move against the market. The degree of correlation between returns on shares in one company and returns on the stock market as a whole can be estimated using dividend and share price data and is captured in a coefficient known as the company's Beta. A company showing higher than average non-diversifiable risk will have a Beta coefficient in excess of one, while a company showing lower than average non-diversifiable risk will have a Beta less than one.
E.12 The cost of equity to the firm can then be calculated according to the basic CAPM formula below:
Re = Rf + Beta.[E(Rm) - Rf],
where Re is the cost of equity finance, Rf is the risk-free rate of return, Beta is the degree of correlation between returns on the company's shares and returns on the stock market as a whole, E(Rm) is the expected return on the market and E(Rm) - Rf is the expected market risk premium or excess return to equities.
E.13 This calculation can be done in real or nominal terms. The two should have identical implications for measuring the financial performance of the enterprise, provided that the inflation rate assumed in the financial forecasts is the same as that implied by the difference between the estimated real and nominal cost of capital.
E.14 One criticism often levelled at the CAPM is that the calculation of the equity premium is based on historic excess returns on equities rather than the returns that investors expected to achieve. Since investors base their decisions today on expectations of returns and their variability in the future, it would appear preferable to look at expectations directly. This is particularly important in the light of evidence that suggests that the risk premium varies over time, so that estimates of historic excess returns may not be a reliable guide to excess returns required in the near future.
E.15 The main alternative model, the Dividend Growth Model (DGM), attempts to alleviate these problems by using the expectations of investors directly. This model can either be applied to the whole stock market, in order to obtain an estimate of the market equity risk premium to be substituted into the CAPM formula discussed above, or can be used directly to estimate the cost of equity for an individual firm.
E.16 In the firm-specific version of the model, the firm's cost of equity is assumed to be equal to the discount rate which, when applied to the expected future dividends on shares in that company, makes the sum of these dividends equal to the current share price. More simply, if it is assumed that dividends are expected to grow indefinitely at an annual rate g, then the cost of equity to the firm can be shown to be given by:
Re = Do/Po + g,
where Do is the dividend paid at time 0 and Po is the share price at time 0.
E.17 The main difficulty with this type of calculation is that it is necessary to form a view about investors' expectations of future dividend growth, and these expectations are usually difficult to elicit with any degree of accuracy. Analysts' forecasts of future dividends on shares in large companies such as BT can be used as an indicator of market expectations of future dividend growth. However, it is questionable whether sufficient independent forecasts are available to provide an accurate estimate of BT's cost of equity. A further problem with this technique is that analysts' forecasts do not typically extend far beyond two years into the future, so that estimates of g are very speculative.
E.18 Nevertheless, since Oftel needs to calculate BT's cost of capital over the next control period, it is important to take account of forward-looking estimates, especially in the light of evidence that the cost of capital tends to change over time. Indicative estimates of BT's cost of equity derived from the DGM are therefore presented alongside CAPM estimates in the following sections.
E.19 In the absence of specific information on the interest rates being paid by the firm in question, the pre-tax cost of debt is typically calculated by adding a small corporate risk premium to an estimate of the risk-free rate of return, as proxied by the return on government debt used in the CAPM calculation.
E.20 In this sub-section, historical estimates of the cost of equity to BT are built up from individual components of the CAPM formula explained above, and compared to forward-looking estimates based upon the DGM approach.
E.21 The nominal risk-free rate of return is typically calculated as the yield on fixed-interest government debt of a certain maturity. The choice of maturity depends upon the time horizon over which the risk-free rate is to be estimated. For Oftel's purposes, this might be the length of the next price control period, ie around 5 years. Gross redemption yields before tax on gilts of this maturity are currently around 7.6% for a zero rate taxpayer. However, from BT's point of view, a more appropriate maturity might be one which corresponds to the average life of its assets. Since this is fairly long, yields on gilts with maturities in excess of 15 years would be an alternative choice for the risk-free rate of return. Gross redemption yields before tax on gilts of this maturity are currently around 8.4% for a zero rate taxpayer.
E.22 The real risk-free rate of return which is consistent with the nominal rate can be estimated from the yields on index-linked gilts of similar maturity. The implied inflation rate expected by investors over the period can then be calculated as the proportionate difference between the two.
E.23 A further complication is that the estimated post-tax risk-free rate and the implied inflation forecast depend upon the tax rate that is assumed for the marginal investor, since the gilt which offers the higher post-tax yield will be different for zero rate taxpayers then for basic rate taxpayers, for example.
E.24 Table E.1 gives an indicative range of the nominal and real risk-free rates of return for gilts of different maturities, together with the implied inflation rate, based on different assumptions about tax rates.
Table E.1 Risk-Free Rates of Return: Indicative Ranges
Nominal Real Inflatio rate*
n
Period: 5 year 15 year 5 year 15 year 5 year 15 year
% per annum
Before inflation risk
adjustment
0% taxpayer 7.6 8.4 3.3 3.7 4.2 4.5
25% taxpayer 8.0 8.8 3.6 4.0 3.2 3.5
After inflation risk
adjustment
0% taxpayer 7.1 7.4 3.3 3.7 3.7 3.6
25% taxpayer 7.5 7.8 3.6 4.0 2.8 2.8
Note: (1) The implied inflation rates are calculated by dividing (1 + nominal %) by (1 + real %), where real and nominal rates are net of tax - as opposed to the rates shown above for the 25% taxpayer, which are gross of tax.
E.25 It is possible that medium- and long-term nominal giltyields incorporate a risk premium over and above short-term yields in order to compensate investors for uncertainty about inflation. In the bottom half of Table E.1, the risk premium has been assumed to be 0.5% for five-year gilts and 1.0% for fifteen-year gilts. Nominal gilt yields and implied inflation rates have been adjusted accordingly, and it has been assumed that the inflation risk premium is negligible for index-linked stock.
E.26 The overall range for the real risk-free rate of return of 3.3% to 4.0% compares with a range of 3.5% to 3.8% used by the Monopolies and Mergers Commission (MMC) in its report on Scottish Hydro-Electric plc (SHE) published in May 1995.
E.27 The market risk premium can be estimated in two main ways, as discussed in the previous sub-section. The standard CAPM approach is to calculate the total return on equities over and above returns on gilts for a given past period. The alternative approach is to use forecasts of investors' required rates of return on equities, as calculated using the DGM. Each of these methods of estimation can give very different answers depending on the period over which the calculations are performed, and depending on whether average excess returns over time are calculated as an arithmetic (simple) or geometric mean.
E.28 Estimates of historic excess returns on equity from the UK, US and Japan range between 8.0% and 9.4%. Estimates calculated over shorter and more recent periods, using the DGM for the US, give a slightly lower range of estimates (6.5% to 7.5%). Equivalent estimates using the DGM for the UK stock market do not exist. Two recent surveys of fund managers in the UK suggest that the risk premium required on equities may be as low as 2.5% to 3.0%. However, these surveys were based upon relatively small smaple sizes, and may not be reliable. The weight of evidence from academic studies at this stage suggests that the market risk premium lies above 4% but below 8%.
E.29 Several recent academic studies suggest that the current size of the equity premium may be lower than that implied by historical estimates from stock market data. Risk premia as high as 8% or 9% do not appear to be consistent with investors' levels of risk aversion, as measured by alternative methods. There is also some evidence to suggest that risk premia vary over time; forecasts of required returns on equities which relate returns on equities to other observable series, such as returns on other types of security, provide estimates towards the bottom end of the 4% to 8% range.
E.30 In price control reviews over the last three years, OFWAT and OFFER have used a market risk premium in the range of 3% to 4%. The MMC used a range of 3.5% to 4.5% in the SHE report. Oftel's initial view is that it would not be justified in considering a market risk premium greater than 6%. In the estimates of BT's cost of capital that follow, a range of 4% to 6% is used.
E.31 The value of BT's equity Beta measures the volatility of returns on BT's shares compared to returns on the stock market as a whole. It will rise with BT's debt/ equity ratio, since a higher level of gearing implies that a given change in profits will have a greater impact on the returns to holders of equity.
E.32 The estimated value of BT's Beta varies depending on the time period over which it is measured and on whether monthly or daily share price information is used. It can also be significantly biased if 'events' produce major changes in Beta which violate the assumptions upon which the CAPM methodology is founded. Examples of such events might be the stock market crash of 1987 and the general elections of 1987 and 1992.
E.33 OXERA and LBS Risk Management Service estimates of BT's equity Beta, using monthly data for the five-year period ending in May and June 1995 respectively, are 0.83 and 0.80. These appear to be robust to the omission of the 1992 general election.
E.34 These estimates of equity Beta relate to BT Group. In the past an estimate of Beta for BT Group has been used as a reasonable proxy for the Beta for the price-controlled activities. However, BT's Beta has risen since the time of the last price control review, probably mainly as a result of the expansion of its non-regulated business. This is likely to be more risky than its price-controlled activities, for two main reasons. Firstly, services in which BT still has a degree of market power will tend to be less risky than services where competition is better developed, since profits from this source will tend to be less volatile. Secondly, basic telephony services are more likely to be 'essential', implying that demand for them is likely to fluctuate by less than average over the cycle.
E.35 This suggests that services that remain outside a tariff basket will tend to be more risky than those within a basket. It is therefore likely that the implicit Beta for BT's price-controlled activities will be lower than that of BT Group. Oftel intends to use a Beta of 0.80 in its estimate of the overall cost of capital for BT Group, but will consider the effect of reducing this in modelling the price controls.
E.36 As an illustrative example, the effect of reducing the value of Beta to 0.60 on the cost of equity is shown in the tables which follow. This figure is within the range of 0.50 to 0.65 used by the MMC for the equity Beta of SHE's electricity distribution business, typically viewed as being of low risk. The true Beta of BT's price-controlled activities is likely to fall somewhere in between 0.60 and 0.80.
E.37 The estimate of the post-tax cost of equity depends on the view taken on the size of any tax advantages to debt to be obtained by offsetting interest payments against corporation tax. Under the UK imputation system with advance corporation tax (ACT), the tax shield afforded by debt is significantly reduced compared to the US, for example. It is also possible that any remaining tax advantage to debt is captured by lenders in the form of higher interest rates.
E.38 In Table E.2, the range of estimates of the post-tax cost of equity includes scenarios in which it is assumed that there is a tax advantage to debt and scenarios where is it assumed that any tax advantage is captured by lenders. The estimates derived from the CAPM for each combination of equity risk premium and equity Beta also reflect the full range of real and nominal risk-free rates (ie with and without an inflation risk adjustment).
Table E.2 BT's Post-Tax Cost of Equity: Indicative Ranges
CAPM DGM
Equity 4 4 6 6
premium
Equity Beta 0.6 0.8 0.6 0.8
% per annum
Nominal Low 7.1 7.9 8.3 9.5 7.3
High 9.4 10.2 10.6 11.8 10.9
Real Low 4.6 5.4 5.8 7.0
High 5.6 6.4 6.8 8.0
E.39 In the table, estimates of BT's cost of equity which have been derived directly from the DGM are shown as a comparison to those built up from the various components of the CAPM discussed above. The range of estimates from the DGM reflects forecasts of annual dividend growth rates from brokers' reports in the range of 3% to 6%, applied to BT's net dividend yield (4.3% to 4.9%) calculated at various points in time.
E.40 Historical evidence suggests that 'blue chip' corporate debt commands a risk premium of approximately one half to one per cent higher than the risk-free rate. In the SHE report, the MMC used a range of 0.3 to 0.7% for the premium. In the indicative estimates of BT's weighted average cost of capital that follow, a 0.5% corporate risk premium has been used. In order to convert the resultant pre-tax cost of debt into a post-tax rate, corporation tax at 33% has been subtracted.
E.41 In the calculation of the post-tax weighted average cost of capital, weights equal to the proportion of debt and equity finance are applied to the post-tax cost of debt and equity in turn. It is usual to use market values of debt and equity in this calculation.
E.42 In those scenarios where it is assumed that there is a tax advantage to debt, it may be argued that BT could reduce its cost of capital by increasing its gearing. BT's current gearing ratio is around 15%. Oftel intends to consider whether BT's weighted average cost of capital should be adjusted downwards to reflect the fact that it may be under-geared. However, initial estimates suggest that even a large increase in gearing from current levels to 50% would decrease the post-tax WACC by at most 0.5 percentage points in real terms and 0.7 percentage points in nominal terms.
E.43 A potentially more important issue is if BT significantly increased its gearing beyond that which might be considered prudent. In such circumstances, any tax advantages from higher gearing might be outweighed by the higher risk premium charged by lenders and a higher cost of equity. Oftel would not allow any increase in the cost of capital as a result of this kind of financial restructuring to affect this or any future price control regime.
E.44 Table E.3 gives an indicative range for BT's post-tax WACC. Again the ranges for each value of Beta and equity risk premium reflect different assumptions about the tax advantages of debt, different risk-free rates, as well as adjustments for higher gearing levels.
Table E.3: BT's Post-Tax WACC: Indicative Estimates
CAPM DGM
Equity 4 4 6 6
premium
Equity Beta 0.6 0.8 0.6 0.8
% per annum
Nominal Low 6.8 7.5 7.9 8.9 6.2
High 9.0 9.6 10.0 11.0 10.2
Real Low 4.3 5.0 5.3 6.3
High 5.2 5.9 6.2 7.3
E.45 The indicative range of estimates of the post-tax WACC for BT Group using the CAPM with a Beta coefficient of 0.8 is 7.5% to 11.0% in nominal terms and 5.0% to 7.3% in real terms. This range may need to be adjusted downwards to give a cost of capital for the price-controlled activities. As an illustrative example, the lower end of the range would fall to 6.8% in nominal terms and 4.3% in real terms if a Beta coefficient of 0.6 were used for the price-controlled activities.
E.46 The estimates of the cost of capital presented above incorporate the returns after tax which are required by investors to induce them to buy or retain shares in BT or to lend the company money. Oftel needs to have a cost of capital estimate that can be compared to pre-tax rates of return in the financial modelling.
E.47 There is no direct method of deriving the company's pre-tax cost of capital for the next price control period from the post-tax estimates presented above, in the absence of information on future cashflows. A standard simplification that can be used to derive an estimate of the pre-tax WACC, which Oftel has used in the past, is to multiply the post-tax cost of equity by
(1 - ACT)/(1 - Tc)
where ACT = marginal rate of advance corporation tax (currently 20%) and Tc = marginal rate of corporation tax (currently 33%). The adjusted pre-tax cost of equity can then be combined with the pre-tax cost of debt using the gearing weights to give an estimate of the pre-tax WACC.
E.48 This adjustment is based upon a number of simplifying assumptions, eg that all profits are paid out as dividends. The correct adjustment will depend on BT's cash flow profile over the forecast period, amongst other things.
E.49 Table E.4 shows a range of estimates for BT's pre-tax cost of capital for an illustrative case using the simplified formula above. It is important to note that the range is only given for a gearing ratio of 15% and does not incorporate the adjustments for desirable gearing that are shown in the previous table. BT's actual pre-tax WACC may be lower or higher than this range, depending upon adjustments for desirable gearing and the cashflow forecasts generated from the financial model.
Table E.4: BT's Pre-Tax WACC: Illustrative Case For 15% Gearing Using Standard Formula (1)
CAPM DGM
Equity 4 4 6 6
premium
Equity Beta 0.6 0.8 0.6 0.8
% per annum
Nominal Low 8.4 9.2 9.6 10.8 8.6
High 11.0 11.8 12.2 13.4 12.4
Real Low 5.2 6.1 6.5 7.7
High 6.4 7.2 7.6 8.8
Table note: (1) See paras. E.47 to E.49 for a discussion of the simplifying assumptions used.
E.50 Under these simplifying assumptions, an indicative range for the pre-tax cost of capital for BT Group (using a Beta coefficent of 0.8) for the next price control period would be 9.2% to 13.4% in nominal terms and 6.1% to 8.8% in real terms. Again, as an illustrative example, the lower end of the range would fall to 8.4% in nominal terms and 5.2% in real terms if a Beta coefficient of 0.6 were used for the price-controlled activities.
E.51 A breakdown of the components of the range of cost of capital estimates for BT is compared with those used by the MMC for Scottish Hydro-Electric's distribution business in Table E.5. It is important to note that the two businesses would not be expected to have the same cost of capital since they have different risk characteristics and gearing ratios. Oftel's wider range also reflects uncertainty at this stage over the size of the equity risk premium, the level of desirable gearing, and the size of the tax adjustment necessary to derive a pre-tax WACC. Oftel hopes to receive submissions on the indicative range for the pre-tax WACC and its components as a result of this consultation exercise.
Table E.5 Oftel Indicative Ranges for Components of Real Pre-Tax WACC: Comparison With MMC View In SHE Report (1)
Oftel MMC
Low High Low High
Real risk-free 3.3% 4.0% 3.5% 3.8%
Equity risk 4.0% 6.0% 3.5% 4.5%
premium
Equity Beta 0.6% 0.8% 0.5% 0.65%
Post-tax cost of 4.6% 8.0% 5.2% 6.7%
equity
Gearing 15.0% 50.0% 8.0% 8.0%
Dewbt premium 0.5% 0.5% 0.3% 0.7%
Post-tax cost of 2.5% 3.0% 3.8% 4.5%
debt(2)
Real Post-Tax 4.3% 7.3% 5.1% 6.5%
WACC
Real Pre-Tax 5.2% 8.8% 6.1% 7.8%
WACC(3)
Notes: (1) Oftel's estimates of the ranges for the post-tax and pre-tax WACC cannot be derived directly from the high and low values of the components, due to different tax and gearing adjustments.
(2) The MMC range for the cost of debt does not distinguish post-tax from pre-tax rates.
(3) Oftel's range of pre-tax WACC estimates depends upon the simplifying assumptions in the text, and does not incorporate the full range of gearing shown in the table.
E.52 There are two main options as to the measure of the rate of return to be used in the financial modelling process: economic or accounting rates of return.
E.53 Rates of return on an investment project are typically calculated as the internal rate of return (IRR). This is the discount rate which equates the revenue streams of a project with the costs of the project. This measure requires an initial and terminal economic value of the asset base (at the start and end of the period) as well as the free cashflows in each year of the period. Rates of return calculated on this basis are directly comparable with the returns required by investors and lenders in order to induce them to supply the necessary funds. Economic rates of return therefore benefit from being directly comparable with BT's cost of capital, as calculated by any of the methods discussed in the previous section.
E.54 Measuring economic rates of return would, however, raise a number of practical difficulties. Firstly, economic rates of return can be very sensitive to the profiles of cashflows, such as the timing of capital expenditure, which it is difficult to forecast with any accuracy. Secondly, the estimate of the terminal value of the asset base in the IRR calculation is fraught with difficulties and is highly subjective. Finally, and perhaps most importantly, price controls should be based upon measures of financial performance which are transparent and understood by BT, its competitors and the wider community. There could be significant potential for confusion if Oftel were to set a price control based upon one measure of profitability (economic rates of return) when BT reports to its shareholders on the basis of another measure (historic cost accounting rates of return on capital employed).
E.55 The alternative to an economic rate of return, measuring profitability as an accounting rate of return, would not however be free from difficulties. Although Oftel would use a definition of the accounting rate of return which would better reflect the economic rate of return than would rates of return on an HCA basis (for reasons discussed below), any differences would in principle require an adjustment to the cost of capital to ensure comparability. But, in making the adjustments to the cost of capital, difficulties would arise, which are similar to those set out above for measuring economic rates of return. Oftel will continue to explore this issue during the price control review.
E.56 Accounting rates of return express the ratio of accounting profit to the contemporaneous value of capital. Both numerator and denominator vary between firms depending on the accounting conventions adopted, including the choice of Current Cost Accounting (CCA) or Historic Cost Accounting (HCA). The main differences relate to the treatment of fixed assets in the balance sheet and the depreciation charge to the profit and loss account. There are three main forms of bias which can prevent an accounting rate of return from being directly comparable to the cost of capital discussed in the section above. These occur when assets are not valued in Modern Equivalent Asset (MEA) terms, when profits are not measured as a 'clean surplus', and when depreciation policies and asset lives used in the balance sheet do not reflect underlying economic values.
E.57 BT uses the conventional HCA basis of accounting, where fixed assets are valued at original purchase cost net of cumulative depreciation. This has the merit that most of its competitors use the same method of asset valuation. However, HCA rates of return are not directly comparable with estimates of BT's cost of capital. The explanation for this is that the net book value of fixed assets in historic cost terms will not in general be equal to the economic value of capital employed. HCA net book values take no account of general price inflation or changes in the relative price of specific assets over the period since they were purchased, and so do not properly measure the cost of the resources employed.
E.58 CCA asset valuations attempt to correct for this effect by valuing fixed assets at the net replacement cost of a Modern Equivalent Asset of the same service capability, allowing for the remaining useful asset life. Since this valuation uses current fixed asset prices, it takes into account general inflation and specific asset price changes that have taken place since the asset was purchased. The CCA rate of return would therefore provide a better approximation to the economic rate of return than would the HCA rate of return.
E.59 When changing the basis of the measurement of rate of return from HCA to CCA, it will be necessary to ensure that the revaluation of assets does not result in windfall gains or losses accruing to the shareholders of BT. For a discussion of this issue in the context of the network charge cap, see Chapter 5. The avoidance of windfall gains or losses would require that profit were measured as a 'clean surplus', ie the revaluation surplus (or deficit) should be reflected in the profit and loss account. The concept of 'clean surplus' is discussed further below.
E.60 In a company balance sheet, depreciation charges represent the charge to profits necessary to recover the loss of asset value which arises as the asset is consumed over its life. For practical reasons, this will typically be on a straight-line depreciation basis, where an equal proportion of the gross book value of the asset is written down in each year of the asset's life.
E.61 However, when using net book values as the denominator in a rate of return calculation, there would be a closer reflection of economic rates of return if the asset lives and depreciation profiles were chosen so as to correspond to the true economic lives and the rate at which the asset is used up over its life. In other words, CCA accounting rates of return would better approximate to economic rates of return if accounting depreciation were more closely aligned to economic depreciation. Since there could be practical difficulties in implementing this adjustment, it is an issue that will be explored further during the price control review.
E.62 Another possible source of bias may be introduced by the accounting measure of profit used in the accounting rate of return calculation and, in particular, the depreciation charge to the profit and loss account. Under CCA procedures, there are two main alternative principles which may be followed: these are operating capability maintenance (OCM) and financial capital maintenance (FCM).
E.63 Under OCM, the HCA depreciation charge is adjusted to take into account changes in the MEA valuation of the asset since the start of the accounting period. Barring unexpected events, this will enable sufficient funds to be put aside before the distribution of profit in each year for the asset to be replaced at the end of its life. In a world where asset prices are rising, an OCM depreciation charge to the profit and loss account will be higher than the corresponding HCA depreciation charge.
E.64 Whilst an OCM approach to depreciation will, as its name suggests, enable a firm to maintain its operating capability, it will not in general give a measure of profit and hence rate of return that is comparable to the firm's cost of capital, even when assets are valued in CCA terms. This is because it neglects the fact that a change in the MEA value of the firm's assets from one year to the next represents a change in the wealth of its shareholders.
E.65 An FCM deprecation charge to the profit and loss account therefore includes an additional holding gain or loss equal to the change in the MEA value of the firm's assets between periods. The resulting profit is sometimes described as a 'clean surplus'. This represents the amount that can be distributed after maintaining the nominal value of the firm's financial capital. Provided the depreciation policy used in the balance sheet is economically justified, and subject to certain other restrictive assumptions, CCA rates of return calculated on this basis can be comparable with the firm's nominal cost of capital.
E.66 A further adjustment is possible to put the calculation into real terms. In order to maintain the real value of shareholders' funds, profits must additionally be reduced by a further 'shareholder adjustment' equal to the rate of inflation over the period multiplied by the value of shareholders' funds at the start of the period. No adjustment is necessary to maintain the real value of debt, since lenders receive a nominal interest rate which already reflects their expectations of inflation.
E.67 In theory, it would be desirable to use economic rates of return in the financial modelling, since these are directly comparable to the cost of capital that was discussed in the previous section of this Annex. However, there are merits in a modelling approach based upon accounting rates of return. If an accounting rate of return were used, it should be measured using CCA conventions (on an FCM basis) rather than HCA, in order to reflect more closely the nature of forward-looking costs. Oftel will need to ensure that the implications of its chosen approach for the HCA rates of return reported in BT's accounts are fully explained, since these are the only measures of BT's financial performance available to and understood by a wider audience.
E.68 The expected future financial performance of BT will be assessed using a financial model. The model will be used to project forward BT's costs, revenues and capital employed for the services within each tariff basket. In general, the more disaggregated a model, the more realistic it is. For this review, where it is proposed that a network price cap will exist side by side with a retail price cap, it is particularly important to model BT's financial performance at the network and retail level.
E.69 Because a large proportion of BT's costs and fixed assets are shared between activities inside and outside the proposed price control baskets, it is essential to expand the coverage of the model to include all those activities where there is the potential for costs or assets to be shared. This might include BT's non-regulated activities as well as non-price-controlled services.
E.70 Based upon the forecasts derived from the model, a range of values of X will be chosen for each price control so as to allow BT to earn an expected rate of return which Oftel regards as acceptable by the end of the control period (subject to the caveat in the introduction to this Annex). This in turn depends upon the view that Oftel takes upon BT's cost of capital for those services and upon the calculation of the rate of return earned on them, as discussed in the previous sections.
E.71 In the March consultative document, Oftel intends to publish the details of the range of X for each tariff basket and sub-cap that it proposes as a result of its financial modelling. At the same time, it will explain the assumptions and parameters underlying the range of price controls, as well as the forecasts of the financial performance that BT is expected to achieve in the activities within each price cap or sub-cap.
E.72 In order to model BT's financial performance, it is necessary to take a view on a number of different variables and parameters. The most important of these are the following:
E.73 Each of the main model input assumptions is discussed in more detail below.
E.74 Market growth is one of the most important determinants of BT's financial performance over the next control period. A given change in the volume of output of a particular service can have a significant effect on profitability if economies of scale are important (see discussion in para. 4.21 of Chapter 4). This is likely to be the case for many of the services supplied over BT's network since, in the short run, the marginal cost of supplying an additional telephone call is negligible. This means that an increase in call revenues may lead to an almost one to one increase in measured profits.
E.75 This makes it crucially important that volume growth assumptions for each service market incorporate the full range of possible outcomes. Preferably, demand growth forecasts should be related in an objective way to the underlying determinants of demand for each product (eg by applying statistical techniques to historical data). These would typically include the market price, the price of substitute and complementary products, an income variable (eg GDP) and any others (which might be captured by a time trend). Depending on the market in question, the market price, which in the past has effectively been that set by BT subject to the constraint imposed by its price cap, will increasingly be determined by the extent of actual or potential competition in the market (see the discussion of Effective Competition in Chapter 3).
E.76 In practice, the main determinants of the strong growth in demand for basic telephony since privatisation (as displayed in Tables 7.4 and 7.5 of Chapter 7) have been, in rough order of importance, rising real incomes, technological innovation (eg the falling price of fax machines) and falling real call prices (as a result of price cap regulation and increasing competition). The assumptions made about these influences will have the biggest impact on forecasts of demand over the next forecast period.
E.77 Alongside a forecast of the total market demand for a given product, a view needs to be taken on what share of that demand will be supplied by BT's network or retail operations. In general, this requires assumptions on the price BT will charge for each product over the price control period (within the constraint imposed by any basket price cap or sub-cap), the prices its competitors charge, and the propensity for consumers to switch supplier for a given price differential.
E.78 Table 7.6 in Chapter 7 shows how BT's market shares in the supply of exchange line connections, inland and international calls have fallen since 1991/2. When considering the factors that have contributed to the loss of market share by BT recorded so far, and projecting this forward into the next price control period, it is useful to draw a distinction between direct and indirect market share loss by BT.
E.79 BT's directly-connected customers will, by definition, obtain their exchange line connection from BT's retail operations. In the absence of additional benefits from cable TV connection or access to broadband services, they will only tend to switch to a different access supplier if they expect their total bill (access and call charges) to fall, given their pattern of calls. Hence, market share loss forecasts for directly connected customers will depend mainly upon the relative price of an average basket of services for customers with different demand profiles purchased from BT Retail, compared to an equivalent basket bought directly from an alternative access supplier.
E.80 This requires a view to be taken on BT's and competitors' pricing strategies and on how quickly different groups of consumers (eg residential and business customers with different demand profiles) will switch access supplier in response to a given expected saving on their average bill. This will change with the advent of number portability.
E.81 In addition to direct market share loss as an access provider and supplier of retail telephony, BT Retail is vulnerable to indirect market share loss in inland and international calls to competing trunk operators (eg Mercury, Energis, ISR operators) which interconnect with BT Network. It would be expected that the loss of market share by BT Retail to indirect competitors would be more closely related to the relative price of BT Retail for the service in question. Assumptions regarding indirect market share loss will therefore depend upon assumptions about BT Retail prices for individual services (subject to the level of the retail price cap), as well as competitors' prices and the propensity to switch for a given price differential. It might be expected that a given price differential would induce higher indirect than direct market share loss, because of customer inertia and the risk involved in switching access providers.
E.82 Once a customer has switched access supplier to another licensed operator (eg a cable company, Mercury), their custom will be lost to BT's retail operations. However the access supplier will still need to purchase access to BT's network for the majority of calls, either through an existing interconnect agreement with BT, by purchasing interconnect components at BT's network component tariff, or by purchasing long-distance conveyance from another operator who in turn pays BT for interconnect at the terminating end.
E.83 Demand for BT Network's interconnect services will then depend on:
(a) BT's share of directly-connected customers; and
(b) BT's market share in the supply of different network components.
The latter will in turn depend upon the tightness of the network price cap and sub-cap on call terminations, the network tariff structure that BT adopts given that constraint and the pricing strategies of competing suppliers of trunk conveyance, amongst other things.
E.84 As well as revenues derived from the demand model discussed above, the financial model needs to be able to forecast in some detail the costs incurred by BT. These will be driven by three main factors. Firstly, higher demand for one or more of BT's services will, in the absence of offsetting efficiency gains, lead to higher derived demand for one or more factor inputs (eg labour, capital). Secondly, the volumes of output produced per unit of input may increase as BT becomes more efficient due to the incentives given by price cap regulation and emerging competition, or because of general industry developments. Thirdly, the prices of factor inputs may change, either due to downward pressure being placed on suppliers' prices by BT or because of technical change or other factors out of BT's control.
E.85 The first of these determinants of costs requires estimates of the relationship between different components of BT's costs and the volumes of different services supplied. These should preferably be broken down by factor inputs, eg switching capacity, numbers of person-hours.
E.86 BT's efficiency has already been discussed in relation to the assessment of BT's relative efficiency at the start of the next control period. What is required is an estimate of how BT's productivity is likely to improve over the next control period in the key service areas in which it operates. The standard measure of efficiency that has been calculated for BT by Oftel in the past is an index of real unit costs, which measures the total cost incurred by BT per unit of output.
E.87 This measure of efficiency has the property that it combines factor price changes with reductions in the volumes of factor inputs consumed. Hence, a tight price control based upon expected real unit cost reductions could be met either by increasing the amount of output produced per unit of factor input, or by exerting downward pressure on the prices paid for inputs to the extent that these are not already sourced from the cheapest suppliers.
E.88 Since privatisation, BT's real unit costs have fallen by around 3.5% per annum on average for a set of services which broadly corresponds to those currently in the PSTN tariff basket. One problem with basing assumptions about productivity gains for the next control period on those achieved historically is that the past may not be a good guide to the future. The potential for further efficiency gains may be less than before if the firm has already implemented 'best practice'. Alternatively, future technological developments may make possible far greater efficiency gains than could have been achieved in the past.
E.89 Comparisons of BT's efficiency with key domestic competitors or comparable overseas operators can help address the first of these problems, by showing the extent that BT still lags behind. As discussed in Chapter 5 and Chapter 7, preliminary results from benchmarking studies suggest that BT's efficiency across its combined Access, Network and Retail Systems businesses may be up to 10% worse than that of the best-performing comparable competitors.
E.90 As referred to above, a large proportion of BT's costs and fixed assets are shared between regulated and non-regulated activities. At a review, there is an incentive on BT to understate its true profitability by allocating an excessive proportion of fixed costs and assets to those activities which are price-controlled.
E.91 To some extent this problem has been alleviated by the implementation of Accounting Separation, which has set out a clear basis upon which joint and common costs should be allocated to Access, Network and Retail businesses. However, in modelling BT's financial performance one important area of concern is in ensuring that new capital expenditure proposed by BT for its price-controlled business over the next control period is, firstly, efficiently incurred and, secondly, is allocated appropriately to BT's price-controlled activities.
E.92 One way of checking that BT's investment proposals incorporate an appropriate allocation across services is to project forward capital expenditure for those activities within a price control basket on a CCA basis. After taking into account the effect of volume growth on demand for new fixed assets (using the cost/ volume relationships discussed above), gross capital expenditure should be equal to depreciation, when this is measured on a CCA basis. This is an additional benefit of forecasting on a CCA basis.
E.93 Oftel intends to explain its proposed treatment of BT's capital expenditure programme over the next control period, as well as other aspects of overhead and fixed asset allocation, in the March consultative document.
Return to this document's table of contents
Go to the Start of this Document