Review of the WACC/discount rate

Review of the WACC/discount rate Electricity Commission Draft decision on Transpower s Auckland 400kV grid investment proposal A report prepared by Marsden Jacob Associates for Mighty River Power 22 June 2006

This report has been prepared in accordance with the scope of services described in the contract or agreement between Marsden Jacob Associates Pty Ltd ACN 072 233 204 (MJA) and the Client. Any findings, conclusions or recommendations only apply to the aforementioned circumstances and no greater reliance should be assumed or drawn by the Client. Furthermore, the report has been prepared solely for use by the Client and Marsden Jacob Associates accepts no responsibility for its use by other parties. CONTACT: Dr. Jeff Washusen jeff.washusen@marsdenjacob.com.au Mr. Jasper Boe Mikkelsen jasper.mikkelsen@marsdenjacob.com.au Financial & Economic Consultants Level 3, 683 Burke Road, Camberwell, Victoria 3124 Tel: (03) 9882 1600 Fax: (03) 9882 1300 Email : economists@marsdenjacob.com.au

TABLE OF CONTENTS Page 1.Introduction and Summary...1 2.Background...4 3.Update of WACC over time...12 4.Use of CC WACC...14 4.1.The CC s approach...14 4.2.Changes needed to the WACC...16 4.2.1. Treatment of tax...16 4.2.2. Nominal vs. real...19 4.2.3. Conversion from nominal post tax to real pre-tax...20 4.3.Estimation of the WACC...21 5.MJA adjustments to CC WACC...25 5.1.PTMRP...26 5.2.Beta...32 5.3.Leverage...34 5.4.Debt Premium...35 5.5.Summary of estimates...38 6.Additional adjustments...40

1. Introduction and Summary 1. Marsden Jacob Associates (MJA) has been requested by Mighty River Power to provide comments on the discount rate to be used for application of the Grid Investment Test (GIT). This request arises from discussion relating to electricity transmission investments in the Electricity Commission s (EC) Draft decision on Transpower s Auckland 400kV grid investment proposal for the purposes of consultation under section III of part F of the Electricity Governance Rules 2003 (26 April 2006). 2. Specifically, Mighty River Power has requested comment in relation to four issues: the impact on the 7.0% real, pre-tax discount rate adopted by the EC arising from changes in interest rates since 2004; whether extrapolation of the WACC value determined by the Commerce Commission (CC) for Unison s electricity lines business (ELB) to a GIT discount rate is appropriate; adjustments that could be made to the CC s Unison WACC value to determine an appropriate WACC for Transpower; and adjustment that could be made to reflect (i) differences between the WACC for a transmission and ELBs; and (ii) differences between the regulatory context of price control (Unison) and application of the GIT. 3. The comments and opinions expressed in this paper are those of MJA and do not necessarily reflect those of Mighty River Power. No part of this submission is confidential and MJA has no objection to it being made publicly available. 4. Our findings and recommendation are summarised below: We note that the EC's focus is not on WACC per se as it would be applied in revenue and pricing determinations. However, once the GIT has been approved, the resulting network investment will be subject to Section 17 of Part F of the Electricity Governance Rules (EGRs), which allows Transpower to recover the investment cost including the cost of capital. We agree with the EC (and its consultants) that the prime function of the discount rate in a GIT analysis is to assist in ranking investment options with differing cost/benefit value streams. The value of the discount rate will make substantial changes to numerical outcome of an NPV analysis. 1

There is no fundamental reason why a GIT discount rate need be the same value as an 'efficient WACC', even though it is sensible to undertake analysis of investment options using a discount rate that has a value that is not too dissimilar to an 'efficient WACC'. However, it is our view that the GIT should use a default discount rate with the same value as an efficient WACC. This is clearly the simplest approach. It is our view that adopting such an approach should assist in improving transparency (and reducing the obvious confusion) that surrounds selection of a WACC/discount rate value in both the GIT and the revenue/pricing process. We believe it would be sensible, and reasonably transparent, to conduct sensitivity analyses using discount rates lower and higher than an 'efficient WACC. We acknowledge that this view is different to that adopted by the ACCC which determined that the regulated WACC should be the minimum value of the discount rate to be applied in the ACCC regulatory test, which has many attributes that are identical to the GIT. Having said that, we also believe it would be prudent for the EC to explicitly commit to determination of an 'efficient WACC' for Transpower, which would be subsequently adopted as the 'default' discount rate in the GIT - especially given that a value of 7% real, pre-tax appears to be above the 'cautious and conservative' values adopted by Australian regulators using parameter values at the high end of plausible ranges. It is our view that an efficient WACC should be: consistent with reasonable capital market expectations that reflect the relatively low risk of Transpower's regulated activities; consistent across Transpower's business/assets, reflecting the relatively low risk of asset stranding for investment decisions subject to detailed regulatory scrutiny (and application of the GIT); and based on average or mid-range estimates of parameter values developed by the EC taking into consideration all available relevant information. On the above basis, we estimate an appropriate efficient WACC for Transpower would be no more than 6.05% real, pre-tax assuming 2

inflation at 3.00% 1. We also suggest the EC use 5.1% and 7.0% as the upper and lower bound discount rates in its sensitivity analysis. We consider this to be a conservative (i.e. erring on the high side) estimate of the WACC. If the result of the GIT analysis is accepted by the EC, Section 17 of Part F of the EGRs guarantees recovery of the approved costs. This form of regulatory guarantee may have the effect of reducing the investment risk faced by Transpower compared to investments not subject to a GIT analysis. The consequence of this is that the asset Beta for Transpower s new investments may well be lower than the asset beta in the context of price control and hence lead to a lower WACC. Assuming a reduction of the asset Beta by 0.1 we estimate a real pre-tax WACC of 5.15%, however more analysis would be needed before such an estimate could be relied on. 1 We assume 3% based on the latest available statistics and forecasts from NBNZ Business Outlook. According to the National Bank NZ survey expectations over the twelve months to May 2007 are for prices to rise 3.3%. Inflation expectations have been over 2.8% since August 2004. Source: National Bank Business Outlook April 2006 : http://www.nbnz.co.nz/economics/outlook/060426/default.htm Inflation expectation data from: http://www.nbnz.co.nz/economics/outlook/nbnz_business_outlook_data.xls ; both accessed 20 June 2006 3

2. Background 5. The EC has responsibilities under the EGRs to develop the GIT and apply it to Transpower s Grid Update Plans (GUPs). 6. Reports prepared by the EC s consultants (Frontier Economics and SAHA International) 2 confirm that the GIT developed by the EC is closely related to the ACCC s regulatory test, 3 which is required to be used in Australia to assess major investment options in all electricity networks. These consultant reports, and the EC s own documents, 4 confirm that the GIT is based on a cost-benefit approach that requires, inter alia, reasonable forecasts of costs and benefits for alternative options that can be compared using a standard NPV analysis with a discount rate that reflects, among other things, the scarcity of capital. 7. The EC has committed substantial time and resources to development of the GIT, approval of Transpower s GUP and application of the GIT. Most interestingly, the EC has taken a very active role in developing alternatives for Transpower to consider - even producing a 'Short-Short List' of alternatives consisting of seven generation options, four demand-side initiatives and four alternative transmission solutions. 5 The EC has also required Transpower to assess the viability of the four specific transmission alternatives ( alternative projects ) using the GIT. 8. In respect of quantifying costs and benefits that can be used as input data in a GIT analysis, we note that paragraph 6.5.6 of the EC draft decision says in the assessment/ranking of alternative transmission investment options: 2 3 4 5 Draft Grid Investment Test Final Draft Discussion Paper, Frontier Economics, June 2004; Discount Rate for Application in Grid Investment Test Final Report, SAHA International, 30 November 2004. The EC (and its consultants) make reference to difficulties encountered with application of the regulatory test in some circumstances. In particular, we note that Frontier Economics has suggested these difficulties can be overcome by more detailed prescription (by the EC) of matters that must be considered in applying the GIT. While we agree that clarity about such matters is essential, we also note it is essential that the mechanics of the GIT be made entirely consistent with the fundamental discipline of Cost Benefit Analysis on which both the GIT and regulatory test are based. It is beyond the scope of this report to address these issues, although, we have done so in a recent paper prepared for submission to the Australian Energy Markets Commission s review of the Regulatory Test principles. (see: Comment on MCE Proposal for a Regulatory Test Rule Change - A report prepared for the Energy Users Association of Australia, Marsden Jacob Associates, 20 June 2006). Explanatory Paper - Grid Investment Test, Electricity Commission, 3 December 2004 (EC GIT Explanatory Paper) and Draft decision on Transpower s Auckland 400kV grid investment proposal for the purposes of consultation under section III of part F of the Electricity Governance Rules 2003, Electricity Commission, 27 April 2006 (EC Draft Decision). EC Draft decision, p. 15. 4

As the future is by definition uncertain, the expected net market benefit (or the expected net market cost) is an expected (or averaged ) quantity. 9. The EC does not advance any arguments to demonstrate why an "averaged" value is any more robust than any other. But this view suggests that a factor in assessing/ranking the proposals (i.e. the discount rate) should also be based on the "averaged" value of parameters used to estimate the discount rate - rather than "cautious/conservative/upper range" values traditionally adopted by regulators/utilities elsewhere when developing their estimates for WACC. 10. The EC makes no comment on this particular issue. In respect of the discount rate, the EC says: 6 6.10.1 The GIT requires the discount rate to be used in all NPV calculations to be: (a) a discount rate determined by the Commission; or (b) if the Commission has not determined a rate, equivalent to a pre-tax real rate of 7%. 6.10.2 In the GIT Explanatory Paper, the Commission set out reasons why a default value of 7% was chosen. 6.10.3 Transpower has applied a discount rate of 7% in its Proposal. 6.10.4 The Commission considers that 7% is an appropriate discount rate to be applied to NPV calculations in this application of the GIT and has therefore applied that rate to the NPV analysis. Under the assumptions used in applying the GIT, this translates to an approximately 10% pre-tax nominal discount rate. This discount rate is subjected to a sensitivity analysis by applying the alternative rates of 5% and 9%. 11. The EC s GIT Explanatory Paper says: 7 The Commission proposed using Transpower s weighted average cost of capital (WACC) as the initial discount rate to be used in the GIT. This was due to concern at the practical difficulties associated with unambiguously determining appropriate figures for particular projects and the limited time available to research alternative private WACC figures. 6 7 EC Draft Decision, p. 43. GIT Explanatory Paper, p. 19. 5

12. However, we note that neither the EC nor Transpower has provided any details of how the 7% figure was derived. There is no reference to the capital asset pricing model (CAPM) in the EC s draft decision, much less any mention of the version of pre-tax CAPM or input parameter values used to derive this figure. 13. In its October 2004 submission on the draft GIT, Transpower offers the following explanation related to derivation of the 7% value: 8 Transpower recommends that initially a long run pre-tax real equivalent of Transpower s cost of capital should be used as the discount rate for the GIT. Transpower has applied a discount rate of 7% as an estimate of this figure in its analysis to date. Ultimately, the Commission should be endeavouring to establish a number of discount rates to be applied to the cost and benefit streams of various investments. 14. Transpower also offers limited information relating to derivation of this value. It was based on: 9 an annual WACC calculation based upon observed rates for government bonds in March/April of the prior year (i.e. it is a backward looking estimate based on actual government bond rates, which is presumably used as a proxy for the risk free rate); the expanded Brennan-Lally version of the CAPM; unspecified assumptions regarding Transpower s leverage and debt margins; factors important to a private investor such as taxation and funding mechanisms being irrelevant meaning that project costs and benefits should be prepared on a pre-tax basis and that the discount rate should be estimated in pre-tax terms; costs and benefits expressed in real terms, with the discount rate also expressed in real terms, to avoid the GIT systematically undervaluing proposals. 15. In respect of application of the discount rate, Transpower offers the following suggestions and recommendations: 10 8 9 10 Submission to the Electricity Commission on the draft Grid Investment Test, Transpower New Zealand Ltd, October, p. 16, Transpower 2004, pp. 16-17. Transpower 2004, pp.. 16-17 6

initially a long-run pre-tax real equivalent of Transpower s cost of capital should be used as the discount rate for the GIT (although Transpower did not make such an estimate); the EC should be endeavouring to establish a number of discount rates to be applied to the cost and benefit streams of various investments which meant that applying a single discount rate will introduce selection bias 11 ; and using a historical measure of the cost of capital creates some problems for the GIT analysis because the point in time nature of the estimate makes it volatile and an alternative approach is to establish a long-run forward view of the discount rate that does not vary year on year due to short-term movements in debt markets. 12 16. In effect, Transpower suggested the EC adopt a discount rate with a value of 7% on the basis that this was pragmatic given the EC s timetable for approval of the first GUP and no other value was available. 17. We note that Frontier Economics used the quote below in respect of the actual value of the Discount Rate: 13 It should be noted that the use of a private rate of return will typically not change project rankings. As noted by Mishan in his seminal text on cost-benefit analysis: '... the use of a market rate of return on private investment, rather than the social rate..., though it will of course affect the magnitudes of the terminal (or present) values, it is not very likely to make any difference to the resultant ranking of the public projects... the alternative investments profiles would have to be markedly different, and the divergence between the ideal [social rate of return] and the market rate of return on private investment would have to the quite startling, for the use of the market rate of return to generate a different ranking of public projects than that which would result from the use of [social rate of return].'. 18. We accept that Mishan s view typically reflects practical outcomes from application of cost-benefit principles to investment analysis provided 11 12 13 Although Transpower makes no suggestions about what range of discount rate values could be adopted, it does note that the GIT involves comparison of projects that involve fundamentally different assets (including large-scale generation, transmission, distribution, renewable energy sources etc. Transpower suggested the 10-year view provided by using Government Bonds as the risk free rate is not a sufficiently long term view, as it covers only 50% of the EC s proposed planning horizon but Transpower did not indicate how a longer term discount rate might be developed. Draft Grid Investment Test Final Draft Discussion Paper, Frontier Economics, June 2004, p. 28 with quote sourced from p. 235, Cost-Benefit Analysis, EJ Mishan, 1976, Praeger Special Studies, New York. 7

adequate attention is given to ensuring these principles also guide preparation of cost/benefit value streams and competing alternative options are reasonably comparable. 19. Frontier also expressed a view that: 14 Under this approach (application of modern finance theory to address uncertainty), the discount rate would be the private discount rate applying to an investor in electricity transmission assets. This is likely to be higher that the regulatory weighted average cost of capital (WACC) applicable to Transpower, which may not reflect the risk of transmission investment to customers. 20. This is consistent with the ACCC's view, which is that the discount rate lower boundary should be the regulated cost of capital. 15 21. The ACCC also said: 16 On the use of either a real, nominal, pre-tax or post-tax discount rate, the ACCC believes that, in line with VENCorp s suggestion, the guiding principle should be that the discount rate must be consistent with the cash flows being discounted. This is consistent with generally accepted finance principles. The ACCC disagrees with Transend that the discount rate used in the regulatory test should reflect the regulatory WACC for the respective TNSPs. The ACCC considers that the discount rate adopted for the purposes of a regulatory test evaluation should be a commercial discount rate in order to ensure network and non-network investments are compared on a competitively neutral basis. The discount rate used in an assessment should be consistent with the opportunity cost of capital of an investment in electricity infrastructure. The ACCC believes that the regulatory WACC might reasonably be considered the lower boundary of the discount rate but not the mean value around which sensitivity testing is conducted. The ACCC has amended the regulatory test to ensure that it the regulatory WACC can only be considered a lower boundary in a regulatory test assessment. 22. This view is not held by the EC s consultant SAHA International, which proffered the view that: 17 14 15 16 17 Draft Grid Investment Test Final draft Discussion Paper, Frontier Economics, June 2004, p. 29. Clause 15(c), ACCC Regulatory Test. Decision - Review of the Regulatory Test for Network Augmentations, 11 August 2004, p 48. Discount Rate for Application in Grid Investment Test Final Report, SAHA International, 30 November 2004, p. 5. 8

the use of an investor centric WACC (taking into account the riskiness of future income streams) is appropriate for the GIT. 23. However, SAHA qualified this view by referring to challenges associated with effort involved in consultation to derive an appropriate rate for a private sector WACC and cautioned that the additional effort required to develop (presumably a formal estimate) of multiple rates may have negligible benefit 18 (in the context of the GIT). 24. Notwithstanding this expression of caution, SAHA noted that the discount rate: 19 has a simple, twofold purpose: o Ranking of projects; and o Ensuring the net benefit of the selected project is positive. and recommended that the EC adopt Transpower s WACC (in effect as a proxy value that reflected a weighting of social and commercial objectives) as the default discount rate. 20 25. As noted above, SAHA in effect, took a bet each way by noting that selection bias will be addressed by sensitivity testing 21 and also noting that development of multiple rates may foster unproductive debate. 26. Our view represents an amalgam of the views expressed by Frontier, SAHA and the ACCC and effectively supports the position adopted by the EC. The discount rate is an important parameter in cost benefit analysis. The derivation of a value for a discount rate requires substantial and inevitable judgement to be exercised. We believe it is important to select a default discount rate for the GIT with reference to an efficient WACC applicable to Transpower and adopt a range of values that might be reasonably judged by the EC to appropriately test the sensitivity of differing cost/benefit value streams to changes in that rate. Accordingly, we would recommend that the 18 19 20 21 SAHA 2005, p. 5 SAHA 2005, p. 9 SAHA s rational for this recommendation was that GIT rankings are not overly sensitive to discount rates; the use of Transpower WACC is consistent with the objectives of the GIT and GPS; the Transpower WACC is a good approximation of such a rate (a weighted average of the social opportunity cost and social rate of time preference) given Transpower s numerous roles in the electricity sector as set out in its Statement of Corporate Intent. We would also note it complies with the GPS by not discriminating against Transpower; its use would be consistent with a suitable level of funding if transmission alternatives were to receive regulated income; and its use is supported on the grounds of pragmatism given the imperatives of grid planning and investment. (see: SAHA 2005, p. 10). SAHA 2005, p. 10 9

EC adopt a default discount rate that is equivalent to its estimate of an efficient WACC applicable to Transpower. 22 27. We note that the EC has not so far considered the cost recovery issues that arise once the GUP is approved. However, Section 17 of Part F of the Electricity Governance Rules (EGRs) Consequence of approval of grid investment or expenditure specifies: 17.1 Recovery of investment costs by Transpower Approved costs incurred by Transpower in relation to an approved economic or reliability investment, or interim grid expenditure approved under rule 16, (irrespective of when they were incurred) are recoverable by Transpower from designated transmission customers on the basis of the transmission pricing methodology and are to be paid by designated transmission customers accordingly. 17.2 No review of investment or expenditure decisions Approval by the Board of grid investment or expenditure may not be revoked or amended except with the consent of Transpower. 17.3 Board to give notification to the Commerce Commission Within not less than 10 Business Days of approving an economic or reliability investment, or interim grid expenditure approved under rule 16, the Board must advise the Commerce Commission of its decision. 28. That is, Section 17 provides an explicit regulatory guarantee for recovery of the costs of Transpower s GUPs (if the EC approves the GUP). This form of regulatory guarantee is consistent with practices in Australia (arising from application of the regulatory test) where the ACCC has generally allowed resulting investments to be rolled into the regulatory asset base without further scrutiny provided the relevant network service provider can 22 We note that the choice of discount rate in a cost benefit analysis is a potentially contentious subject. Lind, R. (1990) A Primer on the Major Issues Relating to the Discount Rate for Evaluating National Energy Options, in R. C. Lind, ed., Discounting for Time and Risk in Energy Policy (Washington, D.C., The Johns Hopkins University Press) provides an overview of problems and issues that may arise. In the simplest theories the choice of a discount rate is straightforward: it is the consumer's rate of time preference equal to the marginal productivity of capital, in which case the market rate of interest is the appropriate rate of discount. Depending the type of project the discount rate could be anywhere between the social cost of capital applicable to the public sector and a commercial cost of capital applicable to private sector nonnetwork investment alternatives. A 'social' cost of capital might be no higher than the risk-free rate (adjusted appropriately for tax and inflation). The WACC of a non-network option investor would be a lot higher, particularly when some of the 'options' could be investments that deliver only a 'window' of value (as an alternative to a 'regulatory guaranteed' investment by Transpower), with recovery of the remaining investment value being in a fully competitive market environment. Such an example might be (say) a large co-gen plant that could defer investment by Transpower for (say) 3-5 years. 10

demonstrate that the actual investment was prudent (and, in effect, consistent with good electricity industry practice ). 23 29. However, we also note that in respect of other investments in regulated energy assets, there have been no specific examples where Australian regulators have re-optimised new investments downward. 24 Regulators have generally accepted (either explicitly or implicitly) that actual expenditure was prudent and efficient and allowed the full actual investment cost to be rolled into the opening regulated asset values even where actual expenditure has been substantially above forecast levels. 25 This also provides a substantial form of regulatory guarantee that the full actual investment cost will be recovered through prices (through both a cost of capital and depreciation). In one case at least, an Australian regulator has gone further by providing an explicit guarantee that all investment costs that were deemed to be prudent at the time the investment decision was made would be recoverable even if the asset was subsequently stranded by changes in usage. 26 30. In the next sections of this paper we suggest appropriate values that could reasonably adopted by the EC to estimate an efficient WACC for Transpower and provide logical arguments to demonstrate why these values are consistent with the EC s contention that inputs to the GIT analysis should be based on averaged values. 23 24 25 26 There has, however, been one notable exception, which was TransGrid s investment in the Sydney MetroGrid project. The investment analysis (i.e. application of the regulatory test) for this project was undertaken by TransGrid to identify a particular option estimated to cost $142.5 million (1999 AU dollars). The actual cost of the project was $276.5 million (nominal) excluding claims against TransGrid that totalled around $40 million. The ACCC s opinion was that, for the MetroGrid project, TransGrid conducted inadequate analysis of the investment choices available to efficiently meet the investment need. In addition, TransGrid failed to respond appropriately to information that the actual project would cost considerably more than envisaged at the time of the regulatory test assessment. Therefore, the ACCC concluded that TransGrid did not demonstrate that all of the investment in the MetroGrid project was prudent; and subsequently re-optimised the allowed cost downward by $30.84 million (nominal) (See, pp. 13-14, NSW and ACT Transmission Network Revenue Cap TransGrid 2004 05 to 2008 09, ACCC, 27 April 2005). By contrast, there have been some cases where initial regulatory asset values were re-optimised downward in application of a depreciated, optimised replacement cost (DORC) asset value methodology. A reasonable (and short) summary of the various regulators treatment of this issue is contained in the Queensland Competition Authority 2005 decision on electricity distribution pricing (See: pp. 58-59, Regulation of Electricity Distribution Final Determination, QCA, April 2005). We also note that this practice (of effectively providing regulatory guarantee for full recovery of initially prudent investment) is consistent with that of UK regulators in the energy and water sectors. Review of Gas Access Arrangements - Final Decision, Victorian Essential Services Commission, October 2002, p. 147. 11

3. Update of WACC over time 31. Before discussing how the WACC may be updated over time, it is useful to briefly set it out in its most basic formulation. The WACC is commonly calculated as the weighted cost of debt and equity: E D WACC = k + ( 1 T) E + D E + D e k d (1) where: E is the market value of equity, D the market value of debt, E+D is the market value of the company, k e the cost of equity, T the effective tax rate, and k d the cost of debt. 32. The cost of debt, k d, should reflect the interest rate that lenders would require for lending their money, i.e. the risk free-rate adjusted to reward lenders for the risk that the borrower will default. The cost of debt is normally estimated as the sum of the risk-free rate and debt premium, i.e. the premium that should be added to the risk free rate to achieve and estimate of the cost of debt. 33. The amount of debt relative to the market value of the firm is called the leverage ratio or level of gearing. As interest is paid before tax, the cost of debt is normally quoted before tax or pre-tax. When calculating the post-tax cost of debt, the tax deductibility of interest payment must be taken into account. To adjust for this effect (commonly referred to as the tax shield) the cost of debt is multiplied by one minus the tax rate in the formula above. 34. According to CAPM, the cost of equity can be calculated as where: k e j f j [ E( R R ] = E( R ) = R + β ) (2) E(R j ) is the expected return on asset j; R f is the risk-free rate; β j measures how sensitive asset j is to movements in the market portfolio; and E(R m ) is the expected return on the market portfolio. [E(R m ) R f ] is the market risk premium, in practice often referred to as the Market Risk Premium (MRP). m f 12

35. We therefore have a wide array of different parameters that may vary over time, including the risk-free rate, beta value, leverage ratio, tax rate and MRP. As the financial market fluctuates, so will many of these input parameters. Hence, a case could be made for re-estimating the cost of capital regularly to ensure decisions are taken on the most up-to-date data. However, to update and collect information on all these parameters is no small task and some of them are unlikely to change much in the short to medium term. This is particularly the case of the MRP. Likewise, there may also be other considerations such as to provide businesses with a stable and predictable framework on which they can base their investment decisions and hence a reluctance to revise estimates too often. 36. Notwithstanding the above, we believe that any estimated discount rate for the GIT should be based on the most recent estimate. That means that as a minimum the most recent risk-free rate and debt premium should be used. In addition, recent benchmark estimates of beta values should be consulted and compared with historic values. We do not believe that the MRP need be reestimated. Likewise the optimal leverage ratio 27 is unlikely to change much over time, although this should ideally be investigated. 27 The optimal leverage ratio is the capital structure that minimises the cost of capital and hence maximises the value of the company. 13

4. Use of CC WACC 37. In our view it would be appropriate for the EC to have regard to the CC s deliberations on the WACC. First, the CC has applied a fairly consistent approach to estimate the WACC across its area of operation. Secondly, much analysis has been conducted in the development of the framework. MJA has on several occasions since 2003 been involved in consultations related to the estimation of the WACC. Each time the WACC has been a hotly-debated topic and over time the CC has made many refinements to the methodology. As such, the CC s approach represents a rational starting point for the estimation of the WACC for the GIT and adoption of the approach by the EC ensures regulatory consistency and certainty across all forms of electricity assets. 28 4.1. The CC s approach 38. The CC adopts the following nominal post-tax definition of WACC: WACC = k ( 1 L) + k (1 T L (3) e d c ) where k e is the cost of equity capital, k d the interest rate on debt capital, T c the corporate tax rate (assumed to be 33%), and L the leverage ratio (or gearing level). 39. In addition, k d is estimated as the sum of the current risk-free rate (R f ) and a premium (p) to reflect marketability and exposure to the possibility of default, i.e., k = R p (4) d f + 40. Thus, the WACC is defined as a weighted average of the cost of equity and the (after tax) cost of debt. 29 41. There is no unique definition of WACC but the definition adopted by the CC is that commonly applied in regulation, valuation and investment decisionmaking in New Zealand. 28 29 The CC has also published Draft Guidelines on the calculation of the cost of capital. The intention of these guidelines is to help parties understand the CC s approach to estimating the cost of capital in performing its regulatory responsibilities. As note in the previous section, as interest is paid before tax, the cost of debt is normally quoted pre-tax. 14

42. Estimation of the cost of equity is the most difficult and most disputed element in the estimation of the WACC. The CC uses the simplified Brennan-Lally form of the CAPM. It is a simplified version because the CC adopts simplifying assumptions with respect to tax that have the effect of reducing that the CAPM to the following formula: 30 k e = R ( 1 T ) + ϕβ, (5) f I e where T I is the average (across equity investors) of their marginal tax rates on ordinary income, φ the market risk premium, and β e the beta of equity capital. 43. This is referred to in New Zealand as the post-tax form of the CAPM in which the market risk premium, MRP, is given by: ϕ = k R 1 T ), (6) m f ( I where k m is the expected rate of return on the market portfolio. 44. According to Lally (2005, p. 9) equation (5) is also used by most ELBs, with Transpower using a modest variant on equation (3) along with a value for T I of 0.20. As noted in section 2, no specific information has been provided on the variant used by Transpower. 45. By using the expression post-tax, the CC means its estimates of the WACC are estimates of the returns that investors would expect to receive from investments in a firm (for which the WACC is calculated) after it has paid tax on its profits. Nevertheless, these returns as estimated by the post-tax CAPM reflect the underlying structure of personal taxes. 46. It is important to recognise which particular form of the CAPM is being used, as different forms have different formulas for the MRP and failure to recognise this can lead to mismatching in estimation of the MRP. The MRP used in the Brennan-Lally framework is generally higher than that used in CAPM used in Australia. 47. Although we acknowledge that the CC s framework model is generally theoretically sound, we note that the CC s application of Brennan-Lally model fails to recognise that New Zealand is now increasingly integrated with world capital markets. The implication being that any assessment based 30 It is assumed that capital gains taxes are zero, that firms attached maximum imputation credits to their dividends, and that shareholders can fully utilise the imputation credits. 15

solely on a domestic model will inflate that WACC estimate if no judgements are made of the likely effects of increased integration. 48. We also note that regulators in Australia and the UK are increasingly moving to the Vanilla WACC formula, which ignores tax impacts within the CAPM, as discussed below. 4.2. Changes needed to the WACC 49. The discount rate specified by the EC is real, pre-tax. This differs from the CC s approach in that it is specified in nominal, post-tax terms. 50. In order to compare the two versions of the WACC, we therefore need to establish some common ground. As we discuss below, conversion between different WACC specifications is no easy task and has been subject to much dispute. This is particularly the case for the treatment of tax. In the following we discuss the treatment of taxation and real and nominal values and provide some examples of simple conversion formula that have been used in other jurisdictions. 4.2.1. Treatment of tax 51. Post-tax forms of the WACC assume that company taxes are treated as a cost separate to the cost of capital. Pre-tax forms of WACC account for the cost of taxation directly in the WACC estimate and hence company tax liabilities are included in the return on capital. 52. A pre-tax WACC will have a higher value than a post-tax WACC. Since taxes are levied on profits, a company must have higher rate of return on its assets pre-tax in order to provide the required post tax returns. In theory, both definitions will yield the same results provided that the definition of cash flows or costs and benefits are consistent with the definition of WACC applied and the tax rate used to gross up the required post-tax return to the required pre-tax return is the effective tax rate paid by the entity. However, practical application suggests that this is far from the case due to the complexity of tax systems and how they are applied. 53. In Australia, regulators have adopted two broad approaches to date: transform the post-tax WACC into a pre-tax WACC (reflecting an assumption about the effective tax rate of the entity), thus making an allowance for tax by using a higher regulated WACC. include an allowance for the cost of tax directly in the cash flows (or revenue) of the regulated entity, based upon an explicit projection of the 16

taxation liabilities for the activity under consideration and use the Vanilla WACC formula which excludes tax impacts from the CAPM. 54. The first methodology has the benefit of computational simplicity and has been advocated on the grounds that it avoids the need to explicitly add into the cost of service calculation an amount to compensate for tax obligations of the business. It is therefore less intrusive, leaving the business to manage its own tax affairs. 55. It does, however, have a number of problems, the most important of which is that no simple transformation method can capture the complexities of the New Zealand tax system. For example, it must take account of rules on the treatment of tax returns to debt, tax returns to equity and detailed provisions of the tax system for capital allowances, investment credits, imputation etc. There has also been substantial controversy about which of the numerous alternative transformation methods provide the best estimate (see below). Moreover, it is very difficult to deduce the assumptions made about the taxation system from the simple transformations, which has further exacerbated the controversy. 56. For these reasons the Vanilla WACC has gained popularity among Australian and UK regulators. The Vanilla WACC has the advantage of (relative) simplicity and (relative) transparency because it allows the cost of tax to be estimated separately and treated externally to the CAPM/WACC formulae. This allows simpler formulae to be employed in the estimation of WACC and more transparent treatment of tax costs, particularly where tax laws allow regulated utilities to gain significant efficient tax benefits. 57. The Vanilla, post-tax WACC approach was adopted by the UK water regulator Ofwat in the lead-up to the first independent price review in 1994 because, at that time, the newly privatised water companies were exempt from income taxes. The Office of the Regulator-General in Victoria (now the Essential Services Commission) adopted the Vanilla WACC approach in the lead-up to the first independent review of electricity distribution prices in 2000 having initially adopted the more complex pre-tax approach for the initial review of gas distribution prices. 58. The Vanilla WACC approach was subsequently adopted by the Queensland Competition Authority in 2000, the ACCC in 2000, the Office of the Tasmanian Energy Regulator in 2003, the UK energy regulator Ofgem in 2004 and the Essential Services Commission of South Australia in 2005. 31 31 Despite adopting a more-or-less common (and relatively simple) approach, differences remain in the presentation of results. For example, the Victorian ESC retains the use of real, post-tax Vanilla WACC 17

59. Nevertheless various conversion formulae exist that try to reconcile pre-tax and post-tax WACC calculations. One of the most simple, called the forward transformation, is: Pre - tax WACC = Post - tax WACC 1 T (7) where T is the effective tax rate. 60. Clearly, to estimate a pre-tax WACC using this method requires a single effective company tax rate. This is problematic as it is difficult to accurately estimate a single effective tax rate that will reflect a company s taxation liabilities. In addition, what we need ideally for the cost benefit analysis is also the tax profile looking forward which will inevitably vary from year to year. On the other hand, the use a post-tax WACC necessarily entails a direct estimation of the tax liabilities that could create additional complexity in the cost-benefit analysis that may be unwarranted given the primary objective of such analysis is to rank alternative options through a process that is not generally highly sensitive to changes in the discount rate. 61. To summarise, there is necessarily a trade off between the additional information requirements and analysis necessary in the post-tax version compared with using the simple transformation shown above. An alternative could be to introduce more complex formulae to attempt to overcome some of the shortcomings of the simple formula. 62. Davis (2004) 32 shows that the process of generating a pre-tax discount rate is complex and liable to error. In addition, to our knowledge, none of the conversion formulae commonly proposed are complex enough to account for all of the effects discussed above and may ultimately simply create additional fruitless discussion. 63. For example, in the lead-up to the 1994 water industry price review in the UK, Ofwat put forward a complex tax wedge model that would take account of capital allowances, inflation and the dividend cover ratio. Considerable debate ensued between Ofwat and the water industry about the inputs in the tax wedge model and in particular the appropriate assumptions about dividend cover and gearing. The debate was never really resolved because Ofwat decided not to use the model in its 1994 review. values, the ACCC and ESCoSA report nominal, post-tax Vanilla values and the QCA reports its results in nominal post-tax terms but uses the name Officer WACC3, which refers to the version (number 3) of the CAPM formula contained in a paper developed for QCA by Professor Bob Officer. 32 Davis (2004), The Design of Regulatory Pricing Models for Access Arrangements: Inflation, Tax and Depreciation Considerations, Accounting Research Journal, Vol 17 No. 1. 18

64. For the purpose of this report we adopt the formula 7 above. 4.2.2. Nominal vs. real 65. The choice to use either a nominal or real WACC depends upon whether the model using the WACC is specified in real or nominal terms. 66. Under the nominal approach, values of all costs (including asset values) are expressed in money of the day terms and a nominal WACC is used to calculate returns on assets. Under a real approach, values of all costs are expressed in constant price terms and a real WACC is used. All other things being equal, the two approaches are mathematically equivalent and the choice is one of preference for modelling. 67. Returns on equity and debt can be converted from real to nominal values (and vice versa) using the relationship between nominal and real interest rates given by Fisher Equation: 1 + nominal WACC = (1 + real WACC) (1 + inflation rate) (8) 68. Different views exist on the appropriateness of real and nominal specifications. An argument for a nominal specification is that many costs are fixed in nominal terms, most particularly depreciation and interest and the tax system is defined in nominal rather than real terms. However, as Transpower notes (in respect of the forecasts needed to undertake the GIT): 33 Real values are easier to estimate, model and evaluate so the approach is more transparent during the assessment phase. However, capital expenditure approved would differ from that included in the GIT, and ultimately recovered from consumers, so transparency is harder to achieve during later phases of the project. Nominal values are more complex to model, and do not completely address transparency concerns during implementation and operation of a project because of the difficulty of isolating nominal price movements. 69. Essentially the same difficulty faces regulators during price reviews, and their responses are (approximately) equally mixed. In Australia, some regulators undertake the price review using real values (e.g. the Victorian Essential Services Commission (ESC)) and some using nominal values (e.g. the Queensland Competition Authority (QCA)). In our view, the ESC s deliberations are easier to understand and follow (i.e. more transparent) than the QCA s, but the ESC requires financial performance information to be 33 Submission to the Electricity Commission on the draft Grid Investment Test, Transpower, October 2004, p. 18. 19

reported in real terms which makes comparisons between forecasts and actuals more confusing (i.e. less transparent). 70. Accordingly, we are inclined to endorse Transpower s view on this particular issue. Use of either nominal or real values WACC, or the discount rate, will make no difference to the outcome of the GIT provided all inputs are expressed in the appropriate terms. The chosen discount rate(s) must be consistent with the cost and benefit streams being analysed. If costs and benefits are expressed in real terms, then the discount rate should also be expressed in real terms, otherwise the GIT will systematically undervalue proposals. 34 4.2.3. Conversion from nominal post tax to real pre-tax 71. To convert a post-tax nominal WACC to a pre-tax version two steps are required: conversion from post-tax to pre-tax and from nominal to real. 72. Assume a 6% post-tax nominal WACC. To calculate a pre-tax real term we would gross up by the factor 1 minus the corporate tax rate. Assuming the tax rate is 33% (i.e. the effective tax rate is identical to the statutory tax rate) this would yield a nominal pre-tax WACC of 8.96%. The next step would be to adjust the nominal pre-tax WACC as follows to derive a real pre-tax WACC: 35 real pre - tax 1 + nominal pre - tax WACC WACC = 1 1+ inflation rate (9) If the expected inflation rate is 3%, then the real pre-tax WACC is 5.78% using the number above. While 3% represents the upper limit of the Reserve Bank s target over the medium term, current inflation is now 3.3%. 36 73. An alternative approach is the reverse transformation, under which the posttax nominal WACC is first deflated to obtain a post-tax real WACC, and then converted to a pre-tax real WACC by grossing by a factor of one minus the tax rate. The reverse transformation results in a lower WACC. Assuming a 6% post-tax nominal WACC as above, the reverse transformation results in a pre-tax real rate of 4.53% which is considerably lower than 5.78%. 74. The confusion created by use of these WACC transformations has not been dealt with well by Australian or UK regulators, and has been one of the 34 35 36 Op Cit, p. 17 This is essentially a re-write of the Fisher equation. See current Policy Target Agreement, 17 September 2002. Current inflation is 3.3% (March quarter 2006). RBNZ A3 Incomes and Prices : http://www.rbnz.govt.nz/statistics/econind/a3/data.html, last updated 18 May 2006, accessed 20 June 2006. 20

drivers for the switch to the Vanilla WACC approach. Rather than explaining this outcome in a transparent manner, Australian regulators have generally ignored the difference and selected the forward transformation value. 75. Neither approach will produce the correct value as they are approximations, but the forward transformation is likely to overstate the true value while the opposite is true of the reverse transformation. 4.3. Estimation of the WACC 76. In the following we summarise the parameters used by the CC to estimate a WACC for Unison s ELB and convert this estimate to pre-tax and post-tax values using the above formulae considering updated parameters where relevant. 77. We note that in order to properly compare the CC s WACC for Unison with the real pre-tax discount rate of 7% we would need to understand how Transpower originally derived its WACC value to begin with, including the formulae and parameter values. However, as noted in previous sections we have been unable to find this information. Transpower has indicated that it derives an estimate of WACC on a yearly basis based upon observed rates for government bonds in March/April of the prior year using the expanded Brennan-Lally version of the CAPM, but this is insufficient to derive an estimate. 37 78. The CC used the following input parameters to derive a WACC for Unison. 37 Submission to the Electricity Commission on the draft Grid Investment Test, Transpower, October 2004, p. 17. 21

TABLE 1: CC ESTIMATES FOR UNISON* Unison Risk-free Rate 6.31% Post-Tax MRP 7.0% Equity Beta 0.667 Asset Beta 0.4 Cost of Equity 8.89% Debt Premium 1.20% Cost of Debt (pre-tax) 7.51% Gearing 40.0% Corporate Tax Rate 33.0% Investor Tax Rate 33.0% Post-tax Nominal WACC 7.35% * Table cells shaded grey are input values. Source: MJA Analysis of CC (2005) 79. As indicated in section 3, we believe the risk-free rate and debt premium should as a minimum be updated. However, as noted below update of the risk-free rate requires little judgement on the basis that regulators (in the UK, Australia and NZ) universally adopt a recent average of the (relevant) multiyear Government Bond Rate as a proxy for the risk-free rate. 38 Considerable judgement is required to establish values for other parameters and it would be inappropriate for us to attempt to judge the CC s position today for Unison or Transpower. Rather, in section 5 we offer our own view and updates to the parameters for Unison. 80. The CAPM provides no guidance as to the appropriate maturity of the riskfree rate. Regulators have typically chosen a maturity that matches: the technical or economic life of the assets used in providing the regulated service, on the basis that this reflects the planning horizon of investors in those assets; or the duration of the regulator s determination ( the regulatory period ), given that the risk-free rate will be adjusted in any subsequent reset. 81. The CC uses the risk-free rate equivalent to the price setting period of 3 years, i.e. a three-year maturity is chosen (as linear interpolation between the two 38 We note that Transpower has expressed a view that a long-run equivalent to Transpower s cost of capital should be used as a discount rate for the GIT. Transpower is also of the view that the 10-year view provided by using Government Bonds as the risk-free rate is not a sufficiently long-term view, as it covers only 50% of the Commissions proposed planning horizon. (See p. 16 and Footnote 9, p. 17, Op Cit). However, Transpower does not explain how it might derive an estimate for a longer term risk-free rate for which no NZ Government Bond market exists. Given the primary objective of a GIT analysis (to rank alternative investment options), we do not see much value in looking beyond current Bond market data. 22