THE COST OF CAPITAL FOR THE 2016 BNE PEAKING PLANT A NOTE PREPARED FOR THE REGULATORY AUTHORITIES SEPTEMBER Cambridge Economic Policy Associates

Transcription

1 THE COST OF CAPITAL FOR THE 2016 BNE PEAKING PLANT A NOTE PREPARED FOR THE REGULATORY AUTHORITIES SEPTEMBER 2015 Submitted by: Cambridge Economic Policy Associates

2 CONTENTS 1. Introduction Context Purpose Methodology for calculating the BNE WACC CEPA initial report Consultation comments CEPA comments Cost of debt CEPA initial report Consultation responses CEPA comments Gearing CEPA initial report Consultation responses CEPA comments Cost of equity CEPA initial report Consultation responses CEPA comments Conclusions Updated ranges Cross-check to recent regulatory consultations Selecting a point estimate IMPORTANT NOTICE This report has been commissioned by the Northern Ireland Authority for Utility Regulation (NIAUR) and the Commission for Energy Regulation (CER). However, the views expressed are those of CEPA alone. CEPA accepts no liability for use of this report or any information contained therein by any third party. All rights reserved by Cambridge Economic Policy Associates Ltd.

3 1. INTRODUCTION CEPA, working in association with Ramboll, has been commissioned by the Northern Ireland Authority for Utility Regulation (NIAUR) and the Commission for Energy Regulation (CER), collectively the Regulatory Authorities (RAs), to provide an estimate of the fixed costs of a Best New Entrant (BNE) peaking plant for the calendar year Context In our initial report to the RA s we provided an estimate of the fixed costs which a rational investor would be likely to incur in constructing and operating a peaking plant to enter the Single Electricity Market (SEM) in A key component of this was an initial estimate of the BNE plant s Weighted Average Cost of Capital (WACC). The ranges for the BNE s WACC proposed in our initial report are summarised in the table below. Table WACC Parameter Range for BNE 2016 CEPA initial report Republic of Ireland Northern Ireland (UK) Low High Low High Cost of Debt 1.00% 3.00% 0.75% 2.25% Risk-free rate 1.00% 2.50% 0.50% 1.50% Equity Risk Premium 4.50% 4.50% 5.00% 5.00% Asset Beta Debt Beta Equity Beta Post-tax Cost of Equity* Taxation 12.50% 12.50% 20.00% 20.00% Pre-tax Cost of Equity Gearing 60.00% 60.00% 60.00% 60.00% Pre-tax WACC Equivalent Vanilla WACC Source: CEPA Note: Cost of Equity is calculated using the average risk-free rate for both high and low cases. This does not affect the mid-point but leads to a narrower range than without this average being used. These proposed ranges for the WACC were commented on extensively by the respondents to the RA s consultation paper. The general view was that the ranges proposed - and the RAs selection of the mid-point of those ranges - resulted in a WACC that was significantly below what was appropriate for a BNE. 1

4 1.2. Purpose In this report we set out our response to the comments raised by respondents specifically on the BNE WACC and provide our updated recommended ranges based on updating the market evidence in our initial report and our review of the responses to the RAs consultation. The document is structured as follows: Section 2 considers the comments raised on the assumptions and methodology that are used to arrive at a BNE WACC; Section 3 focuses on the comments and additional evidence received on the cost of debt; Section 4 focuses on the gearing assumption that should be reflected in the WACC for the BNE plant; Section 5 comments on the responses received on the cost of equity; and Section 6 provides overall conclusions. 2

5 2. METHODOLOGY FOR CALCULATING THE BNE WACC 2.1. CEPA initial report In our initial report to the RAs, we assumed that the BNE investor was an integrated energy utility seeking to raise funding at the corporate level, but which would assess the BNE project risk and return profile on own its merits. Following regulatory precedent, we also assumed the BNE investor would have an investment grade credit rating (with a BBB rating or above) which impacts on the cost of debt assumed for that investor. CEPA has previously noted that key factors influencing the risk profile of the BNE investment include the high fixed costs of the generation plant (which magnify the effect of underlying systematic (price and volume) risks) and the stability of the plant s regulated revenue stream through the capacity payment mechanism ( CPM ) Consultation comments The majority of respondents disagreed with the assumption that the BNE investor would be investment grade rated given the gearing/capital structure which had been proposed in the consultation and the nature of the BNE (see box below). A number of respondents also argued this assumption would discriminate against SEM participants who were not rated investment grade. Box 1 Ratings considerations for a new BNE peaking plant In its response, Energia stated that assuming a BBB rating for the BNE was anticompetitive and it could inhibit new entrants from entering the market impeding the further development of competition, and discriminates against a number of investors. Viridian commissioned the Royal Bank of Scotland (RBS) to undertake illustrative analysis of the possible position of the BNE on Moody s unregulated power companies methodology grid using information available and RBS assumptions. This assessment showed gearing of below 35% would be the indicative level to meet BBB investment grade status using a benchmark greenfield plant and assumptions provided by Viridian of the implied income statement for the peaking plant. However, the RBS analysis also noted that there are a range of other factors, in addition to gearing levels, that will impact on a credit rating assessment. Respondents also stated that the electricity generation market across Europe had seen a dramatic downturn since the BNE WACC was last set in 2013, as highlighted by significant generation asset impairments across Europe, including in Ireland. A number of factors in the view of market participants had substantially increased the risks associated with investment in generation plant and, in particular, no longer supported the 3

6 assumption that the WACC applied to the BNE should be based on the assumption of an integrated utility with an investment grade credit rating. These factors included: increased market and regulatory risk related to the I-SEM; increased exposure to commodity risks; and the more general downturn in the generation market in Europe. As set out in a recent SEM Committee consultation paper, with respect the CPM under the I- SEM, this is expected to take the form of Reliability Options: The reliability holder will bid to receive a basic capacity payment from the Reliability Option Counterparty determined by result of a competitive auction. As the finer details of the CPM and expected outcomes of the auctions are still to be worked out, naturally this can be expected to impact on investor uncertainty. In the consultations we held with market participants, the feedback was also that although the current CPM is a regulated revenue stream for generators, investors would not perceive the BNE as anything like a quasi-regulated investment, nor would the investment be viewed as having a contracted revenue stream. The BNE would instead be viewed as a merchant plant, and one that will sell into an as yet untested I-SEM. Some respondents also stated that the assumptions and the methodology which had been followed to calculate the BNE WACC would also result in an average WACC for a vertically integrated utility (VIU) rather than calculating the marginal project WACC for a BNE peaking plant entering the SEM. In their report commissioned by Electricity Association of Ireland, Frontier Economics ( FE ) 1 suggested that because generation investments can be more risky than other investments in an integrated energy company, an additional generation investment would push up the WACC of the firm on average. To allow for this they suggest that the appropriate benchmark for the BNE should allow for the possibility that the new entrant is a standalone generator, with a lower credit rating than investment grade CEPA comments Methodology principles It is important to highlight the context in which the WACC that is applied to the BNE investment is set. The SEM Committee is seeking to set the price element of the CPM so as to 1 Frontier Economics (2015): Benchmarking the BNE WACC for 2016 a report prepared for the Electricity Association of Ireland 4

7 be sufficient to attract investment in a BNE peaking plant that in long run equilibrium would serve the last MW of demand. In a wholesale electricity market without a regulated capacity payment either an energy only market or a competitive auction based CPM it would instead be competition in the market that would set the clearing price for this last MW of demand, including the return demanded to compensate for the risk of the investment. The SEM Committee in contrast face the challenge that the SEM CPM must be based on a notional investment and, therefore, notional view of the efficient WACC for the investment. CEPA agrees with the consultation respondent views that it should be the marginal project WACC for the generation investment rather than the average WACC for the notional company which is used for the purpose of the BNE calculation. However, the challenge is identifying the right assumptions and methodology to calculate the building blocks of the WACC which are also consistent with the objectives for the CPM. A notional marginal cost of equity for the BNE can be priced using components of the Capital Asset Pricing Model (CAPM) - see Section 4. The challenge with establishing the marginal cost of debt and, therefore, overall the marginal WACC for the BNE project, is that it requires assumptions of how the debt would be priced for the project. In the long run equilibrium the BNE calculation is seeking to approximate, it is the efficient marginal WACC we are seeking to estimate for the peaking plant, not the average WACC for an investment grade utility. Cost of debt methodology As discussed above, we are conscious that the range for the BNE WACC needs to reflect the risk profile and, therefore, marginal project WACC of the BNE, rather than the VIU business as a whole. This is likely to be higher than the average WACC for a VIU given the other activities undertaken by the business. The difficulty is the marginal cost of debt for the BNE project is in this case non-observable. Therefore our historical approach to developing range for the BNE cost of debt has been to cross-check evidence from current spot rates for investment grade BBB rated debt to yields at issue on other investment grade rated bonds, long-term averages of investment grade borrowing costs and an implied all-in cost of debt for the BNE if recent evidence of investment grade debt spreads are added to longer-term estimates of the real risk-free rate (as opposed to spot rates on benchmark government bonds). This methodology has historically resulted in a degree of headroom or aiming up when setting a range for the cost of debt (see Figure 2.1 below) to allow for the 20-year life BNE project potentially not being able to borrow at current spot rates, mean reversion to longer term historic trends of investment grade borrowing costs (if any refinancing is required) and that the marginal cost of debt that is reflected in the project WACC given the underlying risk profile of the generation investment and current and expected average costs of debt for 5

8 Real Yield to Maturity (%) the VIU might be assumed to be higher by a VIU investor than a spot cost of debt derived from current market evidence of an investment grade benchmark. Figure 2.1 BBB investment grade benchmark and precedent on the BNE cost of debt CEPA recommended range for 2013 included headroom over spot rates for the BBB investment grade benchmark Previous BNE decision Allowed 2013 range for UK BNE CoD Eurozone BBB 10yr UK BBB 10yr Source: CEPA analysis of Bloomberg data We believe this approach should be retained for calculating the cost of debt for the 2016 BNE determination. If only a pure spot benchmark is used for the BNE, there may be a risk that in the long run equilibrium the BNE calculation is looking to approximate, the resulting capacity payments would not allow generators (including the BNE but also plant further down the merit order) to earn sufficient revenues to recover their fixed costs including the WACC. An alternative to our historic approach as the FE paper for the Electricity Association of Ireland discusses would be to look at the borrowing costs of a standalone generator as a proxy for the true risk profile of the BNE and, therefore, the required project WACC set by the investor. The CMA s recent draft findings for the energy market investigation in Great Britain (GB) states that evidence from the GB market (Drax) indicates that a stand-alone generator in the GB electricity market would probably only achieve a credit rating of BB (i.e. just below investment grade). 2 Whilst the all-island market (and the setting of the current CPM) is a 2 As presented in a number of responses to the RAs consultation, there is evidence to suggest this may also be the case in the all-island market. 6

9 different environment to the CMA s market investigation, this does provide further weight to the argument that cost of debt figure that is derived from the approach followed in previous BNE calculations, should at least be crossed-checked to evidence other than a spot investment grade benchmark. We present these cross-checks in Section 3. Cost of equity methodology As discussed above, in commenting on the nature of the BNE peaking plant investment, a number of respondents to the consultation also highlighted the challenging environment for European generation since 2013 and the increased market and regulatory risks that are affecting the all-island electricity market now and as the market transitions to a new regulatory framework. Whilst this evidence is useful context to the 2016 BNE determination, we note that for the purposes of setting the required return on equity for the BNE investment, it is only the nondiversifiable systematic risks i.e. risks that are faced by the company that cannot be diversified away from as part of an investor s balanced portfolio of assets that require a return by the investor if applying the CAPM. Based on the evidence presented by the CMA as part of the ongoing GB energy market investigation, and precedent of the range for the beta used in applying the CAPM in previous BNE WACC calculations, we believe that the range for the beta set out in our initial report is likely to be sufficient to remunerate the systematic risk associated with the BNE investment. However, as discussed in Section 5, there is still a question of where within the range the appropriate beta for the generation plant should lie. 7

10 3. COST OF DEBT 3.1. CEPA initial report As discussed in the previous section, in our initial report we relied on a BBB investment grade benchmark deflated by RPI inflation to derive an all-in cost of debt for the BNE. We proposed an initial range for the BNE cost of debt in Northern Ireland of 0.75% % and 1.00% % in Ireland Consultation responses The majority of respondents stated that the proposed ranges for the BNE cost of debt in Ireland and Northern Ireland were too low. One respondent suggested that a spread of 175bps above the risk-free rate would be a very conservative estimate of the borrowing costs that could be faced by the BNE and that a spread of 300bps (over 6-month Libor swap) would be a more appropriate assumption for a merchant generation plant. A number of responses cited analysis commissioned by the Irish Electricity Association by FE who suggested a range for the cost of debt of 2.60% % in Northern Ireland and 1.98%- 4.44% in Ireland. The FE ranges are based on a BBB benchmark 3 that is used to derive the bottom end baseline estimate of the cost of debt and a standalone generator (with a below investment-grade credit rating) for the top end estimate. FE then apply a premium to their baseline cost of debt estimates for Northern Ireland and the Republic of Ireland, a premium for new issuance and carry costs (associated with refinancing) and a forward yield premium on the assumption the SEM Committee will use the 2016 BNE to also set the CPM in A number of respondents noted that given the CPM has historically been indexed to CPI rather than RPI inflation, an estimate of CPI inflation should be used to deflate evidence on the nominal cost of debt to derive an estimate of the real cost of debt (a CPI deflator is used by FE to derive their estimate for the cost of debt for the Northern Ireland BNE). As the CMA s updated WACC paper as part of the energy market investigation shows 4, this can utilise nominal yields deflated by CPI or if real RPI indexed yields are used, the differential between CPI and RPI inflation can be used. 3 An investment grade credit rating is BBB- and above

11 Real yield (%) 3.3. CEPA comments Given the issues discussed in the previous section, we have updated our review of market evidence on the cost of debt Northern Ireland We agree with consultation respondents that where the CPM pot is indexed to CPI it should be CPI that is also used to deflate nominal all-in rates. However, our understanding from discussions with the RAs is that the SEM Committee propose for the forthcoming BNE calculations to use RPI rather than CPI as the indexation basis for the Northern Ireland BNE. Whilst it may be likely that the financial performance of the BNE would be affected by general price inflation, as captured by CPI, the calculation of the WACC should be consistent with the inflation index which is chosen. Therefore, as with our initial report, we have updated evidence on the cost of debt on an RPI basis. Update of current all-in cost of debt evidence Figure 3.1 below shows historic real yields on UK ten-year corporate debt based on a published Bloomberg index and Bank of England reported 10-year break-even inflation. Figure 3.1: Real yields on UK corporate debt of ten-year tenor Source: Bloomberg, Bank of England, CEPA analysis A Based on this analysis, the spot rate on BBB rated debt on the 10 th July 2015 was 0.68% compared to a one-year average of 0.46%. BBB 9

12 Real Cost of Debt (%) An alternative source of evidence is the iboxx indices. Figure 3.2 compares iboxx nonfinancials 10+ A and BBB rated indices, deflated by 10-year break-even inflation. Figure 3.2: Real yields on UK corporate debt of ten-years plus tenor A BBB Source: Markit iboxx, Bank of England, CEPA analysis Based on this analysis, the spot rate on BBB rated debt on the 10 th July 2015 was 1.43% compared to a one-year average of 1.19%. Aiming up from the investment grade spot benchmark As discussed in the previous section, we have also considered evidence of the additional debt premium that would be required over spot rates based on the expectation of a need for refinancing and that the marginal cost of debt assumed by the investor in setting the project WACC is higher than the spot investment grade benchmarks set out above. Table 3.1 derives a cost of debt estimate by adding a range for the debt premium to regulatory precedent on the risk-free rate. As this approach intentionally aims up from current spot rates, it is considered to include any regional, new issuance or forward yield premium rather than these needing to be mechanistically added to the underlying benchmark. As an assumption for the risk-free rate, we use an estimate of 1.25% as has been adopted by the CMA in its draft determination for the Bristol Water price review referral. The current spread to gilt rate on Bloomberg s ten-year BBB bond index is 152bps, the 1-year average 133bps, the 5-year average 168bps and the 10-year average 175bps. We therefore use a potential range for the BNE debt premium of bps noting that one respondent to 10

13 consultation indicated that a debt spread of 175bps was the minimum that would be required to attract investment into the BNE. Table 3.1 Investment Grade benchmarks on Northern Ireland cost of debt Component Low High Risk Free Rate 1.25% Debt premium 1.25% 1.75% Cost of debt 2.50% 3.00% Source: CEPA analysis of Bloomberg and iboxx data Based on the methodology and assumptions presented in Table 3.1, this could imply a cost of debt for the BNE of %. This is above current spot rates but is intended to provide a more conservative assumption (as applied in previous BNE WACC calculations) to reflect potential need for refinancing and forward movements in rates, and because the spot benchmark may well not reflect the true marginal cost of debt that would be reflected in the project WACC for the BNE. Northern Ireland premium In the 2013 BNE calculation, we recommended to the SEM Committee that a 50bps premium be added to the baseline investment grade benchmark for the UK drawing on market evidence that Northern Ireland utility bonds in particular, a NIE bond traded at a premium to comparator utility bonds elsewhere in the UK. Therefore, to continue to assume a premium in the 2016 BNE calculation there would need to be continued market evidence of such a premium today. Our initial report showed that since early 2013 the differential between NIE s 2026 bond and comparator bonds (with similar business activities, credit ratings and time to maturity) had been no more than 20bps and suggested, therefore, that there was no longer clear evidence to support an explicit Northern Ireland premium (see Figure ). 5 The chart updates the analysis from our initial report. 11

14 Figure 3.3 NIE yield compared to GB comparators Source: Bloomberg However, as part of their analysis commissioned by the Electricity Association Ireland, FE also looked at evidence of the yields on a Phoenix Natural Gas (PNG) bond 6 compared to a set of comparator bonds with a similar credit rating and time to maturity. Their analysis showed a consistent Northern Ireland premium in the case of the PNG bonds between 2009 and Given that the company was subject to a Competition Commission (CC) review 7 over this period FE suggest that the most accurate estimate of the Northern Ireland premium based on PNG bonds as the comparator is one which is estimated over the period October 2009 August 2011, which FE estimate to be 69 basis points. We have updated FE s analysis for the PNG bond using the same comparator bonds (excluding one comparator bond for Electricity North West that has now matured) up to the 14 th July 2015 in Figure 3.4 below. 6 GBP275m bond issued by Phoenix Natural Gas Finance Plc at 'BBB+' 7 Competition Commission (2012): Phoenix Natural Gas Limited price determination 12

15 Figure 3.4 PNG bond yields compared to comparable GB traded bonds CC PNGL Draft decision CC PNGL Final decision Source: CEPA analysis of Bloomberg data If the benchmark is the PNG bond rather than the NIE bond this would indicate there is some evidence to support a Northern Ireland premium but based on the time period since the CC investigation this has been lower than the 50bps previously applied for the 2013 BNE calculation (the average premium has been 41bps for the PNG bond since October 2012 or 27bps on average over the last twelve months). In our view, the market evidence no longer supports a Northern Ireland premium of 50bps as was applied in the 2013 BNE decision. However, given some albeit relatively limited continued evidence of a premium for some benchmark comparators (e.g. the investment grade rated PNG bond), a 25bps premium to the cost of debt estimates for a BBB investment grade rated benchmark could potentially be justified. Issuance fees We consider that the 20bps assumption to reflect the transaction costs of the BNE issuing new debt should be sufficient to cover all costs including carry associated with the BNE raising funding at a corporate level over the lifetime of the plant. The CC used the same assumption in its 2013 NIE price determination. Cross-check to wider market evidence on cost of debt Based on our comments in Section 2, we have also reviewed as a cross-check a range of evidence of yields on non-investment grade rated debt. The challenge with sub-investment 13

16 grade is there are no available bond indices available and the spreads to gilt and the BBB investment grade index can vary significantly between issuer and by credit rating (BB+ to B-). Table 3.2 compares benchmarks presented by FE and non-investment grade benchmark bonds of varying maturities that we were able to source 8. As the average time to maturity is shorter than for the benchmark, this would imply that the spreads are conservative (given the upward sloping nature of the yield curve). Table Spreads to iboxx BBB ten year plus index CEPA calculated spread to iboxx CEPA average time to maturity (years) FE calculated spread to iboxx BB+ +12bps bps BB +75bps bps BB- +139bps bps B+ +144bps bps B +243bps bps B- +336bps bps Source: CEPA analysis of Bloomberg and iboxx data & FE Note: These are sterling benchmarks based on fixed coupon bonds. FE judges that a 200bps uplift to be a reasonable uplift on a BBB benchmark to be appropriate and likely to capture that new entrants and existing generators may not be vertically integrated or simply not have the size needed to obtain an investment grade credit rating. They add 200bps to the BBB benchmark to derive the baseline estimate for the top end of their range for both the Northern Ireland and Ireland BNE cost of debt. Based on the evidence above, a 200bps uplift to the BBB benchmark would not appear unreasonable provided it was deemed an appropriate assumption that the marginal cost of debt for the BNE should reflect a credit rating of B+ to B-. However, it is our view that broad single B rated debt is more speculative than would be the case for the BNE investor and we focus on double BB rated debt as the appropriate cross-check to the short and longer-term evidence of the cost of debt already discussed above. We note that a BB rather than B rating also appears to have been adopted by the CMA as the appropriate benchmark for a standalone generator in its GB energy market investigation. Assuming a conservative premium of 100bps to the iboxx BBB investment grade spot benchmark as a basis for the marginal cost of debt estimate for the BNE, this could imply a real cost of debt of c. 2.50%. 8 Our analysis finds a shortage of bonds for these credit ratings with a maturity around ten years, with increasing difficulty as you move down the ratings notches. As such, we focus on a range of maturities to avoid the impact of any one particular outlying bond. The same issue applies to creating a utility-only index, where there are relatively few examples. 14

17 Overall conclusions In Table 3.3 below we bring together our views on the cost of debt for Northern Ireland and present updated ranges. The low end of the range reflects updated analysis of the all-in cost of debt for the BBB 10- year and 10-year plus indices current spot rates, deflated by break-even inflation, with a 25bps Northern Ireland premium and 20bps for new issue fees. This estimate allows for the assumption that the weighted average maturity of the debt is 10-years (but that the BNE could need to put in place debt of a longer maturity) and the possibility that debt costs may rise by 2016 as evidenced by the forward yield curve for gilts. Figure 3.5 Forward rates for 10-year UK nominal gilts Source: Bloomberg We have considered carefully the evidence submitted by market respondents and have concerns as to whether the benchmark BB to B rating bonds proposed (given their unique characteristics to the individual businesses in question) provide a suitably robust reference for the marginal cost of debt of the BNE given the VIU investment grade assumption. However, in previous calculations (as discussed in the previous section) typically the high end of our range has introduced some headroom to capture movements in rates, potential need for refinancing or that the efficient marginal cost of debt to reflect the risk profile of the BNE in the project WACC was higher than current spot rates. Based on the methodology followed in previous BNE calculations of applying evidence of investment grade debt premiums to a longer-term risk free rate estimate this would suggest that the top end of the range could be set to 2.50%-3.00%. On balance, our judgement is that the top end of the range should be set at 3.00% which would allow for some headroom / aiming up above the 10-year investment grade 15

18 benchmark, whilst being consistent with the methodology we have applied in previous calculations to capture the uncertainty of the marginal cost of debt assumption which a BNE investor would seek to reflect in the project WACC. Table 3.3 Updated cost of debt ranges - real Low High Cost of debt 1.70% 3.00% Source: CEPA Ireland Update of the all-in cost of debt evidence Table 3.4 presents evidence on the nominal all-in cost of debt for Bloomberg s European BBB rated 10-year bond index. We note that nominal yields have risen by 80bps since the analysis in our initial report was undertaken in February/March Table 3.4 Evidence of nominal debt yields Benchmark Bloomberg BBB 10yr Spot (13 th July 2015) 2.00% 1-year average 1.69% 3-year average 2.63% 5-year average 3.44% 10-year average 4.50% Source: CEPA analysis of Bloomberg data Table 3.5 below deflates the evidence in Table 3.4. For deflating nominal yields in Europe, we use estimates for HICP inflation. These estimates are the longer term (five years ahead) forecasts provided by the ECB Survey of Professional Forecasters on a quarterly basis. The latest figure for this is 1.8% Table 3.5 Evidence of real debt yields Benchmark Bloomberg BBB 10yr Spot (13 th July 2015) 0.20% 1-year average -0.11% 3-year average 0.83% 5-year average 1.64% 10-year average 2.70% Source: CEPA analysis of Bloomberg data 16

19 Ireland premium and issuance fees As FE highlight in their report, in the 2013 BNE calculation, CEPA applied a specific country risk premium for Ireland which was added to both the cost of debt and the cost of equity estimates through the risk free rate assumption used. This was informed by comparisons of yields on German government and Irish government bonds the analysis of which is reproduced and updated in Figure 3.6 below. Figure 3.6 Irish 10 year gilt minus German 10 year gilt yield Source: CEPA analysis of Bloomberg data Although this spread has narrowed significantly from when we produced our 2013 report, Figure 3.6 shows that it has increased since our initial report produced in February/March. The spot premium is currently 76bps whilst the 1yr average is 86bps. There is therefore evidence, albeit more limited compared to the BNE 2013 calculation, of a country premium for Ireland. The methodology of deriving the cost of debt from the European BBB indexes does not allow for this as this is across European countries. Cross-check to other market evidence As for Northern Ireland, we have cross-checked our analysis by considering the additional debt premium that may be required if the marginal cost of debt assumed by the investor in the project WACC was higher than the investment grade benchmarks set out above. 17

20 Table 3.6 below presents a similar analysis as Table 3.2 but for Euro dominated BB+ to B- bonds compared to Bloomberg s 10-year BBB bond index. Similarly, these cover a range of sectors and maturities, with average maturities being lower than the equivalent benchmark. Table Spreads to Bloomberg BBB corporate ten year index Rating CEPA calculated spread to Bloomberg index CEPA average time to maturity (years) BB+ +57bps 4.2 BB +117bps 4.8 BB- +141bps 5.2 B+ +315bps 5.3 B +456bps 4.9 B- +496bps 5.2 Source: CEPA analysis of Bloomberg data Note: Figures are for Euro-denominated fixed coupon bonds. There are a larger number of benchmark bonds with a lower credit rating than investment grade in our European sample including many more financials. The spread for B rated bonds relative to investment grade ranges from 315bps to 496bps whilst evidence on BB (sub-investment grade) rated bonds (which informs the top end of our range for Northern Ireland) could imply an uplift premium of 100bps or more. Overall conclusions In Table 3.9 below we bring together our views on the cost of debt for Ireland and present updated ranges. The low end of the range reflects updated analysis of the all-in cost of debt (the BBB 10-year current spot rates, deflated by HICP inflation) with a 80bps Ireland premium and 20bps for new issue fees. This estimate also allows for the assumption that the weighted average maturity of the debt is 10-years (but that the BNE could need to put in place debt of a longer maturity) and the possibility that debt costs may rise by 2016 as evidenced by the forward yield curve for Irish 10 year gilts (see Figure 3.7). 18

21 Figure 3.7 Forward rates for 10-year Irish nominal gilts Source: Bloomberg As with Northern Ireland, in addressing comments from market participants, there is a question around whether a benchmark of BB to B rated bonds is robust and appropriate for calculating the marginal cost of debt of the BNE given the VIU investment grade assumption. Applying evidence of investment grade debt premiums to a longer-term risk free rate estimate would suggest that the top end of the range could be set to 3.50%, whilst a crosscheck of deriving an uplift premium using a BB credit rating benchmark would imply a % all-in cost of debt (assuming a bps premium above the lower end of our proposed cost of debt range). Our judgement is that the top end of the range should be set at 3.50% which would allow for some headroom above the 10-year investment grade benchmark and would be consistent with the methodology we have applied in previous calculations and across jurisdictions. Table 3.9 Updated cost of debt ranges for Ireland - real Low High Cost of debt 2.00% 3.50% Source: CEPA 19

22 4. GEARING 4.1. CEPA initial report For regulatory stability purposes and based on guidance from the SEM Committee, we retained a gearing assumption of 60% for the BNE. We presented evidence of gearing levels for other energy companies and noted that the CMA in the GB energy market investigation assumed a gearing level of 20-40% Consultation responses A majority of consultation responses argued that the gearing level proposed in the consultation was too low. Various respondents again supported by analysis undertaken by FE stated that firms with a large proportion of their business represented by generally typically have a much lower gearing rates. Analysis by RBS commissioned by Viridian (see text box in Section 2) also indicated that the gearing level should be reduced and that a 60% assumption was not supported CEPA comments As we have noted in our previous BNE reports, identifying an appropriate gearing assumption for the BNE is inevitably a judgment as the plant is a notional investment in the SEM. We would recommend that the SEM Committee adopt a lower gearing assumption than the 60% level used in the consultation paper for two reasons. Firstly analysis provided by RBS and Viridian would indicate that an investment grade credit rating may not be supported at a 60% gearing level, although as we note in Section 2, there are a range of factors that impact on a credit rating assessment not only gearing levels. Secondly a reduction would be consistent with evidence presented in the CMA GB energy market investigation and other benchmarks that were submitted as part of the responses. We propose a gearing level in the range 20-40% to be consistent with the WACC benchmarks developed by the CMA for its ongoing market investigation. The SEM Committee should note that changes in the gearing assumption only has a marginal impact on the WACC as the equity beta reduces as more expensive equity replaces debt in the notional BNE capital structure. 20

23 5. COST OF EQUITY 5.1. CEPA initial report Our initial report relied on the CAPM to produce an estimate of the cost of equity for the BNE. To derive parameters for the CAPM formula, we reviewed regulatory precedent and market evidence on the risk free rate for both Ireland and Northern Ireland. In Northern Ireland, we proposed a range for the risk free rate of % and an Equity Risk Premium (ERP) of 5.00%. For Ireland, we used a risk-free rate of % together with an ERP of 4.50% in our original report Consultation responses A number of responses found the upper end of ranges for cost of equity parameters to be appropriate, although the lower bounds were seen as too low by some market participants. FE submitted that the use of CMA regulatory precedent for Ireland was inappropriate as the CMA had placed too great a weight on contemporary evidence. In estimating the Total Market Return (TMR), FE looked at historic averages with a one year holding period. Based on this evidence, a TMR of 7.1% for Northern Ireland and 6.8% for Ireland were assumed. Alleged distortions in the risk-free rate for Northern Ireland led to FE using a twenty year average in arriving at a risk free rate of 2.0%. The same figure was used by FE for the risk-free rate for Ireland. Bord Gais Eireann (BGE) and the IWEA argued that the risk-free rate was set too low, with BGE quoting 10 year UK government bonds at yields of 2%. On the beta term, some respondents noted the challenges facing generation plant in both Northern Ireland and Ireland, including uncertainty over I-SEM and additional risks that would be faced by standalone generators. There were comments from participants stating that the overall WACC level was too low, rather than focussing on the cost of debt or cost of equity in particular. Many respondents, who included ESB, PPB, Brookfield and IWEA, felt that the range outlined in the initial report was not reflective of the risks faced in either jurisdiction and were below recent regulatory benchmarks. Bord na Mona felt that the WACC should reflect the risk profile of the market, especially the market within which the asset earns its revenue CEPA comments Northern Ireland We assume a risk-free rate of 1.25% and a ERP of 5.25% based on the CMA s provisional findings for Bristol Water. 9 9 CMA (2015): Bristol Water plc price determination provisional findings 21

24 This implies a TMR of 6.5% and relies on longer term historical evidence of the required market return on equity (ex post and ex ante). We do not agree with FE s proposed TMR as it assumes too short a holding period (1-year) nor does their estimate take account of the wider range of evidence as presented in the CMA s NIE determination and ongoing energy market investigation of lower medium term equity market return expectations. Whilst BGE point to 2% yields on UK government bonds, the yields they quote appear to be nominal and are therefore prior to deflating for inflation. As such these yields are currently negative in real terms, using RPI inflation. We have aimed up from this rate, as we seeking to derive a long-term estimate of the real risk-free rate. We presented a broad range for the asset beta of in our initial report drawing on the draft evidence and findings of the CMA s energy market investigation. This range has been retained by the CMA for its recent investigation findings, including for the stand-alone electricity generator benchmark. However, our initial report also noted that assuming a debt beta of 0.1, a gearing level of 60% and an equity beta range of 1.2 to 1.3 as used in previous BNE calculations, this would imply an asset beta of 0.54 to 0.58 at the notional gearing level. Given this asset beta assumption would reflect the historical view and precedent of the systematic risk faced by the BNE investor and given further the feedback on how investors would perceive the investment as a merchant plant, we propose that the asset beta range should remain unchanged from previous BNE calculations i.e to Given our change to the gearing assumption (as discussed in Section 4) the equity beta has reduced, despite retaining the same range for the asset beta. We present our updated range for the cost of equity in Northern Ireland in Table 5.1. Table 5.1 Updated cost of equity range for Northern Ireland Low High Pre-tax cost of equity 5.71% 7.31% Source: CEPA Ireland The longer term nature of the cost of equity means that there is less likelihood of substantial changes in assumption when compared to the cost of debt. Regulators have tended to adopt a figure towards the top end of the cost of equity range in recent determinations and, based on market data, we have narrowed our range, raising our lower bound for the TMR by 50bps. This increase is through the risk-free rate, where an upward movement and slight steepening in the Irish yield curve has led to this upwards revision. This gives a mid-point for the risk-free rate of 2.0%, which is equal to the FE assumption. The FE TMR assumption of 6.8% lies within our range, albeit above the mid-point. 22

25 The asset beta remains consistent across both Northern Ireland and Ireland for the project as per regulatory precedent. This is discussed in our comments on Northern Ireland. Table 5.2 Updated cost of equity range for Ireland Low High Pre-tax cost of equity 5.06% 7.49% Source: CEPA 23

26 6. CONCLUSIONS 6.1. Updated ranges Based on the evidence and findings presented above, in Table 6.1 below we have summarised our revised ranges for the BNE WACC which we recommend that the SEM Committee look to apply in the 2016 BNE calculation. Table Revised WACC Parameter Range for BNE 2016 Republic of Ireland Northern Ireland (UK) Low High Low High Cost of Debt 2.00% 3.50% 1.70% 3.00% Risk-free rate 1.50% 2.50% 1.25% Equity Risk Premium 4.50% 5.25% Asset Beta Debt Beta Equity Beta Post-tax Cost of Equity (real) 4.43% 6.55% 4.66% 5.98% Taxation % 12.50% 18.28% 18.28% Pre-tax Cost of Equity (real) 5.06% 7.49% 5.71% 7.31% Gearing 20% 40% 20% 40% Pre-tax WACC (real) 4.45% 5.89% 4.90% 5.59% Mid-point (pre-tax, real) 5.17% 5.25% Source: CEPA Our understanding is that the RAs will not include an NI debt premium in the BNE calculation and so Table 6.2 removes 25bps from both the bottom and top-end of the cost of debt range for the NI BNE. 10 In the Budget 2015, it was announced that the UK corporation tax rate would fall from 20% to 19% in 2017, and then to 18% in Our tax estimate reflects the average corporate tax rate over the twenty year economic life of the plant, based on these assumptions. 24

27 Table Revised WACC Parameter Range for BNE RA removal no NI debt premium Republic of Ireland Northern Ireland (UK) Low High Low High Cost of Debt 2.00% 3.50% 1.45% 2.75% Risk-free rate 1.50% 2.50% 1.25% Equity Risk Premium 4.50% 5.25% Asset Beta Debt Beta Equity Beta Post-tax Cost of Equity (real) 4.43% 6.55% 4.66% 5.98% Taxation % 12.50% 18.28% 18.28% Pre-tax Cost of Equity (real) 5.06% 7.49% 5.71% 7.31% Gearing 20% 40% 20% 40% Pre-tax WACC (real) 4.45% 5.89% 4.85% 5.49% Mid-point (pre-tax, real) 5.17% 5.17% 6.2. Cross-check to recent regulatory consultations We are aware both the CER and NIAUR have forthcoming consultations or current consultations on cost of capital in price review determinations. Draft SONI price review determination NIAUR has an ongoing public consultation for the SONI part of which is a proposal for the WACC (see Table 6.3 below). 11 In the Budget 2015, it was announced that the UK corporation tax rate would fall from 20% to 19% in 2017, and then to 18% in Our tax estimate reflects the average corporate tax rate over the twenty year economic life of the plant, based on these assumptions. 25

28 Table 6.3 SONI price review: UR proposed values for controls WACC components Cost of debt 3.20% Cost of equity 6.50% Gearing 55% WACC (Vanilla) 4.69% WACC (Pre Tax) 5.42% Components of the Cost of Equity Risk-free rate 1.50% Asset beta 0.45 Equity beta 1.00 Equity risk premium 5.00% Cost of equity 6.50% Components of the Cost of Debt Risk free rate 1.50% Debt premium 1.50% Issuance costs 0.20% Cost of debt 3.20% Source: NIAUR UR proposed values The proposed cost of debt figures are arguably not comparable to the BNE WACC as we presume that the proposed figure reflects a balance of the cost of new and embedded debt within the company. We have a consistent TMR of 6.5% as NIAUR has proposed for the SONI price control but adopt a slightly lower risk free rate and higher ERP by adopting the point estimates the CMA has used in its draft Bristol Water determination. We adopt a consistent assumption for debt issuance costs as NIAUR has used for the SONI consultation. The asset beta assumption is not comparable between the two sectors as they reflect the different risk profiles of each sector. Overall, although the economy wide parameters of the WACC proposed by NIAUR in SONI s draft determination, namely the risk-free rate, ERP and TMR, are informative for setting the 2016 BNE WACC, the risks that are faced by the notional BNE and SONI businesses are very different and the BNE is also a completely new investment financed with new debt. Together this means that the WACC calculations in each sector are not fully comparable. 26

29 Forthcoming electricity T&D price control The CER has commissioned Europe Economics (EE) to provide an estimate of the WACC for PR14 which will apply in EirGrid and ESB Networks forthcoming price control. The table below summarises EE s draft estimates for the WACC for PR14, before any aimingup on the estimate. Table 6.4 EE overall PR14 WACC estimate before aiming up Source: EE EE have recommended that the CER aim up on the estimates that set out in Table 6.3 with their proposed methodology for doing so based on Monte Carlo analysis. EE derive an implied WACC range that allows for aiming up of 4.4 to 5.0 per cent, and an overall pre-tax WACC point recommendation, after aiming up, of 4.8 per cent. For the BNE calculation, we have proposed a broader range for the risk-free rate than proposed by EE for PR14, while our ERP assumption is slightly lower. However, overall the implied TMR ( %) is similar to what we have proposed for the BNE in the Republic of Ireland, %. As with the regulatory precedent in Northern Ireland, the asset beta assumption is not really comparable between the BNE and PR14 given the different relative risk profiles. Our assumption assumes a BNE peaking plant is a more risky proposition for an investor compared to a price regulated electricity network, but the equity beta is lower once the impact of the lower BNE gearing assumption is taken account of Selecting a point estimate Given the range of evidence provided as part of the consultation, including regulatory precedent in current RA price reviews, we believe there is an argument for the SEM 27

30 Committee using a more conservative methodology than selecting the mid-point from the range, such as use of the 75 th percentile. This would be justified on the basis: of regulatory precedent (NIAUR for example, has adopted a point estimate at the top end of its range for the risk free rate for the SONI draft determination); Europe Economics aiming up from the mid-point of their WACC range for the electricity T&D price control draft determination; that setting the WACC too low for the BNE plant could create the risk that the CPM leads to under investment in the all-island market and does not meet the objective of providing a stable long term entry and exit signal to the market; and there is uncertainty of the true marginal project WACC for the BNE and selecting a point estimate that aims towards the top-end of our proposed range would place greater weight on the marginal WACC for the plant being higher than that implied by a spot investment grade benchmark. Based on the range presented in Table 6.2, this would imply an updated point estimate for the BNE WACC of 5.33%. 28