BEFORE THE PUBLIC UTILITIES COMMISSION OF THE STATE OF CALIFORNIA

Transcription

1 Docket: A Exhibit: SCE- BEFORE THE PUBLIC UTILITIES COMMISSION OF THE STATE OF CALIFORNIA Application of Southern California Edison Company (U -E) For Authority to Establish Its Authorized Cost of Capital for Utility Operations for 01 and to Reset the Annual Cost of Capital Adjustment Mechanism REBUTTAL TESTIMONY OF DR. MICHAEL J. VILBERT IN SUPPORT OF SOUTHERN CALIFORNIA EDISON COMPANY S AUTHORIZED COST OF CAPITAL FOR UTILITY OPERATIONS FOR 01 August, 01

2 Docket: A Exhibit: SCE- TABLE OF CONTENTS I. Introduction and Summary...1 II. Estimation of the Market Risk Premium... III. The Measurement of Financial Leverage and Its Impact on a Regulated Utility s Allowed Return on Equity...1 IV. The Market-to-Book Ratio Test Cannot Be Right... V. The Effect of Decoupling on the Cost of Capital... ii

3 Docket: A Exhibit: SCE- I. INTRODUCTION AND SUMMARY Q1. Please state your name and address for the record. A1. My name is Michael J. Vilbert. My business address is The Brattle Group, 01 Mission Street, Suite 00, San Francisco, CA, USA. Q. Please summarize your background and experience. A. I am a Principal of The Brattle Group, ( Brattle ), an economic, environmental and management consulting firm with offices in Cambridge, Washington, London, San Francisco, Madrid and Rome. Brattle s specialties include financial economics, regulatory economics, and the gas and electric industries. My work concentrates on financial and regulatory economics. I hold a B.S. from the U.S. Air Force Academy and a Ph.D. in finance from the Wharton School of Business at the University of Pennsylvania. I have worked in the areas of cost of capital, investment risk and related matters for many industries, regulated and unregulated alike, in many forums. I have testified before the U.S. Federal Energy Regulatory Commission ( FERC ), Canadian National Energy Board ( NEB ), and before many state/provincial regulatory commissions in the U.S. and Canada. I have previously filed testimony and testified before the California Public Utilities Commission ( CPUC ). Appendix A to this rebuttal testimony contains a more complete description of my professional qualifications. Q. What is the purpose of your rebuttal testimony in this proceeding? A. I have been asked by the Southern California Edison Company ( SCE or the Company ) to respond to direct testimony by Dr. J. Randall Woolridge on behalf of the Division of Ratepayer Advocates ( DRA ), Mr. Stephen G. Hill on behalf of The Federal Executive Agencies ( FEA ), Mr. Daniel J. Lawton on behalf of The Utility Reform Network ( TURN ), and Mr. Michael P. Gorman on behalf of the Energy Producers & Users Coalition ( EPUC ) in the current proceeding, (jointly, the Intervenors ). I address the following issues in my rebuttal testimony: (1) the determination of the appropriate market risk premium ( MRP ) for use in the risk 1

4 Docket: A Exhibit: SCE- positioning model; () the appropriate measurement of financial leverage and its impact on a regulated utility s allowed return on equity; () whether the market-to- book ratio test relied upon by the intervenors indicates whether investors in a utility company expect to earn more or less than their cost of equity; and finally () the effect of decoupling on the cost of capital. Q. Are you sponsoring any appendices? A. Yes. I am sponsoring Appendix A which describes my professional qualifications. I am also sponsoring Appendix B, which contains a revised version of The Brattle Group s decoupling study discussed in Section V below. Q. Were these appendices prepared by you or under your direction? A. Yes, they were. Q. Would you please summarize the main points of your rebuttal testimony? A. Yes. I believe that the intervenors make several errors of judgment in their testimonies which have the combined effect of lowering their estimated returns on equity for SCE, and therefore result in estimates that are too low relative to what investors expect to earn on investments of equivalent risk. These errors can be summarized as the following: 1. The intervenors fail to assess the reasonableness of their return on equity ( ROE ) estimates given the current economic uncertainty: i. The intervenors MRP estimates are unreasonably low given recent ii. increased market volatility and elevated utility yield spreads. The historical MRP values reported by the intervenors are not the figures recommended by Ibbotson, despite their having cited Ibbotson as the source. 1 1 Dr. Woolridge also relies on historical growth rates in his DCF model, (including growth rates for sub-investment grade companies), that do not reflect current conditions for electric utilities. See Woolridge Attachment JRW-, pp. -, for historical growth rates used in his DCF model. Despite Dr. Woolridge s statements at pp. - of his direct testimony that his electric and gas proxy groups have been selected to include only investment grade-rated companies, several of his sample

5 Docket: A Exhibit: SCE-. The intervenors fail to consider that the financial risk of SCE at its regulatory capital structure is higher than the financial risk reflected in the cost of equity estimates for the samples. This leads to an underestimation of the appropriate return on equity for SCE.. The market-to-book ratio test referenced by Mr. Hill and Dr. Woolridge does not indicate whether investors in a utility company expect to earn more or less than their cost of equity.. Decoupling does not have a measurable impact on the cost of capital, in contrast to the arguments of Mr. Hill. Q. How is your testimony organized? A. Section II addresses the intervenors estimation of the market risk premium. Section III discusses the appropriate measurement of financial leverage and its impact on a regulated utility s allowed return on equity. Section IV discusses whether the market- to-book ratio test implemented by the intervenors truly indicates whether investors in a utility company expect to earn more or less than their cost of equity. Section V addresses the effect of decoupling on the cost of capital and concludes my testimony. 1 II. ESTIMATION OF THE MARKET RISK PREMIUM Q. What do you cover in this section of your testimony? A. This section of the testimony covers two broad issues. First, I address the MRP estimates cited in the Hill, Lawton, and Woolridge Testimonies. Specifically, the historical MRP values reported are not the figures recommended by Ibbotson despite the use of Ibbotson as the source. I also review several issues with the individual companies have been rated sub-investment grade over the past years, (March 00 to March 01). These companies are: PNM Resources, Teco Energy, Inc., CMS Energy Corp., and Avista Corp. Therefore, Dr. Woolridge s growth rates and beta estimates are derived from non-investment grade utilities. While I note that the Woolridge Testimony relies on geometric measures for the MRP, I do not fully address the merits of using geometric versus arithmetic MRP averages in my testimony. This issue is discussed further in the testimony of Dr. Paul Hunt.

6 Docket: A Exhibit: SCE- MRP recommendations provided by each witness. Second, I address the impact of the current economic conditions on the MRP. Q. Please summarize the market risk premia used by the intervenors and explain how they were derived. A. The Hill Testimony uses an MRP of percent based on the historical MRP, while the Lawton Testimony uses. percent, which is derived as an average of an historical MRP and a forecasted MRP. The Gorman Testimony uses an MRP of. percent, which is at the high end of Morningstar s range. Finally, the Woolridge Testimony uses.01 percent which is obtained as an average of the historical (arithmetic and geometric) average MRP, Damodaran s forward-looking MRP, survey results, and building block methods. Q. Above you mentioned that the data relied upon by the intervenors was inconsistent with Ibbotson s recommendations. Please explain. A. The testimonies of Mr. Hill, Mr. Lawton, and Dr. Woolridge all use a number different from Ibbotson s recommended historical arithmetic MRP, although they all cite the data provided by Morningstar in the Ibbotson SBBI 01 Yearbook (in the case of Mr. Lawton and Dr. Woolridge), and the Ibbotson SBBI 0 Yearbook in the case of Mr. Hill. Both the Lawton and Woolridge Testimonies use. percent as the average historical arithmetic MRP. However, the 01 version of the Ibbotson text recommends. percent (for the S&P 00) rather than. percent. The Hill Testimony cites an historical arithmetic MRP of.0 percent citing the Ibbotson SBBI 0 Yearbook. However, the 0 version of the Ibbotson text recommends. Hill Testimony, pp. -. Lawton Testimony, p. 1 and Schedule DJL-. Gorman Testimony, p.. Woolridge Testimony, p. and Attachment JRW-, p.. See, Morningstar, Ibbotson SBBI 01 Valuation Yearbook, pp. -, and Appendix C, Table C-1. See also p. of the same text. Hill Testimony, p.. See also Morningstar, Ibbotson SBBI 0 Valuation Yearbook, p..

7 Docket: A Exhibit: SCE- percent. Thus, despite the fact that the Hill, Lawton, and Woolridge Testimonies rely on the Ibbotson SBBI Yearbooks, and Dr. Woolridge references his methodology as the Ibbotson Approach, the relied upon figures are not equal to the corresponding figures recommended by Ibbotson. They are instead lower by 0-0 basis points. The reason for this downward bias is that the testimonies rely on the total return of government bonds rather than the income return. The latter, however, is the measure that Ibbotson and others recommend, because the income return represents the truly riskless portion of the government bond s return. Q. Please explain why it is appropriate to use income returns in the MRP calculation. A. The total return consists of three components: the income return, the capital appreciation return, and the re-investment return. As neither capital appreciation returns nor re-investment returns are free of risk, these components of the total return are not risk-free. As the 01 Ibbotson text concludes: Another point to keep in mind when calculating the equity risk premium is that the income return on the appropriate horizon Treasury security, rather than the total return, is used in the calculation.... The income return is thus used in the estimation of the equity risk premium because it represents the truly riskless portion of the return. [emphasis added] 1 Q1. What specific comments do you have on the Woolridge Testimony s MRP estimate? A1. The Woolridge Testimony cites a number of reports, studies, and data in its Attachment JRW-, p.. However, the testimony relies only on the documents and 1 See Morningstar, Ibbotson SBBI 0 Valuation Yearbook, pp. - and Appendix C, Table C-1 (NB: this table shows a rounded MRP estimate of. percent). Woolridge Testimony, p. 1. See, for example, Morningstar, Ibbotson SBBI 01 Valuation Yearbook, p. ; Leonardo R. Giacchino & Jonathan A. Lesser, Principles of Corporate Finance, Public Utilities Reports, Inc., 0, pp. -, and Roger A. Morin, New Regulatory Finance, Public Utilities Reports, Inc., 00, pp Morningstar, Ibbotson SBBI 01 Valuation Yearbook, p..

8 1 1 1 Docket: A Exhibit: SCE- data presented in Attachment JRW-, p., which originate from the period Therefore, this rebuttal focuses on Attachment JRW-, p.. As the historical average geometric MRP is irrelevant for the purpose of determining a forward- looking MRP, the data on the geometric MRP should be ignored. The theoretical support for this statement is discussed in the rebuttal testimony of SCE Witness, Dr. Paul Hunt. In addition, the Woolridge Testimony chooses to rely on an Ibbotson figure other than what the Ibbotson text recommends, and thus biases Dr. Woolridge s MRP estimate downwards. Table 1 below shows that these corrections to the Woolridge Testimony s calculation of the MRP (specifically, ignoring the geometric mean estimates and correcting the Ibbotson estimates) increase the median MRP estimate by more than 0 basis points, from.01 percent to.1 percent, and increase the mean by approximately 0 basis points, from. percent to. percent. 1 Woolridge Testimony, p..

9 Docket: A Exhibit: SCE- Table 1 1 Summary of 0-1 Equity Risk Premium Studies Revisited (Woolridge Attachment JRW-, p. ) JRW Revised Historical Risk Premium Ibbotson Historical-Arithmetic.0%.% Historical Geometric.% ignore Estimate (median).0%.% Ex Ante Models Damodaran 01 Projection Fundamentals - Implied Estimate (median).%.% from FCF to Equity Model Surveys Estimate (median).00%.00% Building Block Ibbotson & Chen Historical-Arithmetic.%.%.1% Historical Geometric.1% ignore Woolridge 01 Supply Model.% ignore Estimate (median).0%.1% Mean.%.% Median.01%.1% 1 1 The Woolridge Testimony claims that studies by leading academics indicate the forward-looking equity risk premium is actually in the.0% to.0% range. 1 Yet, the forward-looking study by Damodaran relied upon in the Woolridge Testimony estimates the MRP at. percent and only estimates that rely on geometric averages are below the.0 percent that the Woolridge Testimony cites as the upper range. Q1. What other comments do you have on Dr. Woolridge s Attachment JRW-, p.? A1. The lowest MRP estimate of the four considered by Dr. Woolridge is the median estimate from survey information. However, surveys are difficult to perform and interpret reliably. For example, according to a publication by Roger Ibbotson: The consensus [i.e., survey] method might appear to be a very good approach; when using this method, one attempts to obtain the 1 1 JRW data from Woolridge Testimony Attachment JRW-, p.. Revisions include the elimination of geometric averages, using the Ibbotson 1-0 historical arithmetic MRP of.%, calculated in Morningstar, Ibbotson SBBI 01 Valuation Yearbook, p., updating the Ibbotson & Chen 1-0 historical arithmetic MRP to the corresponding 1-0 figure of.1%, (calculated in Ibbotson SBBI 01 Valuation Yearbook, p. ), and finally removing Dr. Woolridge s calculated supply model figure due to lack of underlying data. Woolridge Testimony, p..

10 Docket: A Exhibit: SCE- estimates from the market participants themselves (i.e., the very investors who are setting the market prices). But there are a number of problems with this approach. Most of these investors have no clear opinion about the long-run outlook. Many of them have only very short-term horizons. Individual investors often exhibit extreme optimism or pessimism and make procyclical forecasts, and so following a boom, they can have ERP estimates that exceed 0 percent or 0 percent. Following a recession or a decline in stock market prices, their estimates of the ERP might even be negative. Academics and institutional investors may be more thoughtful, but any survey of their opinions would have to be very carefully designed. I have seen surveys, however, that do not seem to even clarify whether the questionnaire refers to arithmetic mean returns or geometric mean returns. Many surveys also do not make clear whether the ERP to which they refer is the excess return of stocks over government bonds or Treasury bills or some other type of bond. This lack of clarity makes the surveys very difficult to interpret. 1 Q1. What are your comments on the data relied upon by Mr. Hill for the MRP? A1. The Hill Testimony relies on data for the period 1 to 00, which therefore ignores the years 0 and 0. Mr. Hill examines Ibbotson s reported geometric and arithmetic historical averages of. percent and percent for the 1-00 period and finds the upper end reasonable. He therefore uses percent as his estimate of the MRP. 1 In an appendix, Mr. Hill discusses the merits of using the geometric versus the arithmetic MRP. 1 While geometric averages are useful summaries of past returns and useful measures of past performance by professional investment managers, I do not believe they should be used to calculate the cost of capital. The use of arithmetic, not geometric, average MRPs for estimating cost of capital should not be controversial. For example, the Cost of Capital Edition of the Ibbotson SBBI 01 Yearbook, which estimates costs of capital by industry, uses only arithmetic The Equity Risk Premium, Roger G. Ibbotson Professor in Practice, Yale School of Management Chairman, Zebra Capital Management, published in Rethinking the Equity Risk Premium, Research Foundation of CFA Institute, December, 0, (pp. 1- at p. 0). Hill Testimony, pp. -. Hill Testimony, Appendix D.

11 Docket: A Exhibit: SCE- averages. 1 The Ibbotson SBBI Yearbooks are the standard source for historical U.S. risk premiums. R. A. Morin, author of New Regulatory Finance, says that [O]nly arithmetic means are correct for forecasting purposes and for estimating the cost of capital. There is no theoretical or empirical justification for the use of geometric mean rates of return as a measure of the appropriate discount rate in computing the cost of capital 0 A U.K. study states that [T]he key distinction is between compound [geometric] and arithmetic average returns. The latter measure is the appropriate measure to apply in the CAPM and most other asset pricing models... 1 Leading corporate-finance textbooks reach the same conclusion. The only dissenter that I know of is Professor Damodaran. He says that [I]n corporate finance and valuation, the argument for using geometric average premiums is strong. Prof. Damodaran further says that geometric averages are relevant when stock-market returns are mean-reverting; that is, negatively serially correlated. He does not explain why serial correlation matters, however. He cites one 1 study as evidence for mean reversion. However, Professor Damodaran s website states that he no longer use[s] historical premiums in either valuation or corporate finance. Mr. Hill also cites serial correlation as a factor that biases the calculation of the arithmetic-average MRP. However, this argument is not persuasive because the The Ibbotson SBBI 01 Yearbook states that [T]he arithmetic average equity risk premium can be demonstrated to be most appropriate when discounting future cash flows. See p.. R. A. Morin, New Regulatory Finance, Public Utilities Reports, Inc., 00, pp. - and pp See also L.R. Giacchino and J.A. Lesser, Principles of Utility Corporate Finance, Public Utilities Reports, Inc., 0, pp. -. S. Wright, et al, Report on the cost of capital provided to Ofgem, Smithers & Co. Ltd., September 1, 00, p.. See R. A. Brealey, S. C. Myers and F. Allen, Principles of Corporate Finance, th Ed., 0, pp. 1-1; J. Berk and P. DeMarzo, Corporate Finance: The Core, 1 st Ed., 00, p. ; S. A. Ross, R. W. Westerfield and J. Jaffe, Corporate Finance, th Ed, 0, pp. 1-1; Z. Bodie, A. Kane and A. J. Marcus, Investments, th Ed., 0, pp. - and pp A. Damodaran (01), Equity risk premiums (ERP): Determinants, Estimation and Implications The 01 edition, Working paper, NYU Stern School of Business, p.. Damodaran Musings on Markets, Equity Risk Premiums: The 01 Edition, March, 01. Available at Hill Testimony, Appendix D, pp. ii-iv.

12 Docket: A Exhibit: SCE- evidence for mean reversion in U.S. stock returns is weak. For example, the Ibbotson SBBI 01 Yearbook reports negligible serial correlations for annual U.S. stock returns (0.0) and concludes that the equity risk premium that was realized over 1 to the present was almost perfectly free of mean reversion Dimson, Marsh and Staunton ( DMS ) similarly conclude that [T]he mean reversion effect is, at best, of modest magnitude and that for forecasting the long-run equity premium, it is hard to improve on extrapolation from the longest history that is available... There are some statistical issues that could impart an upward bias to the arithmetic- average MRP and could, as a result, justify placing some weight on geometric- average MRPs. However, the weight on the geometric average would be very small for the purpose of calculating a three-year forward cost of equity. Q1. How would you describe the current prevailing global economic conditions? A1. Current market conditions are still very unsettled and volatile. Although the volatility in the financial markets has lessened, economic conditions are not back to normal as measured by their pre-crisis status. For example, macroeconomic factors such as Ibbotson SBBI 01 Valuation Yearbook, p.. The DMS data and methodology are covered in E. Dimson, P. R. Marsh and M. Staunton, Triumph of the Optimists: 1 Years of Global Equity Returns, Princeton University Press, 00. The most recent DMS MRP averages are reported in the Credit Suisse Global Investment Returns Yearbook 01, Credit Suisse Research Institute (DMS 01 Yearbook), Table, p.. DMS s conclusions about mean reversion are at p.. Jacquier, Kane and Marcus ( JKM ) explain why arithmetic averages can be upward-biased forecasts of compounded future returns. (See E. Jacquier, A. Kane and A. Marcus (00), Optimal estimation of the risk premium for the long run and asset allocation: A case of compounded estimation risk, Journal of Financial Econometrics, -. Note Table 1, Figure and Section of this paper). JKM derive adjustments to remove the bias, but the adjustments are small over short-term forecast periods. Also, these adjustments are not materially affected by serial correlation. JKM show in particular that arithmetic means can be upward biased when the forecast period H is long relative to the estimation period T. However, in the above-cited sources, estimation periods, (1 years in DMS, years in the Ibbotson SBBI 01 Yearbook) are long, and the forecast period short, ( years as used by Mr. Hill in his Appendix D, p. iv). Therefore, the JKM adjustment would imply putting a weight of (1 - H/T) on the historical arithmetic mean, and weight H/T on the historical geometric mean. The weights on the arithmetic and geometric means would thus be 1 - /1 =., and /1 =.0, respectively. In other words, the arithmetic average gets about % weight and the geometric average about % weight in calculating a three-year forward cost of equity.

13 Docket: A Exhibit: SCE- unemployment and mortgage foreclosures are still very high by historic standards. Some investors fear that the current period of economic recovery may not continue and that we may enter a global double dip recession, due to fears about the euro and the threat of international financial contagion. At the same time, the deficits at all levels of U.S. government are at high and unsustainable levels, with the potential for rampant inflation inherent in the deficit spending by the U.S. government and by the liquidity injected into the capital markets by the Federal Reserve ( Fed ). Moreover, although the spread between U.S. utility bond yields and government bond yields ( yield spread ) has narrowed from their peak at the height of the crisis, in recent months the spread has again increased and is much larger than before the crisis. A higher than normal yield spread is one indication of the higher cost of capital prevailing in the capital markets. This is especially true for lower-rated bonds, as well as for less risky investments such as investment grade-rated utility bonds. This is, in part, the result of a deliberate policy by the Fed to lower long-term as well as short-term bond yields in an effort to induce investors to move to riskier assets such as stock. 0 In fact, yields on certain inflation indexed Treasury bonds and European Treasury bonds have recently turned negative. 1 In addition, long-term government bond yields are currently depressed relative to both historical levels and forecasts of future interest rates. In my view, this is evidence that the flight to 0 1 For example, since the last major peak in November 00, the spread between the yield on BBB-rated 0-year utility bonds and the yield on 0-year U.S. Government bonds has ranged from a low of 1 basis points to a high of 1 basis points, compared to an historical average of approximately basis points. In addition to the spike in the spreads between utility and government bond yields, the variability in bond yields is also currently high. BBB utility 0-year bond yields have varied from a high of.% to a low of.0% for a high-to-low difference of 1 basis points over the period July 0 to July 01. Historically, variations in BBB utility bond yields have rarely been above 0 basis points in any 1-month period. (Data obtained from Bloomberg). Wall Street Journal Online, Fed s Operation Twist Tangles Treasury Trade, February, 01. For example, a $1 billion five-year U.S. TIPS (Treasury Inflation Protected Securities) sale scheduled for August, 01, drew the strongest demand seen in months as buyers accepted a record-low negative 1.% yield, in the hopes that future inflation would increase their compensation over time. ( U.S. Treasurys Advance; TIPS Sold at Record Negative Yield, WSJ, August, 01, Over the past few years, the annual yields on long-term U.S. government bonds have dropped dramatically. For instance, the historical average annual yield on long-term government bonds was

14 Docket: A Exhibit: SCE- safety continues and may again be increasing, which in turn suggests that the market risk premium is higher today than it was prior to the crisis for all risky investments. This is true even for lower than average risk investments such as regulated utilities. Q1. How does the ongoing turmoil in the financial markets affect the cost of capital for a regulated utility such as SCE? A1. The MRP remains higher than before the start of the crisis in about mid-00. Although economic conditions have improved since the height of the crisis, great uncertainty remains in the capital markets, due to the reasons discussed above and discussed in more detail in Dr. Hunt s testimony. Therefore, while the cost of capital may have decreased somewhat since the height of the credit crisis in 00-00, the DCF and risk positioning models implemented by the intervenors underestimate the cost of capital currently prevailing in the markets. The uncertainty in the financial markets affects the results of ROE estimation models, because both the risk positioning (or CAPM ) and the DCF models are based upon the assumption that economic conditions are stable. Although that assumption is probably never met completely, the degree of turmoil currently is much greater than would normally be the case, so estimating the cost of capital under current conditions is more complicated than it would normally be. Because the uncertainty in the financial markets affects the cost of capital for all companies including a regulated. percent from 1 to 0, but the long-term government bond yield declined to just. percent in 0 (see Morningstar s Ibbotson SBBI 01 Valuation Yearbook). The current yield is now even lower, at approximately. percent (based on the 1-day average of long-term (0-year) U.S. government bond yields as of August, 01). Additionally, the consensus economic projections, according to the Blue Chip Economic Indicators report dated March, 01, p. 1, for the yield on -year U.S. Treasury notes are. percent in 01,. percent on average from 01 to 01, and. percent on average from 01 to 0. These forecasts are substantially higher than the current yield on -year U.S. government notes, which is approximately 1. percent (based on the 1- day average as of August, 01). This highlights the fact that current long-term and medium-term U.S. government bond yields are extremely low, both relative to historical levels, as well as compared to consensus forecasts of future rates. The constant growth DCF model requires that dividends and earnings grow at the same rate for companies that earn their cost of capital on average. This is a subjective judgment, but in my view, the possible collapse of the Euro zone is unprecedented in recent times. 1

15 Docket: A Exhibit: SCE- utility such as SCE, I believe adjustments to the ROE models are necessary and appropriate to recognize the effect of the increased volatility in the capital markets, as well as the unsustainable decline in long-term risk-free interest rates, on the cost of capital. First, this is because the downward bias in the long-term risk-free interest rate will inappropriately lower the sample companies ROE estimates generated by the CAPM method. Second, the increase in market volatility and general risk- aversion of investors implies an increase in the prevailing MRP. Q1. Do the intervenors MRP estimates reflect any adjustments for the impact of current economic conditions? A1. No. The Hill Testimony relies on an historical arithmetic-average MRP without making any explicit adjustments for current economic uncertainty. As a result, his methodology understates the forward-looking MRP. The Lawton Testimony employs an average of forward-looking and historical MRP estimates, which results in an MRP of. percent. While this calculation does reflect the impact of current economic conditions to an extent, Mr. Lawton does not explicitly consider the impact of financial uncertainty on the cost of capital. In fact, Mr. Lawton argues that the recent decline in Treasury and utility bond yields, as well as declining authorized equity returns set by national regulatory authorities, reflects a decline in the cost of capital, and thus should be reflected in lower allowed ROEs for utility companies. While I agree that the cost of capital has declined from the height of the crisis, I believe that looking at Treasury yields alone is misleading. The cost of capital has not fallen as much as Treasury yields would indicate. This can be seen by the fact that spreads on utility bonds remain much higher than was the norm in the recent past. Furthermore, the recent decline in long-term government bond yields justifies making an upwards adjustment to the risk positioning model results, given the negative Academic research finds that investors expect a higher risk premium during more volatile periods. The higher the risk premium, the higher is the required return on equity. For example, French, Schwert, and Stambaugh (1) find a positive relationship between the expected market risk premium and volatility; (see K. French, W. Schwert and R. Stambaugh (1), Expected Stock Returns and Volatility, Journal of Financial Economics, Vol. 1, p. ). Lawton Testimony, p.. 1

16 Docket: A Exhibit: SCE- effect of a decline in the risk-free rate on the cost of equity estimate. The MRP is generally inversely related to the interest rate. Dr. Woolridge relies only on reports and articles from the 0 to 01 period and claims his resulting MRP estimate is consistent with the contemporaneous market risk premiums of several notable services. However, as mentioned above, Dr. Woolridge s use of a median MRP estimate of.01 percent is biased downwards due to his reliance on geometric averages and surveys. Ignoring geometric averages, Dr. Woolridge s table in Attachment JRW-, p., shows an MRP mean of. percent and a median of. percent. Excluding surveys as well, the corresponding mean and median are. percent and. percent respectively. Therefore, Dr. Woolridge s MRP estimates do not sufficiently reflect the impact of current market turmoil. Moreover, Dr. Woolridge cites the recent declines in the S&P 00 VIX as evidence that market volatility is down. While it is true that the VIX has been declining over the last few months, the standard deviation of markets is still higher than during the time period most of Dr. Woolridge s cited research was conducted, which was largely prior to the credit crisis (see Figure 1 below). Roger A. Morin, New Regulatory Finance, 00, pp. 1-1, summarizes several studies and finds that the realized ROE changes by approximately 0 basis points when the yield on government bonds changes inversely by 0 basis points. Woolridge Testimony, p.. Woolridge Testimony, pp. -. 1

17 Docket: A Exhibit: SCE- % Rolling -Month Volatility of U.S. S&P 00 Index Returns % % % U.S. - S&P 00 % % % % % 1% 0% Notes and Sources: Volatility of the previous -month stock index returns Indices downloaded from Bloomberg Figure 1 1 Q1. Please summarize this portion of your rebuttal. A1. The intervenors estimates of the MRP are too low, given the current state of market uncertainty as discussed above. Not making an adjustment to either the risk-free interest rate or the MRP, or both, to reflect the effect of market volatility and low government bond yields on the cost of capital when implementing a risk premium model, e.g., the CAPM or the Empirical CAPM, will result in an underestimation of the appropriate return on equity for the company. Furthermore, the intervenors do not even use the correct historical MRP figures reported by Ibbotson, despite citing Ibbotson as their primary data source. III. THE MEASUREMENT OF FINANCIAL LEVERAGE AND ITS IMPACT ON A REGULATED UTILITY S ALLOWED RETURN ON EQUITY Q1. What do you cover in this section of your testimony? A1. This section of my testimony responds to the intervenors criticism of Dr. Hunt s financial leverage adjustment. Dr. Hunt implemented this adjustment in his direct 1

18 Docket: A Exhibit: SCE- testimony to reflect: (1) the difference in financial risk between the sample companies measured at market value and SCE s market value capital structure; and () the difference in financial risk between SCE s regulatory capital structure and its market value capital structure. This calculation takes the impact of embedded debt into account and the debt equivalence of imputed debt from power purchase contracts. First, I demonstrate that financial leverage should be measured on a market value basis, not a book value basis. There is no debate in the academic literature on this point. Dr. Hunt uses the market value capital structure which I agree is appropriate. Second, I discuss the impact of financial leverage and risk on a regulated utility s cost of equity. Q0. What portions of the intervenors respective testimonies are you addressing in this section? A0. The relevant section in Dr. Woolridge s testimony is Chapter, Section, SCE MV/BV Adjustment (at pages 1 to ). Mr. Hill s discussion of financial leverage is between pages 1 and 1 of his testimony. Mr. Gorman addresses Dr. Hunt s leverage adjustment briefly at page of his testimony, and Mr. Lawton does the same at pages to of his testimony. Q1. What are Dr. Woolridge s and Mr. Hill s main arguments? A1. On behalf of SCE, Dr. Hunt proposed to add an approximate 0 basis point (0. percent) adjustment to the cost of equity estimated from the proxy group to reflect the fact that, firstly, SCE s market value capital structure (. percent debt) has more financial risk than the average market value capital structure of the proxy group (.1 percent debt), and, secondly, that SCE s market value capital structure includes the effect of debt equivalence and other adjustments that must be made to SCE s regulatory capital structure to properly measure the risk of the company. 0 result is that a second adjustment to the ROE estimates is required to properly reflect the higher financial risk that is embodied in SCE s requested regulatory capital The 0 Hunt Direct Testimony, pp

19 Docket: A Exhibit: SCE- structure. 1 Dr. Woolridge and Mr. Hill reject Dr. Hunt s leverage adjustment based on two principal reasons: a) Financial leverage should be measured on a book value basis. Hence, there is no need for the leverage adjustment. b) Dr. Hunt s leverage adjustment would reward equity investors in regulated utilities with an above-market risk-adjusted cost of equity. Q. What evidence do Dr. Woolridge and Mr. Hill offer to reject the financial risk adjustment proposed by Dr. Hunt? A. Although both Dr. Woolridge and Mr. Hill acknowledge that financial leverage increases risk to equity investors and increases the cost of equity, they dispute the notion that financial risks are measured on a market value basis. Instead, Dr. Woolridge argues that [F]inancial publications and investment firms report capitalizations on a book value and not a market value basis and there is no need for a leverage adjustment because there is no change in leverage. Mr. Hill similarly argues that SCE s financial risk is a function of the availability of operating cash flows to meet its fixed charges or debt obligations, and one company at one point in time cannot have two levels of financial risk because the amount of fixed charges (the debt costs) are the same [with a market value or book value capital structure]. Q. What is the fundamental flaw in their arguments? A. The disregard of market value capitalization in measuring a company s financial leverage and risk is a fundamental flaw in Dr. Woolridge s and Mr. Hill s arguments. As I will explain below, the cost of equity estimated from capital markets reflects 1 Hunt Direct Testimony, p. 0. Woolridge Testimony, pp. 1-, and Hill Testimony, pp Woolridge Testimony, p.. Hill Testimony, p. 1. 1

20 Docket: A Exhibit: SCE- both the business risk and financial risk of the company, and financial risk is properly measured by the market value capital structure. Q. Does the use of an estimated ROE based upon market value information conflict with the use of a book value rate base to set rates? A. No. Neither Dr. Hunt nor I are suggesting that the market-derived cost of capital estimates should not be used with a rate base measured at original cost. In California as well as in most U.S. utility regulation, rates are set using the regulated company s rate base which is measured on the basis of the original costs or book value. The book value capital structure embedded in the depreciated rate base is generally different from the market value capital structures of the sample companies used to estimate the cost of equity. The estimated (market-derived) ROEs are applied to the book value rate base, but financial risk inherent in the regulatory capital structure may differ from the financial risk of the sample used to estimate the ROE. To account properly for the difference in financial risk between the ROE estimated from market data and the capital structure of the regulated firm, I agree with Dr. Hunt that the allowed return on equity should be adjusted to reflect the difference in financial leverage, so that equity investors will be given a fair opportunity to earn their cost of equity. Q. Is the adjustment that you and Dr. Hunt recommend a market-to-book ratio adjustment? A. Absolutely not. The leverage adjustment should not be confused with the market-to- book ratio adjustment ( MV/BV ) referred to by Mr. Hill and Dr. Woolridge. understand, what they have in mind is illustrated in the following example. Suppose a regulated company s market-to-book ratio is.0, i.e., the market value of the company s share is twice its book value. If the cost-of-capital estimation models suggest an ROE of percent, then, as I understand Mr. Hill and Dr. Woolridge, they As I Hill Testimony, pp. 1-1, and Woolridge Testimony, pp. 1-1 and pp. 1-. Dr. Woolridge erroneously refers to Dr. Hunt s leverage adjustment as a MV/BV adjustment at p. 1 of his testimony. 1

21 Docket: A Exhibit: SCE- believe Dr. Hunt s adjustment would require that the allowed return must be set at 1 percent (twice the market rate), so that the company can expect to earn percent on its market value. This is NOT what Dr. Hunt and I are recommending. We are instead recommending an adjustment based on differences in financial risk (the after- tax weighted-average cost of capital or ATWACC approach, as discussed further below). All else equal, if two companies, A and B, have the same business risk and thus the same ATWACC, then if company A has a higher market value debt ratio than company B, it has higher financial risk, which should be reflected in a higher ROE. This adjustment does not use the market-to-book ratio at all. Q. How are these two concepts confused? A. Consider first a situation in which the book value and market value for all sample companies are equal. The estimated cost of equity from the sample will reflect the business risk and the financial risk of the sample companies as before. Further assume that the rate base capital structure of the regulated entity differs from the average capital structure of the sample companies. I believe that the intervenors would agree with Dr. Hunt and me that an adjustment would be warranted for the allowed ROE for the regulated company, although the intervenors may or may not agree with the exact adjustment recommended by Dr. Hunt. Q. Why is the situation different if the MV/BV ratio is not equal to 1.0? A. This is the essence of the disagreement between us. Dr. Woolridge and Mr. Hill assert that financial risk is properly measured by the book value capital structure so there is no need for the leverage adjustment. This is incorrect. It is the market value capital structure that matters for measuring financial risk, and a leverage adjustment is required if the rate base capital structure is different from the market value capital structure embedded in the sample companies estimates of the cost of equity. For example, assume that the market cost of debt is. percent and the tax rate is 0 percent. Then, the ATWACC is the same at either a percent ROE with 0 These match the assumptions made by Dr. Hunt in his direct testimony (see Appendix E workpapers). 1

22 Docket: A Exhibit: SCE- percent equity, or an. percent ROE with 0 percent equity. Customers would pay the same for service under either situation. The notion that financial leverage is and should be measured on a market value basis, shared by Dr. Hunt and me, is supported in every textbook on corporate finance of which I am aware. Further, the view is not just an ivory-tower creation. Professional valuation books and guidance advocate the use of market value capital structure. Morningstar, an off-the-shelf cost-of-capital provider, also uses market value capital structure in its cost of capital estimates. 0 Even Dr. Woolridge s text, Applied Principles of Finance, uses market values to illustrate the computation of the overall cost of capital. 1 bankruptcy proceedings. Similar views were also endorsed by legal decisions on Q. Isn t it true that credit rating agencies measure financial risk with reference to book values? A. Yes and no. Credit rating agencies are concerned with the ability of debt issuing entities to pay interest and repay debt. Their interest in the cost of equity capital is its value to bondholders in assuring debt repayment, not its value to shareholders. To ensure credit worthiness, credit rating agencies rely upon accounting information to calculate financial ratios to measure the financial health of a company. Historically, 0 1 There would be a difference if the embedded cost of debt were different than the market cost of debt. See, e.g., Richard A. Brealey, Stewart C. Myers, and Franklin Allen, 0, Principles of Corporate Finance, th edition, McGraw-Hill Irwin, p. ; Stephen A. Ross, Randolph W. Westerfield, and Jeffrey Jaffe, 00, Corporate Finance, th edition, McGraw-Hill Irwin, p. 1; and Mark Grinblatt and Sheridan Titman, 1, Financial Markets and Corporate Strategy, 1 st edition, Irwin/McGraw- Hill, p.. See, e.g., Tom Copeland, Tim Koller, and Jack Murrin, 000, Valuation: Measuring and managing the value of companies, rd edition John Wiley & Sons, p. 0; and Shannon P. Pratt and Alina V. Niculita, 00, Valuing a business: The analysis and appraisal of closely held companies, th edition, McGraw-Hill, pp See, e.g., Morningstar, Ibbotson Cost of Capital 01 Yearbook, p.. J. Randall Woolridge and Gary Gray, Applied Principles of Finance, Preliminary Edition, Penn State University, 00, pp See, e.g., Bernstein, Stan, Susan H. Seabury, and Jack F. Williams, 00, Squaring bankruptcy valuation practice with Daubert Demands, ABI Law Review, p.. 0

23 Docket: A Exhibit: SCE- accounting information is based primarily on historical costs, i.e., book value information. Accounting information is used by the rating agencies partly because it follows the Generally Accepted Accounting Principles ( GAAP ) and is audited by third-party auditors. This allows for consistency between companies when comparing financial performance and to evaluate the credit worthiness of a company. Another rationale for the rating agencies use of accounting information is the stability of accounting information, which is generally not updated more frequently than quarterly. Only the annual statements are fully audited. On the other hand, market value information changes daily. Any credit report based upon market information would be out of date very quickly. Use of accounting data avoids this problem. Stability is both a virtue and a flaw (not timely) in historical-cost-based financial accounting and credit analysis. Since Statement of Financial Accounting Standard No. 1 Fair Value Measurements took effect on and after 00, financial statements have incorporated more and more market value information about a company s assets and liabilities. Similarly, credit rating agencies such as Moody s have also used market value information in their assessment of credit risk. For example, Moody s has stated that some of its measures of corporate default risk are continuously updated and extracted from the equity markets. Q. Can you explain why financial leverage is and should be measured on a market value basis? A. The impact of financial leverage on the cost of equity has been developed since the 1 paper by Prof. Franco Modigliani and Merton Miller ( MM ), two economists who eventually won Nobel Prizes in part for their body of work on the effects of debt See last accessed October, 0. It is stated here that Statement No. 1 is effective for financial statements issued for fiscal years beginning after November 1, 00, and interim periods within those fiscal years. The accounting treatment is now covered in ASC 0. See brochure of Moody Analytics, Research-Risk-Measurement/Quantative-Insight/CreditEdge/CreditEdge-Plus-Brochure.ashx, last accessed October, 0. 1

24 Docket: A Exhibit: SCE- on firm value. One key corollary of the MM theorems and their various extensions is that cost of equity increases as financial leverage increases. Although the exact rate of increase in cost of equity differs by models of capital structure, it is universally accepted that as a firm adds debt, its cost of equity increases as a result. Both Dr. Woolridge and Mr. Hill acknowledge that the cost of equity increases with financial leverage; however, they assert that financial risk is measured on a book value basis. This belief is wrong for two reasons. First, in MM s classic paper and subsequent extensions of their original paper, financial leverage has been consistently measured on a market value basis. This is because MM s basic insight is that, under perfect market conditions, financial leverage does not increase the market value of a firm as long as different combinations of debt and equity can be selected by the investors themselves. To implement such a self-help financial engineering, investors have to be able to buy and sell debt and equity to achieve their desired combination. The prices at which they transact are, by definition, market prices. Second, as a more practical matter, economists generally prefer to use market values because they convey timely information, rather than historical data, about the assets. Business decisions on investment, capital budgeting, and financing are all based on real time market value information. Q0. Could you provide a numerical example to illustrate the impact of debt on cost of equity? A0. As a simple example, think of an investor who takes money out of her savings and invests $0,000 in real estate. The future value of the real estate is uncertain. If the real estate market booms, she wins. If the real estate market goes down, she loses. Figure below illustrates this. Franco Modigliani and Merton H. Miller (1), The cost of capital, corporation finance and the theory of investment, The American Economic Review, Vol., No., pp. 1-. For a modern textbook exposition of the capital structure theories, see Brealey, Myers, and Allen, op cit., Chapter 1. In developing the theory, MM assume that investors can adjust the capital structures of their portfolios at no cost.

25 Docket: A Exhibit: SCE-,000,000,000,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000,000 - Buy Real Estate for $0,000 using only Equity If Real Estate Prices Increase or Fall by %, Gain or Lose %. Equity Investment % Gain in Real Estate Value % Gain In Equity Value % Loss in Real Estate Value % Loss in Equity Value Equity % Appreciation or Depreciation $0,000 $0,000 If Real Estate increases by % : $0,000/$0,000=0% If Real Estate falls by % : $0,000/$0,000=0% Changes in Equity Value: +/-% Figure 1 In Figure below, where the investor finances the purchase using 0 percent equity and 0 percent mortgage, the variability in the investor s equity return is twice that in Figure. The entire fluctuation of percent from rising or falling real estate prices falls on the investor s now-halved $0,000 equity investment. The lesson from the example is obvious; debt adds risk to equity.

26 Docket: A Exhibit: SCE-,000,000,000,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000,000 0 Buy Real Estate for $0,000 with a $0,000 Mortgage If Real Estate Prices Increase or Fall by %, Gain or Lose 0%. $0,000 $0,000 Equity Mortgage Initial Investment % Gain in Real Estate Value 0% Gain In Equity Value % Loss in Real Estate Value 0% Loss in Equity Value Equity Mortgage Change in Value $0,000 $0,000 If Real Estate increases by %: $0,000 - $0,000 = $0,000 $0,000/$0,000=% If Real Estate falls by %: $0,000 - $0,000 = $0,000 $0,000/$0,000=0% Changes in Equity Value: +/-0% Figure Q1. Please provide an example that illustrates why market values are relevant. A1. Suppose in the above example that the investor has invested in real estate years ago. Further assume that accounting depreciation has reduced the book value of the real estate from $0,000 to $,000, and assume the investor has paid off 0 percent of her $0,000 mortgage. Thus, the investor has a remaining mortgage of $0,000 (= 0 percent $0,000). The book value of the investor s equity investment is therefore $,000 (= $,000 - $0,000). To calculate the return on equity if real estate prices rise or fall 0 percent, one needs to know how real estate prices have developed over the past years. For example, if the market value of the real estate is now $00,000, then a 0 percent decrease in the price of real estate ($0,000) is almost equal to the investor s book value equity. However, her market value equity (or net worth) is equal to the value of the real estate minus what she owes on the mortgage. If we assume that the market value of the mortgage equals the unpaid balance of $0,000, then the investor s net worth is $,000 (= $00,000 - $0,000).

27 Docket: A Exhibit: SCE- Therefore, the market return on equity due to a 0 percent decline in real estate prices is only -. percent (= -0,000 /,000), versus a corresponding - percent book return on equity (= -0,000 /,000). Q. Do you dispute Mr. Hill s assertion that a company does not have two levels of financial risk, one based upon the book value capital structure and one based upon its market value capital structure? A. No, of course not. There is only one measure of financial risk for a company and that is the one based upon its market value capital structure. The issue Dr. Hunt and I are addressing is that the average market capital structure of the sample firms used to estimate the cost of equity is different from the capital structure of the investment to which it is applied, i.e., the regulatory capital structure of SCE s rate base. Q. How do you respond to Mr. Hill s claim that financial risk is measured by the sufficiency of the firm s operating cash flows to meet its contractual fixed charges or debt obligations? A. While it is true that a firm s debt obligations are typically defined in book value terms, and a firm s internally-generated operating cash flows are the primary source of debt repayment, a company s market value incorporates investors expectations regarding cash flows and debt repayment. Furthermore, market value also incorporates the value of assets not presently producing cash flows, including those that may be sold to repay debt. Thus, the market value of the firm is also a key determinant of a firm s debt capacity and borrowing cost. Anyone with mortgage borrowing experience knows that, in financing a purchase or refinancing an existing mortgage, the amount of mortgage relative to a house s market value ( loan-to-value ratios ) is critical for the lenders. The same observation applies to corporate lending and borrowing as well. Hill Testimony, p. 1.

28 1 1 1 Docket: A Exhibit: SCE- Q. Please summarize this portion of your rebuttal. A. Financial leverage should be measured on a market value basis, not a book value basis. Increased financial leverage generates higher financial risk, which increases the required rate of return on equity for investors. Therefore, to the extent that the market value capital structures of the sample companies differ among themselves, the estimated ROEs must be adjusted so that they are comparable. Likewise, any difference between the market capital structure relied upon to estimate the ROE and the regulatory capital structure of the target entity must be recognized in the company s allowed ROE. The intervenors arguments that there is no need for such financial leverage adjustments are thus flawed. The leverage adjustment for financial risk should not be confused with the market-to-book ratio adjustment referred to by Mr. Hill and Dr. Woolridge. The relevance of the market-to-book ratio is discussed in the next section. 1 IV. THE MARKET-TO-BOOK RATIO TEST CANNOT BE RIGHT Q. What do you cover in this section of your testimony? A. This section of my testimony addresses whether a market-to-book ratio above 1.0 indicates whether investors in a utility company expect to earn a return on equity in excess of the company s cost of equity (the market-to-book ratio test ). Mr. Hill and Dr. Woolridge believe that the market-to-book ratio test is valid. They thus claim that Dr. Hunt s upwards leverage adjustment to SCE s ROE, which reflects a market-tobook ratio above 1, overstates the appropriate return on equity for the company. I respond to this argument below. Q. What portions of the intervenors respective testimonies are you addressing in this section? A. The relevant section in Dr. Woolridge s testimony is between pages 1 and 1, and 1 to of his testimony, as well as Attachment JRW-. Mr. Hill s discussion of the See Dr. Woolridge s characterization of Dr. Hunt s leverage adjustment at pp. 1- of his testimony. Much of the testimony below relies upon discussion with my Brattle partner, Dr. A. Lawrence Kolbe.

29 Docket: A Exhibit: SCE- relationship between the market-to-book ratio, the expected book return and the cost of equity capital is on pages 1 to 1 of his testimony. Q. What is the market-to-book ratio test? A. The market-to-book ratio is supposed to indicate whether investors in a utility company expect to earn more or less than their cost of capital. In particular, for a utility regulated on a book value rate base, a market-to-book ratio of 1.0 is supposed to indicate an expected rate of return on the book rate base equal to the utility s cost of capital. The test is based on the assumption that the value of a utility s stock equals the present value of the returns on (i.e., earnings) and of (i.e., depreciation) a rate base equal to the net book value of the utility s equity. 0 present value or net present value model. Q. Is the market-to-book ratio test valid? This is the standard A. No. Although the assumption seems reasonable and was widely accepted in the past, it is not correct. Even when it was viewed as correct, there were several caveats. Q. What were the caveats concerning the use of the market-to-book ratio test? A. First, the test could work only for companies that consisted entirely of regulated businesses with a rate base equal to net book value. The test never was believed to work for unregulated businesses. The pattern of cash flows over the life of an unregulated investment is quite different from that of an investment regulated on a net book value rate base. 1 In a competitive equilibrium with inflation, market values will generally exceed book values for unregulated firms. The deviations may be even greater in the actual world. Second, even for (1) a pure-play utility with a rate base equal to net book value, with () a true market asset pricing model that would yield a market-to-book ratio of one 0 1 See, for example, A. Lawrence Kolbe, James A. Read, Jr., and George R. Hall, 1, The Cost of Capital, Estimating the Rate of Return for Public Utilities, The MIT Press, pp. - and pp. -1. See, for example, Stewart C. Myers, A. Lawrence Kolbe and William B. Tye, Inflation and Rate of Return Regulation, Research in Transportation Economics, Volume II., Greenwich, CT: JAI Press, Inc. (1).

30 Docket: A Exhibit: SCE- for such a utility in equilibrium, the regulatory process may act with a lag that leaves market-to-book ratios substantially different from one for long periods of time. Third, even for (1) a pure-play utility with a rate base equal to net book value, with () a true market asset pricing model that would yield a market-to-book ratio of one for such a utility in equilibrium, regulators could not try consciously to target a market-to-book ratio of one in setting the allowed rate of return. The reason is that once investors discovered this policy (whether through public pronouncements or analysis of the results of confidential deliberations), investors would take it into account in pricing the stock. That would change the market-to-book ratio, thereby contaminating the information regulators would need to implement the policy. Regulation that consciously tries to set an allowed rate of return that makes the market-to-book ratio equal to one is circular. This circularity existed even before the market taught us that we could no longer believe in the market-to-book test, and even for companies in circumstances that we would have believed would make the market- to-book test valid. Q0. What has now caused you and other analysts to conclude that the market-to- book ratio test does not work? A0. The stock market has taught us that the true, unknown, model or models that drive stock prices is (are) more complicated than the simple models that give rise to the market-to-book test. That means we can no longer trust that the market-to-book test would actually work even for a pure-play utility regulated entirely on a rate base equal to net book value, in equilibrium. Specifically, the stock market forced many analysts to change their view of the value of the market-to-book ratio for a steady- state, pure play utility with a book value rate base when it crashed in October 1. The stock market bubble of the late s and 000 has only reinforced this conclusion. Here I have in mind my two Brattle partners; Dr. Kolbe, mentioned above, and Professor Stewart C. Myers of MIT.

31 Docket: A Exhibit: SCE- In an attempt to explain how the market's level could change so much in such a short period, Prof. Stewart C. Myers wrote a paper that argues that the stock market is good at getting relative prices right, because a great deal of money can be made in riskless arbitrage if securities are mispriced relative to one another. However, the stock market is not able to get absolute prices right, except in a fuzzy way. The market-to-book ratio purports to be a test of absolute value for utilities. If the stock market can get relative prices right, and if any stock has a reliable test for its absolute value, then all stocks will be priced right relative to it, and all stocks will be priced right in absolute value, too. If this were true, the stock market wouldn't have crashed in October 1, nor would the turn-of-the-century tech bubble have happened. Since those events did happen, the supposed test of absolute value for utilities, i.e., the market-to-book ratio test, must not be valid. The unknown true model(s) of stock market prices in practice must be richer and more complicated than assumed in the simple derivation of the market-to-book test. Q1. Can the other potential problems you mentioned explain current market-to- book ratios in ways that preserve the market-to-book ratio test? A1. No. For example, I believe that in recent years there have been companies that are essentially entirely regulated utilities with market-to-book ratios in the 1. to.0 range. Those numbers are too high to be the result of regulatory lag in, for example, Stewart C. Myers, Fuzzy Efficiency, Institutional Investor, December 1. Nobel laureate Paul A. Samuelson expressed a related view in a letter to Profs. Robert Shiller and John Campbell: Modern markets show considerable micro efficiency (for the reason that the minority who spot aberrations from micro efficiency can make money from those occurrences and, in doing so, they tend to wipe out any persistent inefficiencies). In no contradiction to the previous sentence, I had hypothesized considerable macro inefficiencies, in the sense of long waves in the time series of aggregate indexes of security prices below and above various definitions of fundamental values.... Long swings are long in time but that doesn t get them corrected with increasing confidence on the part of observing scientist. Quoted from Robert J. Shiller, Irrational Exuberance, New York: Broadway Books (001), p., emphases in the original.

32 Docket: A Exhibit: SCE- commissions adjusting the allowed rate of return on equity in response to declining interest rates. Q. Why do you say that, when interest rates have been coming down for quite a while now? Could not it be that for utilities, at least, the basic model still fully explains stock prices and the market-to-book ratios we observe are simply a result of a slow adjustment of allowed rates of return to interest rate declines? A. Unfortunately, such a view is not supportable. Suppose you observe a pure-play utility with a book value rate base and a market-to-book ratio equal to.0. Then investors are paying $ now for stock value that will be brought down to $1 as soon as regulators catch up with the interest rate declines. That amounts to a 0 percent return on the initial investment, which under this assumption must be recovered through the excess of the allowed rate of return over the cost of capital during the years before regulators catch up. Put this way, the notion seems implausible on its face. But we can be more quantitative about why the explanation of regulatory lag is unsupportable. Q. Please explain. A. Assume that the market-to-book ratio test worked, that a cost of capital analyst estimated the cost of equity at percent, and that the relevant commission accepted the estimate and set the allowed rate of return at percent. However, suppose the utility s market-to-book ratio is, which if the market-to-book tests were valid would signal that percent is above the cost of equity. Suppose also that the book value of the utility is expected to grow at a long-term annual rate of. percent. Lastly, suppose that investors expected an extreme form of regulatory lag: regulators will leave allowed rates of return at the current percent level for X years. On the last day of the X th year, regulators will readjust the allowed rate of return down to the cost of equity, so the market-to-book ratio goes down to 1.0 on that day. In short, the assumptions are that (1) investors put up $ now for every $1 of book equity rate base, () earn an allowed rate of return of percent (which by hypothesis is above the cost of capital) on the equity rate base (which grows at. percent per year) for X 0

33 1 Docket: A Exhibit: SCE- years, and () then end up with a stock value equal to only to the book value rate base. Thus, they lose 0 percent of their original investment after X years. If the market-to-book ratio test is assumed valid, the discount rate that makes the present value of these hypothesized returns equal to twice the book value of the stock is the utility s true cost of equity. Figure plots the implied true cost of equity associated with values of X running out to 0 years. As benchmarks, it adds the hypothesized percent allowed rate of return on equity and assumes long-term Treasury bond rates of percent. Market-to-Book Test Implies an Unrealistic True Cost of Equity (CoE) (Allowed RoR on Book Equity = Estimated Cost of Equity = %. M/B Ratio Falls from.0 to 1.0 at the End of the Year Indicated on the X-Axis.) 1% %.0% Benchmarks: 1. Estimated CoE = Allowed RoR =.0%. Treasury Bond Yield =.0%. Implied Cost of Equity % 0% -% -% -1%.0% Conclusion: The view that high M/B ratios merely reflect regulatory lag is invalid. The true cost of equity implied by that view is far too low. True CoE if market-to-book test were valid and if initial M/B ratio of.0 fell to 1.0 at end of year shown on x-axis Year Market-to-Book Ratio Drops to 1.0 (i.e., "X" in the text) Figure Q. Please explain what Figure is showing. A. The curving line indicated by long dashes with boxes (which is blue in color copies of this testimony) plots the true cost of capital as the length of regulatory lag (i.e., X ) grows from three years (the first value shown) to 0 years. With a loss of 0 percent of the original investment due to the end of regulatory lag, X must exceed years for the true cost of equity even to be positive. It takes approximately to 1 years 1

34 Docket: A Exhibit: SCE- before the true cost of equity even equals the very low Treasury bond rate of percent. Since the actual cost of equity must be well above the Treasury bond rate, regulatory lag cannot be the explanation for the market-to-book ratios we actually observe. Q. But suppose investors expect that regulators would never adjust allowed rates of return for the fall in interest rates in recent years. That is, suppose they believe the regulatory lag you just discussed is many decades long. Does that save the market-to-book test? A. No. If investors expected regulators to ignore falling interest rates for many decades, the implied true cost of equity would keep climbing as X gets further into the future, although it always would remain materially below the hypothesized percent estimate of the cost of equity. It would be. percent with an X of 0 years, for example. But saving the market-to-book test by assuming that regulators effectively never react to the fall in interest rates is a cure that is worse than the disease. Nor is such an assumption supported by experience. Allowed rates of return for rate regulated companies were far higher in the s, when interest rates were higher, than they are today. Yet the s are a mere three decades ago. I would submit that it is far more plausible, after the experience of recent years, to believe that we do not understand the way stock prices are set than to believe that (1) we can model the stock price process exactly, but () investors today believe that regulators will ignore the implications of falling interest rates forever. The top two lines in the figure, with small dashes (in green in color copies of this testimony), are the allowed rate of return on equity of percent and the assumed long-term bond rate of percent. Reportedly, even Professor Eugene Fama has reached the conclusion that stocks can sometimes be irrationally priced. See As Two Economists Debate Markets, The Tide Shifts; Belief in Efficient Valuation Yields Ground to Role Of Irrational Investors The Wall Street Journal, October 1, 00, p. A-1. Of course, we cannot be sure whether (1) the market is priced irrationally or () the market is priced rationally but is in accord with some model or set of models we do not yet understand. Either way, however, we can no longer rely on the market-to-book test.

35 Docket: A Exhibit: SCE- Q. Dr. Woolridge argues in his testimony that the reason that market values exceed book values is that the company is earning a return on equity in excess of its cost of equity, and presents evidence in his Attachment JRW- demonstrating that there is a strong positive relationship between expected returns on common equity and market-to-book ratios for public utilities. Do you agree? A. I do not. Mathematically, all else equal, a higher return on equity gives rise to a higher market value of equity, and a higher market-to-book ratio. However, all else is not equal in real life. Dr. Woolridge provides very little information on how Attachment JRW- is created, but if Dr. Woolridge intends for Attachment JRW-, which graphically shows positive correlation between a utility s estimated ROE and its market-to-book ratio, to support his contention, the empirical evidence falls short. The positive correlation simply shows that the price/earnings ( P/E ) ratio is positive for the utility companies. To see this, one can multiply book value of equity by the market-to-book ratios and estimated ROEs (which are the ratio of earnings to book equity value) to obtain market value of the stock and the company s accounting earnings. In other words, the slope of the scatterplot is an estimate of the sample average P/E ratio. A positive P/E ratio is not a surprising result, nor does it provide support to Dr. Woolridge s contention that above-market returns on equity, and no other factors, contribute to the utilities market value exceeding book value. Furthermore, from basic statistics, correlation does not imply a cause-and-effect relationship. There are a number of economic issues with Dr. Woolridge s graphical demonstration. First, Dr. Woolridge s estimated ROEs do not measure the cost of capital. They appear to be accounting returns on book value of equity, which reflect accounting convention. In addition, accounting ROEs do not measure the change in stock value, which is also part of the economic return in owning a stock. Second, lack of time dimension in the graphs does not permit one to interpret the relationship between the two variables as to whether higher ROEs lead to higher market-to-book

36 Docket: A Exhibit: SCE- ratios, or higher market-to-book ratios imply higher business risks and hence higher returns on equity. Lastly, even if economic causality could be established, the bilateral correlation in Attachment JRW- fails to control for other reasons that could contribute to a positive relationship between high ROEs and high market-to-book ratios. Q. What are the other factors that could contribute to higher market-to-book ratios? A. A careful study of the causal relationship between allowed return on equity and market-to-book ratios requires better specification of the regression form, and measurement of the relevant variables. Here I offer a few factors that Dr. Woolridge failed to consider. First, although all the companies in Dr. Woolridge s samples have regulated utility operations, some of the companies have lines of business not subject to regulation. Non-regulated operations could be risker and have growth options that are typically not present in utilities. Second, utilities are only allowed a fair opportunity to earn their cost of capital. Actual returns on and of capital depend on factors outside utilities control, such as fluctuation in consumer demand, supply shocks, weather, regulatory environment, etc. Third, investor demand for safe haven investment during the financial crisis and economic downturn could also boost the market-to-book ratios of utilities. (JRW- does not specify the time frame of the data). Fourth, except for accounting artifacts, estimated accounting returns on equity could also be affected by rate freezes, regulatory lags in adjusting the rates or deviation of other rate components (such as depreciation) from economic reality. All these factors could affect a utility s accounting ROE, but they have nothing to do with the utility s cost of capital. Q. What other comments do you have on Dr. Woolridge s Attachment JRW-? A. Data presented in Attachment JRW- show a number of companies with estimated ROEs below percent, yet with market-to-book ratios above one, some approaching two. If Dr. Woolridge is right, the return on equity on these utilities should be adjusted substantially downward. However, this is inconsistent with Dr. Woolridge s

37 Docket: A Exhibit: SCE- recommended. percent reasonable cost of equity for SCE. Estimated ROEs in excess of 1 percent in the attachment also raise the red flag that these ROEs are not the correct proxy for utilities allowed returns on equity. If Dr. Woolridge s hypothesis is correct, the cost of equity for an all equity utility with a market-to-book ratio of 1 would be in the range of to percent or so (based upon projecting the regression lines in Attachment JRW-), which is similar to the cost of debt. Q. Please summarize this portion of your rebuttal. A. It turns out that stock prices are more complicated than our simple models can encompass. As a result, the market-to-book ratio test lacks a firm conceptual foundation. Moreover, the levels of utility market-to-book ratios observed in recent years are simply too high to be the result of rational pricing based on the present value formula that underlies the market-to-book test. Finally, there is no convincing empirical evidence of a significant positive relationship between expected returns on equity and market-to-book ratios for public utilities. I therefore refute the intervenors main argument that a market-to-book ratio above 1 indicates that a given company is earning a return on equity in excess of its cost of equity. Q0. What do you believe regulators should do about the market-to-book ratio? A0. I believe regulators should focus on setting the allowed return according to the best evidence available and leave the market-to-book ratio to whatever (currently incompletely understood) forces drive the stock prices of the individual sample companies and the market as a whole. V. THE EFFECT OF DECOUPLING ON THE COST OF CAPITAL Q1. Are you one of the authors of the study on the effect of decoupling on the cost of capital referenced by witnesses Widjaha (SDG&E) and Shepherd (SCG) and critiqued by Mr. Hill? A1. Yes. The study is called The Impact of Decoupling on the Cost of Capital: An Empirical Investigation, published March 0. I have attached a revised version of the paper, dated July 01, as Appendix B to this rebuttal testimony.

38 Docket: A Exhibit: SCE- Q. What changes have you made in the revised version of the paper? A. There is only one substantive change in the revised version of the paper. The change is the classification of one company from not decoupled to decoupled, which had the effect of removing the result in the original version that the decoupled companies were associated with a small but statistically significant higher cost of capital in two of the five tests we conducted than companies not decoupled. The results now show that there is no statistically significant difference in the estimated cost of capital for companies with or without decoupling. Q. Does the study version or the revised version of the paper claim that the adoption of decoupling increases the cost of capital as suggested by Mr. Hill? A. No. Mr. Hill is incorrect for at least three reasons. First, from a purely statistical point of view, regression analysis does not prove causation even if there had been a statistically significant difference in the cost of capital of the two groups. Second, it is not illogical on its face to find that companies with decoupling have a higher estimated cost of capital. Although the study did not find that situation, it is perfectly plausible that the circumstances under which decoupling policies are adopted could be circumstances under which the cost of capital would increase substantially, but for the adoption of decoupling. In other words, decoupling may be a regulatory response to increased risk from the implementation of conservation policies. Finally, the revised version shows no difference in the estimated cost of capital in any case. Q. What are Mr. Hill s other criticisms of your decoupling study? A. Mr. Hill suggests that the study a. Is biased because the cost of capital estimates in the study were performed by members of The Brattle Group, Hill Testimony, p.. Hill Testimony, p.. Hill Testimony, pp. -.

39 Docket: A Exhibit: SCE- b. Is based upon gas local distribution companies ( LDCs ) which may have as much as 0 percent of earnings from unregulated operations, c. Was conducted during the 00/00 financial meltdown period and is therefore unreliable, d. Included companies with different forms of decoupling mechanisms making it impossible to determine the effect of decoupling, and e. Uses the ATWACC technique to control for differences in financial risk. Q. Are these criticisms valid? A. No. Some of Mr. Hill s criticisms contain descriptions that are accurate, but he draws incorrect conclusions from those descriptions. Q. Please explain why Mr. Hill s conclusions are not accurate. A. I will take each of Mr. Hill s comments in the order listed above. First, it is true that I performed all of the cost of capital estimates reported in the paper, but whether any company was decoupled or not was not a factor at the time I estimated its cost of capital. All of the estimates are from proceedings in which I was a cost of capital witness and are matters of public record. I made no adjustment for whether a sample company was or was not decoupled. There was no opportunity to inject bias into the analyses because the division of companies into the decoupled and not decoupled categories was done after the estimates had been filed. Second, the test was indeed conducted on gas LDCs, precisely because the gas LDC sample is the closest to a pure play sample of regulated companies that is available. My cut off for inclusion in the sample is at least 0 percent regulated assets or operating income, but most of the sample companies have a much higher percentage than 0 percent. The average is around 0 percent. Third, the effect of the financial meltdown on cost of capital estimates is something which I have also been concerned about, as any review of my testimonies during the period would show. However, I believe that whatever the effect of the financial crisis

40 Docket: A Exhibit: SCE- may have been on the estimates that it would affect the estimates for sample companies similarly. In fact, to the extent that decoupling is expected to reduce risk, it is precisely during a financial crisis that the effect would be most pronounced. In any case, the study did not show a lower cost of capital for decoupled companies. Fourth, Mr. Hill is correct that decoupling measures are not standardized, nor are the sample companies either purely decoupled or not. These are inherent limitations in the data. As we noted in the paper, some sample companies contain subsidiaries that are decoupled and some that are not. It is for that reason that we used an index of decoupling to capture the effect of differences in decoupling status within a holding company. 0 Finally, Mr. Hill s objection to the use of the ATWACC method to control for financial risk is completely without merit. As noted above, there is no dispute in the academic community that the cost of equity estimated by the standard models is affected by both financial risk and market risk and that financial risk is properly measured by the market value capital structure. It would be a mistake to ignore differences in financial risk when testing for the effect of decoupling on the cost of capital, because it would be impossible to know whether any differences in the estimated returns on equity were due to decoupling or financial risk. Nor is it relevant that regulators apply the allowed return on equity to a capital structure measured at book value. The question addressed by the paper was whether the decoupling is associated with a lower estimated cost of capital. The answer to that question is no, there is no evidence to support that supposition. Q. Do you have any concluding remarks about your decoupling study? A. Yes. All of the factual issues raised by Mr. Hill were acknowledged in the study. Nevertheless, the only other study than the one by The Brattle Group of which I am aware that attempts to answer the question empirically of whether decoupling is 0 The index uses the volume of gas distributed by decoupled subsidiaries compared to the total gas distributed by the holding company. The index ranged from zero for a holding company with no subsidiaries with decoupling to 1.0 for a holding company with all subsidiaries decoupled.

41 Docket: A Exhibit: SCE- associated with a reduced cost of capital is a study by Professor Richard Michelfelder of Rutgers University and AUS Consultants. 1 This study reaches the same conclusions as The Brattle Group s study. If decoupling had a large effect on a company s cost of capital, I would expect it to be reflected in the cost of capital estimates, but it was not. The Brattle Group s study concludes that if decoupling reduces the cost of capital, the effect cannot be large because the study was unable to detect any reduction. Q. Is your conclusion that decoupling has no effect on a company s risk? A. No. Decoupling reduces the variability of a company s revenues which is likely to reduce the company s total risk. This is why credit rating agencies generally regard the adoption of decoupling policies as favorable to supporting a company s credit rating. Investors in bonds are concerned about a company s total risk while investors in equity are concerned about a company s systematic risk. In other words, the risk reduction from decoupling is likely to show up in a company s cost of debt even if it does not affect the systematic risk of a company. All else equal, credit ratings will be higher and debt costs lower for companies with decoupling in place. Finally, it is worth noting that decoupling policies serve a valuable purpose. They enhance the effectiveness of conservation policies by alleviating the potential conflict of interest for a regulated company to reduce its sales volumes to achieve conservation even though the company recovers its costs and earns a return through sales. 1 Decoupling: Impact on the Risk of Public Utility Stocks, Richard A. Michelfelder, Clinical Associate Professor of Finance, Rutgers University & Managing Consultant, AUS Consultants, for the Society of Utility Regulatory and Financial Analysts, April 1, 0. For example, the Lawton Testimony cites to several credit rating reports that point to decoupling, amongst other cost recovery mechanisms, as supportive of credit quality for utilities. See, for example, Moody s Investor Services: Cost Recovery Provisions Key to Investor Owned Utility Ratings and Credit Quality; Special Comment (June 1, 0) at 1 and, Moody s Investor Services; Ratings Methodology: Regulated Electric and Gas Utilities (August 00) at, and Moody s Investor Services; Global Credit Research Credit Opinion; Southern California Edison Company (July, 01) at.

42 1 Docket: A Exhibit: SCE- Q. Does this conclude your rebuttal testimony? A. Yes, it does. 0

43 Appendix A to the Page A-1 of 1 APPENDIX A: QUALIFICATIONS OF MICHAEL J. VILBERT Michael Vilbert is an expert in cost of capital, financial planning and valuation who has advised clients on these matters in the context of a wide variety of investment and regulatory decisions. He has testified or submitted testimony on cost of capital, economic damages, the business purpose and economic substance of tax related transactions, valuation of assets in arbitration and the effect of regulatory policy changes on the cost of capital. He received his Ph.D. in Financial Economics from the Wharton School of the University of Pennsylvania, an MBA from the University of Utah, an M.S. from the Fletcher School of Law and Diplomacy, Tufts University, and a B.S. degree from the United States Air Force Academy. He joined The Brattle Group in 1 after a career as an Air Force officer, where he served as a fighter pilot, intelligence officer, and professor of finance at the Air Force Academy. REPRESENTATIVE CONSULTING EXPERIENCE $ Dr. Vilbert served as the consulting expert in several cases for the U.S. Department of Justice and the Internal Revenue Service regarding the business purpose and economic substance of a series of tax related transactions. These projects required the analysis of a complex series of financial transactions including the review of voluminous documentary evidence and required expertise in financial theory, financial market as well as accounting and financial statement analysis. $ In a securities fraud case, Dr. Vilbert designed and created a model to value the private placement stock of a drug store chain as if there had been full disclosure of the actual financial condition of the firm. He analyzed key financial data and security analysts = reports regarding the future of the industry in order to recreate pro forma balance sheet and income statements under a variety of scenarios designed to establish the value of the firm. $ For pharmaceutical companies rebutting price-fixing claims in antitrust litigation, Dr. Vilbert was a member of a team that prepared a comprehensive analysis of industry profitability. The analysis replicated, tested and critiqued the major recent analyses of drug costs, risks and returns. The analyses helped develop expert witness testimony to rebut allegations of excess profits. $ For an independent electric power producer, Dr. Vilbert created a model that analyzed the reasonableness of rates and costs filed by a natural gas pipeline. The model not only duplicated the pipeline=s rates, but it also allowed simulation of a variety of Awhat if@ scenarios associated with cost recovery under alternative time patterns and joint cost allocations. Results of the analysis were adopted by the intervenor group for negotiation with the pipeline. A-1

44 Appendix A to the Page A- of 1 $ For the CFO of an electric utility, Dr. Vilbert developed the valuation model used to support a stranded cost estimation filing. The case involved a conflict between two utilities over the responsibility for out-of-market costs associated with a power purchase contract between them. In addition, he advised and analyzed cost recovery mechanisms that would allow full recovery of the stranded costs while providing a rate reduction for the company=s rate payers. $ Dr. Vilbert has testified as well as assisted in the preparation of testimony and the development of estimation models in numerous cost of capital cases for natural gas pipeline, water utility and electric utility clients before the Federal Energy Regulatory Commission (AFERC@) and state regulatory commissions. These have spanned standard estimation techniques (e.g., Discounted Cash Flow and Risk Positioning models). He has also developed and applied more advanced models specific to the industries or lines of business in question, e.g., based on the structure and risk characteristics of cash flows, or based on multi-factor models that better characterize regulated industries. $ Dr. Vilbert has valued several large, residual oil-fired generating stations to evaluate the possible conversion to natural gas or other fuels. In these analyses, the expected pre- and post-conversion station values were computed using a range of market electricity and fuel cost conditions. $ For a major western electric utility, Dr. Vilbert helped prepare testimony that analyzed the prudence of QF contract enforcement. The testimony demonstrated that the utility had not been compensated in its allowed cost of capital for major disallowances stemming from QF contract management. $ Dr. Vilbert analyzed the economic need for a major natural gas pipeline expansion to the Midwest. This involved evaluating forecasts of natural gas use in various regions of the United States and the effect of additional supplies on the pattern of natural gas pipeline use. The analysis was used to justify the expansion before the FERC and the National Energy Board of Canada. $ For a Public Utility Commission in the Northeast, Dr. Vilbert analyzed the auction of an electric utility=s purchase power agreements to determine whether the outcome of the auction was in the ratepayers= interest. The work involved the analysis of the auction procedures as well as the benefits to ratepayers of transferring risk of the PPA payments to the buyer. $ Dr. Vilbert led a team tasked to determine whether bridge tolls were "just and reasonable" for a non-profit port authority. Determination of the cost of service for the authority required estimation of the value of the authority's assets using the trended original cost methodology as well as evaluation of the operations and maintenance budgets. Investment costs, bridge traffic information and inflation indices covering a year period were utilized to estimate the value of four bridges A-

45 Appendix A to the Page A- of 1 and a passenger transit line valued in excess of $1 billion. $ Dr. Vilbert helped a recently privatized railroad in Brazil develop an estimate of its revenue requirements, including a determination of the railroad=s cost of capital. He also helped evaluate alternative rate structures designed to provide economic incentives to shippers as well as to the railroad for improved service. This involved the explanation and analysis of the contribution margin of numerous shipper products, improved cost analysis and evaluation of bottlenecks in the system. $ For a utility in the Southeast, Dr. Vilbert quantified the company=s stranded costs under several legislative electric restructuring scenarios. This involved the evaluation of all of the company=s fossil and nuclear generating units, its contracts with Qualifying Facilities and the prudence of those QF contracts. He provided analysis concerning the impact of securitizing the company=s stranded costs as a means of reducing the cost to the ratepayers and several alternative designs for recovering stranded costs. $ For a recently privatized electric utility in Australia, Dr. Vilbert evaluated the proposed regulatory scheme of the Australian Competition and Consumer Commission for the company=s electric transmission system. The evaluation highlighted the elements of the proposed regulation which would impose uncompensated asymmetric risks on the company and the need to either eliminate the asymmetry in risk or provide additional compensation so that the company could expect to earn its cost of capital. $ For an electric utility in the Southwest, Dr. Vilbert helped design and create a model to estimate the stranded costs of the company=s portfolio of Qualifying Facilities and Power Purchase contracts. This exercise was complicated by the many variations in the provisions of the contracts that required modeling in order to capture the effect of changes in either the performance of the plants or in the estimated market price of electricity. $ Dr. Vilbert helped prepare the testimony responding to a FERC request for further comments on the appropriate return on equity for electric transmission facilities. In addition, Dr. Vilbert was a member of the team that made a presentation to the FERC staff on the expected risks of the unbundled electric transmission line of business. $ Dr. Vilbert and Mr. Frank C. Graves, also of The Brattle Group, prepared testimony evaluating an innovative Canadian stranded cost recovery procedure involving the auctioning of the output of the province=s electric generation plants instead of the plants themselves. The evaluation required the analysis of the terms and conditions of the long-term contracts specifying the revenue requirements of the plants for their entire forecasted remaining economic life and required an estimate of the cost of capital for the plant owners under this new stranded cost recovery concept. $ Dr. Vilbert served as the neutral arbitrator for the valuation of a petroleum products A-

46 Appendix A to the Page A- of 1 tanker. The valuation required analysis of the Jones Act tanker market and the supply and demand balance of the available U.S. constructed tanker fleet. $ Dr. Vilbert evaluated the appropriate Abareboat@ charter rate for an oil drilling platform for the renewal period following the end of a long-term lease. The evaluation required analysis of the market for oil drilling platforms around the world including trends in construction and labor costs and the demand for platforms in varying geographical environments. PRESENTATIONS AUtility Distribution Cost of Capital,@ EEI Electric Rates Advanced Course, Bloomington, IN, 00, 00. AIssues for Cost of Capital Estimation,@ with Bente Villadsen, Edison Electric Institute Cost of Capital Conference, Chicago, IL, February 00. ANot Your Father=s Rate of Return Methodology,@ Utility Commissioners/Wall Street Dialogue, NY, May 00. AUtility Distribution Cost of Capital,@ EEI Electric Rates Advanced Course, Madison, WI, July 00. ACost of Capital Estimation: Issues and Answers,@ MidAmerican Regulatory Finance Conference, Des Moines, IA, April, 00. ACost of Capital - Explaining to the Commission - Different ROEs for Different Parts of the Business,@ EEI Economic Regulation & Competition Analysts Meeting, May, 00. ACurrent Issues in Cost of Capital,@ with Bente Villadsen, EEI Electric Rates Advanced Course, Madison, WI, 00. ACurrent Issues in Estimating the Cost of Capital,@ EEI Electric Rates Advanced Course, Madison, WI, 00, 00, 00, 00, 0 and 0. ARevisiting the Development of Proxy Groups and Relative Risk Analysis,@ Society of Utility and Regulatory Financial Analysts: th Financial Forum, April 00. ACurrent Issues in Explaining the Cost of Capital to Utility Commissions@ Cost of Capital Seminar, Philadelphia, PA, 00. Impact of the Ongoing Economic Crisis on the Cost of Capital of the U.S. Utility Sector, New York Public Service Commission, Albany, NY, April 0, 00. Impact of the Ongoing Economic Crisis on the Cost of Capital of the U.S. Utility Sector, National A-

47 Appendix A to the Page A- of 1 Association of Water Companies: New York Chapter, Albany, NY, May 1, 00. ARTICLES "Flaws in the Proposed IRS Rule to Reinstate Amortization of Deferred Tax Balances Associated with Generation Assets Reorganized in Industry Restructuring," by Frank C. Graves and Michael J. Vilbert, white paper for Edison Electric Institute (EEI) to the IRS, July, 00. "The Effect of Debt on the Cost of Equity in a Regulatory Setting," by A. Lawrence Kolbe, Michael J. Vilbert, Bente Villadsen and The Brattle Group, Edison Electric Institute, April 00. "Measuring Return on Equity Correctly: Why current estimation models set allowed ROE too low," by A. Lawrence Kolbe, Michael J. Vilbert and Bente Villadsen, Public Utilities Fortnightly, August 00. "Understanding Debt Imputation Issues,@ by Michael J. Vilbert, Bente Villadsen and Joseph B. Wharton, Edison Electric Institute, August 00. Review of Regulatory Cost of Capital Methodologies, (with Bente Villadsen and Matthew Aharonian), Canadian Transportation Agency, September 0. The Impact of Decoupling on the Cost of Capital An Empirical Study, Joseph B. Wharton, Michael J. Vilbert, Richard E. Goldberg, and Toby Brown, Discussion Paper, The Brattle Group, March 0. TESTIMONY Direct and rebuttal testimony before the Alberta Energy and Utilities Board on behalf of TransAlta Utilities Corporation in the matter of an application for approval of its 1 and 000 generation tariff, transmission tariff, and distribution revenue requirement, Docket U0, October 1. Direct testimony before the Federal Energy Regulatory Commission on behalf of Central Maine Power in Docket No. ER , December 1. Direct testimony before the Alberta Energy and Utilities Board on behalf of TransAlta Utilities Corporation for approval of its 001 transmission tariff, May 000. Direct testimony before the Federal Energy Regulatory Commission on behalf of Mississippi River Transmission Corporation in Docket No. RP , March 001. A-

48 Appendix A to the Page A- of 1 Written evidence, rebuttal, reply and further reply before the National Energy Board in the matter of an application by TransCanada PipeLines Limited for orders pursuant to Part I and Part IV of the National Energy Board Act, Order AO-1-RH--001, May 001, Nov. 001, Feb. 00. Written evidence before the Public Utility Board on behalf of Newfoundland & Labrador Hydro - Rate Hearings, October 001, Order No. P.U. (00-00), dated June 00. Direct testimony (with William Lindsay) before the Federal Energy Regulatory Commission on behalf of DTE East China, LLC in Docket No. ER , April 00. Direct and rebuttal reports before the Arbitration Panel in the arbitration of stranded costs for the City of Casselberry, FL, Case No. 00-CA-0-1-L, July 00. Direct reports before the Arbitration Board for Petroleum products trade in the Arbitration of the Military Sealift Command vs. Household Commercial Financial Services, fair value of sale of the Darnell, October 00. Direct testimony and hearing before the Arbitration Panel in the arbitration of stranded costs for the City of Winter Park, FL, In the Circuit Court of the Ninth Judicial Circuit in and for Orange County, FL, Case No. C1-01--, December 00. Direct testimony before the Federal Energy Regulatory Commission on behalf of Florida Power Corporation, dba Progress Energy Florida, Inc. in Docket No. SC , March 00. Direct report before the Arbitration Panel in the arbitration of stranded costs for the Town of Belleair, FL, Case No C1-00, April 00. Direct and rebuttal reports before the Alberta Energy and Utilities Board in the matter of the Alberta Energy and Utilities Board Act, R.S.A. 000, c. A-1, and the Regulations under it; in the matter of the Gas Utilities Act, R.S.A. 000, c. G-, and the Regulations under it; in the matter of the Public Utilities Board Act, R.S.A. 000, c. P-, as amended, and the Regulations under it; and in the matter of Alberta Energy and Utilities Generic Cost of Capital Hearing, Application No. 11, July 00, November 00, Decision 00-0, dated July 00. Written evidence before the National Energy Board in the matter of the National Energy Board Act, R.S.C. 1, c. NB, as amended, (Act) and the Regulations made under it; and in the matter of an application by TransCanada PipeLines Limited for orders pursuant to Part IV of the National Energy Board Act, for approval of Mainline Tolls for 00, RH--00, January 00. Direct and rebuttal testimony before the Public Service Commission of West Virginia, on Cost of Capital for West Virginia-American Water Company, Case No 0-0-W-T, May 00. Direct and rebuttal testimony before the Federal Energy Regulatory Commission on Energy Allocation of Debt Cost for Incremental Shipping Rates for Edison Mission Energy, Docket No. RP0--000, December 00 and March 00. A-

49 Appendix A to the Page A- of 1 Direct testimony before the Arizona Corporation Commission, Cost of Capital for Paradise Valley Water Company, a subsidiary of Arizona-American Water Company, Docket No. WS-0A-0, May 00. Written evidence before the Ontario Energy Board, Cost of Capital for Union Gas Limited, Inc., Docket No. EB-00-00, January 00. Direct and rebuttal testimony before the Pennsylvania Public Utility Commission, Return on Equity for Metropolitan Edison Company, Docket No. R-0001 and Pennsylvania Electric Company, Docket No. R-0001, April 00 and August 00. Expert report in the United States Tax Court, Docket No. -0, th Street Partners, DH Petersburg Investment, LLC and Mid-Atlantic Finance, Partners Other than the Tax Matters Partner, Petitioner, v. Commissioner of Internal Revenue, Respondent, July, 00. Direct and supplemental testimony before the Federal Energy Regulatory Commission, Docket No. ER0--00, on behalf of Mystic Development, LLC on the Cost of Capital for Mystic and Generating Plants Operating Under Reliability Must Run Contract, August 00 and September 00. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER0--000, on behalf of Northwestern Corporation on the Cost of Capital for Transmission Assets, October 00. Direct and rebuttal testimony before the Tennessee Regulatory Authority, Case No , on behalf of Tennessee American Water Company, on the Cost of Capital, November, 00 and April 00. Direct and rebuttal testimony before the Public Service Commission of Wisconsin, Docket No. - UR-, on behalf of Wisconsin Energy Corporation, on the Cost of Capital for Wisconsin Electric Power Company and Wisconsin Gas LLC, May 00 and October 00. Rebuttal testimony before the California Public Utilities Commission, Docket No. A , on behalf of California-American Water Company, on the Cost of Capital, May 00. Direct testimony before the Public Utilities Commission of the State of South Dakota, Docket No. NG-0-01, on behalf of NorthWestern Corporation, on the Cost of Capital for NorthWestern Energy Company=s natural gas operations in South Dakota, June 00. Direct, supplemental and rebuttal testimony before the Public Utilities Commission of Ohio, Case No. 0-1-EL-AIR, Case No. 0--EL-ATA, Case No. 0--EL-AAM, and Case No. 0-- EL-UNC, on behalf of Ohio Edison Company, The Toledo Edison Company, and The Cleveland Electric Illuminating Company, on the cost of capital for the FirstEnergy Company=s Ohio electric distribution utilities, June 00, January 00 and February 00. Direct testimony before the Public Service Commission of West Virginia, Case No. 0-0-W- T, on behalf of West Virginia American Water Company on cost of capital, July 00. A-

50 Appendix A to the Page A- of 1 Direct and rebuttal testimony before the State Corporation Commission of Virginia, Case No. PUE , on behalf of Virginia Electric and Power Company on the cost of capital for its southwest Virginia coal plant, July 00 and December 00. Direct and Supplemental testimony before the Public Utilities Commission of Ohio, Case No. 0- -GA-AIR, Case No. 0-0-GA-ALT, and Case No. 0-1-GA-AAM, on behalf of Dominion East Ohio Company, on the rate of return for Dominion East Ohio=s natural gas distribution operations, September 00 and June 00. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER to Docket No. ER0--00, on behalf of Virginia Electric and Power Company, on the Cost of Capital for Transmission Assets, October 00. Direct and rebuttal testimony before the California Public Utilities Commission, Docket No. A , on behalf of California-American Water Company, on the Effect of a Water Revenue Adjustment Mechanism on the Cost of Capital, October 00 and November 00. Written direct and reply evidence before the National Energy Board in the matter of the National Energy Board Act, R.S.C. 1, c. NB, as amended, and the Regulations made thereunder; and in the matter of an application by Trans Québec & Maritimes PipeLines Inc. ( TQM ) for orders pursuant to Part I and Part IV of the National Energy Board Act, for determining the overall fair return on capital for tolls charged by TQM, December 00 and September 00, Decision RH-1-00, dated March 00. Comments in support of The Interstate Natural Gas Association of America=s Additional Initial Comments on the FERC=s Proposed Policy Statement with regard to the Composition of Proxy Companies for Determining Gas and Oil Pipeline Return on Equity, Docket No. PL0--000, December, 00. Direct and rebuttal testimony on the Cost of Capital before the Tennessee Regulatory Authority, Case No , on behalf of Tennessee American Water Company, March and August 00. Post-Technical Conference Affidavit on behalf of The Interstate Natural Gas Association of America in response to the Reply Comments of the State of Alaska with regard the FERC=s Proposed Policy Statement on to the Composition of Proxy Companies for Determining Gas and Oil Pipeline Return on Equity, Docket No. PL0--000, March, 00 Direct and rebuttal testimony before the California Public Utilities Commission, Docket No. A , on behalf of California-American Water Company, concerning Cost of Capital, May 00 and August 00. Rebuttal testimony on the financial risk of Purchased Power Agreements, before the Public Utilities Commission of the State of Colorado, Docket No. 0A-E, in the matter of the application of Public Service Company of Colorado for approval of its 00 Colorado Resource Plan, June 00. A-

51 Appendix A to the Page A- of 1 Direct and rebuttal testimony before the Federal Energy Regulatory Commission, Docket No. RP , on behalf of El Paso Natural Gas Company, on the Cost of Capital for Natural Gas Transmission Assets, June 00 and August 00. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER0--000, on behalf of Virginia Electric and Power Company, on the incentive Cost of Capital for investment in New Electric Transmission Assets, June 00 Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER , on behalf of Public Service Electric and Gas Company, on the Cost of Capital for Electric Transmission Assets, July 00. Direct and rebuttal testimony before the Public Service Commission of West Virginia, Case No W-t, on behalf of West Virginia-American Water Company concerning the Cost of Capital for Water Utility assets, July 00 and November 00. Direct and rebuttal testimony before the Public Utilities Commission of Ohio, Case No. 0--EL- SSO, on behalf of Ohio Edison Company, The Toledo Edison Company, and The Cleveland Electric Illuminating Company, with regard to the test to determine Significantly Excessive Earnings within the context of Senate Bill No. 1, September 00 and October 00. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER0--000, on behalf of Public Service Electric and Gas Company, on the incentive Cost of Capital for Mid- Atlantic Power Pathway Electric Transmission Assets, November 00. Direct and rebuttal testimony before the Public Service Commission of West Virginia, Case No. 0-1-G-PC, on behalf of Dominion Hope Gas Company concerning the Cost of Capital for Gas Local Distribution Company assets, November 00 and May 00. Written Evidence before the Alberta Utilities Commission in the matter of the Alberta Utilities Commission Act, S.A. 00, c. A-., as amended, and the regulations made thereunder; and IN THE MATTER OF the Gas Utilities Act, R.S.A. 000, c. G-, as amended, and the regulations made thereunder; and IN THE MATTER OF the Public Utilities Act, R.S.A. 000, c. P-, as amended, and the regulations made thereunder; and IN THE MATTER OF Alberta Utilities Commission 00 Generic Cost of Capital Hearing, Application No. 11/Proceeding No.. 00 Generic Cost of Capital Proceeding on behalf of NGTL, November 00. Written and Reply Evidence before the Alberta Utilities Commission in the matter of the Alberta Utilities Commission Act, S.A. 00, c. A-., as amended, and the regulations made thereunder; and IN THE MATTER OF the Gas Utilities Act, R.S.A. 000, c. G-, as amended, and the regulations made thereunder; and IN THE MATTER OF the Public Utilities Act, R.S.A. 000, c. P-, as amended, and the regulations made thereunder; and IN THE MATTER OF Alberta Utilities Commission 00 Generic Cost of Capital Hearing, Application No. 11/Proceeding No.. 00 Generic Cost of Capital Proceeding on behalf of AltaGas Utilities Inc., November 00 and May 00. A-

52 Appendix A to the Page A- of 1 Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER0--000, on behalf of ITC Great Plains, LLC, on the Cost of Capital for Electric Transmission Assets, January 00. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER , on behalf of Green Power Express, LLP, on the Cost of Capital for Electric Transmission Assets, February 00. Written evidence before the Régie de l Énergie on behalf of Gaz Métro Limited Partnership, Cause Tarifaire 0, R-0-00, on the Cost of Capital for natural gas transmission assets, May 00. Direct and rebuttal testimony before the Public Service Commission of Wisconsin, Docket No. 0-UR-, on behalf of Wisconsin Power and Light Company, on the cost of capital for electric and natural gas distribution assets, May 00 and September 00. Direct and rebuttal testimony before the State of New Jersey Board of Public Utilities in the Matter of the Petition of Public Service Electric and Gas Company for Approval of an Increase in Electric and Gas Rates and for Changes in the Tariffs for Electric and Gas Service, B.P.U.N.J. No. 1 Electric and B.P.U.N.J No. 1 Gas Pursuant to N.J.S.A. :-1 and N.J.S.A. :-1.1 and for Approval of a Gas Weather Normalization Clause; a Pension Expense Tracker and for other Appropriate Relief BPU Docket No. GR000, June 00 and December 00. Rebuttal testimony before the Florida Public Service Commission in re: Petition for Increase in Rates by Progress Energy Florida, Inc., Docket No. 000-EI, August 00. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER-1-000, on behalf of Public Service Electric and Gas Company, on the incentive Cost of Capital for the Branchburg-Roseland-Hudson 00 kv Line electric transmission project ( BRH Project ), October 00. Direct and Rebuttal Testimony before the California Public Utilities Commission regarding cost of service for San Joaquin Valley crude oil pipeline on behalf of Chevron Products Company, Docket Nos. A.0-0-0, C , C and C.0-0-0, December 00 and April 0. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER-1-000, on behalf of South Caroline Gas and Electric Company, on the Cost of Capital for Electric Transmission Assets, December 00. Direct testimony before the Oklahoma Corporation Commission, Cause No. PUD 00000, on behalf of Public Service Company of Oklahoma, regarding cost of service for a regulated electric utility, June 0. Direct testimony before the Michigan Public Service Commission, Case No. U-0, on behalf of Michigan Consolidated Gas Company, regarding cost of service for natural gas distribution assets, July 1, 0 A-

53 Appendix A to the Page A- of 1 Direct testimony before the Public Utilities Commission of Ohio, Case No. -1-EL-UNC, In the Matter of the Determination of the Existence of Significantly Excessive Earnings for 00 Under the Electric Security Plan of Ohio Edison Company, The Cleveland Electric Illuminating Company, and The Toledo Edison Company, September 0. Direct and rebuttal testimony before the Federal Energy Regulatory Commission, Docket No. RP , on behalf of El Paso Natural Gas Company, on the Cost of Capital for Natural Gas Transmission Assets, September 0 and September 0. Direct and rebuttal testimony before the Michigan Public Service Commission, In the matter of the application of The Detroit Edison Company, for authority to increase its rates, amend its rate schedules and rules governing the distribution and supply of electric energy, and for miscellaneous accounting authority, Case No. U-1, October 0 and April 0. Direct testimony before the Federal Energy Regulatory Commission, Docket No. RP-1-000, on behalf Tennessee Gas Pipeline Company, on the Cost of Capital for Natural Gas Transmission Assets, November 0. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER , on behalf of the Atlantic Wind Connection Companies, on the Cost of Capital and Cost of Capital incentive adders for Electric Transmission Assets, December 0. Direct and rebuttal testimony before the Public Utilities Commission of the State of California, Docket No. A , on behalf of California Water Service Company, on the Cost of Capital for Water Distribution Assets, April 0 and September 0. Rebuttal testimony before the Public Utilities Commission of the State of California, Docket No. A.-0-01, on behalf of California American Water Company, on Application of California American Water Company (UW) for Authorization to Implement the Carmel River Reroute and San Clemente Dam Removal Project and to Recover the Costs Associated with the Project in Rates, June 0. Initial testimony before the Public Utilities Commission of Ohio, Case No. --EL-UNC, In the Matter of the Determination of the Existence of Significantly Excessive Earnings for 0 Under the Electric Security Plan of Ohio Edison Company, The Cleveland Electric Illuminating Company, and The Toledo Edison Company, July 0. Direct testimony before the Federal Energy Regulatory Commission, Docket No. PA-1-000, on behalf of ITC Holdings Corp. in response to FERC Staff, Office of Enforcement, Division of Audits, Draft Report on the appropriate accounting for goodwill for the acquisition of ITC Midwest assets from Interstate Power and Light Company, July 0. Written direct evidence before the National Energy Board in the matter of the National Energy Board Act, R.S.C. 1, c. NB, as amended, and the Regulations made thereunder; and in the matter of an application by TransCanada PipeLines Limited for orders pursuant to Part I and Part IV of the National Energy Board Act, for determining the overall fair return on capital in the business and services restructuring and Mainline toll application, September 0. A-

54 Appendix A to the Page A-1 of 1 Report before the Arbitrator on behalf of Canadian National Railway Company in the matter of a Submission by Tolko Marketing and Sales LTD for Final Offer Arbitration of the Freight Rates and Conditions Associated with Respect to the Movement of Lumber by Canadian National Railway Company from High Level, Alberta to Various Destinations in the Vancouver, British Columbia Area, October, 0. Rebuttal Evidence before the National Energy Board in the matter of AltaGas Utilities Inc., 0-01 GRA Phase I, Application No. ; Proceeding I.D. 0, October, 0. Direct testimony before the Federal Energy Regulatory Commission, Docket No. ER1--000, on behalf of Public Service Electric and Gas Company on the Cost of Capital and for Incentive Rate Treatment for the Northeast Grid Reliability Transmission Project, October 0. Rebuttal testimony before the Florida Public Service Commission, Docket No. 01-EL, on behalf of Gulf Power, a Southern Company, on the method to adjust the return on equity for differences in financial risk, November 0. Direct testimony before the Federal Energy Regulatory Commission, Docket No. PA-1-000, on behalf of ITC Holdings Corp. regarding a rehearing for FERC Staff, Office of Enforcement, Division of Audits, Report on the appropriate accounting for goodwill for the acquisition of ITC Midwest assets from Interstate Power and Light Company, February 01. Direct testimony before the Michigan Public Service Commission, Case No. U-1, on behalf of Michigan Consolidated Gas Company, regarding cost of service for natural gas distribution assets, April 01. Deposition testimony in Primex Farms, LLC, Plaintiff, v. Roll International Corporation, Westside Mutual Water Company, LLC, Paramount Farming Company, LLC, Defendants, April 01. Deposition testimony in Tahoe City Public Utility District, Plaintiff vs. Case No. SCV Tahoe Park Water Company, Lake Forest Water Company, Defendants, May 01. Direct testimony before the Public Utilities Commission of Ohio, In the Matter of the Determination of the Existence of Significantly Excessive Earnings for 0 Under the Electric Security Plan of Ohio Edison Company, The Cleveland Electric Illuminating Company, and The Toledo Edison Company, Case No. 1-1-EL-UNC, May 01. Joint Rebuttal Testimony before the California Public Utility Commission on behalf of California American Water Company, regarding Application of California-American Water Company (UW) for Authorization to increase its Revenues for Water Service, Application -0-00, and In the Matter of the Application of California-American Water Company (UW) for an Order Authorizing and Imposing a Moratorium on New Water Service Connections in its Larkfield District, Application -0-01, August 01. A-1

55 Appendix B to the Page B-1 of 1 THE IMPACT OF DECOUPLING ON THE COST OF CAPITAL: AN EMPIRICAL INVESTIGATION Original Version March 0 Revised July 01 Joseph B. Wharton Michael J. Vilbert Richard E. Goldberg Toby Brown 1 Copyright 0 The Brattle Group, Inc. 1 The authors are Principals or Senior Associates of The Brattle Group, a firm which provides consulting and expert testimony in economics, finance, and regulation to corporations, law firms, and governments around the world. See The authors also want to thank members of The Brattle Group who provided comments including Larry Kolbe, Hannes Pfeifenberger, Bente Villadsen, Dan Kiernan, Jenny Palmer and Stephanie Schwartz.

56 Appendix B to the Page B- of 1 INTRODUCTION Decoupling is a form of regulated ratemaking that separates cost recovery from changes in the volume of sales for a regulated utility. It originated as a policy response in the s when utilities were first encouraged to develop energy efficiency (EE) programs, which can significantly reduce the consumption of regulated commodities, such as electricity, gas, or water. Research into the costs and benefits of EE technologies has shown that the long-run savings frequently exceed the costs, and EE programs have the additional benefit of producing no harmful emissions. Recently, more states have begun to adopt long-term goals for EE and designated the utilities as the program administrators. Despite the programs being beneficial and cost-effective to society, there is a substantial disincentive for the utilities to pursue the EE programs actively unless they are accompanied by some type of revenue decoupling and incentives. This disincentive begins with the fact that a large share of an electric, gas, or water utility s costs are fixed in the short run and do not vary with the amount of the commodity produced and delivered. Traditional cost-of-service ratemaking collects these base revenues largely through volumetric rates (i.e., per therm, kilowatt hour, or 0 cubic feet). A successful EE program will reduce the volume of sales and simultaneously reduce recovery of the fixed costs, which include the equity returns, so regulated utilities have what is called a throughput disincentive to carry out EE programs. Revenue decoupling (or simply decoupling ) is a ratemaking methodology that severs the link between sales and cost recovery. There are various forms of decoupling, but the basic idea is that if sales exceed forecast, the utility will refund to customers the over recovery of fixed costs. If sales are less than forecast, the utility will be able add a charge to future rates to recover fully its fixed costs. This removes the throughput disincentive and therefore the conflict of interest for utility management. With decoupling in place, managers of regulated investorowned utilities eliminate one major hurdle to pursuing EE programs that regulators determine are in the public interest and maintain their fiduciary duty to protect the interests of their

57 Appendix B to the Page B- of 1 shareholders. However, a new hurdle has recently appeared in the form of a tangible regulatory risk of a lowered allowed return on equity (ROE) for utilities that implement decoupling. In addition to management s conflict of interest in reducing sales volume, EE programs are likely to increase the variability of the utility s revenues because the effectiveness of the EE program adds another element of forecasting error into setting the revenue requirement. As decoupling policies become increasingly prevalent and important, some intervenors and some commission staffs have argued that decoupling, by design, reduces the variability of revenues by decoupling cost recovery from sales volume, which, they argue, translates directly into reduced business risk. Intervenors therefore have also argued that the regulated company s cost of capital be reduced because cost of capital is the payment to investors bearing business and financial risks. In the majority of decisions approving decoupling in the past, regulators did not explicitly conclude that decoupling reduces a utility s cost of capital. However, in a very visible minority, regulators did decrease the allowed ROE, with the reductions generally ranging from to 0 basis points (bps) (0 bps = 1 percentage point). Of more concern is that intervenors in some gas, electric, and water utility rate cases have suggested the decoupling impact on ROE should be as large as 0 to 00 bps. After reviewing a number of decisions where reductions were imposed by the regulators, we find that the reductions, whether moderate or substantial, appear to be based on theoretical arguments without the support of empirical evidence demonstrating an actual relationship between decoupling and the cost of capital. To test this theory, we have completed the first empirical investigation, of which we are aware, on the hypothesis that There are three conditions that are likely to be needed to ensure that a utility s management is enthusiastic about the effect of large scale EE programs on their financial performance. The first is decoupling, as discussed in this paper, and the second is the timely and concurrent recovery of the direct EE costs. Although together these two conditions make the utility financially indifferent, they do not provide the utility with a strong reason to pursue energy efficiency. The third condition is a performance incentive, which gives the management a financial incentive and, when successful, positive news for shareholders. Although utilities operate under a regulatory bargain, investors value growth in earnings and wish to see some other real value when growth is being reduced by public policy directives. See Chapter 0, Brealey, Myers and Allen, Principles of Corporate Finance, th edition, McGraw Hill Irwin, 00. For example, see pp. 1-0 of Phase 1B Testimony of Terry L. Murray on behalf of the Division of Ratepayer Advocates on Return on Equity Adjustments before the California Public Utilities Commission filed October 1, 00 in Docket No. I Page of 0

58 Appendix B to the Page B- of 1 decoupling lowers the cost of capital. This was done on natural gas distribution utilities between 00 and 0. The findings of our analysis do not support the belief that utilities with decoupling have a lower cost of capital compared to utilities without decoupling. Contrary to what some might expect to find, at least on the basis of the opinions of certain intervenors and the minority set of judgments where Commissions reduced allowed rates of return because of decoupling, our results show that the estimated difference in the cost of capital for decoupled versus non-decoupled (or traditional) natural gas distribution utilities was not statistically different from zero. SECTION 1: VOLUMETRIC RATE, DECOUPLING, AND THE THROUGHPUT DISINCENTIVE Under traditional ratemaking, regulated utilities collect their base revenues, including the return on capital they hope to earn, using a combination of fixed charges and volumetric rates. (Volumetric rates are in dollars per kilowatt-hour for electric utilities or dollars per therm for gas utilities. Revenues collected through volumetric rates vary with overall usage. ) Volumetricbased cost recovery does not follow the principle of basing rates on marginal cost because a large part of a utility s costs is fixed and does not vary with the level of output and sales. Traditionally, the fixed charges generate only a small proportion of total revenues and do not fully recover the fixed element of total costs. If actual sales are less than forecast sales used to set rates, the utility will not earn its allowed ROE. This study assesses two different ratemaking approaches and uses the term decoupling to include both: 1) decoupling with general revenue true-up mechanisms and ) straight fixed-variable rates. A related, third form of decoupled ratemaking is a lost revenue adjustment mechanism (LRAM) that focuses specifically on the impacts of EE programs on base revenues. Some discussions of decoupling limit the term to just the first approach and some include all three. There is no agreement about which of the three is best. This paper is only interested in the cost of capital impacts, not the comparative merits of the three types of decoupling. We include the first two decoupling approaches because they have similar impacts in insulating revenues collected from changes in the commodity throughput, other than those driven by increased numbers of customers. The companies in our gas distribution utility sample use only these two forms of decoupled ratemaking. A small percentage of fixed costs are recovered through customer charges and are not part of the decoupling issue. Some fixed costs for large customers are recovered in demand charges and are much less affected by EE programs. Finally, fuel and purchased power are large, variable costs for electric and gas consumers, and are both independent of the throughput disincentive, because revenues and costs go up or down together and changes generally cancel each other out. Page of 0

59 Appendix B to the Page B- of 1 Successful EE programs will reduce sales volume relative to what sales would have been without the EE program and perhaps relative to sales volumes in previous years. The effect of declining sales on the revenue requirement (as well as the effect of increases in prudent costs) will be addressed in the next general rate case when base rates are reset based on actual sales, but the lost fixed cost recovery for the effect of past EE programs will remain. Even with revised volumetric rates in place as a result of a rate case (and taking into account the effect of EE programs), the throughput disincentive immediately comes back into effect, so that between general rate cases, any reduction in sales from EE programs results in a reduction in earned ROE. To counteract the effect of declining sales on earned ROE, the utility may be forced to file more frequent general rate cases, which are time consuming, expensive, and risky, and therefore not a desirable solution. Thus utilities with volumetric rates have a throughput disincentive to carry out more aggressive programs for their customers to reduce energy usage, lower their utility bills, and help meet climate change policy goals. This tension is in contrast with the overall cost effectiveness of the EE programs from several other perspectives. It is regularly shown in filings that a wide variety of utility-administered EE programs are cost effective from society s point of view. Because of their cost effectiveness and emissions-reducing effects, a number of states have adopted formal EE resource standards or similar EE goals. A 0 survey of electric utilities showed that ratepayer-funded EE expenditures increased by over percent between 00 and 00. It is evident that the issue of addressing the throughput disincentive is becoming increasingly important. As of 0, decoupling of one form or another had been approved for natural gas LDCs in about states and is pending in four more. As of 0, decoupling had been approved for electric utilities in about 1 states and is pending in seven more. If energy efficiency is the goal, decoupling is a useful form of regulated ratemaking that separates, or decouples, the collection of base revenues from sales of the product, so utility Summary of Ratepayer Funded Electric Efficiency Impacts, Expenditures and Budgets, Based on CEE/IEE Industry Database, Updated IEE Brief, May 0. American Gas Association, Innovative Rates, Non-Volumetric Rates and Tracking Mechanisms: Current List, as of June 0. Institute for Energy Efficiency, State Energy Efficiency Frameworks, July 0, pp. -; updated for adoption in AZ and moving NV to a different policy. Page of 0

60 Appendix B to the Page B- of 1 profits are not hurt from the reductions in sales (for example, caused by EE programs). Decoupling removes the throughput disincentive and aligns the interests of utility management with the public policy goal of EE, and reduced sales with the management goal of earning the allowed rate of return for investors. SECTION : DECOUPLING AND THE COST OF CAPITAL Decoupling was adopted for three electric utilities in California in the s as an adjunct to EE policies and has continued ever since (except for a brief hiatus during restructuring after 1). California utilities have not had their cost of capital explicitly reduced because of decoupling. However, as decoupling has grown in importance in recent years in other states, a substantial regulatory risk has also emerged in the form of the potential reduction in the allowed ROE. As of 0, about one-fifth of regulatory decisions that we have reviewed related to decoupling for gas and electric utilities have concluded that decoupling does reduce a utility s cost of capital, and accordingly these decisions have reduced the allowed ROE. The reductions in allowed ROEs have ranged from to 0 bps. However, in our review of these cases, we could find no empirical evidence that supported a reduction of any particular magnitude. The other four-fifths of the decisions adopted decoupling without explicitly reducing the cost of capital. Since the subject has gained greater notoriety, regulatory risk may still resurface as an issue when those utilities litigate their cost of capital in future general rate cases. We have developed an approach to investigate whether any reduction in the allowed ROE is warranted, and if so how small or how large should the reduction be. A utility s operating earnings (i.e., earnings before income taxes) are the difference between base revenues (non-fuel) and the sum of all prudent costs (operations and maintenance (O&M), administrative and general (A&G), depreciation, interest). There are several sources of variability in the base revenue stream that can also be eliminated by the decoupling mechanisms analyzed here. EE programs decrease revenues because they decrease sales. Other increases and/or decreases in base revenue are driven by changes in weather, business activity over the business cycle, the net number of new customers, and the number of delinquent bills. By design, decoupling eliminates or significantly weakens the linkage between revenues and the volume sold, independent of the source of variability. Page of 0

61 Appendix B to the Page B- of 1 Decoupling stabilizes revenues, but net income can still vary. Although depreciation and interest expense are relatively stable, other costs can change quickly between rate cases. At times of rapid capital investment, like the present for utilities that are facing significant environmental retrofits, depreciation and interest may also increase rapidly so that general rate cases are frequently required. O&M and A&G costs can rise more than revenue but are reasonably predictable and do not vary as directly with sales as do revenues. Therefore, if decoupling stabilizes the revenue side of the earnings equation, does it stabilize operating earnings as well? This leads directly to the question: if decoupling reduces revenue variability largely independent of the cost situation, does risk go down at the same time? A more targeted question is whether decoupling reduces the non-diversifiable risk that determines the cost of capital in financial markets? The answer is not a simple yes. Because investors can simply avoid certain risks, not all risks or sources of variance in earning affect the cost of capital equally. For example, extremes of weather, including extreme mildness, will cause variance in a single utility s revenues and are a risk factor for that utility s earnings. However, investors can assemble a portfolio of utility stocks from across the U.S. climate zones and mitigate the weather effects of individual stocks in the portfolio. For the portfolio of utility stocks, the effect of weather variations should largely cancel out, removing weather as a source of investment risk, and negating its effect on the cost of capital. Portfolio formation removes other such risks that can be minimized by diversification. Non-diversifiable risks (also known as business risks ) are the risks that remain after diversification, and they are the risks that drive a company s cost of capital because investors must bear them. The distinction between diversifiable risk and non-diversifiable business risk is important to recognize when evaluating the effect of decoupling, or other regulatory policy, on a company s cost of capital. Simply reducing total risk does not imply that the cost of capital has been reduced. The risk reduced must be part of a company s business risk to affect its cost of capital, so only reductions in business risk justify a reduction in a regulated company s allowed ROE. Thus, we believe that there is strong need to address the empirical question of whether the presumed decrease in risk from adoption of decoupling policy results in a measurable decrease in the cost of capital for regulated utilities. Page of 0

62 Appendix B to the Page B- of 1 SECTION : AN EMPIRICAL TEST OF THE EFFECT OF DECOUPLING ON THE COST OF CAPITAL For the nearly five year period from October 00 to June 0 (the study period), we examined a sample of exchange-traded companies in the natural gas distribution sector for which estimates of the cost of capital were available. For these holding companies, we examined each of their regulated gas local distribution company (LDC) subsidiaries and based on published sources we identified whether and when each LDC operated under decoupled versus traditional volumetric rates. We then compared the estimated cost of capital across two mutually exclusive groups that we could observe: regulated utilities with more than 0% share of revenues collected under decoupled ratemaking versus regulated utilities with less than 0% share of traditional volumetric rates. During the study period, co-author Dr. Michael Vilbert and others at The Brattle Group participated in a number of rate proceedings where the cost of capital was estimated for a sample of regulated gas LDCs. Some of the LDCs examined in these proceedings had decoupled rates at the beginning, some did not, and some became decoupled during the period. Hence, the cost-of-capital estimates from these proceedings provide a natural experiment on the impact of decoupling on the cost of capital. The specifics of our study using the estimates from the proceedings are described below. DEVELOPING A SAMPLE OF TRADED NATURAL GAS HOLDING COMPANIES Our analysis is based on a systematically selected sample of holding companies (HCs) that were exclusively or primarily involved in the regulated natural gas distribution industry. Over the study period, Brattle expert witnesses submitted cost of capital testimony 1 separate times based on an evolving sample of six to 1 gas LDC holding companies. Across the 1 dates, over HC-specific estimates of the cost of capital were made and submitted into evidence and subject to scrutiny. In each proceeding, Brattle selected a sample from the universe of all traded natural gas LDC holding companies followed by the investment advisory service, the Value Line Investment Survey, using six sample selection criteria that are designed to remove companies with Page of 0

63 Appendix B to the Page B- of 1 characteristics that could bias the cost of capital estimates. Over the study period, the selection criteria remained the same, but the composition of the sample changed slightly due to changing circumstances of the companies, for example, due to a merger (selection criteria No. ). Thus, we have assembled a record of statistically valid estimates of the changes in the cost of capital for the study period for every holding company in the sample. DETERMINING THE DEGREE OF DECOUPLING FOR COMPANIES IN THE SAMPLE The study has an important dichotomy. Holding companies, not their subsidiaries, have the market information necessary to estimate the cost of capital, because they have publicly traded stock. However, individual, state-regulated subsidiaries, not the holding companies themselves, can be granted decoupled rates by state regulators. Hence, we characterized the degree of decoupling of each holding company by examining the decoupling policies of its subsidiaries after differentiating each state in which the subsidiary operates. To begin, we identified all regulated gas LDCs belonging to each holding company in the sample and then used a combination of primary and secondary sources to identify the subset of those gas LDCs that had decoupled rates during the study period. The 1 holding companies collectively held regulated natural gas LDC subsidiaries as of June 0. We used a broad definition of decoupling and included true-up decoupling schemes and straight fixed-variable (SFV) rates. 1 The sample selection criteria are: 1 1. An investment grade credit rating, i.e., BBB- or better from S&P;. No dividend cuts during the last years;. Revenues or assets > $00 million;. Greater than half of the assets in the regulated line of business, either natural gas distribution or regulated electric operations;. Data available from Value Line, Bloomberg and Moody s; and. No major mergers or acquisitions during the last years. In this report, we use the term subsidiary to refer to the part of a utility that is regulated at the state level. A particular holding company might own two utilities that are separate corporations. Assume the first is located in a single state, while the second has a service territory extending over three states. In our analysis, this holding company would have four subsidiaries. See footnote above. We did not find any gas LDC companies with decoupling in the form of a lost fixed revenue adjustment mechanism (LRAM), which is found frequently in electric utilities. Page of 0

64 Appendix B to the Page B- of 1 Figure 1 shows the distribution in June 0 of true-up and SFV decoupling schemes among the subsidiaries (subs) in our sample. 1 Figure 1 Natural Gas Sample LDC Penetration of Decoupling Mechanisms in June 0 Our evidence is that the decoupling share in our sample is relatively high. In the total population of gas LDCs across the U.S., we estimate there are approximately LDCs with decoupling, 1 out of an estimated population of about 0 LDCs, 1 or a penetration of gas LDCs of about oneseventh. In our sample of subsidiaries of 1 gas LDC holding companies, percent had decoupling policies in place, as shown below in Table Four of the subsidiaries are quite small, so they are not included where a count of LCDs with a specific policy is made, such as in this discussion. They are included in the quantification of the Decoupling Index discussed below in Table. This is based on our examination of published reviews, for example, Review of Distribution Revenue Decoupling Mechanisms, Pacific Economics Group (March 0), State Energy Efficiency Regulatory Frameworks, Institute for Electric Efficiency (January 0). Based on EIA Form 1 data. In counting LDCs, such as in Table 1, we exclude LDCs transporting less than bcf per year in 00. Counting LDCs gives us the penetration share. In estimating the index of decoupling for a holding company at points in time, we do not exclude any subsidiaries because of size. Page of 0

65 Appendix B to the Page B- of 1 Table 1 Natural Gas LDC Decoupling Penetration in June 0 Gas LDC subsidiaries (> bcf) Decoupled Share Decoupled Total 1 1% In Sample 1 % Notes: - The Total row refers to all gas LDCs, counting each one once per state in which it delivered volumes of at least bcf in The In Sample row refers to gas LDCs of 1 publicly-traded companies with a large proportion of operations in gas distribution, using same criteria as above. For the In Sample row, the volume delivery condition removes LDCs from this table. See footnote 1. Not only does decoupling vary by subsidiary, but for some subsidiaries in our sample the status of decoupling has changed over time because state regulators have approved new mechanisms and, in some cases, changed or eliminated decoupling. To reflect this variation, we developed a robust Decoupling Index for each holding company over time. First, for each subsidiary, we made a specific designation of whether or not it was decoupled for each year in the sample period. Second, we aggregated the decoupling scores of the regulated subsidiaries to produce an overall decoupling score for the holding company. In this aggregation, we weighted all of the individual subsidiary scores (i.e., 1 or 0) based on their volume of gas distributed in 00. Table below shows the Decoupling Index or score for each natural gas holding company in the sample in June 0, the end of the study period. Page of 0

66 Appendix B to the Page B-1 of 1 Table Decoupling Index for the Natural Gas Sample in June 0 Company Number of gas LDC subsidiaries Decoupling Scores AGL 0. Atmos Laclede New Jersey Resources Nicor NiSource 0.0 Northwest Natural 0. Piedmont 0. South Jersey Industries Southwest Gas 0. WGL 0.00 Vectren 1.00 Notes: - Each subsidiary is counted once per state in which it operates. - The aggregate decoupling score of a gas holding company is equal to the sum of the 00 gas volumes of its gas LDCs subsidiaries with decoupling (a yes or no proposition at each point in time), divided by the sum of 00 gas volumes of all of its gas LDCs in total. ESTIMATING THE COST OF CAPITAL FOR COMPANIES IN THE SAMPLE The cost of capital is defined as the return investors require before investing in an asset, or comparably, as the expected return available on investments of comparable risk. There are a number of good and frequently used approaches for estimating the cost of capital for a company. One approach, not used here, is to create estimates using historical data to estimate the parameters of a cost of capital model (e.g., estimate the Capital Asset Pricing Model (CAPM) beta using historical stock price returns). Another approach is to compare current stock prices with forward-looking forecasts of cash flows from the business (e.g., the discounted cash flow (DCF) method compares the value of expected future dividends from the stock with current stock prices to estimate the level of expected returns). For this study, we used the DCF approach to estimate the cost of capital to analyze the effect of decoupling. Our reason was that the DCF approach has the advantage of reflecting changes in the cost of capital more quickly than do Page of 0

67 Appendix B to the Page B-1 of 1 other approaches that rely on the analysis of historical data. 1 In the DCF model, the discount rate that makes the present value of expected future dividends equal to the current stock price is the estimate of the cost of equity (COE) for the company. We chose to use a multi-stage version of the DCF model. In this version, security analysts forecasts of company-specific future earnings were used for a five-year first stage of the multistage analysis to estimate expected dividends. 1 Later stages were added to insure that the growth rate for each company eventually settled down to a single long-term growth rate based on the forecast of the long-term U.S. GDP growth rate. In particular, forecast growth rates for years through were linearly trended between company-specific earnings growth rates for the first years and the long-term growth GDP growth rate in effect starting in year. Earnings growth was translated into corresponding expected dividend payments over time. The COE is the information of interest to regulators when they set the allowed ROE for a utility, so our focus is ultimately on whether there is a measurable reduction in the COE from the policy of decoupling. In general, the COE increases not only with increased business risk but also with increased financial risk. 1 Therefore, in testing for an impact on the cost of capital from decoupling, we systematically account for differences in the COE in different holding companies in the sample that arise from different levels of financial risk, which has nothing to do with decoupling. To control for the effect of differences in capital structure (i.e., differences in financial risk) among the sample companies, we converted estimates of the COE into corresponding estimates of the overall after-tax weighted-average cost of capital (ATWACC). 1 The ATWACC measures For example, estimates of beta, the systematic risk parameter in the CAPM, are normally based upon to years of historical data. A policy that changed a company s cost of capital would lead to a change in the company s beta, but that change would not be fully reflected in the beta estimate for to years. We also implemented the test using the simple or constant growth DCF method. The results (not reported here) were not substantially different than those based upon the multi-stage DCF method. Financial risk is related to the degree to which the company s assets are debt financed. The greater the share of debt in the capital structure, the greater the interest that must be paid out of operating revenues before any shareholder earnings are available. To be specific, the after-tax weighted-average cost of capital (ATWACC) is the measure we use. ATWACC is a weighted average of both the equity and debt returns after taking into account the tax deductibility of interest payments. The weights used in the calculation are the market values of debt and equity in the capital structure. See Chapter 0 of Brealey, Myers and Allen, op cit. Page 1 of 0

68 Appendix B to the Page B-1 of 1 the cost of capital for the business itself, while the COE estimate represents the cost of equity capital taking into account the equity-holders additional financial risk from the company s level of debt financing. In other words, the ATWACC captures only the effect of business risk, while the COE is also affected by financial risk. For that reason, we use the ATWACC in our statistical analysis below. STATISTICAL ANALYSIS OF DECOUPLING ON THE COST OF CAPITAL To determine the impact of decoupling on the cost of capital, we carried out the five statistical tests described below. The results of these tests were all in general agreement and collectively demonstrate that the overall impact of decoupling the rates of the subsidiaries on the cost of capital of the holding companies was not statistically different from zero. There was no evidence of a decrease in the cost of capital. The estimates of changes from the five tests ranged from - bps to +1 bps. In none of the tests could we conclude that there was a statistically significant decrease (or counterintuitive increase) in cost of capital. The estimated impacts and associated percent confidence intervals for all five statistical tests are summarized in Figure. Our conclusion that decoupling does not cause a measureable decrease in the cost of capital is shown in Figure by noting that all five of the confidence intervals contain the zero no effect horizontal line. We now proceed to describe the five statistical tests. In the first T-test of the impact of decoupling on the cost of capital, we split the sample at each observation date into one of two groups: the group where the Decoupling Index (which can vary between 0 and 1) is below 0. and the remaining utilities in the group where the Decoupling Index is greater or equal to 0.. We then compare the frequency distribution of the cost of capital estimates (specifically, the ATWACC calculated using a multistage DCF) for each group. If the impact of decoupling on the cost of capital is large, we would expect the costs of capital estimates for the group with a high Decoupling Index to be noticeably lower than those with low Decoupling Index. Figure displays the distribution of returns with a Decoupling Index of greater than 0. (dark blue) on top of the distribution of returns with a Decoupling Index of less than 0. (light blue). For the remainder of the paper, we refer to the group of companies with a Decoupling Index of greater than or equal to 0. as more decoupled and those with a Decoupling Index of less than 0. as less decoupled. We see that any impact of the differences in cost of capital from decoupling between Page 1 of 0

69 Appendix B to the Page B-1 of 1 the groups is smaller than the diversity in cost of capital within each group from other differences. Figure Cost of Capital Distribution for More Decoupled Holding Companies (dark blue) and Less Decoupled Holding Companies (light blue) The first result is that the average cost of capital for the more decoupled group was. percent, and the average cost of capital for the less decoupled group was.1 percent. Hence, we find that the more decoupled group of holding companies had a cost of capital that was slightly higher on average, +1 bps, than the less decoupled group. To examine the statistical significance of this first conclusion, we carried out a two-sample T-test and found that the test gave a percent confidence interval of the difference in average cost of capital between the defined groups that ranged from bps lower (- bps) to bps higher for the more decoupled group. The observed difference in means is not statistically different from zero at this confidence level. 0 In other words, a T-test on the two groups leads to an expectation that any reduction in cost of capital from decoupling is highly likely to be less than bps. 0 This confidence interval was from a T-test based on the assumption that the sample variances of the two groups were equal. The assumption of equal variance makes the percent confidence interval range Page 1 of 0

70 Appendix B to the Page B-1 of 1 Our first statistical test put a percent confidence interval range on the estimated impact from decoupling on the cost of capital that is bps wide. A natural next step is to see if there are patterns to the cost of capital data that can account for enough of the variability in returns in the subgroups to more tightly constrain the assessed limits on the decoupling impact. Figure displays the cost of capital estimates by time period for each of the 1 holding companies in the sample and shows that the cost of capital estimates vary by time period and company. (In Figure, for each time series of cost of capital estimates for one holding company, the bold lines represent the exact periods in which the company s Decoupling Index was high, i.e., 0..) It is evident that there is a great deal of shared time variation 1 in the estimates, in addition to a great deal of company-specific variation. This observation serves to suggest additional statistical tests that should separate out the effects of shared time variation and company-specific variation in the statistical analysis. 1 slightly smaller but does not change the conclusion that the difference in means is not statistically significant. Shared time variation is the variation in the ROE estimates that occur as a result of being done at different points in time. The broken lines in Figure are due to removing some holding companies using the objective criteria and for missing data for some estimation dates. Page 1 of 0

71 Appendix B to the Page B-1 of 1 Figure Estimated Cost of Capital for each Holding Company over Time in Bold Lines when the Decoupling Index was >= 0.00 Four additional statistical tests are done using linear regression analyses. In all four additional tests, we accounted for shared time and company-specific variation through the use of time period-specific and company-specific indicator variables (which are common ways to isolate the impact of such factors from the impact of the quantity you are testing, here the level of decoupling). The use of these Indicator variables accounted for over 0 percent of the variation in the estimated costs of capital in the data. The four tests differ in how the Decoupling Index was treated and whether the prior value of the cost of capital was included in the regression. In two of the tests, the decoupling is represented as an Indicator variable with a value of 1 if decoupling is 0. ( more decoupled ) and 0 if it is < 0.. In the other two tests, decoupling is represented using the actual numerical value of the Decoupling Index. The use of the actual Examination of the regression residuals with a normal quantile-quantile plot and carrying out a Jarque- Bera test did not indicate problems with either non-normality or heteroskedasticity. However, the regression residuals do show some degree of serial correlation so we used Newey-West adjusted standard errors when making inference regarding the statistical significance of the impact of decoupling on the cost of capital. Page 1 of 0

72 Appendix B to the Page B-1 of 1 value of the Decoupling Index is to determine whether the cost of capital estimates vary even if decoupling affects a relatively small or large portion of the assets of the holding companies. To summarize, the four additional tests consist of two regressions with decoupling represented with a Decoupling Index variable carried out both with and without the inclusion of the prior cost of capital, and two regressions with decoupling represented by the Decoupling Index variable carried out both with and without the inclusion of the prior cost of capital in the regression. In the first regression analysis, adding an Indicator for the Decoupling Index being at least 0. into the regression leads to an estimated slight increase of + bps in the cost of capital from being more decoupled (with a Newey-West percent confidence interval ranging from - to + bps). This result is consistent with the results of the initial T-test, because both include the common region from - to + bps in their percent confidence interval range. The impact of being more decoupled is not significantly different from zero. In the second regression analysis, we estimated the impact of decoupling by using the Decoupling Index itself as an explanatory variable in the regression and by taking into account company-specific differences and the variation in the cost of capital over time. This test shows that the slight increase in the cost of capital is + bps for a fully decoupled holding company compared to one with no decoupling (0 Decoupling Index). For the third and fourth regression analyses, we reran the prior two regressions with the additional inclusion of the prior period s cost of capital in the regressions. The prior period s cost of capital was included to address directly some serial correlation which we had seen in regression residuals from the base regression of cost of capital on period and company. The two regressions differ by the representation of decoupling as either an Index or as an Indicator variable. Both results are estimates of slight reductions in the cost of capital of - bps and - bps, for the Indicator with lag and the Index with lag, respectively. We note that the time intervals between the data points were not of uniform length because the underlying cost of capital studies are done when they are needed for regulatory proceedings, not on any fixed For the second regression, the Newey-West percent confidence interval ranges from -0 bps up to +0 bps. This result of no impact remains consistent with the prior two tests, because they all contain the interval from - to +1 bps in their percent confidence intervals. Page 1 of 0

73 Appendix B to the Page B-1 of 1 schedule. We believe the results of these third and fourth regressions still provide important evidence. Both when we included the Decoupling Index being at least 0. into the regression and when we included the Decoupling Index itself into the regressions, we found that the percent confidence interval on the impact of decoupling contained the common region from - to +1 bps found in the prior three tests. This is the common region for all five statistical tests and includes the zero (0 bps) impact of decoupling. In conclusion, these results provide what we consider strong evidence that the overall impact in our data sample on companies cost of capital following decoupling of their rates was not statistically different from zero. The estimated impact and associated percent confidence intervals for all five statistical tests can be seen in Figure. We believe that the non-uniform periods of time between the dates of the cost of capital estimates are not an important issue in this analysis. Capital markets adjust to new information on a daily or even more frequent basis. The real issue is whether there were important changes between the observations to give them a degree of independence. Even to the casual observer, it is clear that the period from 00 to 0 was a period of change in U.S. financial markets, so the observations appear sufficiently independent to us. For the regressions with the prior cost of capital included, HC standard errors were used as described by Long J. S., Ervin L. H., Using Heteroscedasticity Consistent Standard Errors in the Linear Regression Model. The American Statistician,, 1, 000. Page 1 of 0

74 Appendix B to the Page B-0 of 1 Figure Confidence Intervals for Five Statistical Tests for an Impact of Decoupling on the Cost of Capital CONCLUSION Our statistical tests do not support the position that the cost of capital is reduced by adoption of decoupling. If decoupling decreases the cost of capital, these tests strongly suggest that the effect must be minimal because it is not detectible statistically. As decoupling continues to grow in importance, cases will frequently come up where intervenors and commission staff may explore the extent to which decoupling reduces business risk and the utility s cost of capital. To date, in a minority of cases in which decoupling was approved, the utility explicitly had their allowed ROE reduced. Our research leads us to conclude that these reductions are likely to have been implemented without empirical analysis to support the ROE reduction. The results of our analysis show that if such empirical analysis had been done, it is unlikely that it would have supported even the moderate reductions in allowed ROE than were imposed on the utilities. Page 1 of 0