a IN THE MATTER OF
RETURN ON COMMON EQUITY FOR A BENCHMARK UTILITY
DECISION August 26, 1999
Before: Peter Ostergaard, Chair Lorna R. Barr, Deputy Chair Paul G. Bradley, Commissioner Barbara L. Clemenhagen, Commissioner
TABLE OF CONTENTS Page No. 1.0 INTRODUCTION 1 2.0 RATE OF RETURN ON COMMON EQUITY 2 2.1 Rate of Return on Equity Recommendations 2 2.2 Requirements of a Fair Return 3 2.3 Equity Risk Premium Tests 5 2.3.1 Determination of Total Market Risk Premium 5 2.3.1.1 Experienced Equity Market Risk Premium 5 2.3.1.2 Forward Looking Equity Market Risk Premium 9 2.3.2 Determination of Utility Risk Premium 10 2.3.2.1 Relative Risk Measures 10 2.4 Comparable Earnings Test 13 2.5 Commission Determinations 14 3.0 AUTOMATIC ADJUSTMENT MECHANISM 17 3.1 The Sliding Scale 17 3.2 Suggested Improvements 21 3.3 Commission Determinations 22 COMMISSION ORDER NO. G-80-99 APPENDIX A - Appearances APPENDIX B - Index of Witnesses APPENDIX C - List of Exhibits
1.0 INTRODUCTION In its Decision dated June 10, 1994, the British Columbia Utilities Commission (“the Commission”, the BCUC”) established rates of return on common equity (“ROE”) for BC Gas Utility Ltd. (“BC Gas”), Pacific Northern Gas Ltd. (“PNG”), West Kootenay Power Ltd. (“WKP”), and Centra Gas Fort St. John Inc. (“Centra-FSJ”). In the same Decision, the Commission established an automatic adjustment formula for varying the ROEs on an annual basis, based on a forecast of Government of Canada Long Bond yields 1 (“long Canadas”). This formula was later amended by Commission Order No. G-49-97 and included a provision that the formula would not be applied if the forecast of long Canada bond yields fell below 6.0 percent. For each of the years, 1995 through 1999, the ROE for the Utilities has been set using the automatic adjustment formula.
In September 1998, the Commission issued a letter to affected parties, advising them that it appeared the forecast of long Canada bond yields would be below 6.0 percent and asking them if they wished to use the formula for setting the ROE for 1999 in any case. All parties accepted the use of the formula for 1999; however, there were requests from various parties to re-examine the formula to determine if further amendments were required for the year 2000 and beyond. In October 1998, the Commission issued Letter No. L-73-98, advising the above-mentioned Utilities and other interested parties that the formula would be used to set the ROE for 1999 and that, in 1999, the Commission would be establishing a process for a comprehensive review of all issues surrounding the setting of the ROE.
On March 11, 1999, the Commission issued Order No. G-26-99 setting down an oral public hearing into the appropriate Rate of Return on Common Equity and Capital Structure for each of the previously mentioned utilities as well as Centra Gas Whistler Inc. to commence May 31, 1999. However, by way of letters dated March 22 and March 23, 1999, BC Gas, PNG and WKP (collectively “the Utilities”) asked the Commission to reconsider the scope of the hearing and limit the items for consideration to the ROE for a low-risk benchmark utility and to the automatic adjustment formula. They suggested that consideration of the appropriate premium off the benchmark and the appropriate capital structure for each utility should be moved to utility-specific processes. The Utilities suggested that the current capital structures and premiums off the benchmark should continue to apply until such processes took place.
1 For the purpose of the formula, the forecast of long Canadas is equal to the November Consensus forecast of the 10 Year Canada Bond rate, adjusted to reflect the spread between 10 and 30 year Canada bonds for the last six business days in November. The long Canadas chosen to calculate the spread are typically 30 year bonds of similar coupon to current yields, which are actively traded.
1
2 The Utilities also asked that the Commission delay the date for the filing of all evidence to May 17, 1999, and to make such further adjustments to the Regulatory Timetable as necessary to allow for this delay. Finally, the Utilities asked that the hearing proceed on a written rather than on an oral basis.
Commission Staff circulated these letters to the Consumers’ Association of Canada (B.C. Branch) et al. and the Joint Industry Electrical Steering Committee (collectively the “Customers”, individually "CAC (B.C.) et al." and "JIESC") – two parties who had actively intervened in the 1994 hearing – for their comments. By way of letters dated March 31 and April 6, 1999, respectively, the Customers supported the restriction in the scope of the hearing but opposed the request for a written hearing. The Customers stated that a written hearing may be advantageous in circumstances where the facts are known but is inappropriate where the evidence is likely to be technical and contentious.
After considering all the submissions, the Commission determined that an oral public hearing should be established to review the appropriate ROE for a low-risk benchmark utility and to consider processes or formulae that may be employed to improve the determination of the ROE for utilities in future years. Accordingly, on April 7, 1999, the Commission issued Order No. G-38-99 which set down the hearing for June 21, 1999.
The evidentiary portion of the hearing was held in Vancouver and lasted three days during which testimony was presented by Ms. K. McShane on behalf of the Utilities, by Drs. W.R. Waters and R.A. Winter on behalf of the Customers, and by Mr. J. Olson on behalf of the Canadian Gas Association (“CGA”). The evidence of Ms. McShane and Drs. Waters and Winter addressed the determination of the fair rate of return on common equity for a low-risk benchmark utility while the evidence of Mr. Olson addressed the relative competitiveness of utility equities in the investment market. The evidentiary portion of the hearing was followed by Written Final and Reply Argument, filed on July 5 and July 12, 1999, respectively.
2.0 RATE OF RETURN ON COMMON EQUITY 2.1 Rate of Return on Equity Recommendations Both Ms. McShane and Drs. Waters and Winter provided evidence as to the fair rate of return on equity for a low-risk benchmark utility under present circumstances. In her evidence, Ms. McShane concluded that the fair rate of return on common equity for a low-risk benchmark utility is currently 10.75 percent, assuming long Canada bond yields of 5.5 percent (Exhibit 4A, p. 2). This recommendation is based on her application of the Equity Risk Premium test, which suggested a rate of return of 10.5 percent to 10.75 percent, and on her application of the Comparable Earnings test, which suggested a rate of return on
3 equity of 11.5 percent to 12.0 percent. Ms. McShane's recommendation is endorsed by Mr. Olson, in part as a result of his support of the Comparable Earnings test (T1: 167).
In contrast, Drs. Waters and Winter concluded that, at the same long Canada bond yields, the fair rate of return is 8.0 percent to 8.25 percent (Exhibit 6A, p. 6). This recommendation is based solely on their application of the Equity Risk Premium test.
The issues associated with these differing approaches to estimating the fair rate of return on equity for a low-risk benchmark utility are discussed below.
2.2 Requirements of a Fair Return Throughout the hearing, there was substantial debate as to what was meant by a fair rate of return on equity. Both Ms. McShane and Drs. Waters and Winter agreed that there are three standards governing the determination of a fair rate of return on equity (Exhibit 4A, p. 13; Exhibit 6A, p. 15) but disagreed over what was required for these standards to be met.
The three standards are: (1) the rate of return should be commensurate with the rate of return the equity investor could expect to receive from an investment of comparable risk; (2) the rate of return should be sufficient to allow the firm to attract new capital without impairing the position of existing investors, i.e., without diluting the price of existing shares; and (3) the rate of return should be sufficient to maintain the financial integrity of the utility, i.e., the rate of return should be sufficient to allow the utility to attract capital even when market conditions are unfavorable. The Utilities argued that the legislation and legal principles established by the courts set out that public utilities in British Columbia are entitled to have their rates established at a level which will result in the equity invested in utility operations earning, as a minimum, the return that would be earned if the equity were invested in another business with risk characteristics similar to those of the utility (Utilities Final Argument, p. 36). They argued that a utility must be allowed an ROE that will enable it to have earnings which will increase shareholders' equity by an amount similar to the increase achievable from an investment in other equities (Utilities Final Argument, p. 2). Further, they argued that if investors' expectations for the earnings of non-utilities are high, the allowed ROE must reflect those expectations if the comparability standard is to be met (Utilities Final Argument, p. 10).
4 With respect to the capital attraction and financial integrity standards, the Utilities suggested that the Commission must consider the requirements of the market to attract new sums of equity into utilities. Although Ms. McShane indicated that the evidence showed that the financial integrity of Canadian utilities had been maintained to date (T1: 17), she expressed concern that this might not be the case in the future unless allowed rates of return on equity rise. She suggested that increasing globalization of capital markets is providing Canadian investors with more and more opportunities and that, as this occurs, the ability to attract capital will become of greater concern (Exhibit 4A, p. 10, T1: 16). In further support of her view that rates of return on equity must rise if capital is to be attracted by utilities, she noted that the negotiated rates for the Alliance Pipeline and the Maritimes and Northeast Pipeline imply that investors expect returns in the range of 12-13 percent (Exhibit 4B, BCUC IR1, Q9).
These concerns were echoed by Mr. Olson who stated that Canadian companies had trouble keeping investors from outside of Canada (T2: 165).
Although the customers agreed that to attract capital the return must be equal to the returns available from investments of comparable risk (JIESC Final Argument, pp. 1 and 2), Drs. Waters and Winter disputed the notion that rate of return awards should be designed to meet investors' expectations. They suggested that a regulator that adopted, as a goal, the meeting of investors' expectations would be trapped by circularity, since the regulator would be making its decision on the basis of investors' expectations regarding its decision (Exhibit 6C, p. 18). Further, they maintained that if the returns are adequate to attract new investors while maintaining financial integrity, there is no reason for higher awards to be made (JIESC Reply Argument, p. 11). They argued that market-to-book equity ratios in excess of 1.0 clearly demonstrated that the existing allowed return in B.C. of 9.25 percent for a low-risk utility and the recent award for Newfoundland Power and Light are more than sufficient to attract capital without impairing financial integrity (JIESC Final Argument, p. 2; Exhibit 6A, pp. 63-66).
In response, the Utilities argued that neither economic theory nor the real world capital market indicate that competitive returns are those that keep market value close to book value. They stated that setting the utility rate of return to achieve a market-to-book ratio of 1.0 was inconsistent with the objective of regulation, which is to emulate the competitive environment (Utilities Final Argument, p. 5). Indeed, while Ms. McShane agreed that financial integrity could be measured in a narrow sense by whether a utility's shares had a market-to-book ratio of greater than one, she maintained that a broader measure of financial integrity required an assessment of a utility's shares compared with the shares of investments of commensurate risk to the utility (T1: 13). Finally, the Utilities noted that the stock price of Newfoundland Power and Light's parent has declined 17 percent since the Newfoundland and Labrador Board's Decision compared with an average decline of 1.5 percent in other utility stock prices, indicating that the recent award had adversely affected the stock price (Utilities Reply Argument, p. 8). However, the Customers argued
5 that an analysis of the impact of the decision should be confined to a window of one to five days around the event. Further, they noted that at December 31, 1998, five months after the decision, the stock price of Newfoundland Power and Light's parent was more than 90 percent of its July 30, 1998 value (Exhibit 6C, p. 17).
2.3 Equity Risk Premium Tests As indicated above, both Ms. McShane and Drs. Waters and Winter used applications of Equity Risk Premium tests to develop their recommendations. This test assumes that the fair rate of return for a regulated utility is equal to the yield on long Canadas plus a premium to compensate investors for the extra risk associated with holding utility stocks, i.e., an equity risk premium. In addition, a cushion or flotation allowance is added to the return to allow the utility adequate financing flexibility. To apply the test it is necessary to determine the required risk premium for the equity market as a whole and the relative riskiness of the utility compared with the market.
In undertaking their equity risk premium tests, both Ms. McShane and Drs. Waters and Winter assumed a long Canada yield of 5.5 percent and a flotation allowance of no more than 50 basis points. Accordingly, the difference in the two recommendations reflects divergent views regarding the premium investors require above the long Canada yield to be induced to invest in the equity market and divergent views with respect to the relative riskiness of utilities compared with the market. Using both historic and forward-looking data, Ms. McShane concluded that the equity risk premium for the market as a whole is 6.0 percent to 7.0 percent (Exhibit 4A, p. 22) and that the appropriate relative risk adjustment for a benchmark utility is .70 (Exhibit 4A, p. 28). In contrast, using historic data only, Drs. Waters and Winter concluded that the total market risk premium is 4.5 percent (Exhibit 6A, p. 27) and that the appropriate relative risk adjustment is .50. (Exhibit 6A, p. 32).
2.3.1 Determination of Total Market Risk Premium 2.3.1.1 Experienced Equity Market Risk Premium Both Ms. McShane and Drs. Waters and Winter examined historic market risk premiums for Canada and the United States to determine estimates of the expected market risk premium. As a result of her examinations, Ms. McShane found that the experienced risk premiums for the Canadian market showed a range of 4.5 percent to 5.5 percent depending on the data source and period covered while the experienced risk premiums for the U.S. market indicated a risk premium of approximately 7.5 percent to 8.0 percent. Assuming an 80 percent Canada, 20 percent United States weighting, she concluded that the historic data indicated a minimum expected market risk premium of 5.5 percent (Exhibit 4A, p. 18).
6 As a result of their examinations, Drs. Waters and Winter found that the experienced risk premiums for Canada and the United States were 3.4 percent and 5.9 percent, respectively. Assuming a 70 percent Canada, 30 percent United States weighting (T2: 211), they concluded that the historic data indicated that the expected market risk premium is 4.5 percent (Exhibit 6A, p. 27).
Although the witnesses used similar sources for their raw data and placed comparable weights on the Canadian and U.S. data in developing their estimates, the results differed for two major reasons. First, the witnesses disagreed over the historical time period that should be examined. Ms. McShane concentrated on data for the period 1947 to 1998 (Exhibit 4A, Schedule 9). Although she recognized that a longer time period captured a greater variety of economic events, she stated that structural changes which have occurred since 1946 warranted placing primary focus on the post-World War II returns (Exhibit 4A, p. B-4; T1: 78). In contrast, Drs. Winter and Waters examined data for the period 1926 to 1998 (Exhibit 6A, Table 5). Although Dr. Winter recognized the reduced relevance of experience the further back in time an event occurs, he indicated that the need for more data points to offset the volatility of the series justified the use of the longer time series (T3: 308).
Second, the witnesses disagreed about whether the market risk premium calculated from the data should reflect the arithmetic mean or the geometric mean 2 . Ms. McShane used arithmetic averages to calculate the market risk premium from the various series of data she examined while Drs. Waters and Winter used the arithmetic mean of the geometric means of the various holding periods which they examined (T2: 195).
Ms. McShane supported the use of the arithmetic mean to calculate the total market risk premium on the grounds that the arithmetic mean recognizes uncertainty in the stock market (Exhibit 4A, p. B-2) while the geometric mean smoothes out uncertainty (Exhibit 4E, p. 5). Accordingly, she stated that the geometric mean does not account for the variation in year-to-year returns which are of concern to investors when evaluating potential future outcomes. As a result, she indicated that it could lead investors to the conclusion that the least risky investments (i.e., those which experience the least volatility of returns) had the highest cost of equity and the most risky (i.e., those which experience the most volatility of returns) had the lowest cost of equity (T1: 82). In addition, Ms. McShane argued that the arithmetic mean reflects the probability distribution of what investors can expect in the future based on the historic returns and the variability of
2 Drs. Waters and Winter define the two terms as follows: • Arithmetic mean: the sum of all values (rt) in a growth or return series divided by the number (n) of values in the series. • Geometric mean: the nth root of the product of all return or growth ratios (1+r) in the series - 1. (Exhibit 6a, Appendix VIII, Attachment A)
7 those returns (T1: 52). In support of her views, Ms. McShane cited Ibbotson Associates, Stocks, Bonds, Bills and Inflation, 1998 Yearbook, pp. 157-159.
Ms. McShane acknowledged that the geometric mean gives the most meaningful after-the-fact information of what an investor achieved, and that the Association for Investment Management and Research requires that holders of the Chartered Financial Analyst designation present investment performance results on this basis (T1: 53). However, she suggested that this is not relevant to a forward-looking analysis where the exercise is to determine the future required rate of return on equity. She stated that the only way to measure the required rate of return from past returns where there is variability is to assign each annual average return an equal probability of occurring in each future period and that this is represented by the arithmetic mean (T1: 90 and 91).
As indicated above, Drs. Waters and Winter calculated their estimate of the total market risk premium using the arithmetic average of the geometric mean of realized rates of return over five, seven, ten and 73 year holding periods (Exhibit 6A, p. 23; T2: 195; T3: 4326). They gave primary weight to the 73 year holding period for the Canadian data (T3: 427). In support of the use of the geometric mean, they stated that to estimate a utility's cost of capital it is necessary to estimate the rate of return required to induce investors to own the shares of the utility. Since investments in utility shares are investments in long-term assets, they stated that the relevant earning power of the utility is its prospective compound rate of return over a multiyear horizon (Exhibit 6A, pp. 23 and 24). Accordingly, they stated that allowing for a variety of investment horizons is more realistic than adopting the assumption of a single year investment horizon implicit in the arithmetic average (Exhibit 6C, p. 4; T3: 333).
Ms. McShane agreed that the arithmetic mean assumes a sequence of one year horizons (T1: 119) and that the geometric average assumes the holding period over which the average is calculated (T1: 118). However, she maintained that the appropriate time period to consider is one year since investors are likely to reevaluate stock holdings on a year-to-year basis (T1: 81).
In addition, the Utilities argued that the arithmetic mean is the return that equates the expected value of an investment at the end of the investment horizon, in dollar terms, to the beginning investment while the geometric average ignores the relative dollar value of the individual returns included in the average. As a result, the Utilities stated that the application of the geometric average return to an initial investment would significantly understate the expected future value (Utilities Final Argument, p. 40).
Dr. Winter acknowledged the possibility of frequent re-evaluation; however, he stated that when investors are making decisions concerning which investments are worthwhile to hold, they examine all available information on the prospective returns of the stock including the current decision, recent decisions, and the
8 trends in the industry. He suggested that, from these factors, investors establish both the risk and expected return of their investment over their investment horizon and calculate, at least implicitly, the expected return they could achieve over that investment horizon. He stated that the expected return over the investment horizon is the compound return (T3: 346 and 347) or geometric mean. Drs. Waters and Winter stated that the geometric mean provides an exact and unambiguous statement of the investors' required rate of return from one point in time to another (Exhibit 6A, Appendix VIII, p. 1). In contrast, they stated that if investment horizons are longer than one year, the arithmetic average of annual returns is a biased estimate of the investors' expected return over the actual investment horizon (Exhibit 6C, p. 4).
In addition, the Customers argued that the recent trend to purchase index funds rather than managed funds results in individual stocks not being evaluated individually and the contents of the selected stock market index being held for long periods of time. As well, they stated that investor expectations will be affected by the achieved returns on the various indices and that these returns are calculated on a compound or geometric basis over a variety of holding periods (JIESC Reply Argument, p. 4).
Drs. Waters and Winter dismissed the comments by Ibbotson Associates as a justification for relying exclusively on the arithmetic mean. They indicated that the comments were made in the context of the appropriate discount rate for corporations to apply in arriving at a decision regarding whether to commit to a particular capital expenditure project and so weren't relevant to the estimation of the cost of capital (Exhibit 6A, p. 25). However, upon examining Exhibit 26, an extract from Ibbotson Associates 1999 Yearbook, Market Results for 1926–1998, Dr. Winter agreed that the justification for using the arithmetic mean which is put forward in Ibbotson is not confined to corporate projects but includes regulatory proceedings (T3: 339).
Drs. Waters and Winter also argued that the arithmetic mean was not required given the uniqueness of public utilities in terms of the extreme stability of their earnings. They stated that the low variation in utilities' achieved rates of return, relative to variability in market portfolio annual returns, meant that the use of the market portfolio's arithmetic mean would result in end-of-(multi)-period utility shareholders' equity values which were closer than intended to the end-of–period wealth values for the market portfolio. They stated that this would result in the typical Canadian energy utility providing investors with two percentage points or more each year in excess of their required rate of return (Exhibit 6A, p. 26).
Ms. McShane dismissed this reason, arguing that if one factors out the uncertainty in the measure of the market risk premium by using the geometric average, followed by a relative risk adjustment that reflects the greater stability of utility returns compared with the market, the lower risk of utilities has been accounted for twice. (Exhibit 6E, p. 5). The Customers suggested that the same argument could be applied, in reverse, to her use of the arithmetic mean (JIESC Reply Argument, p. 7).
9 2.3.1.2 Forward Looking Equity Market Risk Premium In addition to her estimates of the experienced market risk premium, Ms. McShane also provided estimates of the required market risk premium based on forward looking data. She indicated that this type of analysis was required since sole reliance on historic risk premiums to estimate investor expectations ignores the fact that, in her view, risk premiums vary with investors' perceptions of future interest rates, inflation and growth prospects (Exhibit 4A, p. 18). Ms. McShane stated that the increasing integration of capital markets requires recognition that Canadian companies, including utilities, are competing for capital in a global market (Exhibit 4A, p. 10). Recognition that the performance of the Canadian market has significantly lagged that of the international market means that sole reliance on achieved risk premiums in the Canadian market will understate the expected market risk premium (Exhibit 4A, p. 10).
Ms. McShane stated that a forward looking estimate of investors' expectations for the stock market can be made through the use of a discounted cash flow (“DCF”) cost of equity model (Exhibit 4A, p. 19). The DCF model assumes that the investors' required rate of return is equal to the current dividend yield plus the investors' earnings growth expectations.
Ms. McShane provided two estimates of the market risk premium using DCF analysis. First, for the period January 1989 to December 1998, she estimated monthly DCF costs of equity using the sum of month-end dividend yields for the TSE 300 and the respective IBES five-year normalized earnings growth projections as a proxy for longer-term growth. Regressing her monthly equity estimates against month-end yields on long Canada bonds, she determined that at a long Canada yield of 5.5 percent, the indicated market risk premium is 8.2 percent. Second, she performed a similar analysis over the same period for the U.S. stock market using the S&P 500 as the proxy for the market. Regressing her equity estimates against U.S. long-Treasury bond yields, she found that at a bond yield of 5.5 percent, the indicated market risk premium is 9.4 percent. Weighting the Canadian and U.S. forward looking premiums on the same basis as the experienced risk premiums discussed above, she concluded that the forward looking market risk premium is 8.5 percent at a long Canada yield of 5.5 percent (Exhibit 4A, pp. 20 and 21).
The Customers were critical of the forward looking risk premium estimates for three reasons. First, Drs. Waters and Winter stated that DCF-based risk premium estimates depend, on a one-for-one basis, with the IBES analysts' forecasts incorporated in the estimates. Second, they suggested that these forecasts are generally upwardly biased estimates of future growth rates (Exhibit 6A, p. 41). Although they admitted to having used the IBES forecasts in the past (T3: 375, 405), they stated that they had placed little weight on them. Instead, Dr. Winter indicated that the best estimate of investors' expectations could be inferred from an examination of historical evidence (T2: 217). Third, the Customers suggested that the IBES forecast of
10 growth applies primarily to those companies that are the most followed within the TSE 300 and that the number of companies within the forecast declines over time (T1: 38).
In response to the first criticism, Ms. McShane acknowledged that there are other organizations that put together forecasts and that on the basis of the KPMG 18 th Annual Survey of Economic Expectations (Exhibit 11), the equity market risk premium for Canada is 2.7 percent (T1: 44). However, she indicated that she did not know how the numbers included in the survey had been established so that she did not know if they could be viewed as a measure of investor expectations (T1: 45).
In response to the criticism of overly optimistic growth forecasts, Ms. McShane acknowledged that her forward looking DCF results had been characterized by the Province of Alberta, Electric Utilities Act Independent Assessment Team as being overly optimistic (T1: 49) but noted that there have been studies that indicate that analysts' expectations tend to be accurate (T1: 40, Exhibit 10, p. 36). In addition, she stated that she had considered the fact that analysts' forecasts might be considered too optimistic and so did not recommend an ROE that fully reflected these forecasts (Exhibit 4A, p. 22).
2.3.2 Determination of Utility Risk Premium Based on both the historic and forward-looking risk premiums, Ms. McShane concluded that a conservative estimate of the equity market risk premium is 6.0 to 7.0 percent at a 5.5 percent yield on long Canada bonds. At the same long Canada bond yield, Drs. Waters and Winter concluded that the equity market risk premium is 4.5 percent. However, both sets of witnesses agreed that their results could not be added without adjustment to the long Canada yield to obtain the fair rate of return for a utility. Instead, both parties agreed that the market equity risk premium needed to be adjusted to account for the lower relative risk of a benchmark utility.
2.3.2.1 Relative Risk Measures To determine the appropriate adjustment to the equity market risk premium result for a benchmark utility, Ms. McShane began by considering the recent betas 3 of both gas and electric utilities in Canada. She noted that the beta on the TSE 300 Gas/Electric Utilities Index has risen from 0.35 to 0.55 since 1992. However, Ms. McShane indicated that the 'raw' beta, as shown above, is an inadequate measure for determining the required risk premium for four reasons.
3 A measure of the systematic risk of a security, i.e., the risk that cannot be diversified away by holding a broad portfolio of securities.
11 First, Ms. McShane stated that the regression analysis that she had undertaken shows that a significant portion of the utility risk premium was unexplained by the raw beta. She suggested that this meant that there were other factors explaining the relative risk premium that had been overlooked (Exhibit 4A, p. 25). In particular, she noted that utility stocks are sensitive to interest rates and that when these are included in the analysis, the indicated relative risk adjustment rises to 0.73 (Exhibit 4A, p. 26).
Second, Ms. McShane stated that the trend in standard deviations of market returns of Canadian utilities (as a portfolio) shows that the volatility of utility stocks relative to the market as a whole has risen significantly over time (Exhibit 4A, pp. 26 and 27).
Third, she indicated that a review of the U.S. experienced utility risk premiums compared with experienced market premiums for the entire post-World War II period indicated a relative risk adjustment of 0.77.
Finally, Ms. McShane indicated that the major investment houses rely on 'adjusted' betas 4 rather than 'raw' betas to represent the risk of a stock relative to the market. She stated that a raw beta of 0.55 translates into an adjusted beta of 0.70, virtually identical to the results she obtained from the analysis discussed above (Exhibit 4A, p. 27). Based on this analysis, Ms. McShane indicated that the appropriate relative adjustment factor is 0.70, leading to an indicated benchmark utility risk premium of 4.5 percent (Exhibit 4A, p. 28).
Drs. Waters and Winter criticized the analysis undertaken by Ms. McShane on two grounds. First, they indicated that it was inappropriate to derive the risk premium for utilities using a beta derived from both equity and bond market indices but applied only to the equity market risk premium (Exhibit 6A, pp. 43 and 44). Second, they stated that the use of adjusted betas should be made to industry averages not market averages (Exhibit 6A, Appendix V and Exhibit 6C, p 11).
In response, Ms. McShane stated that her analyses demonstrated that there has been a component of utility returns that is not explained by the raw equity beta but is explained by utility sensitivity to interest rates and that this sensitivity could be captured in a simple and straightforward way by relying on adjusted betas for utilities rather than raw betas (Exhibit 4E, pp. 12 and 13).
Drs. Waters and Winter stated that they estimated the relative risk premium for a benchmark utility by constructing a sample of low-risk non-utilities and a sample of lowest risk utilities and then looking at five
4 A beta which has been adjusted to reflect the view that estimated beta values above the average value of one are likely to be upwardly biased estimates of the true value while beta values below the average value of one are likely to be downwardly biased estimates of the true value. The risk of the overall market is set at 1.0.
12 risk measures for these groups. Three of the risk measures, including the raw beta, address share price volatility while two of the measures addressed share earnings volatility. Based on these measures, Drs. Waters and Winter concluded that the risk exposure of the low-risk utilities group is no more than one-half that of the equity market as a whole. In particular, they noted that the greatest relative risk value shown in the evidence related to the standard deviation of achieved returns and was only 0.54 (Exhibit 6A, p. 32, Table 7C).
In their testimony, Drs. Waters and Winter indicated that some of the data presented in their evidence with respect to relative risk measures was not up to date although they believed the effect was either to overestimate the relative risk of the utilities or to leave the values materially unchanged (T3: 381). When the evidence was updated, it showed that for the ten lowest risk utilities, the share price volatility measures either stayed the same or increased, while the share earnings volatility measures either stayed the same or declined. The increase in the risk measures declined if data for the telephone utilities ("telcos") were removed, companies which Drs. Waters and Winter suggested were questionable benchmarks for present purposes, due to competition and technological developments (Exhibit 37). Based on the updated information, the greatest relative risk value is 0.61 for the ten lowest risk utilities and 0.52 if the telcos are excluded.
The Utilities stated that Drs. Waters and Winter had provided no theoretical or empirical support for the risk-return relationship that they relied on for translating their risk measures into a specific adjustment factor. The Utilities dismissed the analysis of Drs. Waters and Winter as being a simple comparison of the market risk measures of the shares of utilities over past years with the risk measures of shares of other companies in the market. The Utilities stated that this might provide some insight into past measures of market riskiness but does not provide adequate insight into current and future riskiness nor insight into the expectations of investors regarding the circumstances of utilities in the future. Specifically, the Utilities argued that the per share earnings volatility risk measures used by Drs. Waters and Winter are not meaningful measures of risk since the TSE 300 values give equal weight to each and every company in the Index, regardless of size and market value. In addition, they argued that a measurement of riskiness based on the volatility of earnings per share suggests that if the rate of return on equity of utilities were fixed in perpetuity, utilities would be less risky since their earnings per share would not change over time (Utilities Final Argument, pp. 18-22).
13 2.4 Comparable Earnings Test Ms. McShane's testimony contains the results of her application of the Comparable Earnings test. Ms. McShane stated that she undertook the Comparable Earnings test because it provides a measure of the fair return based on the concept of opportunity cost (Exhibit 4A, p. C-1).
The Comparable Earnings test measures the rate of return on equity achieved by non-regulated companies judged to be of similar risk to a low-risk utility over a business cycle. As the initial universe of companies from which to obtain her sample, Ms. McShane began by considering only those that were in consumer oriented industries. She stated that these companies exhibited relatively greater stability of earnings than other companies and were more likely to have a risk profile similar to a utility (Exhibit 4A, p. C-2). She then eliminated companies which did not meet certain other criteria, e.g., those that did not have sufficient historical book and market data to allow her to undertake the test, those with a common equity component of less than $50 million or with fewer than 125,000 common shares traded annually, and those which had cut their dividends by more than 25 percent or had not paid dividends over her measurement time period. As a result of these criteria, she had a sample of 41 companies. The remaining firms were ranked by four risk measures and then arrayed by their composite ordinal ranking. Ms. McShane chose as her final sample those 20 companies representing the lower half of the 41 companies based on their composite risk ranking.
Ms. McShane then measured the earnings for the selected sample over the period 1989 to 1997, a period she indicated represented a business cycle. She found that the annual average return for the period was 11.8 percent while the median of annual averages was 11.2 percent. She then conducted an analysis to determine if the past returns could be seen as reasonably representative of near-term future returns and an analysis of the relative risk of her sample to the risk of a low-risk utility. On this basis, she determined that the Comparable Earnings test based on Canadian data gave a result of 11.25 percent to 11.75 percent.
A similar test undertaken with U.S. industrials and adjusted for U.S./Canadian corporate tax rate differences and the differential risk with Canadian utilities suggested that the appropriate return for a Canadian utility was 12.8 percent.
On the basis of these two tests, Ms. McShane concluded that the Comparable Earnings test suggested that the fair rate of return for a benchmark utility is 11.5 to 12.0 percent.
Drs. Waters and Winter did not undertake a Comparable Earnings test. They stated that the concept of comparable earnings does not necessarily have any relationship with the concept of a fair return and that the measurement of comparable earnings (based on accounting data) provides results which are difficult to
14 compare meaningfully across companies and across time (Exhibit 6A, p. 74). Specifically, they argued that the accounting data upon which the Comparable Earnings test is based are not good surrogates for the economic data needed to measure the fair rate of return to the utility (Exhibit 6A, p. 75). Although Dr. Winter agreed that the accounting profession and securities commissions require that book information be provided to investors in the financial statement of companies, he stated that it was not intended to allow securities commissions to calculate a fair rate of return (T3: 407).
In addition, Drs. Waters and Winter were critical of the test on the grounds that the screens used to choose companies to include in the sample might upwardly bias the results. In particular, they indicated that some corporations may have the ability to earn rates of return in excess of competitive levels either through the exercise of market power or through product innovations and that the use of a dividend screen can cause these companies to be over represented (Exhibit 6A, p. 75).
In response to the latter concern, Ms. McShane stated that the companies in her sample operate in a global environment and that their returns when compared with the returns earned by some of the large low-risk industrials in the United States clearly indicated that these companies do not have market power (T1: 106).
In response to the Customers' dismissal of the Comparable Earnings test, the Utilities stated that the test provides an explicit measure of what non-utilities of comparable risk to the benchmark utility are expected to earn and thus directly links the legal requirement and regulatory principle that utilities be allowed to earn a fair and reasonable return on the capital invested in their utility operations with the earnings of non-regulated businesses (Utilities Final Argument, p. 25). The Utilities stated that examination of market risk premiums and utility adjustment factors allows an estimate of the return that investors expect on the market value of their shares, but does not provide a direct estimate of the earnings they expect the companies to achieve. As a result, they argued that the Commission should not rely exclusively on the market equity risk premium approach to set the rate of return on equity (Utilities Final Argument, p. 4).
Mr. Olson, the witness for the CGA, also supported reliance on the Comparable Earnings test (Exhibit 5A, p. 5). Although he acknowledged that the Comparable Earnings test methodology is messy (T2: 170), and that there is no evidence that Canadian utilities are having trouble attracting investment capital (T2: 171), he expressed concern that tying the rate of return on equity exclusively to long bond yields will lead to poor returns and difficulty in placing issues.
2.5 Commission Determinations As indicated in section 2.2 of this Decision, there are three standards governing the determination of a fair rate of return on equity: the comparability standard, the financial integrity standard, and the capital attraction
15 standard. In this hearing, the Utilities have argued that the rate of return that meets the comparability standard is best measured through the use of the Comparable Earnings test. More specifically, Ms. McShane has stated that the Comparable Earnings test is a key measure of the opportunity cost of capital since it measures the returns achieved and achievable by competitive firms of similar risk (Exhibit 4A, p. 44).
In its 1994 Decision, the Commission expressed concerns that the Comparable Earnings test does not measure the opportunity cost of capital (Decision dated June 10, 1994), which the approved ROE should reflect. Although that Decision did not specifically set out the reasons for the Commission's concerns, it referenced concerns raised by witnesses for customer groups at that hearing. These included measurement concerns related to the use of accounting data and sensitivity to sample selection. The evidence brought forward by the Utilities at this hearing has not alleviated those concerns.
In addition, the Commission believes that the distinction which the Utilities have tried to draw between the fair rate of return under the comparability standard and the fair rate of return under the capital attraction standard is artificial. The fact that capital can be attracted at the approved rate of return is proof the comparability standard is met. Only in the event that alternative investments of comparable risk and higher returns were sufficient in volume to push the market price of capital above the regulated ROE would the utility be unable to attract investment. As a result, the Commission has placed little reliance on the Comparable Earnings test in determining the appropriate rate of return on equity for a low-risk utility.
The Commission also has concerns with respect to the forward-looking DCF-based risk premium tests utilized by Ms. McShane. Although the relationship captured in the DCF methodology has a foundation in financial theory, the Commission believes that the measured ROE that results from the application of the DCF test is extremely vulnerable to the assumptions made with respect to forecast growth. As a result, the confidence placed on the resulting investors’ expected rate of return is limited to the confidence placed on the growth estimates. The evidence brought forward at this hearing suggests a strong potential for upwardly-biased growth estimates resulting from overly optimistic analyst forecasts and a decline over the five-year outlook in the number of companies upon which the estimates are based. As a result, the Commission has reached the conclusion that little reliance should be placed on these growth estimates and the resulting DCF-based estimates of the investors’ required rate of return.
Given these two conclusions, the Commission must make its determination giving primary weight to the evidence regarding the historical market equity risk premium and the relative risk of a low-risk benchmark utility. Although the market equity risk premium methodology also requires numerous estimates, the Commission is satisfied that the test most appropriately reflects the standards of establishing a fair return on equity for a low-risk benchmark utility. The estimates of the historical market equity risk premium
16 presented to the Commission vary depending on the data source used, the time period chosen and the measurement technique used, but lie within a range of 3.4 to 5.5 percentage points for Canada and 5.9 to 8.3 percentage points for the U.S.
The evidence provided by Ms. McShane indicates that the market equity risk premium is best estimated by the arithmetic average of one-year holding period returns, using post-World War II data. In contrast, the evidence provided by Drs. Waters and Winter indicates that the market equity risk premium is best estimated by the arithmetic average of the geometric mean of a variety of five or more year holding periods, using data that includes the pre-World War II period. Thus the divergence in equity premium estimates relates largely to the mode of calculation and the time period employed.
With regard to the method of calculation, the choice of arithmetic versus geometric mean reflects the analyst's assumptions regarding investor behavior, specifically the holding period of the equity. The arithmetic average of one-year returns assumes a one-year holding period. Viewed another way, the investor is assumed to make annual additions or withdrawals as required to keep the principal constant, and the average is the mean of these. By contrast, the geometric mean assumes gains are reinvested and losses are absorbed over the selected time span. Where returns are variable, the arithmetic average will be higher than the geometric average, with the divergence increasing the greater the variance.
In the Commission’s view, investors are likely to have a variety of holding period expectations for both bonds and equities, including one-year. As a result, the Commission is of the view that the market equity risk premium put forward by Ms. McShane, which relies exclusively on a one-year holding period, is likely to be upwardly biased. The estimate put forward by Drs. Winter and Waters, while conditioned by consideration of a number of holding periods, does exclude periods of less than five years and so may be downwardly biased. While placing greater weight on multi-year holding periods, the Commission recognizes that issues raised regarding estimates of the market equity premium have not been definitively resolved. These include the scope of the data base, both as regards time period and coverage, as well as the method of averaging. Assessing the proffered estimates in light of their applicability to current bond yields, the Commission finds the market risk premium to be 5.0 percent.
Similarly, the Commission has been presented with a variety of measures to estimate the risk of a benchmark utility relative to that of the market as a whole. In the Commission's view, no one measure is likely to adequately capture all the components which act to form investors' perceptions of the relative risk of a benchmark utility versus that of the market as a whole. However, the Commission is not convinced that the sensitivity of utility stocks to interest rates should be included when assessing the relative risk of a low-risk utility to the market as a whole nor that the ‘adjusted’ beta has a sufficiently strong theoretical foundation to allow the Commission to rely upon it. At the same time, the Commission notes the variety of
17 evidence presented which indicates that a strict reliance on the ‘raw’ beta may under-estimate the relative riskiness of utilities. On balance, the Commission adopts a relative risk for a benchmark utility of 0.6.
After review of this evidence, the Commission finds that the equity risk premium for a low-risk benchmark utility is 350 basis points when long Canada yields are 5.5 percent. This premium includes a 50 basis point cushion to cover the risk of dilution and the cost of new share issues in ordinary circumstances. Accordingly, the fair rate of return on equity for a low-risk benchmark utility is 9.0 percent when long Canada yields are 5.5 percent.
3.0 AUTOMATIC ADJUSTMENT MECHANISM To establish the annual allowed return on equity for a low-risk, high-grade utility, the Commission currently employs a multi-year regulatory model that includes an automatic adjustment mechanism. All participants in this proceeding supported a multi-year model in favour of, for example, annual hearings.
However, the mechanics of the existing automatic adjustment formula were the subject of some evidence and cross-examination. Most notably, parties disagreed about the appropriateness of the so-called sliding scale – that is, a provision of the automatic adjustment mechanism reflective of an inverse relationship between equity market risk premiums and forecast interest rates – at relatively low rates of interest. As well, parties suggested some other amendments to the automatic adjustment formula to improve its effectiveness and to reflect the current operating environment.
3.1 The Sliding Scale The Utilities in this proceeding have recommended an adjustment formula that changes the benchmark utility equity return by 75 basis points for every 100 basis point change in interest rates. Adjustments would be made in reference to a benchmark utility return of 10.75 percent at a long Canada yield of 5.5 percent. The sliding scale would apply at interest rates both above and below 5.5 percent (Exhibit 4A, p. 39). However, the formula would be capped at 8 percent, and reviewed if long Canada yields rise above that level (Exhibit 4A, p. 41)
In contrast, the Customers have recommended that the allowed return on equity change on a one-to-one basis with the change in long Canada yields as long as the benchmark interest rates remain below 8 percent (JIESC Argument, p. 21).
Both the Utilities and the Customers agree that, when long-term interest rates exceed some level, the equity risk premium shrinks as interest rates go up, and widens as interest rates fall. The willingness of investors
18 to accept smaller differentials in the rates of return between stocks and bonds, at high rates of interest, has been ascribed to the existence of a premium which investors require to compensate them for the risk that future unexpected inflation will erode the purchasing power of their investment over the holding period. This premium has been referred to in this proceeding as the lock-in premium, the purchasing power risk premium, and the interest rate risk premium (T3: 440).
The specific rationale for the purchasing power risk premium, which was not in dispute in this proceeding (Exhibit 6A, pp. 49-51; Utilities Final Argument, pp. 5 and 6), is as follows. The coupon rate for a bond is typically fixed, and reflects both the real rate of return required by investors and the compound rate of inflation that investors expect over the bond’s term to maturity. In addition, bond investors require a purchasing power risk premium to compensate them for the risk that the experienced compound rate of inflation over the term of the bond will exceed the expected rate of inflation at the time the bond was purchased. Equity investors are protected from the risks of unanticipated inflation to the extent that the corporation can respond to inflation by increasing its prices to maintain its profitability in real terms. Investors in the shares of regulated utilities are particularly well protected in this regard.
As a general proposition, the higher the level of inflation anticipated by bond investors, the greater is the risk that actual inflation will exceed expected inflation. As a result, the purchasing power risk premium rises and falls with the level of interest rates, at least above some minimum level of interest rates. It follows that over some range of interest rates the equity risk premium is inversely related to interest rates. This is explicitly recognized under the current sliding scale, which has widened the equity risk premium as interest rates have declined. One area of difference between the Utilities and the Customers is the point where the purchasing power risk premium disappears altogether – that is, where bond investors are sufficiently sure of their ability to forecast inflation that they require no specific compensation for the risk that they will get it wrong. The Customers argue that no lock-in premium is required when benchmark interest rates drop below 10 percent (Exhibit 6A, pp. 50-53), while the Utilities argue that the premium disappears when rates fall to 6 percent (Exhibit 4A, p. 37).
The Customers find support for their position in an empirical analysis of the relationship between equity risk premiums and interest rates over time. This analysis concludes that a non-linear relationship is the best fit for data covering two time periods: 1979 to 1996 and 1980 to 1994. This relationship describes a relatively consistent equity risk premium over the level of bond yields that might be expected in a period of low inflation (Exhibit 6A, pp, 53-57). This suggests that no adjustment of an historically estimated risk premium is necessary to reflect current low interest rates (Exhibit 6A, p. 58).
While agreeing that the non-linear relationship described by the Customers is, in fact, the best statistical fit for the data presented (T1: 64), the Utilities dismiss the relevance of the regression by suggesting that the
19 data used were not an accurate reflection of investor expectations for the period in question, and were drawn from too small a sample of companies (T1: 65).
To this criticism, the Customers concede that the data they have used systematically overstates the cost of capital throughout the period, because these data were developed from a DCF analysis. However, the Customers emphasize that they are not concerned with the levels of the equity risk premiums – which may well be too high – but rather are commenting on the relationship of the equity risk premium to changing interest rates. (T3: 411 and 412).
The Customers also used the recent performance of BC Gas stock as evidence that the sliding scale now used by the Commission is overly generous to the Utilities. In particular, the Customers argue that the inappropriateness of the sliding scale at low rates of interest is demonstrated by the fact that the market-to-book ratio of BC Gas has risen substantially during recent periods when interest rates were falling and equity risk premiums have increased (JIESC Final Argument, p. 3).
The Utilities’ support for a sliding scale is based on two different rationales, depending on the level of interest rates.
Below 6 percent, the Utilities claim that “… other factors impact the size of the stock/bond differential, including (1) differential tax treatment of dividends and capital gains versus interest; (2) changes in the perceived riskiness of the equity market relative to the bond market; (3) investors’ willingness to assume risk (i.e., degree of risk aversion); and (4) cyclical factors, e.g., rising earnings expectations in the face of lower interest rates and vice versa” (Exhibit 4A, p. 37). These four factors are defined by the Utilities as “… general propositions that would explain why there would not necessarily be a one-to-one movement below 6 percent between interest rates and equity returns” (T1: 111). It was not the evidence of the Utilities that equity markets are riskier at lower levels of interest (T1: 112).
At rates above 6 percent, the Utilities link the need for a sliding scale to the existence of a purchasing power risk premium. In support of this position, the Utilities have calculated, for the past five years, the difference between the yields on “nominal” long Canadas and Real Return Canada Bonds, and compared this difference with a proxy for investors’ expectations of inflation. This comparison – allowing for tax and liquidity issues which may mean that the yield on Real Return Bonds is not a true measure of the real cost of capital, and allowing that the Consensus Forecast may not reflect investors’ true perceptions about inflation (T1: 58 and 59) – produces a positive differential at interest rates above 6 percent. This differential, the Utilities argue, is evidence of a purchasing power risk premium (Exhibit 4A, p 35).
20 The Utilities also reference an historical data analysis using a broad sample of low-risk utilities (as distinct from the sample used by the Customers), where an inverse relationship between the equity risk premium and interest rates can be found at “current and recent levels of interest rates” (T1: 69).
In contrast, the Customers argue that using a sliding scale at low levels of interest produces “counter-intuitive and analytically unsupportable” results. Using the National Energy Board's (“NEB”) model as an example, the Customers state that for the NEB model to have legitimacy at current rates of interest (i.e., 5.5 percent), one of two situations must be true: either the risk premium required for the equity market as a whole is roughly 6.6 percent, or utility common equity has a relative risk of approximately 0.8. The Customers conclude that both of these “corollaries” of the NEB model are unreasonable and, therefore, the model itself, and others like it, are fundamentally flawed (Exhibit 6A, pp. 59-62).
Within the range covered by their recommendations, the Customers have argued for a constant relationship between the equity risk premium and interest rates. The Customers have not explicitly recommended a factor by which to adjust the ROE for a given change in interest rates when interest rates are in the range over which they believe that an inverse relationship exists (that is, above 10 percent).
The Utilities have suggested an adjustment factor of 0.75, compared with 0.80 in the current BCUC formula. This figure is developed using two different tests.
In the first test, the Utilities have taken the difference between the 1994 yields on nominal long Canadas (9.25 percent) and Real Return Canadas (4.8 percent), and compared this with investor expectations of inflation (taken as 3.0 percent, the high of the Bank of Canada’s target range). They then performed the same calculation using 1999 bond yield data (5.5 percent nominal, 4.0 percent real), and an inflation forecast of 1.9 percent (taken from the Consensus Forecast). The Utilities assume that the market cost of equity in 1994 incorporated the sum of the real long-term cost of capital, investors’ aggregate view of expected inflation, and a premium for risk. Therefore, the net decline in the cost of equity over the past five years would be limited to the sum of the decline in the real long-term cost of capital (measured by the change in Real Return Bond yields, in this case 4.8 percent less 4.0 percent, equals 0.8 percent) and the decline in investor expectations for inflation (in this case, 3.0 percent less 1.9 percent, equals 1.1 percent).
The Utilities compare this 1.9 percent (1.1 percent plus 0.8 percent) reduction in the required stock market return with the corresponding 3.75 percent (9.25 percent less 5.5 percent) drop in nominal bond yields, and conclude that required equity market returns decrease by roughly 50 basis points (1.9 percent divided by 3.75 percent) for every 100 basis point decline in nominal bond yields. Put another way, the equity market risk premium increased by 1.9 percentage points as bond yields fell from 9.25 percent to 5.5 percent (Exhibit 4A, pp. 35-36).
21 By extending this analysis, the Utilities conclude that this roughly 200 basis point increase in the equity market risk premium translates to a 140 basis point change in the benchmark utility risk premium, using the Utilities’ recommended relative risk adjustment of 70 percent. This 140 basis point change in the utility risk premium, when compared to the 375 basis point change in nominal interest rates, suggests a 35 basis point change in the utility risk premium for every one percentage point change in interest rates (Exhibit 4A, pp. 36 and 37).
The Utilities base a second test on investor expectations of earnings growth. They note that over the 1992 to 1998 period, “during which long Canada yields ranged from 9.5 percent to 5.25 percent, the dividend yield/bond yield relationship for a sample of relatively non-diversified Canadian utilities remained in the narrow range of 0.68–0.74, averaging 0.72. That relationship indicates that, absent a change in long-term growth expectations, the utility cost of equity changes by 72 percent of the change in bond yields” (Exhibit 4A, p. 39).
3.2 Suggested Improvements With the exception of the sliding scale, there were only minor differences between the parties, with both the Customers and the Utilities agreeing that the existing automatic adjustment mechanism is working well.
The Customers recommend that the BCUC continue its process of adjusting annually the allowed ROEs for the subject utilities. They also recommend that a generic cost of equity hearing be held if the long Canada bond yield rises above 8 percent and remains above 8 percent for more than six months (Exhibit 6A, p. 7). The Utilities agree that the same 8 percent figure represented a useful – if not “hard and fast” – point at which to review the automatic adjustment formula (T1: 115). The Utilities’ witness further agreed that interest rates should remain above 8 percent for six months in order to trigger a review of the automatic adjustment mechanism (T1: 116).
Under the heading of Technical Issues, the Utilities raised three areas of consideration with respect to the automatic adjustment mechanism (Exhibit 4A, p. 46):
(1) reliance on the Consensus Forecast for November for estimates of the subsequent year’s 10-year Canada yield; (2) rounding the resulting ROE estimate to the nearest 25 basis points; and (3) adjusting the 10-year Canada forecast to a 30-year equivalent by adding the observed spread between 10- and 30-year Canadas for the last seven trading days in November. With respect to the first and second points, the Utilities conclude that the Consensus Forecast should continue to form the basis for the subsequent year’s interest rate forecast. As well, the Utilities conclude
22 that not rounding the final ROE allowance would suggest “spurious accuracy”, and since there is no inherent bias to the rounding, there is no reason to discard the existing rounding procedure (Exhibit 4A, p. 47).
With respect to the third point, the Utilities suggest some changes to the current mechanism. Specifically, the Utilities argue that using seven trading days in November to estimate the spread between 10- and 30-year bonds creates two problems. First, the short observation period is more likely to produce anomalies than would a longer observation period. Second, since forecasts generally reflect recent experience, the estimate of the spread should be based on a period prior to the forecast, rather than subsequent to it. For these reasons, the Utilities have recommended adjusting the period over which the 10-to 30-year spread is measured to capture all the trading days in the October preceding the forecast (Exhibit 4A, p. 47)
Dr. Waters described the changes proposed by the Utilities in this area as “reasonable”. As well, the witness used the same description when asked about the acceptability of retaining other aspects of the automatic adjustment formula (T3: 432).
3.3 Commission Determinations In the Commission’s view, the evidence adduced at this hearing concerning the automatic adjustment mechanism raised a single aspect of disagreement between the Utilities and the Customers: the appropriateness of a sliding scale when long Canada rates are below 10 percent. At long Canada rates above 10 percent, the Customers and the Utilities agree on the existence of a purchasing power risk premium. The Commission accepts this and believes that a sliding scale is appropriate over the range of double-digit interest rates.
The Utilities and the Customers presented conflicting evidence on the appropriateness of the sliding scale for single-digit interest rates. Considering first the range of rates below 6 percent, the Utilities agree with the Customers that no purchasing power risk premium exists at this level. However, the Utilities held that four other factors support the use of a sliding scale, even at very low rates of interest.
The Commission is not persuaded by the Utilities’ evidence in this latter respect, noting that Ms. McShane was able to say only that the four factors are “general propositions that would explain why there would not necessarily be a one-to-one movement between interest rates and equity returns.” This, the Commission believes, is some distance from demonstrating that an inverse relationship does exist between these two elements. The Customers’ evidence showing both why a one-to-one relationship is likely to be correct at low interest rates, and the problems associated with applying a sliding scale at very low rates of interest is
23 more credible. Therefore, the Commission does not accept that a sliding scale is appropriate when long Canada rates are below 6 percent.
The use of a sliding scale for rates between 6 and 10 percent is a more troubling question to the Commission. Simply put, the Commission is not convinced that the evidence before it in this hearing is sufficient to establish with certainty the nature of the relationship between equity risk premiums and long Canada yields over this range of interest rates. More exactly, the Commission: (1) has reservations about the historic data set used the Customers’ witnesses; (2) was not shown the historic data described by the Utilities’ witness; and (3) believes that the Utilities’ analysis covering the post-1994 period is inconclusive because of data limitations.
Having identified this concern, the Commission believes that it would be prudent to retain the sliding scale for interest rates above 6 percent at this time. If a purchasing power risk premium exists between 6 percent and 10 percent, failing to have a sliding scale within that range could produce inadequate returns for the Utilities and result in capital attraction difficulties. From its position of limited information, the Commission believes that the safest course is to maintain the inverse sliding scale. The Commission anticipates that better information on the relationship between equity risk premiums and single-digit interest rates will be available when the Commission next considers this issue.
In considering the size of the adjustment factor to use when interest rate levels trigger the sliding scale (i.e., higher than 6 percent), the Commission is satisfied that the 80 percent figure now in use is appropriate going forward. This decision is based on two factors. First, the Commission is not persuaded that the evidence adduced at this hearing is conclusive in suggesting a more appropriate expansion/contraction factor than the one currently employed. Second, the 80 percent factor has been adequate to protect utility integrity within this range.
Outside of the sliding scale, the Utilities suggested only a single change to the existing formula: that the period over which the 10- to 30-year spread be measured to capture all of the trading days in the October preceding the forecast. The Customers accepted this change. The other technical features of the current formula were accepted by both the Utilities and Customers.
Both the Customers and the Utilities agreed in this hearing that the evidence and recommendations that they have provided should be considered robust to the extent that long Canada yields fail to exceed 8 percent for a six-month period. The Commission accepts this in principle, but believes that such interest rate levels should trigger an assessment of the need for a return on equity hearing, rather than a hearing itself.
24 Therefore, the Commission directs that the current automatic adjustment formula shall continue to be employed, with the following exceptions:
(1) at forecast long Canada yields of 6.0 percent or below, the equity risk premium for a low-risk benchmark utility will be fixed at 350 basis points; (2) at forecast long Canada yields of contraction/expansion factor (i.e., the sliding scale) of 0.8 of the difference in forecast long Canada yields shall be retained and shall be driven off a low-risk benchmark utility ROE of 9.5 percent; (3) to determine the forecast long Canada yield, the period over which the 10- to 30-year spread is to be measured shall be redefined as all the trading days in the October preceding the November Consensus forecast; and (4) the Commission will canvass interested parties on the need for a review of the automatic adjustment formula when long Canada rates exceed 8.0 percent for a period of at least six months. DATED at the City of Vancouver, in the Province of British Columbia, this 26 _____Original signed by:____________________
greater than 6.0 percent, the current th day of August, 1999. Peter Ostergaard Chair _____Original signed by:____________________ Lorna R. Barr Deputy ChairChair _____Original signed by:____________________ P.G. Bradley Commissioner _____Original signed by:____________________ Barbara L. Clemenhagen Commissioner
B R I T I S H C O L U M B I A U T I L I T I E S C O M M I S S I O N
O R D E R N U M B E R G-80-99 SIXT H F L O O R, 9 0 0 H O W E STR E E T, B O X 2 5 0 V A N C O U V E R, B . C . V 6 Z 2 N 3 C A N A D A T E L E P H O N E : (6 0 4) 6 6 0 - 470 0 w e b s it e : h tt p : / / w w w . b c u c . c o m B C T O LL F R E E : 1- 8 0 0 - 6 6 3 -13 8 5 F A C SI M IL E : (6 0 4) 6 6 0 -110 2
IN THE MATTER OF the Utilities Commission Act, R.S.B.C. 1996, Chapter 473 and a Generic Hearing into the Rate of Return on Common Equity for a Benchmark Utility BEFORE: P. Ostergaard, Chair ) L.R. Barr, Deputy Chair ) P.G. Bradley, Commissioner ) August 26, 1999 B.L. Clemenhagen, Commissioner ) O R D E R WHEREAS: A. In its Decision dated June 10, 1994, the Commission established rates of return on common equity (“ROE”) for BC Gas Utility Ltd. (“BC Gas”), Pacific Northern Gas Ltd. (“PNG”), West Kootenay Power Ltd. (“WKP”), and Centra Gas Fort St. John Inc. (“Centra FSJ”); and B. In the same Decision the Commission established a mechanism for varying the ROE on an annual basis; and C. The Commission, by Order No. G-49-97, amended the mechanism to correct for certain problems and to make it more consistent with the practices of other regulatory bodies; and D. In December of each year from 1995 through 1998, the Commission reviewed the financial criteria and issued Letters informing BC Gas, PNG, WKP and Centra FSJ about the approved ROE for each subsequent year based on the original and amended mechanism; and E. In 1998 there were requests from various parties to re-examine the mechanism to determine if further amendments were required. The Commission, by Letter No. L-73-98, advised interested parties that the Commission, in 1999, would be establishing a process for a comprehensive review of all issues surrounding the setting of the ROE; and F. The 1994 ROE Decision anticipated a review of the mechanism; and G. On March 11, 1999, the Commission issued Order No. G-26-99 setting down an oral public hearing into the appropriate ROE and Capital Structures for BC Gas, PNG, WKP, Centra FSJ and Centra Gas Whistler Inc. to commence May 31, 1999; and H. By way of letters dated March 22, 1999 and March 23, 1999, BC Gas, PNG and WKP (“the Utilities”) asked the Commission to reconsider the scope of the hearing and limit the items for consideration to the ROE for a low-risk benchmark utility and to the automatic adjustment formula; and I. The Utilities also asked that the Commission delay the date for the filing of all evidence to May 17, 1999, to make such further adjustments to the Regulatory Timetable as necessary to allow for this delay, and to proceed on a written rather than an oral basis; and . . . /2
B R I T I S H C O L U M B I A U T I L I T I E S C O M M I S S I O N
O R D E R N U M B E R G-80-99 2 J. Comments on this request were received from the Consumers’ Association of Canada (B.C. Branch) et al. and the Joint Industry Electrical Steering Committee supporting the restriction in the scope of the hearing but opposing the request for a written hearing; and K. The Commission considered all the submissions and set down an oral public hearing into the appropriate ROE for a low-risk benchmark utility and into future processes or mechanisms that may be employed to improve the determination of ROE for the Utilities in future years; and L. The public hearing proceeded as scheduled on Monday, June 21, 1999 in Vancouver. NOW THEREFORE, the Commission directs as follows: 1. The equity risk premium for a low-risk benchmark utility is 350 basis points when long Canada yields are 5.5 percent. This premium includes a 50 basis point cushion to cover the risk of dilution and the cost of new share issues in ordinary circumstances. Accordingly, the fair rate of return on equity for a low-risk benchmark utility is 9.0 percent when long Canada yields are 5.5 percent. 2. The current automatic adjustment formula shall continue to be employed, with the following exceptions: (a) at forecast long Canada yields of 6.0 percent or below, the equity risk premium for a low-risk benchmark utility will be fixed at 350 basis points; (b) at forecast long Canada yields of greater than 6.0 percent, the current contraction/expansion factor (i.e., the sliding scale) of 0.8 of the difference in forecast long Canada yields shall be retained and shall be driven off a low-risk benchmark utility ROE of 9.5 percent; (c) to determine the forecast long Canada yield, the period over which the 10- to 30-year spread is to be measured shall be redefined as all the trading days in the October preceding the November Consensus forecast; and (d) the Commission will canvass interested parties on the need for a review of the automatic adjustment formula when long Canada rates exceed 8.0 percent for a period of at least six months. DATED at the City of Vancouver, in the Province of British Columbia, this 26 th day of August 1999. BY ORDER Original signed by: Peter Ostergaard Chair Orders/ROE ’99 Decision
APPENDIX A Page 1 of 1 APPEARANCES
P. MILLER British Columbia Utilities Commission, Counsel K. DUKE C.B. JOHNSON BC Gas Utility Ltd. J. LUTES Pacific Northern Gas Ltd. J. CHRISTIAN Canadian Gas Association R.B. WALLACE Joint Industry Electricity Steering Committee R.J. GATHERCOLE Consumers' Association of Canada (B.C. Branch) et al. [British Columbia Old Age Pensioners' Organization, Federated Anti-Poverty Groups of B.C., Senior Citizens' Association of B.C., West End Seniors' Network, Kootenay-Okanagan Electric Consumers’ Association, End Legislated Poverty and the Tenants' Rights Coalition] R.B. HOBBS West Kootenay Power Ltd. J.D.V. NEWLANDS Fording Coal Ltd. F.S. JAMES British Columbia Hydro and Power Authority D.W. EMES Commission Staff C.B. LUSZTIG ALLWEST COURT REPORTERS LTD. Court Reporters & Hearing Officer
APPENDIX B Page 1 of 1
INDEX OF WITNESSES
BC Gas Utility Ltd., Pacific Northern Gas Ltd., and West Kootenay Power Ltd. — Panel Canadian Gas Association — Panel The Joint Industry Electricity Steering Committee and the Consumers’ Association of Canada (B.C. Branch) et al. — Panel
KATHLEEN C. MCSHANE JOHN E. OLSON WILLIAM R. WATERS RALPH A. WINTER
APPENDIX C Page 1 of 3 LIST OF EXHIBITS Exhibit No.
B.C. Utilities Commission Orders No. G-26-99 dated March 11, 1999 and No. G-38-99, dated April 7, 1999 Letters of Intervention West Kootenay Power Ltd. Affidavit of Publication, dated June 17, 1999 BC Gas Utility Ltd. Affidavit of Publication, dated June 23, 1999 Pacific Northern Gas Ltd. Affidavit of Publication, dated June 22, 1999 Prepared Testimony on behalf of BC Gas Utility Ltd., Pacific Northern Gas Ltd. and West Kootenay Power Ltd. by Kathleen C. McShane, dated May 1999 BC Gas Utility Ltd. Response to B.C. Utilities Commission Information Request, dated June 1999 BC Gas Utility Ltd. Response to Joint Industry Electricity Steering Committee and Consumers' Association of Canada (B.C. Branch) et al. Information Request, dated June 1999 BC Gas Utility Ltd. Response to Fording Coal Ltd. Information Request, dated June 1999 Rebuttal Testimony of Kathleen C. McShane, dated June 1999 Submissions of the Canadian Gas Association and Direct Testimony of John E. Olson Response to Joint Industry Electricity Steering Committee and Consumers' Association of Canada (B.C. Branch) et al. Information Request by John E. Olson for the Canadian Gas Association, dated June 14, 1999 Canadian Gas Association Information Request No. 1, Question 6 Amended Table Written Evidence of Dr. William R. Waters and Dr. Ralph A. Winter, dated May 1999 Responses of Drs. Waters and Winter on behalf of the Joint Industry Electricity Steering Committee to West Kootenay Power Ltd., Pacific Northern Gas Ltd. and BC Gas Utility Ltd. Information Request No. 1 and B.C. Utilities Commission Staff Information Request No. 1 Rebuttal Evidence of Dr. William R. Waters and Dr. Ralph A. Winter, dated June 1999 Errata sheet with respect to Direct Evidence of Dr. William R. Waters and Dr. Ralph A. Winter, dated May 26, 1999 BC Gas Utility Ltd. Information Request of Drs. Waters and Winter, dated June 18, 1999 Table 1, Equity Risk Premium Estimates of the Fair Return on Equity on a High-grade Utility IBES Global Estimates, The Number of Global Company Estimates Maintained in the IBES Database as of December 1998
1 2 3A 3B 3C 4A 4B 4C 4D 4E 5A 5B 5C 6A 6B 6C 6D 7 8 9
LIST OF EXHIBITS (Cont'd.)
Article by V.K. Chopra entitled “Why So Much Error in Analysts’ Earning Forecasts?” dated November/December 1998 KPMG, 18 th Annual Canadian Survey of Economic Expectations, dated 1999 Copy of Appendix D entitled “Equity Market Risk Premium” to the Province of Alberta Electric Utilities Act, Independent Assessment Team, Volume X “Cost of Capital in Power Purchase Arrangements” Draft Final Report: 31 Rebuttal Testimony of John E. Olson on behalf of Trailblazer Pipeline Company Dr. Waters “Calculation of the Risk Premium for a Low Risk Utility, PNG February 1992” Dr. Waters “Calculation of the Risk Premium for a Low Risk Utility, BCUC March 1994” Dr. Waters “Calculation of the Risk Premium for a Low Risk Utility, BCUC May 1999” Comparison of Dr. Waters Calculation Equity Market Risk Premium and Risk Premium for a Lowest Risk Utility, 1992-1999 Extract from Written Evidence of Dr. William R. Waters on Fair Rate of Return, dated March 1994 Transcript pages 813–815 of Written Evidence of Dr. Waters in 1994 Written Evidence of William R. Waters before the National Energy Board Hearing, dated September 1994 Article entitled “The Behavior of Equity and Debt Risk Premiums” by John D. Finnerty and Dean Leistikow Excerpt from the article by John D. Finnerty and Dean Leistikow (noted above Exhibit 21) Excerpt from BC Gas Inc. 1998 Annual Report Written Evidence of William R. Waters before the National Energy Board, dated December 14, 1984 Copied Globe and Mail Newspaper articles, dated February 20, 1999 Ibbotson Associates 1999 Yearbook, Market Results for 1926–1998 Excerpt from Principles of Corporate Finance, Fifth Edition Excerpt from Written Evidence of Drs. Waters and Winter Decision Tree
APPENDIX C Page 2 of 3 Exhibit No.
10 11 st May, 1999 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
LIST OF EXHIBITS (Cont'd.) Excerpt from Written Evidence of William R. Waters in B.C. Utilities Commission March 1994 Proceedings Article entitled “Using Analysts’ Growth Forecasts to Estimate Shareholder Required Rates of Return”, by Robert S. Harris Excerpt from Written Evidence of William R. Waters and Ralph A. Winter, Newfoundland and Labrador Board of Commissioners of Public Utilities, dated April 15, 1998 Direct Evidence of William R. Waters in the BC Gas Utility Ltd. April 1992 Hearing before B.C. Utilities Commission Written Evidence of William R. Waters in the March 1994 Generic Hearing before the B.C. Utilities Commission Excerpt from Financial Theory and Corporate Policy by Copeland/Weston Excerpt from Pre-filed Testimony of Dr. Ralph A. Winter before the Ontario Energy Board, Centra Gas Hearing on August 16, 1993 Excerpt from the National Energy Board Reasons for Decision relating to TransCanada Pipelines Limited, dated June 1989, and excerpt from the transcripts of Dr. Waters’ Evidence Response to Undertaking, given at Transcript page 383 to update Tables 7A, 7B and 7C, dated June 30, 1999 Response to Undertaking, given at Transcript page 390 to update Global Beta Values to December 31, 1998, dated June 30, 1999 Analysis referenced on page 4 of Rebuttal Testimony of Kathleen McShane Return on Equity, Bond Yields at Decision date and Implicit Risk Premiums Bank of Canada Review, Summer 1996
APPENDIX C Page 3 of 3 Exhibit No. 30 31 32 33A 33B 34 35 36 37 38 39 40 41