DECISION QUEBEC RÉGIE DE L’ÉNERGIE D-2007-116 R-3630-2007 October 15, 2007 PRESENT: Richard Carrier, B. Sc. (Econ.), M.A. (Econ.) Gilles Boulianne, B. Sc. (Econ.) Richard Lassonde Commissioners Gaz Métro Limited Partnership (Gaz Métro) Applicant and Interveners whose names appear on the following page Decision Application to modify the tariffs of Gaz Métro Limited Partnership effective October 1, 2007 D-2007-116, R-3630-2007, 2007 10 15 2 […] 4. TOPICS ADDRESSED IN HEARING 4.1 RATE OF RETURN 4.1.1 MODELS USED The experts heard used different approaches and models to estimate the rate of return on the shareholder’s equity of Gaz Métro. The IGUA’s expert, Dr Booth, uses the traditional capital asset pricing model (CAPM) and a two-factor model involving the market risk premium and the risk premium for Canada longterm bonds1. Gaz Métro’s expert, Dr Chrétien, applies the Fama-French model to portfolios of corporate securities that have similar characteristics to those of a benchmark distributor. He also uses a modified CAPM2. The CAPM is represented by the following equation: K = Rf + β*(Rm - Rf) This equation represents the rate of return (K) an investor expects to earn on an investment in a security having a certain risk. The expected return for that security equals the return that could be earned on a risk-free investment (Rf) plus a risk premium. The premium for the security evaluated is proportional to the market risk (Rm - Rf) that is estimated using the return (Rm) earned on a diversified portfolio of securities. The relationship between the market risk and the risk associated with the security is expressed by the beta (β) factor. The Fama-French model used by Dr Chrétien has been developed since 1993 and is increasingly used in finance. Based on the evidence, however, its use in regulatory milieus to fix the return of a public utility is a first in Canada and has only been used once in the United States3. Even though the two models differ, the objective of both is to estimate the return an investor expects to earn on an investment in securities having a certain risk. The main difference 1 2 3 Exhibit C-8.10-ACIG, Evidence of Laurence D. Booth, July 2007, pages 54 and 55 Exhibit B-4-Gaz Métro-7, document 8, page 44 Exhibit B-11-Gaz Métro-7, document 8.4, page 1 D-2007-116, R-3630-2007, 2007 10 15 3 between the two approaches is in the method used to express that risk. In the Fama-French model, the security’s risk is based on three explicative variables, i.e. the market, value and size risk premiums, and not on a single variable, i.e. the market risk premium, as is the case in the CAPM. Moreover, the model proposed by Dr Chrétien is based on a different evaluation of the risk-free rate (Rf) than that normally used with the CAPM. Dr Chrétien proposes the following equation: K = Rfadj + β(market) * λ (market) + β(size) * λ (size) + β(value) * λ (value) Dr Chrétien states that the Fama-French model emulates the market returns better than the CAPM. He says that is verifiable particularly for securities having a lower than average risk and for “value” securities, regardless of the market considered, i.e. in North America and in other international markets. Dr Booth says that model is very controversial in finance and submits various articles published in specialized literature to support his opinion4. The Régie does not retain the Fama-French model for establishing the rate of return in this decision. The application of that model to regulated enterprises has not been sufficiently examined to date to be used as a basis for fixing the rate of return of a distributor. Based on the evidence, regulatory application has only been discussed in a very few specialized literature articles. Nevertheless, the evidence submitted by Dr Chrétien shows the model’s ability to produce meaningful robust statistical results to explain the returns earned in the past by corporate portfolios, mainly in the gas distribution industry. Dr Booth states that the statistical performance of the model is not the primary factor to consider. Instead, the model’s consistency with fundamental finance principles should be examined, in particular in terms of analysis of the risk factors that may influence the return on securities. In the Régie’s view, a more in-depth analysis of these issues is required before recognizing and applying the Fama-French model for determining a reasonable rate of return. Based on the evidence, the Régie also notes, among other things, the inherent difficulty in developing series of data over a sufficiently long period of time and the inclusion in the reference portfolios of enterprises whose activities sometimes go significantly beyond distribution or “regulated” activities. The Régie also questions the inclusion of income trusts in the portfolios retained and the possible impact thereof on the estimate of the value risk factor. Lastly, the Régie questions the possible circularity phenomenon that may result from using portfolios of regulated enterprises to estimate the size and value parameters. While a 4 Exhibits C-8.30, C-8.31 and C-8.32-ACIG, Answers to commitment number 2. D-2007-116, R-3630-2007, 2007 10 15 4 number of these difficulties may also apply with the CAPM, it appears they may be more acute in the Fama French model because of the nature of the explicative factors considered. With respect to the modified CAPM used as a complement by Dr Chrétien and the multifactor model used by Dr Booth, the Régie only uses the information provided by the results presented to validate its assessment of the rate of return. For these reasons, the Régie mainly retains the CAPM for its decision. It is the approach retained by the Régie in its past decisions and the most common approach in Canada. This model is recognized or used in finance milieus and by most experts who testify before regulatory bodies. However, the use of this model involves substantial difficulties that the Régie addresses in greater detail in the following sections. 4.1.2 MARKET RISK PREMIUM The CAPM requires establishing the market risk premium used to fix the premium of a benchmark enterprise or distributor. Dr Chrétien presents a market risk premium of 6.43%, which is based on average returns obtained using the arithmetic average from 1951 for the Canadian market and from 1927 for the American market5. To do this, he uses the weightings retained by the Régie in Decision D-99-1506, i.e. 60% for Canadian data and 40% for American data. He also justifies taking the American data into account by the increasingly greater integration of the two economies7. Dr Booth says the integration of the two economies should reduce, not increase, the risk premiums. He refers to various Canadian and American data in his testimony, but gives special weight to Canadian data. He also mentions the need to look at data for shorter periods as well as data over long periods of time. Dr Booth presents estimates for the periods starting in 1926 and in 1957 by using the arithmetic and geometric averages and the ordinary least squares analysis method. He recommends a market risk premium of 5.0%. Based on the evidence in the application, the Régie estimates the market risk premium using the same weightings for Canadian and American data as in 1999. In its view, the opening up of markets offers investors various investment alternatives that it is necessary to reflect in 5 6 7 Exhibit B-11-Gaz Métro-7, document 8.1, page 2 Decision D-99-150, application R-3428-99, page 10 Exhibit B-11-Gaz Métro-7, documents 8.27 and 8.30 D-2007-116, R-3630-2007, 2007 10 15 5 establishing a reasonable rate of return. The Régie also maintains the establishment of the market risk premium on the basis of the arithmetic average of market returns. However, the choice of reference periods for establishing the risk premium raises certain issues. In effect, the average may be substantially different depending on the starting year and the series of data retained. In particular, since 1999, the statistics show a significant decrease in the average returns, notably on the averages for shorter periods. The drop in securities values in 2001 and 2002 partially explains this phenomenon. In this context, the Régie places greater importance than before on the long period averages. Based on the evidence in the application, the Régie establishes the market risk premium in a range varying from 5.40% to 5.90%. 4.1.3 RISK OF BENCHMARK DISTRIBUTOR Dr Booth and Dr Chrétien then estimate the risk of a benchmark distributor, i.e. a public utility with a low risk. A benchmark distributor’s risk is measured by the beta factor, which represents the risk differential between the benchmark distributor and the market in general. The establishment of a beta factor is one of the biggest difficulties in applying the CAPM. These difficulties involve establishing reference portfolios that are representative of the risk of regulated enterprises and getting valid series of data to make a robust estimate. Dr Booth points out the problems related to the impact of including data of a few large enterprises in the samples, such as the heavy weighting of Nortel in the TSX index a few years ago. He states that the resulting bias is significant when estimating the betas of all regulated enterprises if the betas of those large enterprises differ significantly from the norm. He presents various estimates based on recent data, but says it is necessary to use judgement and proposes establishing the beta of a benchmark distributor based on the historical average, which he estimates to be between 0.45 and 0.55. Dr Chrétien presents a beta of 0.53 based on various reference portfolios used in connection with the application of the Fama-French model. D-2007-116, R-3630-2007, 2007 10 15 6 He also presents a modified beta of 0.588 in applying the modified CAPM. The objective of the modified beta is to take account of the empirical research showing the trend of betas to converge towards one (1). On the other hand, Dr Booth says the betas of regulated enterprises converge towards the average of the betas of their group and not towards one (1). After review, the Régie maintains the position expressed in Decision D-2003-93 to the effect that the betas of regulated utilities converge towards their own average and not towards the market average, which by definition, equals one (1)9. Even though it is a determining factor in the application of the CAPM, it remains difficult to objectively infer the value of the beta based on the market data for the enterprises retained in the samples. Based on the evidence in the application, the Régie establishes the beta of a benchmark distributor in a range of 0.50 to 0.55. 4.1.4 RISK-FREE RATE The CAPM requires establishing a risk-free rate to which is added the enterprise’s risk premium. The risk-free rate used is normally the rate for Government of Canada 30-year bonds established based on the specialized publication of Consensus Economics Inc. Dr Chrétien proposes an improvement to the approach used in 1999, called the modified risk-free rate. He submits the use of the projected reference year rate as the basis for establishing the reference return is arbitrary in that the rate for the year in question is not representative of the average of those rates over a long period. As the risk premium is established using long period averages, he recommends using the long period average of Government of Canada bonds with a term of more than ten (10) years to establish the reference return and then adjust the resulting risk premium for the 2008 year based on the elasticity factor of 75-25 as established in the 1999 Decision. By retaining the 6.41% rate proposed by Dr Chrétien and the 4.78% risk-free rate in the application, the resulting adjustment of Gaz Métro’s risk premium would be approximately 41 basis points. Dr Booth submits that it is not appropriate to correct the 1999 risk premium using a modified risk-free rate because the fair and reasonable rate of return allowed by the Régie in 1999 was based on an assessment that took account of all the information available at the time. He is not aware of any other regulatory body that has used the modified risk-free rate concept. He recommends the Régie continue to use the unmodified risk-free rate for establishing a 8 9 Exhibit B-4-Gaz Métro-7, document 8, page 53 Application R-3492-2002, phase 1, Decision D-2003-93, page 73 D-2007-116, R-3630-2007, 2007 10 15 7 reasonable return but that, in its assessment of the risk premium, it take account of all the relevant factors on the basis of the evidence before it, including, among others, the relationship between the risk premium and the rate on long-term bonds. The Régie finds that the positions expressed by the two experts are not irreconcilable. According to the Régie, the application of the CAPM involves an additional difficulty when the evaluation of the return occurs in a period when current government bond rates are substantially different from the average long period rate. As the risk premium is calculated over long periods and represents the difference between the arithmetic average of the market returns and the government bond returns, that premium is therefore basically representative of the conditions that prevailed over that period. An adjustment is therefore required in the Régie’s assessment when the conditions of the bond market depart from that average. In short, Dr Chrétien proposes an explicit adjustment of the risk-free rate in such a case whereas Dr Booth relies on the judgement of the experts and the regulator. In the Régie’s view, the modified risk-free rate approach is an interesting means of facilitating the application of the CAPM. The Régie feels this initial discussion of the issue merits further review. However, the discussion would not significantly change the reasonable rate of return to which the shareholder is entitled. Unless explicitly treated as Dr Chrétien proposes, the problem has to be taken into account in the assessment of the facts as Dr Booth says. In this application, the Régie increases the results produced by the CAPM by 40 basis points. 4.1.5 OTHER MARKET ACCESS COSTS The issue and other costs of accessing the markets were not reviewed in detail in this application. They are 0.30% for the enterprise in Dr Chrétien’s evidence and 0.50% in Dr Booth’s. The Régie maintains Gaz Métro’s issue and other costs of accessing markets at 30 basis points as established in Decision D-99-11. 4.1.6 GAZ MÉTRO’S RISK There was ample evidence of Gaz Métro’s risk compared to the risk of a benchmark distributor and the evolution thereof since 1999. In Dr Carpenter’s opinion, that risk has increased since 1999 notably on account of the substantial increase and greater volatility of natural gas prices. He also says the enterprise’s D-2007-116, R-3630-2007, 2007 10 15 8 risk is increased by the performance incentive mechanism introduced in 2001. He concludes that Gaz Métro has a greater risk than a benchmark distributor and that an increase of 50 basis points is justified. Dr Booth states that Gaz Métro’s business risk is greater than its peers because of the composition of its clientele. However, he states that a higher capitalization ratio and greater risk coverage through numerous deferral accounts offsets the business risk, with the result that the enterprise’s overall risk is in the average. Another IGUA witness states that the risk related to the composition of the clientele has lessened since 1999. He also states that the performance incentive mechanism does not increase the enterprise’s risk at all and provides an opportunity to earn a higher return, if justified by performance, and provides various protective measures that make it comparable to a cost-based regime. Based on the evidence in the application, the Régie feels the enterprise’s overall risk has increased since 1999. The Régie bases this conclusion on the much higher average supply price and the impact of that increase, particularly with respect to the competitive position in relation to fuel oil and electricity10. The Régie is concerned about the loss of industrial volumes and the relative stagnation of total sales since 1999. The Régie also notes the greater volatility of supply prices. However, this is partially compensated by the utilization of derivatives in connection with an approach the broad parameters of which have been authorized beforehand by the regulator. In the Régie’s view, the enterprise’s risk has not increased significantly because of the introduction of the performance incentive mechanism. The mechanism has been adjusted twice since 2001. While it is premature to judge the impact of the changes made to the regime in 2007, the opportunity of taking account of the enterprise’s required revenue each year for fixing the tariffs is maintained, which is similar in this regard to traditional costbased regimes. The Régie evaluates the enterprise’s overall risk as higher than average but takes account in its assessment of the greater coverage of those risks through deferral accounts and a slightly higher capitalization ratio than average. In the Régie’s view, the enterprise’s risk has increased since 1999 and is greater than a benchmark distributor. Based on the evidence in the application, the Régie feels the 10 Exhibit B-16-Gaz Métro-2, document 2, page 3 D-2007-116, R-3630-2007, 2007 10 15 9 greater risk justifies setting the risk premium in the order of 25 to 35 basis points higher than a benchmark distributor. 4.1.7 RESULTS OF CAPM Taking account of all the preceding conclusions with respect to the application of the CAPM, the resulting return for the distributor would be in a range varying from 8.43% to 9.08%. 4.1.8 AUTOMATIC RISK PREMIUM ADJUSTMENT FORMULA Based on the evidence, the simple update, with no other adjustment, of the enterprise’s risk premium using the automatic rate of return adjustment formula in force since 1999 would produce a rate of return on shareholder’s equity of 8.91%. 4.1.9 OTHER MODELS As previously mentioned, the Régie uses the information provided by the results of the modified CAPM used by Dr Chrétien and the multi-factor model used by Dr Booth to validate its assessment of the rate of return. 4.1.10 COMPARISON WITH AMERICAN DISTRIBUTORS During the hearing, comparisons of the returns allowed Canadian regulated utilities with their American counterparts were heard at length. Management representatives of both Gaz Métro and IGUA explained the related issues for the industry in general to the Régie. In this application, the Régie decides the reasonable rate of return applicable to Gaz Métro’s shareholder. It does not decide on the current debate within the industry. However, it indicates the factors that had an influence on its assessment. In the Régie’s view, even though rates of return allowed in the United States are clearly higher on average than those allowed in Canada, the evidence does not make it possible to conclude there is any prejudice to or unfair treatment of the distributor. The applicant has not demonstrated that the opportunities offered on the American market are comparable, in D-2007-116, R-3630-2007, 2007 10 15 10 particular in terms of risk. On the one hand, Dr Chrétien’s evidence shows that the risk premiums associated with the portfolios of American enterprises were higher in the past than those of portfolios of Canadian enterprises11. On the other hand, Dr Carpenter’s evidence does not convincingly establish that the enterprise’s risks are comparable to the risks of the American enterprises used in comparison. The evidence does not make it possible to compare the overall differences that may exist in the institutional, economic and financial contexts of the two countries and their impact on the opportunities they provide for investors. Mention was made of the results of the Concentric Energy Advisors study commissioned by the Ontario Energy Board in 2006. The study refers to, among other things, the fact the returns allowed in Ontario and the United States reversed starting in 1997, i.e. since the introduction of the automatic rate of return adjustment formulae based on the variation in the long-term bond rates. These observations will merit further discussion in the future by the proponents of each contention. 4.1.11 CONCLUSION Under the terms of its empowering legislation, the Régie must determine the reasonable return on the distributor’s rate base. This hearing reviewed a new approach for establishing the return on the shareholder’s equity, i.e. the Fama-French approach, incorporating additional explicative factors to take account of the impacts of size and value in the assessment of the risk premium. The new approach is not retained for purposes of this decision because of the lack of work or research, in both academic and regulatory milieus, supporting its use for fixing the rate of return of a regulated enterprise. The Régie retains the results produced by the Capital Asset Pricing Model (CAPM) as a reference basis. However, this hearing raised certain issues relating to the objective application of the CAPM for determining a reasonable rate of return in the present context. In its opinion, the Régie indicates the nature of those issues for which more elaborate evidence would be useful for the future. Lastly, it presents the content of its conclusions on the application of the CAPM. The Régie uses the results of the other risk premium-type models as information. 11 Exhibit B-4-Gaz Métro-7, document 8, page 40 D-2007-116, R-3630-2007, 2007 10 15 11 Given the results of the CAPM, the result the automatic adjustment formula in force since 1999 would have produced and the Régie’s conclusions about the distributor’s greater risk since 1999, the Régie considers it is reasonable to place the return allowed the enterprise at the high end of the range of the results produced by the CAPM. Based on the evidence in the application and for all the reasons given, the Régie fixes the return on the shareholder’s equity of Gaz Métro at 9.05% effective October 1, 2007 on the basis of a risk-free rate of 4.78%, which corresponds to an implicit risk premium of 4.27%. Furthermore, on the basis of an equity ratio of 38.5% and the debt cost presented in the application, the Régie establishes the average cost of capital on the rate base at 7.68% and the prospective cost of capital at 6.69%. The Régie also renews, effective for the 2009 fiscal year, the automatic rate of return adjustment formula on the terms and conditions established in Decision D-99-11. […]