"[O]ne of the biggest challenges facing regulators is to encourage the benefits of competition while protecting electric consumers from excessive rates that produce windfall profits for shareholders."
"For example, over the past few years, California lOUs have asked for a higher allowed return on common equity (ROE) to compensate shareholders for the additional risk associated with competition in the electric generation sector."
The Independent Energy Producers Association (IEP) had asked the California Public Utility Commission (CPUC) to consider a new approach to regulating the cost of capital for California electric utilities. Susan Stratton Morse testified on behalf of the IEP. The IEP offered this premise: Unless utilities disaggregate the risks of providing unbundled services, along with rates and the allowed shareholder returns, customers of investor-owned utility companies (lOUs) will end up overpaying for certain services, such as T&D, while underpaying for others, such as generation. Moreover, utility generation services would gain an unfair advantage over generation-only competitors. In a November 1994 decision, the CPUC agreed the IEP and stated that any future unbundling of services should consider an unbundling of the ROE and rates for such services.
The authors rely on telephone and natural gas utility data as a proxy for electric generation and T&D because no market information on common stocks is available for stand-alone electric utilities engaged solely in generation or solely in T&D. The authors calculate ROE differences between certain telephone and natural gas utilities presumed to operate in competitive or non-competitive markets, and assume that electric generation and T&D were both financed with the same mix of debt and equity, implying no difference in financial risk. They also rely upon a decision from the Federal Energy Regulatory Commission (FERC) now several years old to provide quantitative support. That reliance raises questions.
Morse et al. state that, "at a minimum, the ROE (return on common equity) for generation is at least 15 percent higher than the overall ROE for the combined generation/T&D utility." They also assert that their analysis, using proxy natural gas and telephone utilities (em and calculations using the capital asset pricing model (CAPM) (em implies an ROE differential of 175 basis points for Southern California Edison Co.
In essence, the authors claim that "the challenge in unbundling the cost of capital comes not in proving the theory to be fair and equitable, but in determining how and when to do it." Nevertheless, they provide no evidence to confirm their assumption that differentials in investment risk (the sum of business and financial risks) derived from calculated ROEs for telephone and natural gas proxy utilities should equal whatever may be the differential in ROE business risk between electric generation and T&D.
An ROE set by a regulatory agency represents a substitute for an ROE determined by common stock investors. Thus, one can obtain a preliminary indication of the relative ROE risk difference between different kinds of utility services by reference to market-determined, price-earnings multiples (P/E) and comparisons of market-to-book ratio (M/B). Table 1 presents a tabulation of those ratios, as derived from the March 1996 C.A. Turner Utility Reports.
The data reveal that investors are willing to invest significantly more for each dollar of earnings, and to pay a higher premium over book value for the common stocks of telephone and natural gas utilities, than for electric and water utilities. These two widely employed market benchmarks (P/E and M/B) strongly suggest that investors believe an investment in the common stock of electric and water utilities (with lower P/E's and M/B's) is similar but more risky than an investment in the common stock of telephone and natural gas companies. Further, the higher ratios of the telephone and natural gas companies for March 1996 do not represent an isolated phenomenon. Their P/E's and M/B's have remained higher for several years.
Only a refined analysis may reveal whether business risk accurately tracks investment risk differences, or whether financial risk differences between different kinds of utilities explain a conclusion of lower investment risk for electric and water utilities compared to telephone and natural gas utilities. For instance, the risk of rate regulation makes up one
of the more significant elements
of business risk, especially in
an emerging competitive environment.
Morse et al. assume that the RHCs (regional Bell holding companies) and LDCs (local gas distribution companies) essentially still operated in a monopoly environment, while AT&T and natural gas pipeline companies operated in a workably competitive environment. They calculate investment ROE differentials using a CAPM analysis. Their calculations revealed a higher investment risk ROE for AT&T than for the RHCs, and a higher investment risk ROE for pipelines than for LDCs. However, M/B and P/E comparisons do not confirm those conclusions. And more than that, every market model employed to estimate the market required investment risk ROE, such as CAPM (as employed by the authors), must rest upon the Efficient Market Hypothesis (EMH). If the market is not efficient (em and there is significant academic support to show that the market may not be efficient (em then CAPM computations are questionable.
The authors expect the reader to believe that differentials in risk indicated by ROEs derived from CAPM computations between competitive AT&T and "monopoly" RHCs, and competitive pipelines and "monopoly" LDCs, equal the market's perceived business risk differential between electric generation and T&D. However, an analysis of betas and common equity ratios over a relevant time period indicate that the market finds competition less risky than continued exposure to price regulation for many of the proxy "monopoly" utilities.
Table 2 shows financial risk (common equity ratios) and beta comparisons for AT&T, the average for Moody's eight gas distribution companies, C.A. Turner Pipelines, and Edison International (formerly SCEcorp), the parent of Southern California Edison, for 1995 and 1985.
The average beta for the pipelines was higher at 1.13 in 1985, when they operated in a monopoly environment, compared to 0.98 in 1995, when they operated in a competitive environment. The average betas for the RHCs and the Moody's eight LDCs also declined between 1985 and 1995. Yet, in 1995 both are exposed to at least some competition. The beta for Edison International also declined, yet its principal subsidiary faces greater business risk today. Table 2 also shows that the AT&T beta rose modestly, but that rise is attributable to an increase in financial risk through a substantial decline in
its common equity ratio (em from 58 percent in 1985 to about
40 percent in 1995. Thus, based upon CAPM computations that take into account differences in financial risk, the increase in beta for AT&T is not attributable to an increase in business risk, despite AT&T's having moved to a competitive environment.
Investment risk represents the sum of business and financial risk. Beta presumably captures elements of both. A change in beta may indicate a change in either business or financial risk, or both. To demonstrate the point, consider the fact that AT&T's beta was 0.80 in 1985 (just after the AT&T break-up and the creation of seven HCs), but rose to 0.85 by 1995. The authors assert:
[D]uring the last decade the telephone industry was deregulated and long-distance carriers were exposed to competitive risks. ROEs for long-distance carriers increased steadily due to increased competition, while ROEs for the regional Bell operating companies (the regulated arm) remained relatively constant.
It is unclear which ROEs the authors relied upon to support their statement. The fact is, however, that the RHCs have seen their achieved ROEs rise dramatically since the AT&T breakup, to a current average of more than 23 percent. AT&T's ROE also stands currently at about 23 percent, in spite of massive write-offs unrelated to the telephone industry. The market appears to prefer the risk of competition compared to the risk of original-cost, ROE price regulation, as evidenced by the fact that telephone P/E's and M/B's now appear significantly higher than prior to competition. When financial risk difference is taken into account, the beta comparisons between 1985 and 1995 lead to the same conclusion: Namely, in spite of the introduction of competition, investment risk stands lower today than 10 years ago (em not higher, as implied by the authors.
vs. Business Risk
One method that addresses the impact of financial risk on the estimated cost of common equity can be found in an article from 1987 written by Brigham, Gapenski, and Aberwald.1 The article states that for each one percentage point change in debt ratio between 40 and 50 percent, there is a change of 7 to 15 basis points
in cost rate. The cost rates for long-term debt and equity move in the same direction (although not in tandem).
For simplicity, assume the cost rate change is 10 basis points for each 1 percentage change in common equity ratio. For AT&T, the cost rate change between 1985 and 1995 related only to financial risk should equal 180 basis points, given an 18-percentage-point drop in the common equity ratio. Yet, AT&T's 1995 beta of 0.85 is 0.05 higher than its 1995 beta of 0.80. If one can assume a 7-percent risk premium in 1995 in the CAPM computation, then the beta-derived ROE investment cost rate should be only 35 basis points higher (7% x. 0.5). Compare this increase of 35 basis points, attributable to higher business risk according to the authors, with the increase in financial risk of 180 basis points inherent in AT&T's higher common equity cost rate for 1995. These data suggest a perceived decrease in business risk attributable to competition, even though investment risk rose overall. That higher overall investment risk can be attributed to the rise in financial risk related to the decline in common equity ratio (em from 58 to about 40 percent.
Similar computations for the seven RHCs also fail to confirm the authors' conclusions. For example, based upon an assumed equity-risk premium of 7 percent and the current beta difference between AT&T and the average for the RHCs of 0.12 (0.85 - 0.73), the ROE investment risk difference is 84 basis points (0.12 x 7%). However, when one takes into account the incremental difference of 10 percentage points in common equity ratio (50 percent for the RHCs versus 40 percent for AT&T), there is implied a greater net difference in financial risk (em 100 basis points more for AT&T. This higher financial risk exceeds the investment risk differential of 84 basis points implied solely by the difference in beta. Since investment risk represents the sum of business and financial risk, the end result (relying on beta difference) indicates that the business risk for AT&T stands lower than the average risk for the RHCs, in spite of AT&T's exposure to more competitive risk than the RHCs. This result suggests a conclusion opposite to that advanced by the authors. It implies that the market perceives a greater risk in regulation than in competition.
Moreover, if beta is flawed, as some suggest, or if the market is not efficient, then flaws will also appear in the authors' quantification of a higher business risk for generation and a lower business risk for T&D.
The Trouble With Beta
There is reason to believe beta is flawed. The R2 term (coefficient of determination of the regression analysis which gives rise to beta) for a utility beta generally ranges between less than 0.10 to 0.30. Thus, 70 to 90 percent of the stock price movement used to develop the betas have nothing to do with company-specific risk. In fact, Fama and French have concluded2 that beta seems to have no role in explaining the average returns on New York, American, or NASDAQ stock exchanges for 1963-1990. Reporting on that work prior to publication, the New York Times found other experts in agreement: "[F]inance experts say that Professors Fama and French have presented the most conclusive evidence against beta." The Times confirmed that conclusion from Fama: "'The fact is,' Professor Fama said in a recent telephone interview, 'Beta as the sole variable explaining returns on stocks is dead."3 However, Fischer Black later rebutted Fama and French, saying: "Beta is a valuable investment tool if the line is as steep as the CAPM predicts. . . . The evidence that prompts such statements implies more uses for beta than ever."4
All this suggests that analysts should avoid relying exclusively on any single model for cost of common equity, including the CAPM. Nevertheless, we do not mean to say that the CAPM cannot help quantify an ROE or ROE differentials between utilities or between electric generation and T&D. However, we do suggest using other tools to minimize the estimating error. Keep in mind that every method or model employed to estimate ROE involves subjective judgment: Every method is flawed in one way or another.
"Pure Plays" and Benchmarks
Morse et al. assert that bond-rating criteria published by Standard & Poor's (S&P) reflect a belief by S&P that electric generation is more risky than T&D. Of course, S&P may be correct that the majority of business risk faced by electric utilities arises from the generation function. However, S&P does not offer any direct quantification of the ROE generation differential vis-a-vis T&D. Nor does S&P offer any comparisons of ROE differentials between electric and gas utilities, or between electric and telephone companies (the two proxies employed by the authors).
The authors also rely upon a finding by the FERC that granted market-based rates for electric wholesale generators which, it is claimed, are at least an approximation for "pure play" generation companies. Specifically, the authors cite the cases of Ocean States Power I5 and II.6
In the Ocean States cases, the authors state that the FERC-approved, market-based rates included an ROE 15 percent higher than the FERC's generic benchmark ROE for electric utilities (i.e., 115 percent of generic ROE). However, the opinion of the FERC on common equity cost rates is often disputed, even by its own staff. Thus, FERC staff recently relied upon a risk-premium method to estimate the market-required cost rate for common equity based upon state ROE findings. State findings over time have been higher than FERC findings, according to FERC staff.
More important, however, is the fact that the Ocean States analysis appears out of date.
In January 1992, after it had issued the Oceans States I opinion, the FERC abolished its Generic Benchmark Rate of Return on Common Equity for electric utilities.7 According to the FERC, in its Ocean States II decision, the generic rate had not produced many of the benefits envisioned for it in 1984. Also, the FERC has since rejected the method used in its generic ROE formula.
The generic ROE was derived using a discounted cash flow (DCF) model with a single-stage growth rate factor. In each of its quarterly calculations from 1984 through 1991, the FERC had applied essentially the same single-stage growth rate. Subsequently, several FERC staff members testified that the DCF model was unreliable when used by itself and added in a risk-premium calculation to determine the appropriate ROE. Moreover, since 1994 the FERC has adopted a DCF model with a two-stage growth factor, claiming the single-stage model produced unreliable results.
For example, consider this statement made by the FERC in Opinion No. 396, issued in May 1995:
"In Ozark, however, [we] found that a projection limited to five years, with no evidence of what is anticipated beyond that point, is not consistent with the DCF model and cannot be relied on in a DCF analysis."8
In short, the method relied upon by the authors to determine ROE for a "pure play" generating company (em one that is 115 percent of the FERC generic benchmark ROE, as described in Ocean States I & II (em is based upon 1) old precedent, 2) a generic benchmark approach later abandoned by the FERC, and 3) a DCF model with a single-stage growth factor that the FERC has since found unreliable.
Adding Up the Evidence
Undoubtedly, it may prove desirable to unbundle ROE to distinguish the difference in business risk between electric generation and T&D. Generation may carry a higher business risk than T&D.
Nevertheless, the authors have failed to provide any reliable quantification of whatever, if any, is the ROE differential between electric generation and electric T&D. Indeed, it appears the authors have proved, based upon beta, that the market believes competition imposes a smaller business risk than does traditional, rate-base regulation, when the ROE differentials between different types of telephone carriers and different types of natural gas companies are used as a proxy for ROE differential between T&D and generation. The only other conclusion that can be reached is either that beta is flawed, or the market is not efficient.
Moreover, the quantification offered by the authors, relying as it does on past FERC decisions, is supported by outdated and abandoned methods. Finally, even if the quantifications offered by the authors were correct, they offered no clear proof that ROE investment risk differentials between different types of telephone and natural gas companies represent the ROE business risk differential between electric generation and T&D.
The authors ask the question, "Rate Unbundling: Are We There Yet?" Based on their evidence, the answer is, "Not by a long shot." t
Joseph F. Brennan is chairman of the board of AUS Consultants, headquartered in Moorestown, NJ. He has testified on the subject of rate of return in more than 400 utility cases over the past 30 years. J. Robert Malko, Ph.D., is a professor of finance at Utah State University, in Logan, and an associate of AUS Consultants. He is the former chief economist of the Wisconsin Public Service Commission. Messrs. Brennan and Malko gratefully acknowledge assistance by Frank J. Hanley, president of AUS Consultants, in the preparation of this article.
Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.