PRESENTATION TO THE REGIE DE L'ÉNERGIE FILE R-3492-2002 By: Dr. Lawrence Kryzanowski Dr. Gordon Roberts March 13, 2002 Part II RATE OF RETURN ON COMMON EQUITY OR ROE § We rely primarily on the Equity risk premium test § Under this method, the recommended ROE is the risk-free rate estimate + the risk premium estimate § When using historical data, we deal with three important issues: • First, what is the proper annual average? - If returns were IID or identically and independently distributed over time, and horizon was 1 year, then it would be the arithmetic mean - Unfortunately, returns are not IID due to mean reversion/aversion and regime shifts, & horizons are long-term; yesterday we heard “market volatility is not constant” - Thus, a weighted-average of the geometric and arithmetic means, which places more weight on the geometric mean, is more appropriate - To be conservative, we use an equal-weighted average of these two types of means • “In practice, two combinations of risk -free rates and equity-risk premia are seen: (1) long-term risk -free rates plus geometric means or (2) short-term risk -free rates plus arithmetic means. Nothing in the theory of the CAPM dictates the use of these parameters; they are artifacts of practice. A recent survey of leading American corporations and financial institutions suggests greater use of the geometric-mean/long-term risk -free rate approach.” • This recent study is: R. F. Bruner, K.M. Eades, R.S. Harris and R. Higgins, Best practices in estimating the cost of capital: Survey and synthesis, Financial Practice and Education (Spring/Summer 1998). § Second, what is the proper time period? • Again, if returns are IID, then longer time periods are better since the precision of the estimate increases with more data points • When returns are not IID, the use of a long time period gives equal weight to observations from a regime that is no longer valid as it does to observations from a regime that is currently valid • When returns are for a market proxy whose construction is not consistent over time, this also argues in favor of using a shorter time -period • We conclude that the time period since 1956 is best for Canada but we also utilize both longer and shorter time periods § Third, realized and expected risk premiums can differ • Many studies show that the historical realized risk premium exceeds the risk premium that was expected over that period for both Canada and the US • Equity investors realized more than they expected while bond investors realized less than they expected • Thus, the use of historical estimates generates an upwardly biased estimate of expected or going-forward risk premia § We estimate an equity market risk premium of 4.70% • To emphasize the conservative nature of this estimate, please note that it is near the highest weighted risk premium of 4.76% reported in our Schedule 8, and does not reflect the large negative risk premium for 2002 § We use a DCF test using historical and future estimates of dividend growth at the market level to check the reasonableness of our estimate • We again conclude that our estimate is conservatively high § We use various other estimates to determine where our estimate falls in the possible range of values. These include: • Expected risk premia estimates of various academics and practitioners – almost all of these are below our 4.7% • Surveys of Canadian investment professionals (Mercer; Watson Wyatt, 5.5%, 6%, 7.8%, 7.9%) that imply at most a 3% risk premium over long Canada bonds which is considerably lower than our 4.7% estimate • These sources strongly suggest that future equity risk premia will be lower than in the past, & may even be nil or negative. In other words, there has been a regime shift downwards in the risk premium • The US equity risk premium experience also suggests that our estimate is generous § We provide numerous fundamental changes that imply a decreasing equity risk premium, such as: • Marked reduction in trade costs over the last century, which has lowered the equity risk premium by about 1% • Achievement of market diversification at lower cost through the development of various index products • Fuller and less costly diversification of domestic market risk through international diversification • However, we make no adjustment to our market risk premium estimate for these fundamental changes § We estimate the relative investment riskiness or beta of HQ DIST at 0.5 • We conclude that this estimate is conservatively high since: - The trend in utility betas is downwards to zero ? Not upwards towards 1 § § § § By multiplying our market risk premium estimate of 4.70% by our estimate of HQ DIST's beta of 0.5, we obtain the own risk premium for HQ of 2.35% We add 10 basis points to compensate for potential equity issuance costs This yields our return on equity recommendation for HQ DIST of 8.45% This is 245 basis points over our long Canada forecast of 6.0% CRITIQUE OF THE EVIDENCE SUBMITTED BY DR. MORIN § Beta estimation problems: • Beta estimation problem #1 – Use of Value Line Betas • If the betas of Canadian utilities have been moving towards 1 at a reversion rate of onethird as Dr. Morin has argued over the years, how come the rolling five-year average beta is moving towards zero and not 1 for our sample of utilities? • Dr. Morin misinterprets the 1986 study by Kryzanowski and Jalilvand. Like the study by Gombola & Kahl, this study provides support for the regression tendency of utility betas to regress toward the grand utility mean and not towards the market average of 1 • Thus, since Dr. Morin basically uses the sample average utility beta as his estimate of the beta for HQ DIST, no further adjustment is required • Unadjusting his adjusted betas yields a beta estimate of 0.51, which is almost identical to ours § § § Beta estimation problem #2 – Further upward beta adjustment via the ECAPM We show that the further adjusted beta becomes 0.75 by placing a 75% weight on Dr. Morin's adjusted beta of 0.67 & the remaining weight on the market beta of 1 Thus, a raw beta of 0.51 becomes 0.67 after round-one adjustments, and becomes 0.75 after round-two adjustments § § § Not only is the ECAPM devoid of any theoretical basis as is generally the case for financial models but Dr. Morin provides no peer -reviewed support for its empirical validity or use in practice The ECAPM allegedly adjusts for early findings that the estimated risk-free rate exceeds its realized value when its realized value is proxied by the Treasury Bill rate. However, regulatory rate setting uses the higher long Canada rate Thus, how many times do we need to adjust for the same problem? § § § § § Beta estimation problem #3 – No downward beta adjustment when using the US market risk premium It is well known in finance that different beta estimates are obtained using different market proxies Our evidence shows that the beta of interlisted Canadian utilities using the U.S. index is lower than that using the Canadian index Thus, when using the higher US equity risk premium, Dr. Morin should have used a lower beta In integrated markets, such as for the interlisted shares of Hydro if it was a public company, the same firm has the same return in both markets § Market risk premium estimation problems • Market risk premium problem #1 – Dr. Morin only uses historical arithmetic mean returns - Counter to the underlined IR response by Dr. Morin, many textbooks on finance & scientific journals advocate a major or exclusive role for the geometric mean in computing the costs of capital - Dr. Morin gives the geometric mean no role; we give it an equal role § § Market risk premium problem #2- Choice of return series for determining the market risk premium If Dr. Morin believes that the longest time period is best, why did he not extend the time periods of some known studies either forward or backward in time when such was possible? § We document the impact of extending the series & for the correction of apparent calculation errors on his estimates § We also calculate the sizeable impact from adding 2002 realized returns to his estimates § § Two of the 6 studies used in his estimation of the market risk premium use a DCF analysis His DCF analyses on Canadian and U.S. markets are extremely unreliable for such reasons such as: • His constant growth discounted dividend valuation model assumes that the growth of dividends, earnings, book value & stock prices are identical and constant forever • Although the growth of earnings and dividends are assumed to be equal for this model to be valid, he obtains, for example, his estimate of the growth rate for Canadian dividends of 10.3% by averaging the projected dividend growth rate of 5.4% with the projected earnings growth rate of 15.1%. We do not consider 5.4% and 15.1% to be approximately equal! § § § § If the long-run annual rate of inflation is assumed to be 3%, then Dr. Morin is estimating that his sample of Canadian firms is expected to grow forever at an annual rate of 7.3% These forever dividend and earnings growth rates greatly exceed expectations for the growth rate in the Canadian economy Dr. Morin's 5-year expected growth rate makes no adjustment for the historic low in the earnings of Corporate Canada in 2001. High 5-year growth rates from a low base are seldom good as forever forecasts Dr. Morin's analysis makes no adjustment for the known upward bias in bottom-up analyst forecasts. This bias can be as high as 20% for the next year's earnings forecast, & the inflated forecasts tend to grow for more distant forecasts § Market risk premium problem #3 – Use of DCF estimates of fair return as check of ROE recommendations • Circularity is a particular problem for DCF use for individual firms in regulated industries; analysts base their expectations on the rate of return allowed by regulatory bodies • The DCF model assumes that returns are set competitively. We provide evidence that such is not the case, since investors have reaped annual free lunches form their utility investments in Canada • Furthermore, the 2001 study by Richard Bernstein and Lisa Kirscher finds that the S&P Utility Index outperformed the Nasdaq index since the inception of Nasdaq in 1971; a higher annual return while incurring less risk § Comparison of the recommended rates of return against those generated using adjustments formulas • The formulas embody the risk -free and the required risk premium deemed appropriate by various regulators for determining the rate of return on equity for an average utility • Using our forecasted rate of 6% in five adjustments formulas produces an average risk premium of 363 basis points with a relatively narrow range of 345 to 381 basis points § § § The resulting average rate of return on equity of 9.62% is lower than Dr. Morin's recommendation of 10.5 to 11% and higher than our recommendation of 8.45% However, the average recommended ROE drawn from these regulatory formulas is a generous upper bound This average return needs to be adjusted downwards to, first, reflect the relatively lower risk of HQ DIST, & second to reflect the fact that these formula were set in a higher risk premium regime. Our evidence presents a large body of argument showing that the equity risk premium is expected to be considerably lower in the future than it was in the past COMMENTS BY WARREN BUFFET ABOUT DERIVATIVES § DERIVATIVE INSTUMENTS WERE “FINANCIAL WEAPONS OF MASS DESTRUCTION” § THEY “POSED A “POTENTIALLY LETHAL” THREAT TO THE ECONOMY UNDERLYING REASON § § REFLECTED HIS FRUSTRATION AT BEING UNABLE TO EXTRACT BERKSHIRE FROM GENERAL RE SECURITIES LATTER HAS PORTFOLIO OF “EXOTIC” PRODUCTS THAT WAS PROVING IMPOSSIBLE TO UNWIND QUICKLY § § § VERY LONG DATED PRODUCTS WITH TERMS IN EXCESS OF 10 YEARS IN COMPARISON TO AVERAGE 5 YEAR TERM OF CREDIT DERIVATIVES WITH TOTAL RETURN CHARACTERISTICS § E.G., SECURITIES WHOSE PERFORMANCE LINKED TO 30-YEAR YIELD CURVE & OTHER VOLATILE BENCHMARKS CREDIT DERIVATIVES MARKET § § RELATIVELY SMALL REGULATORS OCCASIONALLY EXPRESS PUBLIC CONCERN ABOUT: • LACK OF TRANSPARENCY IN THIS MARKET • INABILITY TO MEASURE EXPOSURE ACCURATELY