APPENDIX B
CRITIQUE OF EVIDENCE PRESENTED BY DR. MORIN
1.0
Introduction
The purpose of this critique is to show that the «base» equity risk premium estimates of a fair rate of return made by Dr. Morin are quite close to our own recommendation. What causes Dr.
Morin’s final return recommendation to be higher than our fair return recommendation are the «bells and whistles» which he applies to his base estimates. Is it a coincidence that each of the adjustments made by Dr. Morin raises his return recommendation? All other things equal, we would expect that some of the adjustments would increase and some would decrease the return estimate. This, however, is not the case. We shall show that these add-ons have no economic basis and are merely attempts to inflate his return recommendation. Once they are removed and the base risk premium estimates of a fair return on equity is compared with our estimate in this hearing, the true picture of what is a fair return emerges.
2.0 Risk Premium Evidence
Our risk premium estimate above long Canadas is considerably less than the estimates produced by Dr. Morin for a number of substantive reasons. These relate to the correct beta estimate and the size of the market risk premium.
2.1
Beta Estimates
The beta estimate used by Dr. Morin in this hearing is too high. The witness uses Value Line betas which adjust for the tendency found by Blume (1975) that the average U.S. equity beta regresses in the long term toward the market beta of 1.0.
1
That is, firms with betas less than one tend
1
M.E. Blume, 1975, "Betas and Their Regression Tendencies", Journal of Finance , Vol. 10, No. 3, pp. 780-
1
to increase toward the market beta of one and firms with betas greater than one tend to decrease to one over time. When Blume ran the regression of the betas for the period 1948-54 against the same stock's betas for 1955-61, he arrived at the following results:
Install Equation Editor and doubleclick here to view equation.
where ß i2
stands for the beta on stock i in the later period (1955-61) and ß i1
stands for the beta for
stock i in the earlier period (1948-54). This relationship implies that the beta in the later period is
.343 + .677 times the beta in the earlier period. The equation lowers high values of beta and increases low values of beta. The adjustment of 1/3 plus 2/3 of the raw beta is evident from the above equation estimated by Blume.
The Blume study was, however, performed on a large sample of New York Stock Exchange common shares. It relates, therefore, to the average security in the U.S. equity market. The study was not performed exclusively on utility stocks as we shall do in this section.
Schedule B1 presents rolling beta estimates for 16 Canadian utilities over 12 separate 5-year periods between January 1984 and December 1999. The beta estimates which incorporate the
October 1987 stock market crash are shaded. The betas appear quite constant from period to period with a general depression in the estimates once the October 1987 stock market crash data is included in the estimates . If we now repeat Blume's analysis using the 1995-99 and 1990-94 periods, we get the following results:
Install Equation Editor and doubleclick here to view equation.
795.
2
where ß i2
represents stock i's beta in the 1995-99 period and ß i1
stock i's beta in 1990-94. These results suggest a regression tendency towards an overall beta of .582 (i.e., .378/.650) which is slightly higher than the mean of the utility sample in Schedule B1 over the 1999 period. It appears, therefore, that Canadian utility betas tend to regress toward their grand utility mean and not 1.0. R.
Gombola and D. Kahl have confirmed this result for U.S. utilities. They also showed by analysing individual stocks, that there is a tendency of betas to regress towards their mean value.
2
Despite the fact that Blume’s results do not apply to regulated Canadian utility stocks, Dr.
Morin uses this upward adjustment (by adopting the Value Line betas) to provide an upward bias in his recommended return on equity for TransÉnergie. That is, instead of the adjusted beta of .65 used
by Dr. Morin, he should have used the unadjusted beta of .48 [= (.65 -.33)
×
1.5].
As a check on his beta value of .65 for TransÉnergie, Dr. Morin goes on to examine the business risk (unlevered) betas for 10 Canadian utilities. In doing so, he makes the ‘conservative’ assumption that TransÉnergie’s business risk is one standard deviation below the Canadian utility average unlevered beta. Assuming Hydro-Quebec’s non-consolidated equity ratio of 26.7%, he then derives a levered equity ratio of .64 using the formula on page 19 of his evidence. Schedule B2 shows the unadjusted betas for the same 10 Canadian utilities. The average unlevered unadjusted beta for the sample is .16 with a standard deviation of .08, so that one standard deviation below the average produces an unadjusted unlevered beta of .08 for TransÉnergie. Using the leverage formula on page 19 then produces an unadjusted levered beta for TransÉnergie of .30 (= .08/.267) which is less than half of the estimate achieved by Dr. Morin.
For illustrative purposes, using Dr. Morin’s market risk premium of 6.6%, the upward adjustment in beta of .17 (= .65-.48) produces an upward bias in the allowed return on equity of 1.12%
(=.17
×
6.6%) for TransÉnergie.
2
M.J. Gombola and D.R. Kahl, 1990, «Time Series Processes of Utility Betas: Implications for Forecasting
Systematic Risk», Financial Management, Autumn, pp 84-93 .
3
2.2
Market Risk Premium
The second substantive point of contention is with the market risk premium estimate of 6.6% used by Dr. Morin. Appendix D of our evidence discusses at length some of the problems encountered in estimating the market risk premium from experienced return data. It points out that the arithmetic rate of return is always higher than the geometric rate of return. This suggests that it is important to use the correct estimation procedure to calculate the excess rate of return.
Putting aside his U.S. evidence for the moment, Dr Morin relies on two Canadian data sources. The Hatch and White data covers the period from 1950-87 and produces a market risk premium over that period of 6.9%. If we update that data to 1999 using the Report on Canadian
Economic Statistics 1924-99 data published by the Canadian Institute of Actuaries (CIA), Schedule
B3 shows that the arithmetic mean Canadian market risk premium was .30% over the 1988-99 period and -.87% based upon the geometric mean. When these results are used to update the Hatch and
White data, the Canadian market risk premium over the 1950-99 period is within the range 5.04%-
5.32%, with a point estimate of 5.18% which is well below the 5.5% estimate calculated by Dr.
Morin.
3
The second Canadian data series used by Dr. Morin is the CIA data over the entire 1924-99 period and using this data, he estimates the market risk at 5.8%. It is important to recognize that even though Dr. Morin uses data prior to 1956, there is no consistent equity market data in Canada for the pre-1956 period. The TSE300 index only goes back to 1956, so that earlier data is usually obtained by splicing together series that may have very little in common. Also, the data on fixed income securities is of poor quality, because of the existence of interest rate controls and the absence of a functioning Canadian money market to price securities. It was not until the 1953-54 reforms introduced by the Bank of Canada that an active secondary market for short term Canada bonds even developed. Moreover, the introduction of the dividend tax credit in 1948 meant that equity data
3
Response to Question 28, part a, of the Industrial Coalition.
4
before this period reflected a fundamentally different tax system than that which currently exists. It is mainly for these reasons that the period after 1956 is most useful.
If old data is deemed useful, Siegel "The Equity Premium: Stock and Bond Returns Since
1802," Financial Analysts Journal (January/February 1992) has pushed U.S. data back to 1802! He concludes:
"The magnitude of the excess return on equity, especially during this century appears excessive relative to the behaviour of other macroeconomic variables. In the future, the real return on fixed income assets may be closer to the historical (sic) norm of 3 to 4 per cent. While stock returns will probably continue to dominate bond returns, they will not do so by nearly as wide a margin as they have done over the passed 65 years."
Siegel's work showed that the real return on equity has been around 6.5% throughout the period
1802-1989. In an article in Business Week ("These are the Good Old Days for Bonds", December 6,
1993), the author, Christopher Farrell, refers to Siegel's work and notes that during the last three decades of the 19th century which had similar characteristics to today's economic environment, stocks outperformed bonds, but not by much. After adjusting for inflation, stocks returned 8.5% annually and bonds 6.57% over the thirty year period. What has caused the high excess return during the CIA years was the depressed real return on bonds. Clearly, since 1980, real returns on bonds have not been depressed, which leads to the conclusion that the 1924-1999 period may be anomalous.
As further support for the narrowing of the market risk premium, MIT economist Olivier
Blanchard found a long-term decline in the equity premium since the late 1940s in the U.S.
4
When inflation increased in the late 1970s, the real return to bonds was "hammered" and the equity risk premium widened. Then, when inflation slowed in the late 1980s, real bond returns increased sharply and the equity premium narrowed. What this implies is that any recognition of the
4
Business Week , "Stocks vs. Bonds: Equities Have the Edge", March 7, 1994.
5
integration between of U.S. and Canadian capital markets must recognize that the U.S. market risk premium has decreased with the lower rates of inflation in the recent past and that this pattern should persist into the future.
As we shall show, 80% of the evidence used by Dr. Morin to derive his Canadian market risk premium relies upon U.S. data. Specifically, he first employs the Ibbotson Associates data which shows a 7.8% U.S. market risk premium to make inferences about the Canadian market risk premium.
Dr. Morin also applies a DCF analysis to the TSE using Value Line data. The projected growth for the Value Line common stocks is found to be in the range 5.7-14.4%. Using a dividend yield of 1.5%, he concludes that the expected return on the aggregate equity market is in the range of 7.2-15.9%, for a mid-point of 11.6%. He then adjusts the spot dividend yield to an expected dividend yield and applies his quarterly dividend adjustment to achieve a DCF-based estimate of
12.0%.
5
At a current long Canada bond yield of 6.1%, the implied risk premium is about 6.0%.
What this suggests is that with a long run inflation rate of approximately 3%, Dr. Morin believes that the real long term rate of growth for his sample of Canadian equities is approximately
8.6% (=11.6% - 3.0%). Dr Morin fails to recognize that the growth rate in the constant growth version of the DCF model relates to the forward infinite (long-term) horizon - not to the next year, and not to the next five years alone. If his Canadian equity sample of firms grow at a real rate of
8.6% and the Canadian economy grows at a real rate of 4%, an obvious contradiction exists. Since the Canadian economy is not expected to grow over the long term (or even the short term) at this excessively high rate, it must be that his Canadian equity sample of firms also cannot grow, in the long term, at this rate. Simply put, if this sample of firms did grow in the long term at this rate, they would be the Canadian economy!
5
The CRTC in Telecom Decision CRTC 92-9, page 70, stated «The Commission considers the use of the quarterly DCF model with an annual rate base inappropriate.» We urge the Regie to also disallow this unsupported upward adjustment.
6
Dr. Morin applies the same DCF approach to the U.S. aggregate equity market using a projected growth rate in the range of 7.2% - 12.1%. Again, this excessive growth rate assumption cannot realistically be applied as the long term growth factor required in the DCF model. However,
Dr. Morin still uses this unrealistic growth assumption which is inconsistent with the DCF model to inflate his risk premium.
Using 5 guides, of which two (or 40%) are based upon U.S. evidence, Dr. Morin concludes that the Canadian market risk premium is 6.6% and the CAPM equity risk premium for TransÉnergie is 4.3% (=.65
×
6.6%).
2.3
Empirical CAPM
Earlier, in Section 2.1 we discussed the unjustified upward adjustment of beta employed by
Dr. Morin based upon the tendency for the average U.S. security’s beta to regress toward one in the long term -- even though this was shown not to be the case for utility stocks both in Canada and the
U.S. In spite of this, Dr. Morin employs a second unjustified upward beta adjustment, referred to as his Empirical CAPM (ECAPM). What this is in fact is yet another unsupported beta adjustment in which he multiplies his previously upward adjusted beta by .75 and then adds .25 to the product.
This results in a total adjustment of beta in his ECAPM of .50 times the original unadjusted beta plus .50. If, for example, the correctly estimated beta was .48, Dr. Morin's ECAPM would use .48
times .50 plus .50, or .74. This is more than a 50% increase in the measured beta. It is our view that this adjustment is unsupported since Dr. Morin is using the return on long Canadas in his CAPM and
ECAPM and not the lower 90 day T-Bill rates as would be consistent with the pure CAPM. At the same time, it should be recognized that the multiple adjustments to the measured beta of .48 result in an increase of 1.58% in Dr. Morin’s return recommendation for TransÉnergie.
We should take note, moreover, that the ECAPM beta is multiplied by the market risk premium of 6.6% which is 40% U.S. based so that Dr. Morin’s ECAPM equity risk premium of
7
4.9% (=.74
×
6.6%) is not only biased upward due to the multiple beta adjustments, but also biased because of the use of a U.S. based market risk premium.
Dr. Morin then uses two purely U.S. data sets in which he estimates both a prospective and historical risk premium from each set of data. The first sample is based upon U.S. electric utilities, where he derives a prospective equity risk premium of 3.4% and an historical average risk premium over the sample period of 5.2%. The second test is based upon U.S. natural gas utilities, where he derives a prospective equity risk premium of 3.9% and an historical average risk premium over the sample period of 5.6%.
We can now summarize the risk premium results presented by Dr. Morin in a similar way to that presented on pages 27-28 of his evidence.
CAPM
ECAPM
U.S. Electric Utility Prospective
U.S. Electric Utility Historical
U.S. Natural Gas Prospective
U.S. Natural Gas Historical
AVERAGE
Equity Risk Premium U.S. Percentage
4.3% 40%
4.9% 40%
3.4%
5.2%
3.9%
5.6%
4.6%
100%
100%
100%
100%
80%
Although the Canadian and U.S.economies are partially integrated at present and as such,
Canadian market risk premiums already incorporate these effects, there are serious impediments to the full integration of the two financial markets. For example, the monetary policies of both countries have been different, as have the roles of debt and equity financing. In Schedule B4 is a graph of the difference between U.S. and Canadian three month Treasury bill yields, since January 1952. Two observations are important. First, the level of Canadian Treasury bill yields has generally been above
8
those in the U.S. over the last 15 years, reaching a maximum spread of over 560 basis points in late
1992. This is because of the tough anti-inflation policy adopted by the Bank of Canada. This illustrates the different monetary policies recently adopted by the two countries. Second, the behaviour of interest rates in the two countries has often times been dramatically different. For example, in the 1970's Canada yields reached a maximum of 281 basis points less than those in the
U.S. This difference in volatility affects the risk of different types of debt instruments. The higher volatility of Canadian interest rates means that they are perceived as being riskier than in the U.S., and as a result, long term rates are higher. The implication of this is that adding an historic U.S.
spread to a high Canadian rate, as is implicitly done by Dr. Morin does not make any sense. It is like counting apples and oranges.
As long as the Bank of Canada controls the Canadian money supply, the financial risk of instruments denominated in Canadian dollars may not be comparable to instruments denominated in other currencies. This is because of the different risks associated with both the expected level of inflation and the uncertainty surrounding that level, as well as differences in short-term interest rates brought about by divergent monetary policies.
It should also be pointed out that the spread in Schedule B4 is based upon debt instruments, where there is considerable integration between the U.S. and Canadian markets. The same cannot be automatically implied for equity instruments. This is because of the current 25% restriction on foreign investment within the portfolios of Canadian mutual and pension funds for these funds to be eligible as registered retirement savings plan (RRSP) investments.
6
Also, the dividend tax credit on Canadian equity securities makes foreign equity investments unattractive to personal investors.
The result is that there is no natural relationship between U.S. and Canadian equity yields, since capital does not flow as freely to arbitrage values as in the bond and money markets. The resulting spreads are more volatile.
6
The percentage is scheduled to increase to 30% in 2001.
9
In Canada, equity income is given very favourable tax treatment in the hands of ordinary investors to compensate for the corporate income tax. Hence, the double taxation of equity income at both the corporate and the individual level is mitigated by giving individuals a dividend tax credit.
The U.S., by contrast, continues to tax all investment income, whether interest or dividends, at the same rate. The result is that the required return on U.S. equity instruments is bid up, to compensate for their extra tax burden, relative to Canadian equity securities.
For example, in the U.S. we might observe a bond yield of 10% and estimate an equity yield of 15%. With a 50% tax rate, the after-tax yields would be 5% and 7.5% respectively, so that the investor would require a pre-tax equity risk premium of 5% and an after-tax equity risk premium of
2.5%. In Canada, by contrast we might observe a bond yield of 10% and an estimated equity yield of 'only' 10%. However, with a 50% tax rate on interest income and an effective tax rate of only
25% on equity income (because of the dividend tax rate), the after-tax yields are 5% and 7.5% respectively.
The result of the differential tax regimes is that a pre-tax equity risk premium of zero in
Canada could equate to the same after-tax risk premium in the U.S. of 2.5%. Hence, although the after-tax risk premiums and yields could be identical in the U.S. and Canada, the before-tax yields, that we observe in the market, would be entirely different. Moreover, given the tax structure, everything else held constant, U.S. equity investments will always be higher than for similar firms in Canada. This reason alone, aside from the observed volatility in spreads on debt instruments, is enough to render irrelevant any observations on U.S. equities, not adjusted for tax differences.
3.0 Summary
Dr. Morin provides only risk premium evidence in this hearing, focussing on alternative measures of the market risk premium in Canada and the U.S. and multiple beta adjustments to achieve his equity risk premium estimate of 4.6%. Setting aside the beta adjustments and the U.S.
market risk premium estimates which comprise 80% of his market risk premium evidence, Dr.
10
Morin’s estimate of a fair return on equity is remarkably consistent with our estimate.
The forecasted long Canada rate used by all witnesses in this hearing is 6.00%. At the same time, there is also little difference in the Canadian market risk premium estimates over a period in which consistent and relevant Canadian data is available, i.e., 1956-99. The range is 4.5-5.0% which already incorporates the partial integration achieved with the U.S. market. Finally, when one looks to the beta, or systematic risk, there is no evidence that the beta for a high grade utility has been, or will be, greater than .50 over the test period. Hence, the witnesses in this case are quite close in their
"bare bones" estimates -- approximately 8.25-8.50%. The major differences between estimates among the experts in this hearing are the result of the bells and whistles which the company witness adds to his actual fair return estimate.
11
ROLLING BETAS
FIRM 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
BCE INC
BCT TEL
QUEBEC TEL
NEWTEL
BRUNCOR
MARITIME TT
ISLAND TEL
MEAN TELCOS
1998 1999
0.361
0.368
0.370
0.357
0.480
0.432
0.520
0.477
0.608
0.630
0.989
1.240
0.348
0.290
0.328
0.349
0.548
0.642
0.812
0.739
0.731
0.757
0.975
0.900
0.424
0.351
0.269
0.250
0.296
0.211
0.552
0.421
0.616
0.572
0.880
0.721
0.410
0.417
0.375
0.405
0.559
0.470
0.569
0.568
0.585
0.348
0.539
0.438
0.364
0.380
0.400
0.412
0.545
0.432
0.577
0.336
0.377
0.427
0.775
0.758
0.374
0.367
0.402
0.332
0.359
0.263
0.376
0.274
0.357
0.603
0.785
0.780
0.262
0.260
0.250
0.249
0.189
0.216
0.534
0.441
0.591
0.524
0.710
0.603
0.363
0.348
0.342
0.336
0.425
0.381
0.563
0.465
0.552
0.552
0.808
0.777
MARITIME ELEC
TRANSALTA
FORTIS
CDN UTIL
BC GAS
MEAN GAS/ELEC
0.355
0.383
0.405
0.396
0.536
0.672
0.321
n/a n/a n/a n/a n/a
0.204
0.233
0.284
0.271
0.377
0.451
0.491
0.588
0.585
0.462
0.536
0.285
0.345
0.280
0.230
0.271
0.402
0.377
0.563
0.537
0.390
0.310
0.484
0.320
0.409
0.418
0.413
0.382
0.456
0.475
0.466
0.501
0.561
0.634
0.616
0.530
0.497
0.528
0.522
0.493
0.425
0.444
0.570
0.627
0.562
0.474
0.479
0.338
0.362
0.368
0.371
0.363
0.439
0.484
0.482
0.563
0.525
0.470
0.529
0.368
12
PAC N GAS
TRANSCDA P
TRANS MNT
WESTCOAST
MEAN PIPELINES
MEAN OVERALL
0.467
0.362
0.449
0.478
0.404
0.543
0.305
0.492
0.286
0.443
0.573
0.492
0.694
0.657
0.616
0.550
0.492
0.385
0.549
0.538
0.489
0.338
0.544
0.238
0.684
0.757
0.662
0.665
0.796
0.588
0.525
n/a n/a n/a n/a n/a
0.597
0.723
0.683
0.667
0.522
0.550
0.562
0.557
0.611
0.531
0.453
SCHE
DULE
B1
0.261
0.611
0.625
0.603
0.590
0.554
0.517
0.485
0.529
0.462
0.437
0.523
0.330
0.425
0.424
0.416
0.408
0.462
0.447
0.518
0.507
0.525
0.504
0.667
0.565
13
SCHEDULE B2
CANADIAN UNLEVERED UNADJUSTED UTILITY BETAS
Company
BC Gas Inc
Canadian Natural Resources
Canadian Utilities ‘B’
Fortis Inc.
Great Lakes Power Inc.
NS Holdings Power Inc.
Pacific Northern Gas Ltd.
TransAlta Corp.
TransCanada Pipe.
Westcoast Energy
Unadj. Betas
0.33
0.71
0.26
0.26
0.18
0.48
0.26
0.63
0.56
0.41
Equity Ratios
0.39
0.47
0.34
0.41
0.61
0.42
0.38
0.48
0.25
0.28
Unlev. Unadj. Betas
0.13
0.33
0.09
0.10
0.11
0.20
0.10
0.30
0.14
0.11
AVERAGE
STANDARD DEV.
0.41
0.17
0.40
0.10
0.16
0.08
Source: RAM-11
14
SCHEDULE B3
CANADIAN INSTITUTE OF ACTUARIES DATA USED TO UPDATE
HATCH AND WHITE SERIES
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
-1.43
32.55
-0.18
14.53
28.35
14.98
-1.58
31.71
COMMON
CANADA
INDEX
CANADA
LONG
BONDS
11.08
21.37
-14.80
12.02
10.45
16.29
3.34
24.43
MARKET
RISK
PREMIUM
0.64
5.08
-18.14
-12.41
13.07
22.88
-10.46
26.28
14.29
17.45
14.13
-7.15
-14.51
9.67
10.28
-11.75
14.05
-2.47
-15.72
38.86
ARITH. MEAN
GEOM. MEAN
.30
-.87
15
16
SCHEDULE B4