20190610McKenzie Exhibit 3 Schedule 2.pdf

o
o
o
2
aJ
4
5
6
7
8
9
10
11
Schedule 2
Page 1 of45
I. DESCRIPTION OF OUANTITATIVE ANALYSES
a. What is the purpose of this schedule?
A. Exhibit No. 3, Schedule 2 presents capital market estimates of the cost of
equity for the jurisdictional electric and natural gas utility operations of Avista Corp.
("Avista" or "the Company"). First, I will briefly summarize the concept of the cost of
equity, along with the risk-return tradeoff principle fundamental to capital markets. Next, I
describe my applications of the Discounted Cash Flow ("DCF"), the Capital Asset Pricing
Model ("CAPM"), the empirical form of the CAPM ("ECAPM"), a risk premium analyses
based on allowed returns for electric utilities, and reference to expected rates of refurn for
electric utilities. This exhibit also presents a market-based test to my utility quantitative
analyses by applying the DCF model to a group of low risk non-utility firms.
A. Overview
a. What fundamental economic principle underlies any evaluation of
investors'required return on equity (*ROE"X
A. The fundamental economic principle underlying the cost of equity concept is
the notion that investors are risk averse. In capital markets where relatively risk-free assets
are available (e.g., U.S. Treasury securities), investors can be induced to hold riskier assets
only if they are offered a premium, or additional retum, above the rate of retum on a
risk-free asset. Since all assets compete with each other for investor funds, riskier assets
must yield a higher expected rate of return than safer assets to induce investors to hold them.
Given this risk-return tradeofl the required rate of return (k) from an asset (i) can be
generally expressed as:
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 1 ol 45
t2
13
t4
15
t6
l7
18
t9
20
2t
22
1
2
aJ
4
5
6
7
8
9
10
11
t2
l3
t4
15
t6
t7
18
t9
20
2t
22
,/.J
24
Schedule 2
Page2 of 45
pr: ft1+Rpi
where: Ri: Risk-free rate of refurn, and
RP;: Risk premium required to hold riskier asset i.
Thus, the required rate of return for a particular asset at any point in time is a
function of: l) the yield on risk-free assets, and2) its relative risk, with investors demanding
correspondingly larger risk premiums for assets bearing gteater risk.
a. Is there evidence that the risk-return tradeoff principle actually operates
in the capital markets?
A. Yes. The risk-return tradeoff can be readily documented in segments of the
capital markets where required rates of return can be directly inferred from market data and
where generally accepted measures of risk exist. Bond yields, for example, reflect investors'
expected rates of return, and bond ratings measure the risk of individual bond issues.
Comparing the observed yields on govemment securities, which are considered free of
default risk, to the yields on bonds of various rating categories demonstrates that the risk-
refurn tradeoff does, in fact, exist.
a. Does the risk-return tradeoff observed with fixed income securities
extend to common stocks and other assets?
A. It is widely accepted that the risk-retum tradeoff evidenced with long-term
debt extends to all assets. Documenting the risk-return tradeoff for assets other than fixed
income securities, however, is complicated by two factors. First, there is no standard
measure of risk applicable to all assets. Second, for most assets - including common stock -
required rates of retum cannot be directly observed. Yet there is every reason to believe that
investors exhibit risk aversion in deciding whether or not to hold common stocks and other
assets, just as when choosing among fixed-income securities.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 2 ot 45
o
o
O
o
Schedule 2
Page 3 of45
a. Is this risk-return tradeoff limited to differences between firms?
A. No. The risk-return tradeoff principle applies not only to investments in
different firms, but also to different securities issued by the same firm. The securities issued
by a utility vary considerably in risk because they have different characteristics and
priorities. As noted earlier, long-term debt is senior among all capital in its claim on a
utility's net revenues and is, therefore, the least risky. The last investors in line are common
shareholders. They receive only the net revenues, if any, remaining after all other claimants
have been paid. As a result, the rate of refurn that investors require from a utility's common
stock, the most junior and riskiest of its securities, must be considerably higher than the
yield offered by the utility's senior, long-term debt.
a. What are the challenges in determining a just and reasonable ROE for a
regulated enterprise?
A. The actual return investors require is unobservable. Different methodologies
have been developed to estimate investors' expected and required return on capital, but all
such methodologies are merely theoretical tools and generally produce a range of estimates,
based on different assumptions and inputs. The DCF method, which is frequently
referenced and relied on by regulators, is only one theoretical approach to gain insight into
the retum investors require; there are numerous other methodologies for estimating the cost
of capital and the ranges produced by the different approaches can vary widely.
a. Is it customary to consider the results of multiple approaches when
evaluating a just and reasonable ROE?
A. Yes. In my experience, financial analysts and regulators routinely consider
the results of alternative approaches in determining allowed ROEs. It is widely recognized
that no single method can be regarded as failsafe; with all approaches having advantages and
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 3 of45
o
1
2
aJ
4
5
6
7
8
9
10
l1
t2
13
t4
15
16
t7
18
19
20
2l
22
23
24o
Schedule 2
Page 4 of 45
shortcomings. As the Federal Energy Regulatory Commission ("FERC") has noted, 'oThe
determination of rate of return on equity starts from the premise that there is no single
approach or methodology for determining the correct rate of return."l Similarly, a
publication of the Society of Utility and Regulatory Financial Analysts concluded that:
Each model requires the exercise of judgment as to the reasonableness of the
underlying assumptions of the methodology and on the reasonableness of the
proxies used to validate the theory. Each model has its own way of
examining investor behavior, its own premises, and its own set of
simplifications of reality. Each method proceeds from different fundamental
premises, most of which cannot be validated empirically. Investors clearly do
not subscribe to any singular method, nor does the stock price reflect the
application of any one single method by investors.2
o
1
2
J
4
5
6
7
8
9
10
11
t2
13 As this treatise succinctly observed, "no single model is so inherently precise that it can be
14 relied on solely to the exclusion of other theoretically sound models."3 Similarly, New
l5 Regulatory Finance concluded that:
16 There is no single model that conclusively determines or estimates the expected
17 return for an individual firm. Each methodology possesses its own way of examining
18 investor behavior, its own premises, and its own set of simplifications of reality. Each
19 method proceeds from different fundamental premises that cannot be validated empirically.
20 Investors do not necessarily subscribe to any one method, nor does the stock price reflect the
2l application of any one single method by the price-setting investor. There is no monopoly as
22 to which method is used by investors. In the absence of any hard evidence as to which
I Northwest Pipeline Co., Opinion No. 396-C, 8 1 FERC 'u 61 ,036 at 4 (1997).
2 David C. Parcell, The Cost of Capital - A Practitioner's Guide, Society of Utility and Regulatory Financial
Analysts (2010) at 84.3Id.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 4 ol 45
o
o
o
Schedule 2
Page 5 of45
method outdoes the other, all relevant evidence should be used and weighted equally, in
order to minimize judgmental error, measurement error, and conceptual infirmities.a
Thus, while the DCF model is a recognized approach to estimating the ROE, it is not
without shortcomings and does not otherwise eliminate the need to ensure that the "end
result" is fair. The Indiana Utility Regulatory Commission has recognized this principle:
There are three principal reasons for our unwillingness to place a great deal
of weight on the results of any DCF analysis. One is . . . the failure of the
DCF model to conform to reality. The second is the undeniable fact that
rarely if ever do two expert witnesses agree on the terms of a DCF equation
for the same utility-for example, as we shall see in more detail below,
projections of future dividend cash flow and anticipated price appreciation of
the stock can vary widely. And, the third reason is that the unadjusted DCF
result is almost always well below what any informed financial analysis
would regard as defensible, and therefore require an upward adjustment
based largely on the expert witness's judgment. In these circumstances, we
find it difficult to regard the results of a DCF computation as any more than
suggestive.5
a. Has FERC recently proposed to modify its practices to recognize these
realities?
A. Yes. In Opinion No. 531, FERC recognized the potential for any application
of the DCF model to produce unreliable results.6 Considering the potential for DCF results
to be distorted and in light of prevailing conditions in capital markets, FERC stated that it
had "less confidence that the IDCF model] . . accurately reflects the equity returns
a Roger A. Morin, New Regulatory Finance, Pub. Util. Reports, Inc. (2006) at429.
s Ind. Michigan Power Co., Cause No. 38728, I l6 PUR4th, l, l7-18 (IURC 8/2411990).
6 Coakleyv. Bangor Hydro-Elec. Co., OpinionNo.53l, 147 FERC nil,234 atP 4l (2014).
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 5 oi 45
1
2
J
4
5
6
7
8
9
10
1t
t2
13
t4
l5
l6
17o
18
I9
20
2t
22
,!.)
o
I
2
J
4
5
6
7
8
9
Schedule 2
Page 6 of45
necessary" to attract capital.T FERC recently announced its intention to end its practice of
relying exclusively on the DCF model,8 concluding that:
ln short, we intend to give equal weight to the results of the four financial models in
the record, instead of primarily relying on the DCF model. In relying on a broader range of
record evidence to estimate [a utility's] cost of equity, we ensure that our chosen ROE is
based on substantial evidence and bring our methodology into closer alignment with how
investors inform their investment decisions.e
The four models referenced by FERC to be used in estimating the cost of equity are
the DCF, CAPM, expected eamings, and risk premium approaches. As FERC concluded in
its Briefing Orders, the primary reason for averaging the results of the DCF, CAPM,
Expected Earnings, and Risk Premium methods to evaluate a just and reasonable ROE "is
that investors use those models, in addition to the DCF methodology, to inform their
investment decisions."l0 I submiffed testimony in Docket Nos. EL16-64-002 and
ELl5-45-000, which were both proceedings subject to the Briefing Orders, recommending
that FERC abandon sole reliance on the DCF model and give explicit consideration to the
results of the CAPM, expected earnings, and risk premium methodologies in evaluating a
just and reasonable ROE. I agree with FERC's conclusion that "providing four different
approaches to estimating the cost of equity . . . reduces the risk associated with relying on
7 Id. at P 145. FERC confirmed these findings in Opinion No. 531-B (Coakley v. Bangor Hydro-Elec. Co.,
Opinion No. 531-8, 150 FERC tT61,165 at P 84 (2015), and more recently in Opinion No.55l (Ass'n of
Businesses Advocating TariffEquity v. Midcontinent Indep. Sys. Operator,lrc., Opinion No. 551, 156 FERC fl
61,234 atP 122 (2016).
8 Coakley v. Bangor Hydro-Elec. Co., Order Directing Briefs, 165 FERC 'lT 61,030 (2018) ("Coakley Briefing
Order"); Ass'n of Businesses Advocating Tariff Equity v. Midcontinent Indep. Sys. Operator, Inc., Order
Directing Briefs, 165 FERC 1T61,1l8 (2018) C'MISO Briefing Order") (together, "Briefing Orders").
e Coakley Briefing Order at P 15.
r0 Coakley Briefing Order at P 44 (emphasis original); MISO Briefing Order at 46 (emphasis supplied).
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 6 of45
o
o
10
ll
t2
l3
t4
15
t6
t7
18
o
o
Schedule 2
PageT of45
only one model; that is, the risk of misidentifying the just and reasonable ROE by relying on
a flawed cost of equity estimate."ll
a. What does the above discussion imply with respect to estimating the cost
of common equity for a utility?
A. Although the cost of common equity cannot be observed directly, it is a
function of the retums available from other investment alternatives and the risks to which
the equity capital is exposed. Because it is unobservable, the cost of equity for a particular
utility must be estimated by analyzing information about capital market conditions
generally, assessing the relative risks of the company specifically, and employing various
quantitative methods that focus on investors'current required rates of return. These various
quantitative methods typically attempt to infer investors'required rates of return from stock
prices, interest rates, or other capital market data.
B. Comparable Risk Proxv Group
O. How did you implement quantitative methods to estimate the cost of
common equity for Avista?
A. Application of quantitative methods to estimate the cost of equity requires
observable capital market data, such as stock prices. Moreover, even for a firm with
publicly traded stock, the cost of equity can only be estimated. As a result, applying
quantitative models using observable market data produces an estimate that inherently
includes some degree of observation error. Thus, the accepted approach to increase
confidence in the results is to apply multiple quantitative methods such as the DCF and
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 7 ol 45
1
2
J
4
5
6
7
8
9
10
o
ll
t2
13
14
15
t6
t7
18
19
20
2l
o rr Coakley Briefing Order at P 38; MISO Briefing Order at P 40.
Schedule 2
Page 8 of45
CAPM to a proxy group of publicly traded utility companies that investors regard as risk-
comparable.
a. What specific proxy group of utilities did you rely on for your analyses?
A. In order to reflect the risks and prospects associated with Avista's
jurisdictional utility operations, my analyses initially applied the following criteria to
identiff a proxy group of other utilities:
l. lncluded in the Electric Utility Industry groups compiled by The
Value Line Investment Survey ("Value Line").
2. Corporate credit ratings from Standard & Poor's Corporation ("S&P")
and Moody's Investors Service ("Moody's") corresponding to one
notch above and below the Company's current ratings. For S&P, this
resulted in a ratings range of BBB-, BBB, and BBB+; for Moody's
the range was Baa3, Baa2, or Baal.
3. Value Line Safety Rank of "2" or "3".
4. No ongoing involvement in a major merger or acquisition that would
distort quantitative results.
5. No cuts in dividend payments during the past six months and no
announcement of a dividend cut since that time.
a. What other publicly traded utility is relevant in establishing a proxy
group?
A. Although it has not yet been included in Value Line's electric utility industry
groups, investors also regard Algonquin Power & Utilities, Inc. ("Algonquin") as having
operations comparable to those of other electric utilities in the proxy group. Algonquin is a
North American diversified generation, transmission, and distribution utility with
approximately $10 billion in total assets. Algonquin provides regulated utility services to
over 750,000 customers in Arizona, Arkansas, California, Georgia, Illinois, Iowa, Kansas,
Massachusetts, Missouri, New Hampshire, Oklahoma, and Texas. Algonquin completed its
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 8 of45
o
2
aJ
4
5
6
7
8
9
10
11
t2
13
t4
15
t6
t7
18
o
a
t9
20
2t
22
z)
24
25
26
27
o
Schedule 2
Page 9 of45
acquisition of Empire District Electric Company ("Empire District") on January 1,2017,
which more than doubled its size.l2 A majority of Algonquin's revenues, earnings, and
assets are related to its regulated U.S. utility operations.l3 In addition, Algonquin reports
interim and annual consolidated financial statements in U.S. dollars, its dividend is
denominated in U.S. dollars, and its common shares are listed on the New York Stock
Exchange. Investors would regard Algonquin as a comparable investment alternative that is
relevant to an evaluation of the required rate of return.
a. Is there another publicly traded utility that is relevant in developing the
proxy Group?
A. Yes. Emera, Inc. ("Emera") should also be considered in evaluating
investors'required rate of return forAvista. Headquartered in Halifax, Nova Scotia, Canada,
Emera is primarily engaged in electricity generation, transmission, and distribution; gas
transmission and distribution; and utility energy services and serves approximately 2.5
million customers. Emera completed its acquisition of TECO Energy on July l, 2016.
While Emera is currently included in Value Line's "Power Industry" sector, Value Line also
reported that as a result of the addition of TECO Energy's regulated utilities in Florida and
New Mexico, "the percentage of profits coming from regulated businesses rises to more than
900h."14
12 Empire District was included in Value Line's electric utility indushy group prior to its merger with
Algonquin.
13 For example, Algonquin reported that during 2018 regulated utility operations accounted for 85% of total
revenues, 85oh of pre-tax eamings (ex. corporate losses), and 64% of total assets. Approximately 96% of
Algonquin's consolidated revenue arrd 93o/o of assets are attributable to operations in the U.S.
https:/iwrvw.sec.eov/cgi-bin/viewer?action:view&cik:117.1 169&accession number-000 ll4036l-19-
0041 l6&xbrl t),pe:v#.
14 The Value Line Investment Survey, Emera Inc. (Mar.24,2017).
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 9 of45
I
2
J
4
5
6
7
8
o
9
t0
l1
t2
13
t4
t5
16
t7
l8
o
1
2
J
4
5
6
7
8
9
Schedule 2
Page 10 of45
Similarly, CFRA highlighted Emera's primary focus on electric utility
operations and classified Emera in its "Electric Utilities" industry group,l5 and Emera
reports as an "Electric Utility" under the Standard Industrial Classification Code (4911).16
Thus, investors would regard Emera as a comparable investment alternative that is relevant
to an evaluation of the required rate of refurn for the Company. Emera's operations are
dominated by its U.S.-based utilities in Florida, Maine, and New Mexico, which together
accounted for approximately 67oh of consolidated net income in 2018.17
Applying the criteria outlined above results in a proxy group of 22 utilities. I
refer to this set of comparable companies as the "Utility Group."
a. How did you evaluate the risks of the Utility Group relative to Avista?
A. My evaluation of relative risk considered four objective, published
benchmarks that are widely relied on in the investment community. Credit ratings are
assigned by independent rating agencies for the purpose of providing investors with a broad
assessment of the creditworthiness of a firm. Ratings generally extend from triple-A (the
highest) to D (in default). Other symbols (e.g., "BBB+") are used to show relative standing
within a category. Because the rating agencies' evaluation includes virtually all of the
factors normally considered important in assessing a firm's relative credit standing,
corporate credit ratings provide a broad, objective measure of overall investment risk that is
readily available to investors. Although the credit rating agencies are not immune to
rs CFRA, Emera Incorporated, Quantitative Stock Report (lun.24,2017). CFRA, founded as the Center for
Financial Research and Analysis, is one of the world's largest providers of institutional-grade independent
equity research, acquired the equity and fund research arm of S&P in October 2016.16 See, €.g., Emera, Inc., 2017 SEC Form 40-F.
https:ii rvu ri'. sec. eov/Archives/edeaildatai I I 2 72,18i000 I I 93 I 25 I 8 I 0 I 807/d5 5 543 8d40f .htnr.
17Emera, Inc.,20l8FinancialStatementsatNote4. WhileEmeraannouncedtheplannedsaleofitsMaine
utility operations on March 25, 2019, this transaction is small in relation to Emera's total business, with the
sale price representing approximately 4olo oftotal assets.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 't 0 of 45
o
o
o
10
ll
t2
l3
t4
15
t6
t7
18
t9
Schedule 2
Page I 1 of45o
I
2
aJ
4
5
6
7
8
9
criticism, their rankings and analyses are widely cited in the investment community and
referenced by investors. Investment restrictions tied to credit ratings continue to influence
capital flows, and credit ratings are also frequently used as a primary risk indicator in
establishing proxy groups to estimate the cost of common equity.
While credit ratings provide the most widely referenced benchmark for investment
risks, other quality rankings published by investment advisory services also provide relative
assessments of risks that are considered by investors in forming their expectations for
common stocks. Value Line's primary risk indicator is its Safety Rank, which ranges from
"1" (Safest) to "5" (Riskiest). This overall risk measure is intended to capture the total risk
of a stock, and incorporates elements of stock price stability and financial strength. Given
that Value Line is perhaps the most widely available source of investment advisory
information, its Safety Rank provides useful guidance regarding the risk perceptions of
investors.
The Financial Strength Rating is designed as a guide to overall financial strength and
creditworthiness, with the key inputs including financial leverage, business volatility
measures, and company size. Value Line's Financial Strength Ratings range from "A++"
(strongest) down to "C" (weakest) in nine steps. Finally, Value Line's beta measures a
utility's stock price volatility relative to the market as a whole. A stock that tends to respond
less to market movements has a beta less than 1.00, while stocks that tend to move more
than the market have betas greater than 1.00. Beta is the only relevant measure of
investment risk under modern capital market theory and is widely cited in academics and in
the investment industry as a guide to investors' risk perceptions. Moreover, in my
10
11ot2
13
t4
15
t6
t7
18
19
20
2t
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 11 ot 45
o
22
Schedule 2
Page 12 of 45
experience Value Line is the most widely referenced source for beta in regulatory
proceedings. As noted in New Regulatory Finance:
Value Line is the largest and most widely circulated independent investment
advisory service, and influences the expectations of a large number of
institutional and individual investors. . . .Value Line betas are computed on a
theoretically sound basis using a broadly based market index, and they are
adjusted for the regression tendency of betas to converge to 1.00. 18
How do the overall risks of your proxy group compare with Avista?
Table I compares the Utility Group with Avista across four key indicators of
10 investment risk:
o
1
2
aJ
4
5
6
7
8
9
a.
A.
11
t2
t3
I4
15
16
t7
18
t9
20
2I
TABLE 1
COMPARISON OF RISK INDICATORS
Value Line
Credit Rating
S&P
BBB+
BBB
Moodv's
Baa2
Baa2
Safety
Rank
2
2
Flnancial
Strensth
Br-r
A
Beta
0.64
0.65
oUtility Grorp
Avista
a. What does this comparison indicate regarding investors'assessment of
the relative risk associated with your Utility Group?
A. As shown above, Avista's BBB rating from S&P indicates slightly greater
risk that the average for the proxy group, while its Baa2 Moody's rating is identical to the
average for the Utility Group. Similarly, the average Value Line Safety Rank for the Utility
Group is the same as that assigned to the Company. With respect to Value Line's Financial
Strength rating, the average value for the Utility Group indicates slightly more risk than for
Avista. Meanwhile, Avista's beta value is virtually identical to the average for the proxy
group. Considered together, this comparison of objective measures, which consider a broad
18 Roger A. Morin, New Regulatory Finance, Pub. Util. Reports, lnc. (2006) at7l
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule2, Page 12o145
a
o
Schedule 2
Page 13 of45
spectrum of risks, including financial and business position, and exposure to firm-specific
factors, indicates that investors would likely conclude that the overall investment risks for
Avista are generally comparable to those of the firms in the Utility Group.
C. Discounted Cash Flow Analyses
O. How are DCF models used to estimate the cost of equity?
A. DCF models attempt to replicate the market valuation process that sets the
price investors are willing to pay for a share of a company's stock. The model rests on the
assumption that investors evaluate the risks and expected rates of return from all securities
in the capital markets. Given these expectations, the price of each stock is adjusted by the
market until investors are adequately compensated for the risks they bear. Therefore, we can
look to the market to determine what investors believe a share of common stock is worth.
By estimating the cash flows investors expect to receive from the stock in the way of future
dividends and capital gains, we can calculate their required rate of return. That is, the cost
of equity is the discount rate that equates the current price of a share of stock with the
present value of all expected cash flows from the stock. The formula for the general form of
the DCF model is as follows:
1
2
J
4
5
6
1
8
9
10
o l1
12
t3
t4
15
16
t7
Po
Dl- 0l7J
D" D. P.t!t-trtr(l+k")' (l+l-,)' (l+k")'
18
t9
20
2t
where Po : Current price per share;
Pt: Expected future price per share in period t;
D1 : Expected dividend per share in period t;
k": Cost of common equity.
o
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 1 3 of 45
Schedule 2
Page 14 of 45
a. What form of the DCF model is customarily used to estimate the cost of
equity in rate cases?
A. Rather than developing annual estimates of cash flows into perpetuity, the
DCF model can be simplified to a "constant growth" form:le
Po
Dl
k"-g
where Po: Current price per share;
D1 : Expected dividend per share in the coming year;
k": Cost of equity;
g : Investors'long-term growth expectations.
The cost of equity (K") can be isolated by rearranging terms:
D,k"=4+g
This constant growth form of the DCF model recognizes that the rate of return to
stockholders consists of two parts: 1) dividend yield (Dr/Po), and 2) growth (g). In other
words, investors expect to receive a portion of their total return in the form of current
dividends and the remainder through price appreciation.
a. What steps are required to apply the DCF model?
A. The first step in implementing the constant growth DCF model is to
determine the expected dividend yield (Dr/Po) for the firm in question. This is usually
calculated based on an estimate of dividends to be paid in the coming year divided by the
current price of the stock. The second step is to estimate investors' long-term growth
re The constant growth DCF modet is dependent on a number of assumptions, which in practice are never
strictly met. These include a constant growth rate for both dividends and earnings; a stable dividend payout
ratio; the discount rate exceeds the growth rate; a constant growth rate for book value and price; a constant
earned rate ofreturn on book value; no sales ofstock at a price above or below book value; a constant price-
earnings ratio; a constant discount rate (i.e., no changes in risk or interest rate levels and a flat yield curve); and
all of the above extend to infinity.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 14 of 45
o
1
2
aJ
4
5
6
7
8
9
t0
11
12
l3
t4
l5
t6
t7
l8
t9
o
o
20
o
Schedule 2
Page l5 of45
expectations (g) for the firm. The final step is to sum the firm's dividend yield and estimated
growth rate to arrive at an estimate of its cost of equity.
a. How was the dividend yield for the Utility Group determined?
A. Estimates of dividends to be paid by each of these utilities over the next
twelve months, obtained from Value Line, served as Dr. This annual dividend was then
divided by a 30-day average stock price for each utility to arrive at the expected dividend
yield. The expected dividends, stock prices, and resulting dividend yields for the firms in
the Utility Group are presented on page 1 of Exhibit No. 3, Schedule 5.
a. What is the next step in applying the constant growth DCF model?
A. The next step is to evaluate long-term growth expectations, or '0g", for the
firm in question. In constant growth DCF theory earnings, dividends, book value, and
market price are all assumed to grow in lockstep, and the growth horizon of the DCF model
is infinite. But implementation of the DCF model is more than just a theoretical exercise; it
is an attempt to replicate the mechanism investors used to arrive at observable stock prices.
A wide variety of techniques can be used to derive growth rates, but the only "gi' that
matters in applying the DCF model is the value that investors expect.
a. What are investors most likely to consider in developing their long-term
growth expectations?
A. Implementation of the DCF model is solely concerned with replicating the
forward-looking evaluation of real-world investors. In the case of utilities, dividend growth
rates are not likely to provide a meaningful guide to investors' current growth expectations.
This is because utilities have significantly altered their dividend policies in response to more
accentuated business risks in the industry, with the payout ratios falling significantly from
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 1 5 of 45
o
1
2
J
4
5
6
7
8
9
10
1l
t2
13
t4
15
t6
t7
18
19
20
2t
22
ZJo
Schedule 2
Page 16 of45
historical levels. As a result, dividend growth in the utility industry has lagged growth in
earnings as utilities conserve financial resources to provide a hedge against heightened
uncertainties.
A measure that plays a pivotal role in determining investors' long-term growth
expectations are fufure trends in earnings per share ("EPS"), which provide the source for
future dividends and ultimately support share prices. The importance of earnings in
evaluating investors' expectations and requirements is well accepted in the investment
community, and surveys of analytical techniques relied on by professional analysts indicate
that growth in earnings is far more influential than trends in dividends per share ("DPS").
The availability of projected EPS growth rates also is key to investors relying on this
measure as compared to future trends in DPS. Apart from Value Line, investment advisory
services do not generally publish comprehensive DPS growth projections, and this scarcity
of dividend growth rates relative to the abundance of earnings forecasts attests to their
relative influence. The fact that securities analysts focus on EPS growth, and that DPS
growth rates are not routinely published, indicates that projected EPS growth rates are likely
to provide a superior indicator of the future long-term growth expected by investors.
a. Do the growth rate projections of security analysts consider historical
trends?
A. Yes. Professional security analysts study historical trends extensively in
developing their projections of future earnings. Hence, to the extent there is any useful
information in historical patterns, that information is incorporated into analysts' growth
forecasts.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 1 6 of 45
o
1
2
aJ
4
5
6
7
8
9
10
11
t2
13
t4
l5
t6
t7
18
t9
20
2l
22
o
o
o
Schedule 2
PagelT of45
a. Did Professor Myron J. Gordon, who pioneered the DCF approach,
recognize the pivotal role that earnings play in forming investors'expectations?
A. Yes. Dr. Gordon specifically recognized that "it is the growth that investors
expect that should be used" in applying the DCF model and he concluded:
A number of considerations suggest that investors may, in fact, use earnings
growth as a measure of expected future growth.2o
a. Are analysts' assessments of growth rates appropriate for estimating
investors'required return using the DCF model?
A. Yes. In applying the DCF model to estimate the cost of common equity, the
only relevant growth rate is the forward-looking expectations of investors that are captured
in current stock prices. lnvestors, just like securities analysts and others in the investment
community, do not know how the future will actually turn out. They can only make
investment decisions based on their best estimate of what the future holds in the way of
long-term growth for a particular stock, and securities prices are constantly adjusting to
reflect their assessment of available information.
Any claims that analysts' estimates are not relied upon by investors are illogical
given the reality of a competitive market for investment advice. The market for investment
advice is intensely competitive, and securities analysts are personally and professionally
motivated to provide the most accurate assessment possible of future growth trends. If
financial analysts' forecasts do not add value to investors' decision making, then it is
irrational for investors to pay for these estimates. Those financial analysts who fail to
provide reliable forecasts will lose out in competitive markets relative to those analysts
whose forecasts investors find more credible. The reality that analyst estimates are routinely
20 Myron J. Gordon, The Cost of Capital to a Public Utility, MSU Pub. Util. Studies (1974) at89.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 17 of 45
2
J
4
5
6
1
8
9
o
10
ll
t2
13
t4
15
16
t7
t8
t9
20
2t
22
23
o
1
2
J
4
5
6
7
8
9
Schedule 2
Page 18 of45
referenced in the financial media and in investment advisory publications (e.9., Value Line)
implies that investors use them as a basis for their expectations.
While the projections of securities analysts may be proven optimistic or pessimistic
in hindsight, this is irrelevant in assessing the expected growth that investors have
incorporated into current stock prices, and any bias in analysts' forecasts - whether
pessimistic or optimistic - is irrelevant if investors share analysts' views. Earnings growth
projections of security analysts provide the most frequently referenced guide to investors'
views and are widely accepted in applying the DCF model. As explained in New Regulatory
Finance:
Because of the dominance of institutional investors and their influence on
individual investors, analysts' forecasts of long-run growth rates provide a
sound basis for estimating required returns. Financial analysts exert a strong
influence on the expectations of many investors who do not possess the
resources to make their own forecasts, that is, they are a cause of g [growth].
The accuracy of these forecasts in the sense of whether they turn out to be
qqrregt is not an issue here-as lqqg as fpy reflect widely held .21
a. Have regulators also recognized that analysts'growth rate estimates are
an important and meaningful guide to investors'expectations?
A. Yes. The Kentucky Public Service Commission has indicated its preference
for relying on analysts'projections in establishing investors' expectations:
2r Roger A. Morin, New Regulatory Finance, Pub. Util. Reports, Inc. (2006) at 298 (emphasis added).
22 Kentucley Lltilities Co., Case No. 2009-00548 (Ky PSC Jul. 30, 2010) at 30-3 I .
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 18 of 45
o
o
o
l0
1l
t2
13
t4
15
16
t7
18
t9
20
2t
22
23
24
25
26
27
KU's argument concerning the appropriateness of using investors'
expectations in performing a DCF analysis is more persuasive than the AG's
argument that analysts' projections should be rejected in favor of historical
results. The Commission agrees that analysts'projections of glowth will be
relatively more compelling in forming investors' forward-looking
expectations than relying on historical performance, especially given the
current state of the economy.22
o
Schedule 2
Page 19 of45
Similarly, FERC has expressed a clear preference for projected EPS growth rates in
applying the DCF model to estimate the cost of equity for both electric and natural gas
pipeline utilities:
Opinion No. 414-4 held that the IBES five-year growth forecasts for each
company in the proxy group are the best available evidence of the short-term
growth rates expected by the investment community. It cited evidence that (l)
those forecasts are provided to IBES by professional security analysts, (2)
IBES reports the forecast for each firm as a service to investors, and (3) the
IBES reports are well known in the investment community and used by
investors. The Commission has also rejected the suggestion that the IBES
analysts are biased and stated that "in fact the analysts have a significant
incentive to make their analyses as accurate as possible to meet the needs of
their clients since those investors will not utilize brokerage firms whose
analysts repeatedly overstate the growth potential of companies."23
The Public Utility Regulatory Authority of Connecticut has also noted that "there is
not growth in DPS without growth in EPS," and concluded that securities analysts'growth
projections have a greater influence over investors' expectations and stock prices.2a In
addition, the Regulatory Commission of Alaska ("RCA") has previously determined that
analysts' EPS growth rates provide a superior basis on which to estimate investors'
expectations:
We also find persuasive the testimony . . . that projected EPS returns are more
indicative of investor expectations of dividend growth than historical growth
data because persons making the forecasts already consider the historical
numbers in their analyses.25
The RCA has concluded that arguments against exclusive reliance on analysts' EPS groMh
rates to apply the DCF model "are not convincing."26
23 Kern River Gas Transmission Co.,126 FERC tT 6l,034atP 121 (2009) (footnote omitted).
2a Deciston, Docket No. 13-02-20 ($ept.24,2013).
25 Regulatory Commission of Alaska, U-07-76(8) at 65, n. 258.
26 Regulatory Commission of Alaska, U-08-157(10) at 36.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 19 of45
2
aJ
4
5
6
7
8
9
10
1l
t2
l3
t4
o
l5
l6
t7
18
t9
20
2t
22
23
24
25
26
o
1
2
J
4
5
Schedule 2
Page20 of45
a. What are security analysts currently projecting in the way of growth for
the firms in the Utitity Proxy Group?
A. The projected EPS growth rates for each of the firms in the Utility Group
reported by Value Line, IBES ,27 and Zacks Investment Research ("Zacks") are displayed on
page2 of ExhibitNo.3, Schedule 5.
a. How else are investors' expectations of future long-term growth
prospects often estimated for use in the constant growth DCF model?
A. In constant growth theory growth in book equity will be equal to the product
of the eamings retention ratio (one minus the dividend payout ratio) and the earned rate of
return on book equity. Furthermore, if the earned rate of refurn and the payout ratio are
constant over time, growth in eamings and dividends will be equal to growth in book value.
Despite the fact that these conditions are seldom, if ever, met in practice, this "sustainable
growth" approach may provide a rough guide for evaluating a firm's growth prospects and is
frequently proposed in regulatory proceedings.
The sustainable growth rate is calculated by the formula, g: br*sv, where "b" is the
expected retention ratio, 'or" is the expected eamed refurn on equity, o's" is the percent of
common equity expected to be issued annually as new common stock, and "v" is the equity
accretion rate. Under DCF theory the "sv" factor is a component of the growth rate
designed to capture the impact of issuing new common stock at a price above, or below,
book value. The sustainable, "br*sv" growth rates for each firm in the Utility Group are
27 Formerly yBlE/S International, Inc., IBES growth rates are now compiled and published by Thomson
Reuters.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 20 of 45
o
o
6
7
8
9
l0
l1
t2
13
t4
l5
16
17
18
t9
20
o
o
Schedule 2
Page 2l of 45
summarized on page 2 of Exhibit No. 3, Schedule 5, with the underlying details being
presented on Exhibit No. 3, Schedule 6.28
a. Are there signilicant shortcomings associated with the'6br*sv" growth
rate?
A. Yes. First, in order to calculate the sustainable growth rate, it is necessary to
develop estimates of investors' expectations for four separate variables; namely, "b", "r",
"s", and "v." Given the inherent difficulty in forecasting each parameter and the difficulty of
estimating the expectations of investors, the potential for measurement error is significantly
increased when using four variables, as opposed to referencing a direct projection for EPS
groMh. Second, empirical research in the finance literature indicates that sustainable
growth rates are not as significantly correlated to measures of value, such as share prices, as
are analysts' EPS growth forecasts.2e The "sustainable growth" approach was included for
completeness, but evidence indicates that analysts' forecasts provide a superior and more
direct guide to investors' growth expectations.
a. What cost of equity estimates were implied for the Utility Group using
the DCF model?
A. After combining the dividend yields and respective growth projections for
each utility, the resulting cost of equity estimates are shown on page 3 of Exhibit No. 3,
Schedule 5.
28 Because Value Line reports end-of-year book values, an adjustment factor was incorporated to compute an
average rate ofreturn over the year, which is consistent with the theory underlying this approach.
2e Roger A. Morin, New Regulatory Finance, Pub. Util. Reports, Inc., (2006) at307.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 21 ol 45
1
2
J
4
5
6
7
8
9
10
o
o
11
t2
13
t4
15
T6
17
18
t9
I
2
J
4
5
6
7
8
9
10
11
t2
13
t4
15
I6
17
18
t9
20
Schedule 2
Page22 of45
a. In evaluating the results of the constant growth DCF model, is it
appropriate to eliminate illogical low or high-end values?
A. Yes. In applying quantitative methods to estimate the cost of equity, it is
essential that the resulting values pass fundamental tests of reasonableness and economic
logic. Accordingly, DCF estimates that are implausibly low or high should be eliminated
when evaluating the results of this method.
a. How did you evaluate DCF estimates at the low end of the range?
A. I based my evaluation of DCF estimates at the low end of the range on the
fundamental risk-return tradeoff, which holds that investors will only take on more risk if
they expect to earn a refurn to compensate them for the greater uncertainty. Because
common stocks lack the protections associated with an investment in long-term bonds, a
utility's common stock imposes far greater risks on investors. As a result, the rate of return
that investors require from a utility's common stock is considerably higher than the yield
offered by senior, long-term debt. Consistent with this principle, DCF results that are not
sufficiently higher than the yields available on less risky utility bonds must be eliminated.
a. Have similar tests been applied by regulators?
A. Yes. FERC has noted that adjustments are justified where applications of the
DCF approach produce illogical results. FERC evaluates DCF results against observable
yields on long-term public utility debt and has recognizedthat it is appropriate to eliminate
estimates that do not sufficiently exceed this threshold.3o FERC affirmed that:
The purpose of the low-end outlier test is to exclude from the proxy group
those companies whose ROE estimates are below the average bond yield or
are above the average bond yield but are sufficiently low that an investor
would consider the stock to yield essentially the same return as debt. In
30 See, e.g., Southern Califurnia Edison Co.,l3l FERC fl 61,020 at P 55 (2010) ("Socal Edison").
Exhibit No. 3
Case No. AVU-E-'19-04
A. McKenzie, Avista
Schedule2, Page22ol45
o
O
o
2l
22
23
24
o
Schedule 2
Page23 of45
public utility ROE cases, the Commission has used 100 basis points above
the cost of debt as an approximation of this threshold, but has also considered
the distribution of proxy group companies to inform its decision on which
companies are outliers. As the Presiding Judge explained, this is a flexible
test.3 1
a. What interest rate benchmark did you consider in evaluating the DCF
results for Avista?
A. As noted earlier, the S&P and Moody's ratings forAvista are BBB andBaa2,
respectively, which fall in the triple-B rating category. Accordingly, I referenced average
yields on triple-B utility bonds as my benchmark in evaluating low-end results. Monthly
yields on Baa bonds reported by Moody's averaged 4.81 percent over the six months ending
April2019.
O. What else should be considered in evaluating DCF estimates at the low
end ofthe range?
A. As indicated earlier, it is generally expected that long-term interest rates will
rise as the Federal Reserve normalizes monetary policies. As shown in Table 2 below,
forecasts of IHS Global Insight and the Energy Information Administration ("EIA") imply
an average triple-B bond yield of approximately 5.6 percent over the period 2019-2023:
3r Martha Coakley et al., v. Bangor Hydro-Electric Company, et al., Opinion No. 53 1, 147 FERC \61,234 atP
t22 (2014).
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 23 of 45
o
1
2
J
4
5
6
7
8
9
10
11
12
13
t4
15
16
17
18
O
Schedule 2
Page24 of45
I
2
TABLE 2
IMPLIED BAA BOND YIELD
Baa Yield
2019-23
433%
5.55%
o
o
Projected Aa Utility Yield
IHS Global Insight(a)
ErA(b)
Average 4.94%
J
4
5
6
Current Baa-Aa Yield Spread(c)0.68%
Implied Baa Utility Yield 5.620/0
(a) IHS Global InsiCht, Long-Term Macro Forecast - Baseline (Apr.30, 2019).
(b) Energy Information Administration, Annual Enerry Outlook 20 I 9 (I an. 24,
2019).
(c) Based on monthly average bond yields from Moody's Investors Service for
the six-month period Nov. 201 8 - Apr. 201 9.
The increase in debt yields anticipated by IHS Global Insight and EIA is also
supported by the widely-referenced Blue Chip Financial Forecasts ("Blue Chip"), which
projects that yields on corporate bonds will climb on the order of 100 basis points through
2023.32
a. What does this test of logic imply with respect to the DCF estimates for
the Utitity Group?
A. Adding a 100 basis-point premium to the historical and projected average
utility bond yields implies a low-end threshold on the order of 5.8 percent to 6.6 percent. As
highlighted on page 3 of Exhibit No. 3, Schedule 5, after considering these tests and the
distribution of individual estimates, I eliminated low-end DCF estimates rangingfrom -2.9
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 24 oi 45
7
8
9
l0
11
o
t2
32 Wolters Kluwer, Blue Chip Financial Forecasts (Dec. 1,2018)
o
Schedule 2
Page25 of45
percent to 6.6 percent. Based on my professional experience and the risk-return tradeoff
principle that is fundamental to finance, it is inconceivable that investors are not requiring a
substantially higher rate of return for holding common stock. As a result, consistent with the
threshold established by historical and projected utility bond yields, these values provide
little guidance as to the returns investors require from utility common stocks and should be
excluded.
O. What else should be considered in evaluating DCF estimates at the low
end ofthe range?
A. While FERC has historically relied on a 100 basis point spread over public
utility bond yields as a starting place in evaluating low-end values, reference to a static test
ignores the implications of current low bond yields. Specifically, the premium that investors
demand to bear the higher risks of common stock is not constant. As I demonstrate later in
my testimony, equity risk premiums expand when interest rates fall, and vice versa. Given
that bond yields have remained uncharacteristically low, this inverse relationship implies a
significant increase in the equity risk premium that investors require to accept the higher
uncertainties associated with an investment in utility common stocks versus bonds. As a
result, using a fixed premium of 100 basis points over public utility bond yields will vastly
understate the threshold for investors'minimum required return on utility stocks.
a. Do you also recommend excluding estimates at the high end of the range
of DCF results?
A. I typically recommend the exclusion of high end estimates that are clearly
implausible. In this case, I eliminated a high-end DCF estimate of 16.3 percent. When
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 25 of 45
I
2
aJ
4
5
6
7
8
9
l0
o
11
t2
13
t4
15
16
t7
18
t9
20
2t
22
O
Schedule 2
Page26 of45
compared with the balance of the remaining estimates, this value is unreasonable and should
be excluded in evaluating the results of the DCF model for the Utility Group.
Beyond this, the upper end of the DCF results is set by a cost of equity estimate of
15.8 percent. While a 15.8 percent cost of equity estimate may exceed the majority of the
remaining values, low-end DCF estimates in the 7.0o/o range are assuredly far below
investors' required rate of return. Taken together and considered along with the balance of
the results, the remaining values provide a reasonable basis on which to frame the range of
plausible DCF estimates and evaluate investors'required rate of return.
a. What cost of equity is implied by your DCF results for the Utility Group?
A. As shown on page 3 of Exhibit No. 3, Schedule 5 and summarized in Table 3,
below, after eliminating illogical low-end values, application of the constant growth DCF
model resulted in the following cost of equity estimates:
TABLE 3
DCF RESULTS _ UTILITY GROUP
o
1
2
J
4
5
6
7
8
9
10
11
t2
13
I4
15
t6
t7
18
t9
o
Average
t0.0%
r0.0%
9.2%
8.9%
Midpoint
It.4%
tt.5%
t0.0%
r0.2%
20
D. Capital Asset Pricine Model
a. Please describe the CAPM.
A. The CAPM is a theory of market equilibrium that measures risk using the
beta coefficient. Assuming investors are fully diversified, the relevant risk of an individual
asset (e.g., common stock) is its volatility relative to the market as a whole, with beta
reflecting the tendency of a stock's price to follow changes in the market. A stock that tends
Exhibit No. 3
Case No. AVU-E-'19-04
A. McKenzie, Avista
Schedule 2,Page 26 of45
o
Growth Rate
Value Line
IBES
Tacl*s
br*sv
o
Schedule 2
Page27 of45
to respond less to market movements has a beta less than 1.00, while stocks that tend to
move more than the market have betas greater than 1.00. The CAPM is mathematically
expressed as:
&:Rr+ft1R.-Rr)
where \ : required rate of return for stock j;
Rr: risk-free rate;
R-: expected retum on the market portfolio;
and,
p1 : beta, or systematic risk, for stock j.
Like the DCF model, the CAPM is an ex-ante, or forward-looking model based on
expectations of the fufure. As a result, in order to produce a meaningful estimate of
investors'required rate of return, the CAPM must be applied using estimates that reflect the
expectations of actual investors in the market, not with backward-looking, historical data.
a. Why is the CAPM approach an appropriate component of evaluating the
cost of equity for Avista?
A. The CAPM approach (which also forms the foundation of the ECAPM)
generally is considered to be the most widely referenced method for estimating the cost of
equity among academicians and professional practitioners, with the pioneering researchers
of this method receiving the Nobel Prize in 1990. Because this is the dominant model for
estimating the cost of equity outside the regulatory sphere, the CAPM provides important
insight into investors' required rate of return for utility stocks, including Avista.
a. How did you apply the CAPM to estimate the cost of common equity?
A. Application of the CAPM to the Utility Group based on a forward-looking
estimate for investors' required rate of return from common stocks is presented on Exhibit
No. 3, Schedule 7. In order to capture the expectations of today's investors in current capital
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 27 ot 45
1
2
J
4
5
6
7
8
9
o
10
l1
t2
l3
t4
15
l6
t7
t8
t9
20
2t
22
23
24
25o
1
2
J
4
5
6
7
8
9
10
11
t2
13
t4
15
16
l7
r8
t9
20
Schedule 2
Page28 of 45
markets, the expected market rate of return was estimated by conducting a DCF analysis on
the dividend paying firms in the S&P 500.
The dividend yield for each firm was obtained from Zacks, and the growth rate was
equal to the average of the earnings growth projections for each firm published by Value
Line, IBES, and Zacks with each firm's dividend yield and growth rate being weighted by
its proportionate share of total market value. Based on the weighted average of the
projections for the individual firms, current estimates imply an average growth rate over the
next five years of 10.7 percent. Combining this averuEe growth rate with a year-ahead
dividend yield of 2.5 percent results in a current cost of common equity estimate for the
market as a whole (R') of 13.2 percent. Subtracting a 3.1 percent risk-free rate based on the
average yield on 30-year Treasury bonds for the six months ending April2019 produced a
market equity risk premium of 10.1 percent.
a. What was the source of the beta values you used to apply the CAPM?
A. As I did in the development of my proxy group discussed above, I relied on
the beta values reported by Value Line, which in my experience is the most widely
referenced source for beta in regulatory proceedings.
0. What else should be considered in applying the CAPM?
A. Financial research indicates that the CAPM does not fully account for
observed differences in rates of return attributable to firm size. Accordingly, a modification
is required to account for this size effect. As explained by Momingstar:
One of the most remarkable discoveries of modern finance is the finding of a
relationship between firm size and return. On average, small companies have
higher returns than larger ones. . . . The relationship between firm size and
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 28 ol 45
o
o
2t
22
23
o
o
Schedule 2
Page29 of45
refurn cuts across the entire size spectrum; it is not restricted to the smallest
stocks.33
According to the CAPM, the expected return on a security should consist of the
riskless rate, plus a premium to compensate for the systematic risk of the particular security.
The degree of systematic risk is represented by the beta coefficient. The need for the size
adjustment arises because differences in investors' required rates of refurn that are related to
firm size are not fully captured by beta. To account for this, researchers have developed size
premiums that need to be added to CAPM cost of equity estimates to account for the level of
a firm's market capitalization in determining the CAPM cost of equity.34 Accordingly, my
CAPM analyses incorporated an adjustment to recognize the impact of size distinctions, as
measured by the average market capitalization for the firms in the Utility Group.
a. Is this size adjustment related to the relative size of Avista as compared
with the proxy group?
A. No. I am not proposing to apply a general size risk premium in evaluating a
fair and reasonable ROE for the Company and my recommendation does not include any
adjustment related to the relative size of Avista. Rather, this size adjustment is specific to
the CAPM and merely corrects for an observed inability of the beta measure to fully reflect
the risks perceived by investors for the firms in the proxy groups. As FERC has recognized,
"This type of size adjustment is a generally accepted approach to CAPM analyses."3s
33 Momingstar, 2015 lbbotson SBBI Classic Yearbook atp.99 (footnote omitted).
3a Originally compiled by lbbotson Associates and published in their annual yearbook entitled, Stocks, Bonds,
Bills and Inflation, these size premia are now developed by Duff & Phelps and presented in its l/aluation
Handbook- Guide to Cost of Capital.
35 Opinion No. 531-8, 150 FERC'1T61,165 at P I l7 (2015).
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 29 of 45
o
1
2
aJ
4
5
6
7
8
9
10
l1
t2
13
t4
15
t6
t7
18
t9
o
2
J
4
5
Schedule 2
Page 30 of45
a. What cost of equity is indicated for the Utility Group using the CAPM
approach?
A. As shown on page I of Exhibit No. 3, Schedule 7, after adjusting for the
impact of firm size the CAPM approach implied average and midpoint cost of equity
estimates of 10.4 percent and 10.1 percent, respectively, for the Utility Group.
a. Did you also apply the CAPM using forecasted bond yields?
A. Yes. As discussed earlier, there is widespread consensus that interest rates
will increase over the period when the rates established in this proceeding will be in effect.
Accordingly, in addition to the use of current bond yields, I also applied the CAPM based on
the forecasted long-term Treasury bond yields developed based on projections published by
Value Line, IHS Global Insight and Blue Chip. As shown on page 2 of Exhibit No. 3,
Schedule 7, incorporating a forecasted Treasury bond yield for 2019-2023 implied an
average cost of equity of 10.5 percent for the Utility Group, with a midpoint of 10.3 percent.
E. Empirical Capital Asset Pricine Model
a. How does the ECAPM approach differ from traditional applications of
the CAPM?
A. Empirical tests of the CAPM have shown that low-beta securities earn returns
somewhat higher than the CAPM would predict, and high-beta securities earn less than
predicted. In other words, the CAPM tends to overstate the actual sensitivity of the cost
of capital to beta, with low-beta stocks tending to have higher returns and high-beta
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 30 of 45
O
o
o
6
7
8
9
10
ll
t2
13
t4
15
t6
t7
18
19
20
o
Schedule 2
Page 31 of45
stocks tending to have lower risk returns than predicted by the CAPM.36 This is
illustrated graphically in the figure below:
FIGURE 1
CAPM _ PREDICTED VS. OBSERVED RETURI\S
Retum
Predicted
o
oo o o o o Observed
Lowbeta assets High beta assets
0 1.0 Beta
Because the betas of utility stocks, including those in the Utility Group, are generally
less than 1.0, this implies that cost of equity estimates based on the traditional CAPM would
understate the cost of equity. This empirical finding is widely reported in the finance
literafure, as summarized in New Regulatory Finance:
As discussed in the previous section, several finance scholars have developed
refined and expanded versions of the standard CAPM by relaxing the
constraints imposed on the CAPM, such as dividend yield, size, and
skewness effects. These enhanced CAPMs typically produce a risk-return
relationship that is flatter than the CAPM prediction in keeping with the
actual observed risk-return relationship. The ECAPM makes use of these
empirical relationships. 3 7
36 Because the betas of utility stocks, including Avista, are generally less than 1.0, this implies that cost of
equity estimates based on the traditional CAPM would understate the cost of equity.
37 Roger A. Morin, New Regulatory Finance, Pub. Util. Reports, Inc. (2006) at 189.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 31 ot 45
1
2
J
4
o
o
O9
Rr
o 5
6
7
8
9
10
11
t2
13
t4
l5
9O o
o
1
2
J
4
5
6
7
8
9
10
11
t2
13
l4
t5
I6
t7
l8
t9
20
2t
Schedule 2
Page32 of45
As discussed in New Regulatory Finance, based on a review of the empirical
evidence, the expected return on a security is related to its risk by the ECAPM, which is
represented by the following formula:
&: Rr+ 0.25(R- - Rr) + 0.75[B;@. - Rr)]
Like the CAPM formula presented earlier, the ECAPM represents a stock's required return
as a function of the risk-free rate (Rr), plus a risk premium. ln the formula above, this risk
premium is composed of two parts: (1) the market risk premium (R- - Rr) weighted by a
factor of 25o/o, and (2) a company-specific risk premium based on the stocks relative
volatility [(F)G,n - Rr)] weighted by 75%. This ECAPM equation, and its associated
weighting factors, recognizes the observed relationship between standard CAPM estimates
and the cost of capital documented in the financial research, and corrects for the understated
retums that would otherwise be produced for low beta stocks.
a. Is the use of the ECAPM consistent with the use of Value Line betas?
A. Yes. Value Line beta values are adjusted for the observed tendency of beta to
converge toward the mean value of 1.00 over time.38 The purpose of this adjustment is to
refine beta values determined using historical data to better match forward-looking estimates
of beta, which are the relevant parameter in applying the CAPM or ECAPM models.
Meanwhile, the ECAPM does not involve any adjustment to beta whatsoever. Rather, it
represents a formal recognition of findings in the financial literature that the observed risk-
return tradeoffillustrated in Figure 1 is flatter than predicted by the CAPM. In other words,
even if a firm's beta value were estimated with perfect precision, the CAPM would still
38 See, e.g., Marshall E. Blume, Betas and Their Regression Tendencies, Journal of Finance (Jun. 1975), pp.
185-195.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 32 oI 45
o
o
o
o
Schedule 2
Page 33 of45
understate the return for low-beta stocks and overstate the return for high-beta stocks. The
ECAPM and the use of adjusted betas represent two separate and distinct issues in
estimating returns.
a. Have other regulators relied on the ECAPM?
A. Yes. The ECAPM approach has been relied on by the Staff of the Maryland
Public Service Commission. For example, Staff Witness Julie McKenna noted that "the
ECAPM model adjusts for the tendency of the CAPM model to underestimate retums for
low Beta stocks," and concluded that, "I believe under current economic conditions that the
ECAPM gives a more realistic measure of the ROE than the CAPM model does."3e The
staffof the Colorado commission has recognizedthat, "The ECAPM is an empirical method
that attempts to enhance the CAPM analysis by flattening the risk-return relationship,"ao and
relied on the exact same standard ECAPM equation presented above.al The Regulatory
Commission ofAlaska has also relied on the ECAPM approach, noting that:
Tesoro averaged the results it obtained from CAPM and ECAPM while at the
same time providing empirical testimony that the ECAPM results are more
accurate then [sic] haditional CAPM results. The reasonable investor would
be aware of these empirical results. Therefore, we adjust Tesoro's
recommendation to reflect only the ECAPM result.a2
The Wyoming OfIice of Consumer Advocate, an independent division of the
Wyoming Public Service Commission, has also relied on this same ECAPM formula in
estimating the cost of equity for a natural gas utiliry as have witnesses for the Office of
3e Direct Testimony and Exhibits of Julie McKenna, Maryland PSC Case No. 9299 (Oct. 12, 2012) at9.
a0 Proceeding No. l3AL-0067G, Answer Testimony and Schedules of Scott England (July 31,2013) at 47
4t Id. at 48.
a2 Regulatory Commission of Alaska, OrderNo. P-97-004(l5l) (Nov. 27,2002) at145.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 33 of 45
2
J
4
5
6
7
8
9
l0
o
o
l1
t2
13
l4
15
t6
l7
18
19
2t
20
1
2
aJ
4
Schedule 2
Page 34 of 45
Arkansas Attorney General.a3 More recently, the Montana Public Service Commission
determined that "[t]he evidence in this proceeding has convinced the Commission that the
Empirical Capital Asset Pricing Model ("ECAPM") should be the primary method for
estimating . . . the cost of equi[/" for a gas distribution utility under its jurisdiction.aa
a. What cost of equity estimates were indicated by the ECAPM?
A. My applications of the ECAPM were based on the same forward-looking
market rate of return, risk-free rates, and beta values discussed earlier in connections with
the CAPM. As shown on page I of Exhibit No. 3, Schedule 8, applying the forward-looking
ECAPM approach to the firms in the Utility Group results in an average of 1 1.3 percent and
a midpoint of 11.1 percent.
As shown on page 2 of Exhibit No. 3, Schedule 8, incorporating a forecasted
Treasury bond yield for 2019-2023 implied an average ECAPM cost of equity of 11.4
percent for the Utility Group, with the midpoint being 11.2 percent.
F. Risk Premium Approach
a. Please briefly describe the risk premium method.
A. The risk premium method of estimating investors' required rate of return
extends to common stocks the risk-return tradeoff observed with bonds. The cost of equity
is estimated by first determining the additional return investors require to forgo the relative
safety of bonds and to bear the greater risks associated with common stock, and by then
adding this equity risk premium to the current yield on bonds. Like the DCF model, the risk
premium method is capital market oriented. However, unlike DCF models, which indirectly
a3 Docket No. 30011-97-GR-17, Pre-Filed Direct Testimony of Anthony J. Ornelas (May 1, 2018) at 52-53;
Docket No. 17-071-U, Direct Testimony of Marlon F. Grffing, PH.D. (May 29,2018) at33-35.
aa Montana Public Service Commission, Docket No. D2017.9.80, Order No. 7575c (Sep. 26, 2018) at P 114.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 34 ot 45
o
o
o
5
6
7
8
9
l0
1l
t2
13
t4
15
t6
t7
18
t9
2t
20
o
Schedule 2
Page 35 of45
impute the cost of equity, risk premium methods directly estimate investors'required rate of
return by adding an equity risk premium to observable bond yields.
a. Is the risk premium approach a widely accepted method for estimating
the cost of equity?
A. Yes. The risk premium approach is based on the fundamental risk-return
principle that is central to finance, which holds that investors will require a premium in the
form of a higher return in order to assume additional risk. This method is routinely
referenced by the investment community and in academia and regulatory proceedings, and
provides an important tool in estimating a fair ROE forAvista.
a. How did you implement the risk premium method?
A. I based my estimates of equity risk premiums for electric utilities on surveys
of previously authorized ROEs. Authorized ROEs presumably reflect regulatory
commissions' best estimates of the cost of equity, however determined, at the time they
issued their final order. Moreover, allowed ROEs are an important consideration for
investors and have the potential to influence other observable investment parameters,
including credit ratings and borrowing costs. Thus, this data provides a logical and
frequently referenced basis for estimating equity risk premiums for regulated utilities.
a. Is it circular to consider risk premiums based on authorized returns in
assessing a fair ROE forAvista?
A. No. In establishing authorized ROEs, regulators typically consider the results
of alternative market-based approaches, including the DCF model. Because allowed risk
premiums consider objective market data (e.g., stock prices, dividends, beta, and interest
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 35 of 45
o
1
2
J
4
5
6
7
8
9
10
11
t2
l3
t4
15
t6
t7
18
t9
20
2t
22
o
1
2
aJ
4
5
6
7
8
9
10
11
I2
t3
t4
15
t6
t7
18
t9
20
21
Schedule 2
Page 36 of 45
rates), and are not based shictly on past actions of other regulators, this mitigates concerns
over any potential for circularity.
a. How did you implement the risk premium approach using surveys of
allowed rates of return?
A. The ROEs authorized for electric utilities by regulatory commissions across
the U.S. are compiled by Regulatory Research Associates and published in its Regulatory
Focus report. On page 3 of Exhibit No. 3, Schedule 9, the average yield on long-term public
utility bonds is subtracted from the average allowed rate of return on common equity for
electric utilities to calculate equity risk premiums for each year between 1974 and 2}l83s
Over this 45-year period, these equity risk premiums for electric utilities averaged 3.75
percent, and the yield on public utility bonds averaged 8.19 percent.
a. Is there any capital market relationship that must be considered when
implementing the risk premium method?
A. Yes. There is considerable evidence that the magnitude of equity risk
premiums is not constant and that equity risk premiums tend to move inversely with interest
rates. In other words, when interest rate levels are relatively high, equity risk premiums
narrow, and when interest rates are relatively low, equity risk premiums widen. The
implication of this inverse relationship is that the cost of equity does not move as much as,
or in lockstep with, interest rates. Accordingly, for a I percent increase or decrease in
interest rates, the cost of equity may only rise or fall some fraction of 1 percent. Therefore,
when implementing the risk premium method, adjustments may be required to incorporate
45 Yield averages reported by Moody's are for seasoned bonds with a remaining maturity of 20 years or more.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 36 of 45
o
o
o
o
Schedule 2
Page37 of45
this inverse relationship if current interest rate levels diverge from the average interest rate
level represented in the data set.
a. Has this inverse relationship been documented in the financial research?
A. Yes. This inverse relationship between equity risk premiums and interest
rates has been widely reported in the financial literature. As summarized by New Regulatory
Finance:
Published studies by Brigham, Shome, and Vinson (1985), Harris (1986),
Harris and Marston (1992, 1993), Carleton, Chambers, and Lakonishok
(1983), Morin (2005), and McShane (2005), and others demonstrate that,
beginning in 1980, risk premiums varied inversely with the level of interest
rates - rising when rates fell and declining when rates rose.a6
Other regulators have also recognized that, while the cost of equity trends in the same
direction as interest rates, these variables do not move in lock-step.a7 This relationship is
illustrated in the figure on page 4 of Exhibit No. 3, Schedule 9.
a. What are the implications of this relationship under current capital market
conditions?
A. Current bond yields are lower than those prevailing over the risk premium
study periods. Given that equity risk premiums move inversely with interest rates, these
lower bond yields also imply an increase in the equity risk premium that investors require to
accept the higher uncertainties associated with an investment in utility common stocks
versus bonds. In other words, higher required equity risk premiums offset the impact of
declining interest rates on the ROE. This relationship is illustrated in the figure below,
a6 Roger A. Morin, New Regulatory Finance, Pub. Util. Reports, Inc. (2006) at 128.
a1 See, e.g., Califomia Public Utilities Commission, Decision 08-05-035 (May 29,2008); Entergy Mississippi
Formula Rate Plan FRP-6, https://www.entergy-mississippi.com/userfiles/content/price/tariffs/eml_frp.pdf (last
visited Apr. 15, 2019); Martha Coakley et al.,l47 FERC lJ 61,234 atP 147 (2014).
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2,Page37 of 45
1
2
J
4
5
6
7
8
9
10
l1
o
t2
l3
t4
15
16
17
18
t9
2t
20
22
o
1
2
J
4
Schedule 2
Page 38 of45
which is based on three-year rolling averages for the utility bond yields and risk premiums
shown on page 3 of Exhibit No. 3, Schedule 9.
FIGURE 2
INVERSE RELATIONSHIP
a. What cost of equity is implied by the risk premium method using surveys
of allowed ROEs?
A. Because risk premiums move inversely with interest rates and current bond
yields are significantly lower than the average over the study period, it is necessary to adjust
the average equity risk premium over the study period to reflect the impact of changes in
bond yields. Based on the regression output between the interest rates and equity risk
premiums displayed on page 4 of Exhibit No. 3, Schedule 9, the equity risk premium for
electric utilities increased approximately 43 basis points for each percentage point drop in
the yield on average public utility bonds. As illustrated on page 1 of Exhibit No. 3,
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 38 of45
o
o
5
6
7
8
9
10
ll
12
13 o
16. 0?
14. 0?
1,2 . O*
r-0. 0*
B .0t
6.0t
4.OZ
2.O%
0 .0t
,4"rtrorrtrr"*rdPr*ordrrroro+r""orrrort
-Utility
Bond Yield
-trqgity
Risk Preruium
o
Schedule 2
Page 39 of45
Schedule 9, with the yield on average public utility bonds for the six months ending April
2019 being 4.41 percent, this implied a current equity risk premium of 5.38 percent for
electric utilities. Adding this equity risk premium to the yield on Baa utility bonds of 4.81
percent produces a current cost ofequity of 10.19 percent.
a. What cost of equity was produced by the risk premium approach after
incorporating forecasted bond yields?
A. As shown on page 2 of Exhibit No. 3, Schedule 9, incorporatinga forecasted
yield for 2019-2023 and adjusting for changes in interest rates since the study period implied
an equity risk premium of 5.03 percent for electric utilities. Adding this equity risk premium
to the average implied yield on long-term Baa public utility bonds for 2019-2023 of 5.62
percent resulted in an implied cost of equity of approximately 10.65 percent.
G. Expected Earninss Approach
O. What other analyses did you conduct to estimate the cost of common
equity?
A. As noted earlier, I also evaluated the cost of common equity using the
expected earnings method. Reference to rates of retum available from alternative
investments of comparable risk can provide an important benchmark in assessing the refurn
necessary to assure confidence in the financial integrity of a firm and its ability to attract
capital. This expected eamings approach is consistent with the economic underpinnings for
a fair rate of return established by the U.S. Supreme Court in Bluefield and Hope.
Moreover, it avoids the complexities and limitations of capital market methods and instead
focuses on the returns earned on book equity, which are readily available to investors.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 39 of 45
I
2
aJ
4
5
6
7
8
9
10
o
1l
t2
l3
t4
15
t6
t7
18
t9
20
2t
22
o
1
2
J
4
5
6
7
8
9
Schedule 2
Page 40 of45
a. What economic premise underlies the expected earnings approach?
A. The simple, but powerful concept underlying the expected earnings approach
is that investors compare each investment alternative with the next best opportunity. If the
utility is unable to offer a retum similar to that available from other opporrunities of
comparable risk, investors will become unwilling to supply the capital on reasonable terms.
For existing investors, denying the utility an opportunity to earn what is available from other
similar risk alternatives prevents them from earning their opportunity cost of capital. Such
an outcome would violate the Hope and Bluefield standards and undermine the utility's
access to capital on reasonable terms.
a. How is the expected earnings approach typically implemented?
A. The traditional comparable eamings test identifies a group of companies that
are believed to be comparable in risk to the utility. The actual earnings of those companies
on the book value of their investment are then compared to the allowed return of the utility.
While the traditional comparable earnings test is implemented using historical data taken
from the accounting records, it is also common to use projections of refurns on book
investment, such as those published by recognized investment advisory publications (e.9.,
Value Line). Because these returns on book value equity are analogous to the allowed return
on a utility's rate base, this measure of opportunity costs results in a direct, "apples to
apples" comparison.
Moreover, regulators do not set the returns that investors earn in the capital markets,
which are a function of dividend payments and fluctuations in common stock prices, both of
which are outside their control. Regulators can only establish the allowed ROE, which is
applied to the book value of a utility's investment in rate base, as determined from its
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 40 of 45
o
o
o
10
11
t2
l3
t4
15
t6
t7
18
l9
20
2t
22
23
o
Schedule 2
Page 41 of45
accounting records. This is directly analogous to the expected earnings approach, which
measures the return that investors expect the utility to earn on book value. As a result, the
expected earnings approach provides a meaningful guide to ensure that the allowed ROE is
similar to what other utilities of comparable risk will earn on invested capital. This expected
earnings test does not require theoretical models to indirectly infer investors'perceptions
from stock prices or other market data. As long as the proxy companies are similar in risk,
their expected eamed returns on invested capital provide a direct benchmark for investors'
opportunity costs that is independent of fluctuating stock prices, market-to-book ratios,
debates over DCF growth rates, or the limitations inherent in any theoretical model of
investor behavior.
O. What rates of return on equity are indicated for utilities based on the
expected earnings approach?
A. For the firms in the Utility Group, the year-end returns on common equity
projected by Value Line over its forecast horizon are shown on Exhibit No. 3, Schedule 10.
As I explained earlier in my discussion of the br*sv growth rates used in applying the DCF
model, Value Line's refurns on common equity are calculated using year-end equity
balances, which understates the average return earned over the year.a8 Accordingly, these
year-end values were converted to average returns using the same adjustment factor
discussed earlier and developed on Exhibit No. 3, Schedule 6. As shown on Exhibit No. 3,
Schedule 10, Value Line's projections for the Utility Group suggest an average ROE of
approximately 10.7 percent, with a midpoint value of 11.4 percent.
a8 For example, to compute the annual return on a passbook savings account with a beginning balance of
$1,000 and an ending balance of $5,000, the interest income would be divided by the average balance of
$3,000. Using the $5,000 balance at the end of the year would understate the actual retum.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 41 oI 45
o
I
2
J
4
5
6
7
8
9
10
l1
t2
13
14
15
16
t7
18
l9
20
21
o
I
2
J
4
5
6
7
8
9
10
1l
t2
13
t4
15
t6
17
18
t9
20
2t
22
23
24
Schedule 2
Page 42 of45
II. NON-UTILITY BENCHMARK
a. What other proxy group did you consider in evaluating a fair ROE for
Avista?
A. Consistent with underlying economic and regulatory standards, I also applied
the DCF model to a reference group of low-risk companies in the non-utility sectors of the
economy. I refer to this group as the 'Non-Utility Group".
a. Do utilities have to compete with non-regulated firms for capital?
A. Yes. The cost of capital is an opportunity cost based on the returns that
investors could realize by putting their money in other altematives. Clearly, the total capital
invested in utility stocks is only the tip of the iceberg of total common stock investment, and
there are a plethora of other enterprises available to investors beyond those in the utility
industry. Utilities must compete for capital, not just against firms in their own industry but
with other investment opportunities of comparable risk. lndeed, modem portfolio theory is
built on the assumption that rational investors will hold a diverse portfolio of stocks, not just
companies in a single industry.
a. Is it consistent with the Bluefteld and Hope cases to consider investors'
required ROE for non-utility companies?
A. Yes. The cost of equity capital in the competitive sector of the economy form
the very underpinning for utility ROEs because regulation purports to serve as a substitute
for the actions of competitive markets. The Supreme Court has recognized that it is the
degree of risk, not the nature of the business, which is relevant in evaluating an allowed
ROE for a utility. The Bluefield case refers to "business undertakings attended with
comparable risks and uncertainties." It does not restrict consideration to other utilities.
Similarly, the Hope case states:
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 42 ol 45
o
o
o
o
Schedule 2
Page 43 of 45
By that standard the retum to the equity owner should be commensurate with
returns on investments in other enterprises having corresponding risks.ae
As in the Bluefield decision, there is nothing to restrict 'oother enterprises" solely to the
utility industry.
a. Does consideration of the results for the Non-Utility Group make the
estimation of the cost of equity using the DCF model more reliable for Avista?
A. Yes. The estimates of growth from the DCF model depend on analysts'
forecasts. It is possible for utility growth rates to be distorted by short-term trends in the
industry or by the industry falling into favor or disfavor by analysts. Such distortions could
result in biased DCF estimates for utilities. Because the Non-Utility Group includes low
risk companies from other industries, it helps to insulate against any possible distortion that
may be present in the results for a particular sector.
a. What criteria did you apply to develop the Non-Utility Group?
A. The comparable risk proxy group was composed of those U.S. companies
followed by Value Line that:
1) pay common dividends.
2)have a Safety Rank of "1".
3) have a Financial Strength Rating of "A" or greater.
4) have a beta of 0.75 or less.
5) have investment grade credit ratings from S&P and Moody's.
a. How do the overall risks of this Non-Utility Group compare with the
Utility Group and Avista?
A. As illustrated in Table 4 below, while the average beta for the Non-Utility
Group suggests greater risk, the average credit ratings, Safety Rank, and Financial Strength
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 43 ol 45
1
2
J
4
5
6
7
8
9
10
o
ll
t2
13
t4
15
t6
l7
18
19
20
2t
22
23
24
o ae Federal Power Comm'n v. Hope Natural Gas Co., 320 U.S. 391 (1944) ("Hope").
I
2
J
4
Schedule 2
Page 44 of 45
rating for the Non-Utility Group all indicate less risk than for Avista and the proxy group of
utilities.
TABLE 4
COMPARISON OF RISK INDICATORS
Value Line
Credit Rating Safety
S&P
A-
BBB+
BBB
Moodv's
A3
Baa2
Baa2
Flnancial
Strensth
A+
B++
A
Beta
0.73
0.64
0.6s
o
O
Non-Utility Group
Utility Group
Avista
Rank
1
2
2
5
6
7
8
9
When considered together, a comparison of these objective measures, which consider a
broad spectrum of risks, including financial and business position, relative size, and
exposure to company-specific factors, indicates that investors would likely conclude that the
overall investment risks for the Utility Group and Avista are greater than those of the firms
in the Non-Utility Group.
The 17 companies that make up the Non-Utility Group are representative of the
pinnacle of corporate America. These firms, which include household names such as Coca-
Cola, General Mills, and Walmart, have long corporate histories, well-established track
records, and exceedingly conservative risk profiles. Many of these companies pay dividends
on a par with utilities, with the average dividend yield for the group approaching 3 percent.
Moreover, because of their significance and name recognition, these companies receive
intense scrutiny by the investment community, which increases confidence that published
growth estimates are representative of the consensus expectations reflected in common stock
prices.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 44 of 45
10
12
11
13
t4
15
t6
t7
18
o
o
Schedule 2
Page 45 of 45
a. What were the results of your DCF analysis for the Non-Utility Group?
A. I applied the DCF model to the Non-Utility Group using analysts EPS growth
projections, as described earlier for the Utility Group, with the results being presented in
ExhibitNo.3, Schedule 11. As summarized in Table 5, below, application of the constant
growth DCF model resulted in the following cost of equity estimates:
TABLE 5
DCF RESULTS - NON-UTILITY GROUP
1
2
aJ
4
5
6
7
8
9
Growth Rate
Value Line
IBES
Zaclrs
Average
10.2%
t0.t%
9.6%
Midpoint
11.1%
105%
9.7%
o
As discussed earlier, reference to the Non-Utility Group is consistent with
established regulatory principles. Required returns for utilities should be in line with those
of non-utility firms of comparable risk operating under the constraints of free competition.
Because the actual cost of equity is unobservable, and DCF results inherently incorporate a
degree of error, cost of equity estimates for the Non-Utility Group provide an important
benchmark in evaluating a fair and reasonable ROE. Given the lower risk of the Non-Utility
Group, these DCF results suggest that the 9.9 percent requested ROE for Avista's utility
operations is a conservative estimate of a fair return.
Exhibit No. 3
Case No. AVU-E-19-04
A. McKenzie, Avista
Schedule 2, Page 45 of 45
t0
1l
t2
l3
t4
l5
o